Research Article

The Male Determinant of Self-Incompatibility in Brassica

See allHide authors and affiliations

Science  26 Nov 1999:
Vol. 286, Issue 5445, pp. 1697-1700
DOI: 10.1126/science.286.5445.1697


In the S locus–controlled self-incompatibility system of Brassica, recognition of self-related pollen at the surface of stigma epidermal cells leads to inhibition of pollen tube development. The female (stigmatic) determinant of this recognition reaction is a polymorphic transmembrane receptor protein kinase encoded at the S locus. Another highly polymorphic, anther-expressed gene, SCR, also encoded at the Slocus, fulfills the requirements for the hypothesized pollen determinant. Loss-of-function and gain-of-function studies prove that the SCR gene product is necessary and sufficient for determining pollen self-incompatibility specificity, possibly by acting as a ligand for the stigmatic receptor.

In self-incompatible plants of the genus Brassica, self-related pollen grains are recognized and prevented from germinating by interaction with the epidermal cells of the stigma, the receptive surface of the female reproductive organ. This self-incompatibility (SI) phenomenon is genetically controlled by a complex and polymorphic locus (1, 2). Among the genes at this S locus is a pair of sequence-related genes, the cell wall–localized S-locus glycoprotein (SLG) gene and the plasma membrane–spanning receptor protein kinase (SRK) gene, both of which are expