Loss of Pollen-S Function in Two Self-Compatible Selections of Prunus avium Is Associated with Deletion/Mutation of an S Haplotype–Specific F-Box Gene

Abstract

Recently, an S haplotype–specific F-box (SFB) gene has been proposed as a candidate for the pollen-S specificity gene of RNase-mediated gametophytic self-incompatibility in Prunus (Rosaceae). We have examined two pollen-part mutant haplotypes of sweet cherry (Prunus avium). Both were found to retain the S-RNase, which determines stylar specificity, but one (S₃′ in JI 2434) has a deletion including the haplotype-specific SFB gene, and the other (S₄′ in JI 2420) has a frame-shift mutation of the haplotype-specific SFB gene, causing amino acid substitutions and premature termination of the protein. The loss or significant alteration of this highly polymorphic gene and the concomitant loss of pollen self-incompatibility function provides compelling evidence that the SFB gene encodes the pollen specificity component of self-incompatibility in Prunus. These loss-of-function mutations are inconsistent with SFB being the inactivator of non-self S-RNases and indicate the presence of a general inactivation mechanism, with SFB conferring specificity by protecting self S-RNases from inactivation.

INTRODUCTION

Self-incompatibility (SI) prevents inbreeding in a wide range of flowering plants (reviewed in De Nettancourt, 2001). The mechanism can be either gametophytic or sporophytic depending on the genetic control of the pollen SI phenotype. In sweet cherry (Prunus avium), a member of the Rosaceae cultivated for its fruit, SI and cross-incompatibility between various pairs of cultivars was attributed to a multiallelic S locus, expressed gametophytically (Crane and Lawrence, 1929). As in all other SI systems studied, the S locus of cherry is considered to comprise at least two parts, one expressed in the style and a complementary one expressed in the pollen, as originally proposed by Lewis (1949).

Initially the sporophytic SI of Brassica and the gametophytic SI of the Solanaceae were the most intensively studied at the molecular level, but recently several important breakthroughs have been made in the Rosaceae, which appears to have the same SI mechanism as the Solanaceae. The molecular identification of the pistil components of SI has revealed three distinct mechanisms in the sporophytic SI system of Brassica and in the gametophytic SI systems of the Solanaceae/Rosaceae/Scrophulariaceae and of Papaver (De Nettancourt, 2001). In the Solanaceae, the Rosaceae, and the Scrophulariaceae, the stylar part codes for a ribonuclease (S-RNase) (McClure et al., 1989; Broothaerts et al., 1995; Bošković and Tobutt, 1996; Xue et al., 1996). Ribonuclease activity of S-RNases is needed to inhibit the growth of pollen tubes carrying an S allele that matches an S allele present in the style (Huang et al., 1994; Royo et al., 1994), supporting a model for this type of gametophytic SI in which pollen RNA is degraded in an incompatible interaction.

The pollen component (pollen-S) in Brassica has been identified as a ligand for a receptor kinase located in the stigma (reviewed in Brugière et al., 2000). Candidates for the pollen-S gene of RNase-based SI have recently been identified in Antirrhinum of the Scrophulariaceae (Lai et al., 2002), Prunus (Entani et al., 2003; Ushijima et al., 2003; Yamane et al., 2003), and Petunia of the Solanaceae (Qiao et al., 2004; Sijacic et al., 2004). These are all S-linked F-box genes (SLF or SFB) and have implicated ubiquitination in the SI reaction. In this article, we follow the use of the abbreviation SFB (for S haplotype–specific F-box) in cherry (Yamane et al., 2003). Additional F-box genes (SLFL1-3) have been identified in Prunus S haplotypes, although these exhibit relatively low levels of polymorphism (Entani et al., 2003; Ushijima et al., 2003).

Two self-compatible cherry selections raised at the John Innes Institute (Lewis and Crowe, 1954) offer a unique opportunity for studying the molecular basis of loss of pollen-S function at a rosaceous S locus. Gain-of-function or loss-of-function transgenic experiments in Prunus tree species are hindered by the lack of an efficient transformation system and a long juvenile period. The self-compatible selections, JI 2420 and JI 2434, were obtained from the nominally incompatible cross of the cultivars Emperor Francis (S₃S₄) × Napoleon (S₃S₄, X-irradiated pollen). Both selections carry pollen-part mutations but retain normal stylar function (Matthews and Lapins, 1967; Matthews, 1970). Recently, the genotypes of JI 2420 and two accessions of JI 2434 were clarified, and the basis of their self-compatibility was studied using stylar ribonuclease analyses and test crosses (Bošković et al., 2000). In JI 2420, self-compatibility is attributed to a mutant S₄ allele, denoted S₄′, where the prime symbol indicates the loss of pollen-S function (Lewis and Crowe, 1954). Similarly, in JI 2434, the self-compatibility is attributed to a mutant S₃ allele, S₃′.

Self-compatible pollen-part mutants obtained by selfing or by crossing nominally incompatible plants using x-irradiated pollen have been described not only in P. avium, but also in Oenothera organensis of the Onagraceae (Lewis, 1949, 1951), Petunia inflata (Brewbaker and Shapiro, 1959), and Nicotiana alata (Pandey, 1965). Brewbaker and Natarajan (1960) found that all pollen-part mutants studied in P. inflata carried a centric fragment. Centric fragments, which can result from x-irradiation, are parts of a chromosome including the centromere that segregate randomly at meiosis. The centric fragments in the P. inflata pollen-part mutants carried a duplicated S allele, resulting in some pollen grains being heteroallelic. The self-compatibility of heteroallelic pollen had been advanced by Lewis and Modlibowska (1942) to explain the self-compatibility of a tetraploid pear (Pyrus communis)—pollen grains of genotype S₁S₂ succeed on S₁S₁S₂S₂ styles, whereas S₁S₁ and S₂S₂ pollen fails. The cross of the tetraploid as pollen parent on to the original diploid succeeds, whereas the reciprocal cross fails, indicating a breakdown of pollen-part function. This interaction between the two S alleles in heteroallelic pollen has been termed competitive interaction (Lewis, 1943). Brewbaker and Natarajan (1960) also concluded that the pollen-part mutants of Oenothera and Prunus, studied by Lewis, could have resulted from a duplication of an S allele. However, Lewis (1961) concluded that some pollen-part mutations of Oenothera should be the result of a mutation or deletion of the pollen part and stated that most of his arguments were also valid for the pollen-part mutant selections of Prunus.

In recent years, molecular analyses have been performed on pollen-part mutants generated by irradiation mutagenesis in Solanum tuberosum (Thompson et al., 1991) and N. alata (Golz et al., 1999, 2001) of the Solanaceae. In both studies, DNA gel blot analysis with S-RNase probes showed that some of the self-compatible mutants obtained had an extra S allele, either carried on a centric fragment or translocated to another chromosome. In some cases, no additional S-RNase allele was detected, although there was evidence for a chromosomal duplication, presumably including the pollen part but not the S-RNase (Golz et al., 2001). No evidence of a deletion was found in either study.

The cherry pollen-part mutations in JI 2420 and JI 2434 could, we surmised, have resulted from duplication of a pollen-S gene giving heteroallelic pollen or from a mutation in the pollen-S gene. Thus, the S₃′ mutant allele of JI 2434 could either be a result of a duplicate S₄ pollen-S allele (dS₄) giving rise to S₃dS₄ pollen that would be compatible on an S₃S₄ style or else a mutation of the S₃ pollen-S gene itself. Similarly, explanations of self-compatibility in the S₄′ mutant JI 2420 are a duplicate S₃ pollen-S allele (dS₃) resulting in S₄dS₃ pollen or a mutation of the S₄ pollen-S gene itself. Bošković et al. (2000) found no evidence for a duplicated S-RNase in either mutant selection with ribonuclease activity assays.

A molecular analysis of these cherry mutant S alleles has now been made possible after the cDNA cloning of the S-RNases for the S₃- and S₄-haplotypes (Sonneveld et al., 2001). We demonstrate that there is no evidence of a duplication in either mutant haplotype, but rather a deletion or mutation affecting the haplotype-specific SFB gene, leading to loss of function. These findings provide further evidence for the pollen-S function of SFB genes, and we discuss how these unique loss-of-function mutations can be reconciled with current models of RNase-based SI.

RESULTS

Genetic Analysis of Pollen-Part Mutations

Pollen-part mutations can theoretically result from duplication of a pollen-S allele or from a mutation of the pollen-S gene itself. Heteroallelic pollen can arise when a duplicated pollen-S gene has been translocated to the chromosome bearing the other S allele, or to a non-S chromosome, or it could be carried on a centric fragment. In the first case, the pollen-part mutation will be linked in coupling to one particular S allele; in the other two cases, the pollen-part mutation will segregate independently from the S locus. A mutation in the pollen part itself would always be associated with one S allele.

To confirm that the pollen-part mutation in JI 2420 is linked in coupling with S₄, the progeny of a cross between a pollen parent deriving its self-compatibility from JI 2420 but not having S₃ (Lapins, S₁S₄′) and a female parent with the S₁ allele in common (Erika, S₁S₃) was analyzed. If the mutation was not linked with S₄ (i.e., in the case of a duplication of the S₃ pollen-S allele elsewhere in the genome), both types of pollen (S₁ and S₄), when carrying the duplicated S₃ pollen-S allele, would be heteroallelic and able to grow on the style of the female parent even if it has the S₁ allele in common (Tables 1 and 2).

Table 1.

Erika (S₁S₃) × Lapins (S₁S₄′) Expected Gamete Genotypes and Seedling Genotypes (and S-RNase Phenotypes) if S₄′ Results from a Duplication of Pollen-S₃ (dS₃) Not Linked in Coupling with S₄

Female/Male	S1	S4	S1dS3	S4dS3
S1	X	S1S4 (S1S4)	S1S1dS3 (S1)	S1S4dS3 (S1S4)
S3	X	S3S4 (S3S4)	S1S3dS3 (S1S3)	S3S4dS3 (S3S4)

X indicates incompatibility of pollen.

Table 2.

Erika (S₁S₃) × Lapins (S₁S₄′) Expected Gamete Genotypes and Seedling Genotypes (and S-RNase Phenotypes) if S₄′ Results from a Deletion/Mutation of the S₄ Pollen-S Allele (for Convenience Indicated by S₄′) or from a Duplication of Pollen-S₃ (dS₃) Linked in Coupling with S₄

Female/Male	S1	S4′ or S4dS3
S1	X	S1S4′ or S1S4dS3 (S1S4)
S3	X	S3S4′ or S3S4dS3 (S3S4)

X indicates incompatibility of pollen.

Analysis of stylar ribonucleases showed that 23 seedlings had bands for S₁ and S₄ (S₁S₄′ seedlings), and 18 seedlings had bands for S₃ and S₄ (S₃S₄′ seedlings) (data not shown). None of the seedlings showed a single band in the S₁ position. Consistent rejection of the S₁ pollen confirms that the mutation of JI 2420 is linked in coupling with S₄ and cannot be attributed to a duplication of the S₃ pollen-S allele on a centric fragment or on a non-S chromosome.

Likewise, to confirm that self-compatibility in JI 2434 (S₃′S₄) is linked in coupling with S₃, a progeny from a cross between a pollen parent deriving self-compatibility from JI 2434 but not having the S₄ allele (9239-3, S₁S₃′) and a female parent with the S₁ allele in common (Van, S₁S₃) was raised and analyzed. The presence or absence of S₁S₁′ genotypes (single S₁ band) in the progeny would indicate whether self-compatibility of JI 2434 is linked with S₃ (Tables 3 and 4).

Table 3.

Van (S₁S₃) × 9239-3 (S₁S₃′) Expected Gamete Genotypes and Seedling Genotypes (and S-RNase Phenotypes) if S₃′ Results from a Duplication of Pollen-S₄ (dS₄) Not Linked in Coupling with S₃

Female/Male	S1	S3	S1dS4	S3dS4
S1	X	X	S1S1dS4 (S1)	S1S3dS4 (S1S3)
S3	X	X	S1S3dS4 (S1S3)	S3S3dS4 (S3)

X indicates incompatibility of pollen.

Table 4.

Van (S₁S₃) × 9239-3 (S₁S₃′) Expected Gamete Genotypes and Seedling Genotypes (and S-RNase Phenotypes) if S₃′ Results from a Deletion/Mutation of the S₃ Pollen-S Allele (for Convenience Indicated by S₃′) or from a Duplication of Pollen-S₄ (dS₄) Linked in Coupling with S₃

Female/Male	S1	S3′ or S3dS4
S1	X	S1S3′ or S1S3dS4 (S1S3)
S3	X	S3S3′ or S3S3dS4 (S3)

X indicates incompatibility of pollen.

S allele genotyping of 46 seedlings by PCR using consensus primers for the first intron of cherry S-RNases showed that they segregated into two classes (example shown in Figure 1): 17 with bands for S₃ and S₁ (S₁S₃′ seedlings) and 29 with a single S₃ band (S₃S₃′ seedlings). The absence of a class with a single S₁ band (S₁S₁′ seedlings) indicates that the pollen-part mutation of JI 2434 (S₃′S₄) is linked in coupling with S₃ and cannot be attributable to a duplication of the S₄ pollen component on a centric fragment or on a non-S chromosome.

Figure 1.

Segregation Analysis of S Alleles to Test Cosegregation of the JI 2434 Pollen-Part Mutation with the S₃ Allele.

Genomic DNA of seedlings of Van (S₁S₃) × 9239-3 (S₁S₃′) was amplified by PCR using consensus primers revealing length polymorphism of the first intron of cherry S-RNases. Samples are as follows: ER, Early Rivers (S₁S₂); Victor (S₂S₃) (standards for S₁ and S₃, respectively); Van (S₁S₃); S₁S₃′ selection, 22 seedlings.

DNA Gel Blot Analysis of S-RNase Regions in Pollen-Part Mutants

Because both pollen-part mutant selections appeared to have the mutation linked with a particular S allele, DNA gel blot analysis with an S-RNase probe was performed to look for evidence of a rearrangement in the regions flanking the appropriate S-RNases.

Digestion of genomic DNA with both restriction enzymes EcoRI (Figure 2A) and HindIII (Figure 2B) resulted in a single restriction fragment in the normal S₄ position for JI 2420 (S₄S₄′) (i.e., the S₄′ band was indistinguishable from S₄). In the HindIII digest, the S₃′ restriction fragment of JI 2434 was in the same position as the S₃ fragment from the parents and the standard Van (S₁S₃). In the EcoRI digest, however, the S₃′ restriction fragment of JI 2434 appeared to be slightly smaller than the S₃ fragment. This indicated an alteration in the region flanking the S₃-RNase of JI 2434.

Figure 2.

DNA Gel Blot Analysis of the S₃′ and S₄′ Mutants Using an S₁-RNase cDNA Probe to Test for Rearrangements in the Regions Flanking the S₃- and S₄-RNase Genes, Respectively.

(A) EcoRI digest of parents and pollen-part mutants. EF, Emperor Francis (S₃S₄); Nap, Napoleon (S₃S₄); 2420, JI 2420 (S₄S₄′); 2434, JI 2434 (S₃′S₄); Van (S₁S₃); ML, Merton Late (S₁S₄).

(B) HindIII digest of parents and pollen-part mutants. EF, Emperor Francis (S₃S₄); Nap, Napoleon (S₃S₄); 2420, JI 2420 (S₄S₄′); 2434, JI 2434 (S₃′S₄); Van (S₁S₃); ML, Merton Late (S₁S₄).

(C) EcoRI digest of the progeny Van (S₁S₃) × JI 2434 (S₃′S₄). Van (S₁S₃); 2434, JI 2434 (S₃′S₄); S₃S₃′ selection; S₃S₄ selection; S₁S₃′ selection; S₁S₄ selection; ML, Merton Late (S₁S₄).

(D) SstI digest of parents and S₃′ mutant. EF, Emperor Francis (S₃S₄); Nap, Napoleon (S₃S₄); 2434, JI 2434 (S₃′S₄); S₃S₃′ selection; Van (S₁S₃); ML, Merton Late (S₁S₄).

Low stringency post-hybridization washes allowed the detection of cross-hybridization of the S₁-RNase cDNA probe to both the S₃ and S₄ alleles.

Testing four previously genotyped selections from the cross Van (S₁S₃) × JI 2434 (S₃′S₄) confirmed the shift of the S₃′ restriction fragment in the EcoRI digest. As shown in Figure 2C, the S₃S₃′ selection shows two bands, confirming that the S₃′ fragment has a lower molecular weight than the normal S₃ fragment, indicating a sequence change in the region flanking the S₃-RNase of JI 2434. The estimated sizes of the S₃ and the S₃′ fragments are ∼13 and 11 kb, respectively.

When 11 more restriction enzymes were tested and fragment sizes estimated (Table 5), none of the enzymes showed a fragment size shift for S₄′ compared with S₄. A band shift for S₃′ was, however, found with KpnI, PaeI, PstI, SstI, and XbaI, indicating a major sequence rearrangement flanking the S₃-RNase. Whether the rearrangement in the S₃′-haplotype is in the 5′ or 3′ flanking region of the S₃-RNase was investigated by inclusion of an enzyme cutting at the 5′ end of the S₃-RNase sequence (PstI) and an enzyme cutting toward the 3′ end of the S₃-RNase sequence (BamHI). Because a band shift was found for PstI, but not for BamHI, it appeared that the rearrangement had occurred downstream of the S₃-RNase gene. The SstI digest (Figure 2D) shows that the breakpoint of the rearrangement in the S₃′-haplotype must be within ∼3.5 kb of the S₃-RNase gene.

Table 5.

Restriction Enzymes Used for DNA Gel Blot Analysis with S-RNase Probe and Estimated Fragment Sizes for Each of the S Alleles Present

Estimated Size of Restriction Fragments (kb)a
Allele	BamHIb	BclI	EcoRIc	HindIII	KpnIcd	PstIce	NcoI	NdeI	PaeIc	ScaId	SstIc	SspId	XbaIc	EcoRI/HindIII
S1	14.5	7.5	∼2	7.5	21.0	23	14.5	6.5	11.5	5.5	11.5	∼2	12	∼2.0
S3	9.5	6.0	13	8.0	20.0	7	8.0	9.5	7.5	∼4.0	5.0	∼3 + 3.5	22	7.5
S3′	9.5	6.0	11	8.0	6.5	15	8.0	9.5	16.5	∼4.0	6.5	∼3 + 3.5	20	7.5
S4	>30.0	7.5	∼2	5.5	10.5	∼3	4.5	17.0	13.0	6.5	12.5	∼1	8	∼2.0
S4′	>30.0	7.5	∼2	5.5	10.5	∼3	4.5	17.0	13.0	6.5	12.5	∼1	8	∼2.0

^aSizes of the bands were estimated by comparison with the 5-kb DNA ladder (Invitrogen). Please note that the size fragments under 5 kb and over 30 kb could not be estimated accurately.

^bRestriction site at 3′ end of S₃-RNase (just before C5).

^cRestriction enzyme reveals a fragment size shift for S₃′.

^dRestriction site in S₃-RNase between C3 and RC4 (SspI), between C2 and C3 (KpnI), and just after RC4 (ScaI); the probe does not always detect both resulting fragments.

^eRestriction site at 5′ end of S₃-RNase (just after C1).

The DNA gel blot analyses of the S₁S₃′ and the S₃S₃′ selection confirmed that JI 2434 does not have a duplicated S₄-RNase gene, in accordance with previous reports (Bošković et al., 2000). Similarly, there is no evidence of a duplicated S₃-RNase gene in JI 2420.

Cloning of the S₄-Haplotype–Specific SFB Sequence

To study the haplotype-specific SFB genes, recently identified as candidates for the pollen-S gene in Prunus (Entani et al., 2003; Ushijima et al., 2003; Yamane et al., 2003), the S₄-haplotype–specific SFB (S₄-SFB) genomic sequence was cloned by genomic PCR using degenerate primers. The S₃-haplotype–specific SFB (S₃-SFB) sequence was available from the EMBL database (Yamane et al., 2003).

The deduced amino acid sequence of the putative S₄-SFB was aligned with other Prunus SFB sequences (Figure 3). It shows high sequence similarity with the other sequences (74.7 to 81.5% amino acid identity) and contains the expected F-box motif and two variable regions toward the 3′ end (Ushijima et al., 2003).

Figure 3.

Alignment of the Predicted Amino Acid Sequence of the S₄-Haplotype–Specific SFB with Other Prunus SFB/SLF Sequences.

Seven SFB sequences of Prunus (Pa, P. avium; Pd, P. dulcis; Pm, P. mume) aligned using ClustalW. Residues highlighted in gray denote divergence from the observed consensus, and dashes represent gaps. The F-box region and two variable regions Va and Vb (as indicated in Ushijima et al., 2003) are boxed.

Cosegregation of the cloned putative S₄-SFB sequence with the S₄ allele was confirmed in a progeny segregating for S₄ using allele-specific primers (Figure 4). Seedlings showing amplification had previously been found to carry the S₄ allele by stylar ribonuclease activity assays (Bošković et al., 1997) and S-RNase allele-specific PCR (Sonneveld et al., 2001).

Figure 4.

Segregation Analysis of the Putative S₄-Haplotype–Specific SFB Sequence to Test Cosegregation with the S₄-RNase Allele.

PCR amplification with specific primers for the putative S₄-specific SFB sequence is shown for genomic DNA of parents and a representative sample of seedlings of the cross Bradbourne Black (S₃S₅) × Merton Late (S₁S₄): BB, Bradbourne Black; ML, Merton Late; 12 seedlings. Presence (+) or absence (−) of the S₄-RNase allele is indicated underneath the lanes. The bands amplified by the internal control primers included in each PCR are indicated with PAL.

S-RNase and SFB Allele-Specific PCR

To check for duplications, deletions, or other rearrangements in the pollen-part mutant selections, allele-specific PCR was performed with S₃- and S₄-specific primers for both S-RNase and SFB sequences. Several conclusions can be drawn from the allele-specific PCR amplifications shown in Figure 5 and summarized in Table 6.

Figure 5.

Genomic PCR Amplification of Parents and Pollen-Part Mutants with Specific Primers for the SFB and S-RNase Genes of the S₃- and S₄-Haplotypes.

The primers used are specific for S₃-haplotype–specific SFB (A), S₃-RNase (B), S₄-haplotype–specific SFB (C), and S₄-RNase (D). Samples are as follows: EF, Emperor Francis (S₃S₄); Nap, Napoleon (S₃S₄); 2420, JI 2420 (S₄S₄′); S₄′S₄′ selection; 2434, JI 2434 (S₃′S₄); S₃S₃′ selection; S₃′S₃′ selection; Van (S₁S₃); ML, Merton Late (S₁S₄); water control. The bands amplified by the internal control primers that were included in each PCR are indicated with PAL.

Table 6.

Presence (+) or Absence (−) of Genomic PCR Amplification in Parents and Pollen-Part Mutants with Specific Primers for the S₃- and S₄-RNase and -SFB Genes as Shown in Figure 5

Lane	Cultivar/Selection	S-Genotype	S3-SFB	S3-RNase	S4-SFB	S4-RNase
a	Emperor Francis	S3S4	+	+	+	+
b	Napoleon	S3S4	+	+	+	+
c	JI2420	S4S4′	−	−	+	+
d	A53	S4′S4′	−	−	+	+
e	JI 2434	S3′S4	−	+	+	+
f	9239-1	S3S3′	+	+	−	−
g	JI 2434 AH	S3′S3′	−	+	−	−
h	Van	S1S3	+	+	−	−
i	Merton Late	S1S4	−	−	+	+

Amplification with the S-RNase primers for parents, standards, and pollen-part mutant selections is consistent with their stylar SI response and with the DNA gel blot analysis described above (Figures 5B and 5D).

Amplification with the SFB primers is as expected for parents and standards. For PCR amplification with the SFB primers for the S₄′ mutant, the results of the homozygous S₄′S₄′ selection are informative (Figures 5A and 5C). The selection clearly shows a band for S₄-SFB but not for S₃-SFB. This indicates that, if SFB is the pollen-S gene, self-compatibility in the S₄′ mutant cannot be attributed to a duplication of the S₃ pollen-S allele. This leaves a mutation of the S₄ pollen-S allele as the most likely explanation for the pollen-part mutation. If this is the case, the mutation clearly does not affect PCR amplification and does not visibly alter the product size because the S₄′ PCR product is indistinguishable on the gel from the product for the S₄ allele.

For the S₃′ mutant, again the results of the homozygous selection, S₃′S₃′, are informative (Figures 5A and 5C). This shows no band for the S₃- or the S₄-SFB, indicating that if SFB is the pollen-S gene, the S₃′ mutation is not a duplication of the S₄ pollen-S allele but a mutation of the S₃ pollen-S allele itself. The nature of the mutation cannot be deduced from these results, but it must affect at least one of the primer binding sites of the S₃-SFB sequence or otherwise interfere with PCR amplification.

DNA Gel Blot Analysis with SFB Probes

DNA gel blot analyses with S₃- and S₄-SFB probes were performed to confirm the PCR results described above and to characterize the pollen-part mutant haplotypes further. This approach should indicate whether most of the S₃-SFB sequence is missing in the S₃′ mutant. In addition, it could also check for rearrangements in the regions flanking the S₄-SFB gene in the S₄′ mutant.

The results of the DNA gel blot analysis with specific SFB probes are shown in Figure 6. Each of the parents shows a restriction fragment for the S₃- and the S₄-SFB sequence (Figures 6A and 6B). The S₄-SFB probe also weakly hybridizes to the S₁ allele of the standards (Figure 6A). With the S₄-SFB probe (Figure 6A), the homozygous S₄′S₄′ selection shows a restriction fragment in the same position as the parents and standards. Therefore, with the restriction enzyme used (DraI), there is no evidence of a rearrangement in the S₄′-haplotype. Probing with the S₃-SFB probe (Figure 6B) reveals that the S₃-SFB sequence, at least that included in the probe, is entirely missing in the S₃′ mutant, suggesting that the pollen-part mutation involves a deletion of most of the S₃-SFB sequence. Sequencing of the probe confirmed that it represents the S₃-SFB sequence because it matched the S₃-SFB sequence in the database, which has also been shown to give haplotype-specific hybridization signals in DNA gel blot analysis (Yamane et al., 2003). These DNA gel blot data are consistent with the genomic PCR data in Figure 5.

Figure 6.

DNA Gel Blot Analysis of the S₃′ and S₄′ Mutants Using SFB Haplotype-Specific Probes.

Genomic DNA was digested with DraI and probed with S₄-haplotype–specific SFB probe (A) and S₃-haplotype–specific SFB probe (B). The stringent washing conditions resulted in an allele-specific hybridization signal with the S₃-SFB probe; the S₄-SFB probe also cross-hybridized weakly to the S₁-SFB sequence. Samples are as follows: EF, Emperor Francis (S₃S₄); Nap, Napoleon (S₃S₄); 2420, JI 2420 (S₄S₄′); S₄′S₄′ selection; 2434, JI 2434 (S₃′S₄); S₃S₃′ selection; S₃′S₃′ selection; Van (S₁S₃); ML, Merton Late (S₁S₄).

Structure of the S₃- and S₃′-Haplotypes

The result of an attempt to amplify the region between the S-RNase and the SFB gene in the S₃-haplotype is shown in Figure 7. Allele-specific primers for each of the genes were used in all four possible combinations (Figures 7A to 7D). Only the PCR that included the S₃-RNase forward and the S₃-SFB forward primers (Figure 7A) gave a product, of ∼8 kb, exclusively in samples containing the S₃ allele. The result of the S₃ intergenic PCR reveals the relative orientation of the genes, as well as the distance between them: the S₃-SFB gene is located ∼6.5 kb downstream of the S₃-RNase gene, in opposite transcriptional orientation as shown in Figure 8. This is consistent with the detection of a rearrangement in the S₃′-haplotype downstream of the S₃-RNase gene (Table 5) that results in loss of the S₃-SFB gene. A similar attempt in the S₄-haplotype failed, presumably because the distance between the two genes in the S₄-haplotype is too great to be amplified under the PCR conditions used.

Figure 7.

Intergenic PCR for the S₃-Haplotype Using S₃-RNase and S₃-SFB Specific Primers in Four Possible Combinations to Determine Distance between and Transcriptional Orientation of the Genes.

The primers used on genomic DNA are S₃-RNase forward with S₃-haplotype–specific SFB forward (A), S₃-RNase forward with S₃-haplotype–specific SFB reverse (B), S₃-RNase reverse with S₃-haplotype–specific SFB forward (C), and S₃-RNase reverse with S₃-haplotype–specific SFB reverse (D). Samples are as follows: ER, Early Rivers (S₁S₂); Col, Colney (S₅S₆); EF, Emperor Francis (S₃S₄); Nap, Napoleon (S₃S₄); 2434, JI 2434 (S₃′S₄); S₃S₃′ selection; water control. Only combination (A) shows specific amplification for accessions with the S₃ allele.

Figure 8.

Preliminary Restriction Maps of the S₃- and S₃′-Haplotypes of Cherry of Napoleon (S₃S₄) and JI 2434 (S₃′S₄).

Restriction enzyme positions and distances were inferred from restriction fragment size estimates on DNA gel blots (Napoleon [S₃S₄] and JI 2434 [S₃′S₄]) (Table 5), except for the regions of known sequence. Genes are represented by black boxes, with white boxes for introns. Arrows underneath gene boxes indicate the direction of transcription as deduced from the intergenic PCR (Figure 7). The breakpoint of the rearrangement in the S₃′-haplotype was deduced to be ∼3.5 kb downstream of the S₃-RNase; sequence downstream of the breakpoint (dotted line) is linked to S₃ in the progenitor. Restriction enzyme abbreviations are as follows: Ba, BamHI; E, EcoRI; K, KnpI; Nd, NdeI; Ps, PstI; Sc, ScaI; Sp, SspI; St, SstI.

A preliminary restriction map of the S₃- and S₃′-haplotypes, of Napoleon (S₃S₄) and JI 2434 (S₃′S₄), respectively, is shown in Figure 8. The restriction enzyme sites flanking the S-RNase genes are based on the DNA gel blot analysis (Table 5). A major sequence rearrangement downstream of the S₃-RNase in the S₃′-haplotype includes a deletion of the S₃-SFB sequence. The breakpoint of the rearrangement is ∼3.5 kb from the S₃-RNase.

We examined whether other S locus F-box genes (SLFLs) are affected by the rearrangement in the S₃′-haplotype. Three SLFL genes (SLFL1-3) have been identified in two haplotypes of P. mume, and the relative gene order (also in relation to S-RNase and SFB) was found to be conserved (Entani et al., 2003). In these two haplotypes, only SLFL2 is located downstream of the S-RNase and upstream of SFB. If the gene order at the Prunus S locus is conserved, SLFL2 is therefore most likely to be affected by the rearrangement in the S₃′-haplotype. SLFL2-specific primers amplified a product in Napoleon and in the S₃′S₃′ homozygous selection, and sequencing of the product showed that they were almost identical. The sequence is ∼96% identical to the two P. mume SLFL2 sequences and ∼67% identical to other Prunus SLFLs, indicating that the SLFL2 gene is not deleted in the S₃′-haplotype.

The SLFL2 product was used as a probe in DNA gel blot analysis. Two genomic digests, DraI and SstI, confirmed the presence of an SLFL2 fragment in the S₃′-haplotype. The DraI digest gave a monomorphic band of ∼2 kb (data not shown), but the SstI digest revealed limited haplotype-specific polymorphism (Figure 9). The S₃′ fragment is approximately the same size as that of S₃, indicating that the genomic region around the SLFL2 gene is not rearranged (Figure 9).

Figure 9.

DNA Gel Blot Analysis of the S₃′ and S₄′ Mutants Using an SLFL2-Specific Probe.

Genomic DNA was digested with SstI and probed with a cherry SLFL2 probe. Samples are as follows: EF, Emperor Francis (S₃S₄); Nap, Napoleon (S₃S₄); 2420, JI 2420 (S₄S₄′); S₄′S₄′ selection; 2434, JI 2434 (S₃′S₄); S₃S₃′ selection; S₃′S₃′ selection; Van (S₁S₃); ML, Merton Late (S₁S₄); Victor (S₂S₃); Vic (S₂S₄); ER, Early Rivers (S₁S₂); MH, Merton Heart (S₃S₆); MG, Merton Glory (S₄S₆); Mermat (S₁S₆).

Cloning of the S₄′-Haplotype–Specific SFB

If SFB is the pollen-S gene in cherry, a mutation affecting the S₄-SFB is expected because no duplication of the S₃-SFB was found in the S₄′-haplotype. When the coding region of the S₄′-SFB gene was cloned and the sequence aligned with the S₄-SFB sequence of the progenitor (Figure 10), a deletion of 4 bp was found toward the 3′ end of the S₄′ sequence (position 742 to 745) in the variable region A (as defined in Ushijima et al., 2003), which would result in a frame shift in translation. From that point onwards, the corresponding S₄′ protein contains 16 amino acids distinct from the S₄-SFB predicted protein sequence and then a premature stop codon would result in a truncated protein. Because the variable regions of the protein are altered significantly and the protein is truncated (317 versus 375 amino acids), the S₄′-SFB protein is unlikely to function properly.

Figure 10.

Alignment of Partial Genomic DNA Sequences and Deduced Amino Acid Sequences of the S₄- and S₄′-Haplotype–Specific SFB Genes.

Nucleotide (nt) and predicted amino acid (aa) sequences representing P. avium SFB sequences for S₄ and S₄′. Putative intron sequence within the 5′ untranslated region is presented in lower case and indicated by asterisks. Twelve amino acids are missing at the 3′ end of the S₄ sequence. Residues highlighted in gray denote divergence from the original S₄-SFB sequence. The F-box region and the two variable regions Va and Vb as identified by Ushijima et al. (2003) are boxed. A 4-bp deletion in the S₄′-SFB sequence at position 742 to 745 in the first variable region leads to the deletion of a single Tyr residue (#) and a subsequent frame shift in translation. Premature stop codons present in the S₄′-SFB sequence are highlighted in black.

DISCUSSION

A genetic analysis of two self-compatible cherry pollen-part mutants has confirmed that both selections have self-compatibility linked with one particular allele (i.e., S₃ in JI 2434 [denoted S₃′] and S₄ in JI 2420 [denoted S₄′]), confirming the conclusions of Bošković et al. (2000). A molecular characterization of S locus genes has shown that, in both mutant haplotypes, the S-RNase gene is intact, but each has a mutation affecting the haplotype-specific SFB gene. No evidence of a duplication was found in either mutant selection. In the S₃′-haplotype, a rearrangement has been found downstream of the S-RNase gene, with the breakpoint of the rearrangement ∼3.5 kb from the S-RNase gene. In this rearrangement, the S₃-haplotype–specific SFB gene, which is located ∼6.5 kb downstream of the S₃-RNase gene in the progenitor, is deleted. In the S₄′-haplotype, a 4-bp deletion in a variable region of the S₄-haplotype–specific SFB gene leads to a shift in the translation reading frame and premature termination of the protein.

Self-Compatibility Is Associated with Loss of Function of SFB

Prunus SFB genes have been reported as good candidates for the pollen component of SI. They are pollen-expressed genes with haplotype-specific polymorphism identified at the S locus in almond (P. dulcis) (Ushijima et al., 2003) and Japanese apricot (P. mume) (Entani et al., 2003). The amino acid identity between different alleles is comparable to S-RNase alleles (Entani et al., 2003; Ushijima et al., 2003), and like S-RNases, they show a trans-specific pattern of evolution (Ikeda et al., 2004). Our finding that two self-compatible pollen-part mutants of cherry are associated with a loss of function of the haplotype-specific SFB gene provides additional support that the SFB gene is the pollen-S gene in Prunus.

Loss of function of the S₃-SFB gene in the S₃′-haplotype is the result of a deletion of the gene (Figure 8). A variant of the S₄-SFB protein is still likely to be produced in the S₄′ mutant, but the protein is truncated as the result of a premature stop codon (Figure 10). The C terminus of the S₄′-SFB protein will have 16 amino acids of novel sequence and 58 amino acids missing. These alterations are likely to have a significant effect on the protein structure and function. Even though the F-box domain at the N terminus may be unaffected, the variable domains at the C terminus, which are thought to be involved in specific interactions with proteins recruited for degradation (see later), would be changed or lost. The mutant protein is therefore unlikely to perform its normal specific function, and the mutation can be regarded as a loss-of-function mutation. During the preparation of this article, Ushijima et al. (2004) reported identical findings for the S₄′-haplotype. They speculate that the truncated S₄′ protein, which lacks the two variable regions at the C terminus, may still function partially. Their conclusions are different from ours, as discussed later.

Previously, pollen-part mutations causing self-compatibility in plants with RNase-based SI have been found to be associated with duplications of S alleles (Brewbaker and Natarajan, 1960; Pandey, 1967; Golz et al., 1999, 2001). If SFB is indeed the pollen-S gene, the S₃′ mutant characterized in this article would be an unambiguous example of loss of function of the pollen-S gene for an RNase-based SI system.

It is curious that the only two irradiation-induced pollen-part mutants identified in the Rosaceae, studied in this article, appear to be the result of loss-of-function mutations, whereas at least five pollen-part mutants of P. inflata (Brewbaker and Natarajan, 1960) and seven pollen-part mutants of N. alata (Golz et al., 1999, 2001), resulting from similar experiments in the Solanaceae, were attributable to duplications of an S allele, often on a centric fragment. Pandey (1965) also found that 34 pollen-part mutants of N. alata carried a centric fragment. This led Golz et al. (2001) to suggest that deletions of the pollen-S gene are incompatible on any style with S-RNases. If their conclusion is valid, then finding a mutation/deletion of SFB in two pollen-part mutants would indicate that SFB is not the pollen-S gene.

However, it may be that certain S-bearing chromosomes in solanaceous species are particularly prone to breaking as a result of ionizing radiation. In these studies, it was often found that after irradiation of, for example, an S₃S₆ plant, most pollen-part mutants had a duplication of the S₃ allele, indicating that the mutation rate for some alleles is higher than for others. Also, centric fragments carrying the S locus may be frequent in solanaceous species because the S locus is known to be located at the centromere in Petunia (Entani et al., 1999), which may not be true for the S locus in Prunus. Lewis (1961) points out that the absence of the prime-type mutation in the studies of Brewbaker and Natarajan (1960) is not unexpected when taking into account the number of irradiated pollen grains involved and assuming that the mutation rate for the prime-type mutation in Petunia is the same as in his studies of Oenothera. However, it is currently unknown if Oenothera has RNase-based SI.

It should be noted that a loss-of-function approach was not attempted in the transgenic experiments with S locus F-box genes reported in P. inflata (Sijacic et al., 2004) and Antirrhinum/P. hybrida (Qiao et al., 2004b), in which an extra copy of an S locus F-box gene was shown to cause breakdown of SI because of competitive interaction. This may have resulted from the assumption that these F-box proteins inactivate non-self S-RNases, in which case a loss of function should lead to pollen being incompatible on any style with S-RNases. The mode of action of the RNase-mediated SI reaction in the Rosaceae may not be fundamentally different from that in the Solanaceae or Scrophulariaceae because it has recently been suggested they have a common origin based on phylogenetic analyses of S-RNase sequences (Igic and Kohn, 2001; Steinbachs and Holsinger, 2002). It will be interesting to find out whether knockout mutants of S locus F-box genes in the Solanaceae, which is more amenable to transformation than the Rosaceae, also become self-compatible. This would be another way to demonstrate that specific S locus F-box genes encode the pollen component of RNase-mediated SI in this family, in addition to the gain-of-function experiments reported so far.

Organization of the S Locus

The preliminary restriction map of the S₃-haplotype (Figure 8) shows that the SFB gene is located downstream of the S-RNase gene in opposite transcriptional direction, as was found in other Prunus S haplotypes, although distances may vary (Entani et al., 2003; Ushijima et al., 2003; Yamane et al., 2003). This organization is also found for the AhSLF-S₂ and the S₂-RNase gene in Antirrhinum (Lai et al., 2002). By contrast, the two genes determining stigma and pollen specificity at the sporophytic S locus in Brassica are found in various relative orientations (Boyes et al., 1997; Cui et al., 1999; Watanabe et al., 2000). If additional S locus haplotypes continue this pattern, the convergent arrangement of the two genes at the S locus in families with RNase-based SI may be of functional significance.

In previous studies of pollen-part mutants in the Solanaceae, no breakpoint of rearrangements could be identified at a molecular level (Thompson et al., 1991; Golz et al., 1999, 2001). The size of the P. avium genome (2C = 686 Mbp) is only twice that of Arabidopsis (2C = 343 Mbp), whereas genomes of solanaceous species are much larger (e.g., N. alata 2C = 4753 Mbp; P. hybrida 2C = 3283 Mbp) (Bennett and Leitch Angiosperm DNA C-values database, release 4.0, January 2003, http://www.rbgkew.org.uk/cval/homepage.html). The smaller size of the Prunus genome may have increased the likelihood of detecting an irradiation-induced breakpoint. The relatively small genome size makes Prunus a suitable model for studies of the genomic organization of the S locus.

A wide range of mutations can be obtained from x-irradiation, such as deletions, translocations, inversions, and even point mutations. The rearrangement in the S₃′-haplotype appears to be relatively major, and preliminary analysis suggests a deletion. Sequencing of the S-RNase/SFB intergenic PCR product for the S₃-haplotype and comparing it to sequences downstream of the S₃-RNase in the S₃′-haplotype, obtained from an ∼4.5-kb SstI inverse PCR fragment, identified the breakpoint of the rearrangement ∼3 kb from the S-RNase (data not shown). Primers based on the new flanking sequence in the S₃′-haplotype amplified a PCR product of unique size in samples with S₃ and S₃′ and cosegregated with the S₃ allele in a set of 48 seedlings from the mapping progeny Napoleon (S₃S₄) × P. nipponica (S_aS_b) (Bošković and Tobutt, 1997) (data not shown), suggesting that the new flanking sequence originates from the S₃ locus. A deletion type rearrangement is therefore more likely than a translocation.

Other S locus F-box genes do not appear to be affected by the rearrangement. In particular, the presence of the SLFL2 gene, which is located downstream of the S-RNase in other Prunus S haplotypes, including S₄ of cherry (Ushijima et al., 2004), flanked by the same restriction sites as in the S₃-haplotype, suggests that the S₃′ rearrangement is not very extensive and may not affect other genes. The simplest interpretation is that the S₃′-haplotype has a deletion extending from ∼3 kb downstream of S-RNase to a point in between SFB and SLFL2. However, this cannot be certain without further analysis of the entire S₃- and S₃′-haplotypes. No evidence of a rearrangement has been found by DNA gel blot analysis in sequences flanking the S-RNase and the SFB genes in the S₄′-haplotype and the 4-bp deletion in the coding region of the S₄′-SFB gene appears to be the only significant difference in the mutant S₄-haplotype.

Loss of Function of SFB and Models of RNase-Mediated SI

Although the mechanism of the RNase-based SI reaction is still largely unknown, the model currently favored is the RNase inhibitor model, variously proposed and developed by McClure et al. (1989), Thompson and Kirch (1992), Kao and McCubbin (1996), and Luu et al. (2000) (2001). Because pollen tubes appear to take up S-RNases indiscriminately (Luu et al., 2000), they must have a mechanism to inactivate the cytotoxic action of non-self S-RNases. In the original inhibitor model, the pollen component is assumed to be an RNase inhibitor inside the pollen tube, which is able to inhibit all S-RNases except the one of corresponding S genotype. In this model, pollen-S encodes a protein that has the two functions of inhibiting non-self S-RNases and conferring specificity. Luu et al. (2001) proposed a modification of the inhibitor model, in which the two functions of S-RNase inhibition and allele-specific recognition are carried by two separate proteins, a general RNase inhibitor, not necessarily encoded by the S locus, and pollen-S. Binding of pollen-S to its corresponding S-RNase would protect it from being inhibited by the general RNase inhibitor. The S-RNase therefore remains active and inhibits pollen tube growth. This version allows for deletions of pollen-S, in contrast with the original inhibitor model as pointed out by Golz et al. (2001), although to date there have been no unambiguous reports of pollen-S deletions in plants with RNase-based SI.

The recent findings of F-box proteins as candidates for the pollen-S component in three families with RNase-mediated SI have suggested that ubiquitin-mediated protein degradation by the 26S proteasome is involved in the mechanism to inactivate non-self S-RNases (Lai et al., 2002; Entani et al., 2003; Ushijima et al., 2003; Sijacic et al., 2004). Biochemical studies in Antirrhinum have supported the role of ubiquitin-mediated protein degradation in compatible pollinations (Qiao et al., 2004a). They showed a nonallelic physical interaction of the S locus F-box protein AhSLF-S₂ with S-RNases. The SLF proteins were also found to interact with other proteins involved in the ubiquitin-mediated protein degradation pathway. In addition, proteasomal inhibitors blocked compatible pollinations but had little effect on incompatible pollinations. As pointed out by Zhou et al. (2003), it is difficult to establish whether the Antirrhinum AhSLF gene is the ortholog of SFB in Prunus, so whether studies in Antirrhinum apply to Prunus remains to be seen.

In all three families, it has been proposed that the polymorphic S-locus F-box protein (SLF or SFB) is an inactivator of non-self S-RNases by recruiting non-self S-RNases as a substrate for destruction through the 26S proteasome pathway (Lai et al., 2002; Entani et al., 2003; Ushijima et al., 2003, Sijacic et al., 2004). A different, allele-specific interaction (of unknown nature) of the F-box protein with self S-RNases would prevent these from being degraded. This is in accordance with the original inhibitor model in which the pollen-S component carries the two functions of S-RNase inhibition and allele-specific recognition. Ushijima et al. (2004), who recently reported that the cherry S₄′ and P. mume S_f alleles encode truncated SFB proteins, attempt to reconcile their findings with this model by proposing that the truncated proteins lack the specific S-RNase interaction domain but retain a general S-RNase binding domain, leading to inactivation of all S-RNases, including the self S-RNase.

However, in the original inhibitor model, a loss of function of the haplotype-specific F-box gene should lead to universally incompatible pollen because pollen tubes would lack a mechanism to inhibit S-RNases. In this article, we have provided evidence of loss of function of SFB in at least one self-compatible mutant of cherry. Therefore, if SFB is the pollen-S gene in Prunus, it cannot have the role of inactivator of non-self S-RNases.

It seems more likely that SFB proteins provide specificity to the inactivation of S-RNases effected by a general inactivation mechanism present in pollen tubes. This would fit the two-component inhibitor model (Luu et al., 2000). Deletions of the SFB gene would then result in pollen tubes able to inactivate all S-RNases, leading to self-compatibility. In that case, SFB proteins should only prevent self S-RNases from being degraded (not recruit non-self S-RNases for degradation), and so it is difficult to understand the function of their F-box domain. The F-box domain suggests that SFB may act as part of an E3 ubiquitin ligase complex in targeted protein degradation via the 26S proteasome. However, E3 ubiquitin ligase activity has not been demonstrated for SFB, and it may function in an unexpected way. As discussed, Qiao et al. (2004) provide evidence that ubiquitin/26S proteasome activity is essential in compatible but not in incompatible interactions in Antirrhinum. This suggests that a general inhibitor may use the ubiquitin/26S proteasome pathway of protein degradation for the inactivation of non-self S-RNases, but that in incompatible interactions, in which self S-RNases are protected from being degraded, the ubiquitin/26S proteasome pathway is not involved.

Sims and Ordanic (2001) first suggested that ubiquitination and protein degradation may play a role in the SI reaction in P. hybrida. In a yeast two-hybrid assay, they identified a pollen-expressed protein (PhSBP1) with a RING-finger domain that binds to S-RNases in a non-allele–specific manner. Many proteins with a RING-finger domain participate in E3 ubiquitin ligase complexes, like F-box proteins, and they suggest that PhSBP1 is a candidate for the general inhibitor of S-RNases, the existence of which was proposed in the two-component inhibitor model of Luu et al. (2000). It may be that a RING-finger domain protein with E3 ubiquitin ligase activity exists also in Prunus as a general inactivator of S-RNases.

It is clear that other components are involved in the complex interactions of the SI reaction, such as perhaps an S-RNase binding RING-finger domain protein homologous to PhSBP1 as just discussed or other pollen-expressed F-box proteins that are known to be encoded by the S locus (Lai et al., 2002; Entani et al., 2003; Ushijima et al., 2003; Sijacic et al., 2004), although Qiao et al. (2004) report that other F-box proteins identified in Antirrhinum do not interact physically with S-RNases. Further biochemical investigations using the mutants identified in this article and their progenitors are needed to shed light on the mechanism of S-RNase inactivation and the precise role of SFB in protecting self S-RNases from inactivation.

Self-Compatibility and Cherry Breeding

JI 2420, the S₄′ mutant, has been used successfully in breeding programs combining self-compatibility with high fruit quality. One of its seedlings, Stella, was the first self-compatible cherry cultivar released, and nearly all other self-compatible cherry cultivars that are now available derive from Stella. JI 2434 has not been used extensively in breeding, but it may be in the parentage of Alex. Although the recorded parentage of Alex indicates a distinct mutant selection, JI 2538, the S₃′ allele detected (Sonneveld et al., 2003) displays the same EcoRI shift on DNA gel blots with the S-RNase probe as the S₃′ allele of JI 2434 (data not shown).

Self-compatibility is an important agronomic character in sweet cherry, and a PCR-based test can now be developed for the mutant S haplotypes, so seedlings can be selected for self-compatibility soon after germination. Previously, this had to be done by detecting the S-RNases in these haplotypes, which do not differ from the normal S₃ and S₄ alleles, and self-compatibility had to be deduced from the parentage of the seedlings (Bošković et al., 1997, 2000; Sonneveld et al., 2001, 2003).

METHODS

Plant Material

Two cherry (Prunus avium) pollen-part mutant selections raised at the John Innes Institute from the nominally incompatible cross of the cultivars Emperor Francis (S₃S₄) × Napoleon (S₃S₄, x-ray pollen) were used in this investigation, JI 2420 and JI 2434, with mutations affecting the S₄ and the S₃ alleles, respectively (Lewis, 1949; Matthews and Lapins, 1967; Matthews, 1970). The particular clones used were JI 2420 with genotype S₄S₄′ and JI 2434 EM with genotype S₃′S₄ (Bošković et al., 2000). The JI 2434 AH clone, which was subsequently found to be a possible self of JI 2434 EM with the genotype S₃′S₃′ (data not shown), was used as an example of an S₃′S₃′ homozygote.

To confirm that the pollen-part mutation in JI 2420 is linked with S₄, the progeny of Erika (S₁S₃) × Lapins (S₁S₄′) [Van × Stella (JI 2420 × Lambert)] was scored in 1999 using the stylar ribonuclease assay of Bošković and Tobutt (1996). To confirm that self-compatibility in JI 2434 is linked with S₃, the backcross Van (S₁S₃) × 9239-3 (S₁S₃′) [Van × JI 2434 (S₃′S₄)] was made in 2000 and the progeny genotyped by PCR using consensus primers for the first intron of cherry S-RNases (Sonneveld et al., 2003).

The S₄- and S₄′-SFB genes were cloned from the cultivars Inge (S₄S₉) [and later also Napoleon (S₃S₄)] and Sonata (S₃S₄′), respectively. Sonata, Lapins (S₃S₄′) × [Van × Stella (S₃S₄′)] is derived from JI 2420 via Lapins and Stella. Cosegregation of the putative S₄-SFB sequence with S₄ was tested in 44 seedlings of family 9007, Bradbourne Black (S₃S₅) × Merton Late (S₁S₄).

The following samples were included in DNA gel blot and S allele-specific PCR analyses: JI 2420 (S₄S₄′), JI 2434 (S₃′S₄), Emperor Francis (S₃S₄), and Napoleon (S₃S₄), along with standards for S₃ and S₄, Van (S₁S₃) and Merton Late (S₁S₄), respectively. In addition, for some of the analyses, four selections from family 9239 [Van (S₁S₃) × JI 2434 (S₃′S₄)] with genotypes S₁S₃′, S₃S₃′, S₁S₄, and S₃S₄, or only the S₃S₃′ selection, were used. A selection homozygous for S₄′, A53 [Stella (S₃S₄′) × self] (Bošković et al., 1998), and one homozygous for S₃′ (JI 2434 AH) were also included.

All plant material was grown at East Malling Research (UK), except for the progeny of Erika × Lapins, which was supplied by BAZ (Ahrensburg, Germany).

DNA Extraction

Cherry genomic DNA for DNA gel blotting was extracted from winter buds essentially as described by Dellaporta et al. (1983). DNA for genomic PCR and cloning was extracted from two dormant buds using a scaled down cetyl-trimethyl-ammonium bromide extraction method (Doyle and Doyle, 1987) with the addition of 1% (v/v) β-mercaptoethanol and 2% (w/v) polyvinyl pyrollidone (PVP 40) to the extraction buffer.

DNA Gel Blotting

Four micrograms of genomic DNA was digested overnight with the enzymes listed in Table 5 (S-RNase blots), DraI (SFB blot), or DraI and SstI (SLFL blots). The fragments were separated on 0.8% agarose gels for ∼24 h at 25 V and blotted onto a positively charged nylon membrane according to the manufacturer's instructions (Roche Molecular Biochemicals, Indianapolis, IN).

DIG Labeling of Probe

To detect S₁-, S₃-, and S₄-RNases, an S₁-RNase cDNA product from the C2 to the C5 region (420 bp) was used as a probe. The region covered by the S₁ probe has 82.4% sequence similarity to both S₃- and S₄-RNase. The probe was labeled with Digoxigenin-11-dUTP (DIG-dUTP; Roche Molecular Biochemicals) by PCR with the consensus primer pair PaConsII-F and PaConsII-R (Sonneveld et al., 2003). The PCR reaction mix contained the following: 1× PCR buffer, 2 mM magnesium chloride, 50 μM dATPs, dCTPs, and dGTPs, 39.6 μM dTTPs, 10.5 μM DIG-dUTPs, 0.2 μM of each of the primers, 1.25 units/50 μL reaction Taq DNA polymerase (Qiagen, Crawley, UK), and ∼20 to 100 pg plasmid DNA containing S₁-RNase cDNA (Sonneveld et al., 2001). Cycling conditions were as follows: initial denaturing of 2 min at 95°C, 10 cycles of 10 s at 95°C, 30 s at 57°C, and 2 min at 72°C, 20 cycles of 10 s at 95°C, 30 s at 57°C, and 2 min plus 5 s/cycle at 72°C, followed by 7 min at 72°C.

S₃- and S₄-SFB genomic PCR products were used as the probes to detect the S₃- and the S₄-SFB, respectively. The SFB-specific primers listed below under “Genomic PCR with S-RNase and SFB-Specific Primers” were used to amplify the regions from genomic DNA. The labeling reaction was the same as for the S-RNase probe, except for the annealing temperature of the PCR, which was 55°C.

SLFL2-specific primers designed from the P. mume S₁ and S₇ SLFL2 sequences (AB092625 and AB092626), SLFL2-forward (5′-TGGCAACSTTGAGCAAATTTTCTG-3′) and SLFL2-reverse (5′-CTATCTCATCGTCGTCGTCTTC-3′), were used for amplification of a probe from genomic DNA of Napoleon.

Hybridization and Chemiluminescent Detection of DIG-Labeled Probe

Membranes were hybridized according to the DIG system User's Guide for Filter Hybridizations (Roche Molecular Biochemicals), with 3 μL probe per mL DIG Easy Hyb hybridization buffer. Low stringency post-hybridization washes of membranes that were hybridized with the S-RNase probe allowed the detection of cross-hybridization of the S₁-RNase probe to other S-RNase alleles: two washes of 5 min each with 2× SSC, 0.1% SDS at room temperature, followed by two washes of 15 min each with 0.5× SSC, and 0.1% SDS at 68°C. For membranes hybridized with the SFB probes, more stringent washes were used to achieve an allele-specific hybridization signal: two washes of 5 min each with 2× SSC, 0.1% SDS at room temperature, followed by two washes of 15 min each with 0.1× SSC, and 0.1% SDS at 68°C. For the chemiluminescent detection, the substrate CDP-Star (Sigma-Aldrich, Poole, UK) was used.

Cloning S₄- and S₄′-Haplotype-Specific SFB Genes

Sequence homology analysis was performed using four SFB genes identified in P. dulcis (AB092966, AB092967, AB079776, and AB081648; Ushijima et al., 2003) and three homologous SLF genes identified in P. mume (AB092621, AB092622, and AB092645; Entani et al., 2003) using the DNAstar Megalign (Madison, WI) software. Degenerate primers FBOX5′A (5′-TTKSCHATTRYCAACCKCAAAAG-3′) and FBOX3′A (5′-WATTGAGWAARRSYAAASTTTCTA-3′) were designed to anneal within conserved regions identified upstream of a putative intron within the 5′ untranslated region and towards the 3′ end of the coding sequence, respectively.

Amplification products representing genomic SFB clones were generated using proofreading KOD DNA polymerase (Invitrogen, Paisley, UK) in PCR using 100 ng of genomic DNA as template in a 30 μL reaction mix (1× KOD buffer, 0.2 mM dNTP, 1 mM magnesium sulfate, 0.5 μM forward and reverse primer, and 1 unit of KOD polymerase). PCR cycling conditions were 95°C for 2 min followed by 10 cycles of 94°C for 30 s, 60°C for 60 s with a reduction in temperature of 1°C per cycle, 68°C for 90 s then, 25 cycles of 94°C for 30 s, 50°C for 60 s, 68°C for 90 s, and a final cycle of 68°C for 10 min.

Amplification products were size fractionated by electrophoresis and DNA extracted from agarose using the QIAEX II kit (Qiagen). Amplification products generated from each cultivar were subsequently cloned into the vector pCR4-TOPO (Invitrogen) and transformed into TOP10 chemically competent cells (Invitrogen). Plasmid DNA was then prepared using a mini-spin kit (Qiagen) and inserts sequenced using M13 forward and reverse primers and internal primer, FBOX360 (5′-AGAATTTCAATGGTCTCTTTTTTCC-3′). DNA from four separate colonies containing S₄′-SFB was sequenced to substantiate fully the data collected.

Sequences for SFB sequences for S₄ and S₄′ have been submitted to the EMBL database, and accession numbers are AY649872 and AY649873, respectively. The S₄ sequence was initially cloned from the cultivar Inge (S₄S₉), but we later obtained an identical sequence from Napoleon (data not shown). Our S₄ sequences are identical to the S₄-SFB sequence now in the database (Ikeda et al., 2004).

The following sequences were used for the alignment of Figure 3: EMBL/GenBank accession numbers PaSFB4 (AY649872), PaSFB3 (AB096857), PaSFB6 (AB096858), PdSFBb (AB092967), PdSFBc (AB079776), PmSLF7 (AB092622), and PmSFB9 (AB092645).

Primers specific for the putative S₄-SFB sequence (see next section) were used in genomic PCR to test cosegregation with the S₄ allele in the family 9007 Bradbourne Black (S₃S₅) × Merton Late (S₁S₄), as described previously for the S₄-RNase specific primers (Sonneveld et al., 2001).

Genomic PCR with S-RNase and SFB-Specific Primers

PCR analysis of various selections and cultivars with S₃-RNase and S₄-RNase specific primers and an internal control was performed as described by Sonneveld et al. (2001) (2003); annealing temperatures of 60°C (S₄-RNase) and 63°C (S₃-RNase) were used. Specific primers for the S₃- and the S₄-SFB sequences (S₃-SFB, EMBL database accession number AB096857; S₄-SFB, AY649872) were designed: PaSFB3-F (S₃-SFB forward) 5′-CCACAATTTGAACGTCAGAAC-3′; PaSFB3-R (S₃-SFB reverse), 5′-GATTTCGCCATATCTCATGAC-3′; PaSFB4-F (S₄-SFB forward), 5′-TTGAACGTTTGGTCGACC-3′; PaSFB4-R (S₄-SFB reverse), 5′-TACACAAATGCAGATCTCCTATG-3′. The PAL internal control primers described by Sonneveld et al. (2003) were also included in the PCR with SFB primers. PCR conditions for the SFB primers were as described for the S-RNase allele-specific primers, with an annealing temperature of 54°C.

Intergenic PCR

PCR conditions were based on the Qiagen protocol for long PCR products. Approximately 100 ng of genomic DNA was used in a reaction mix containing the following: 1× PCR buffer (Qiagen), a final concentration of 2 mM magnesium chloride, 0.2 mM dNTPs, 0.2 μM of each of the primers, and 1.25 units/25 μL reaction Taq DNA polymerase (Qiagen). Cycling conditions were as follows: initial denaturing of 2 min at 94°C, 10 cycles of 10 s at 94°C, 1 min 30 s at 55°C, and 6 min at 68°C, 25 cycles of 10 s at 94°C, 1 min 30 s at 55°C, and 6 min plus 10 s/cycle at 68°C.

Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession numbers AY649872 and AY649873.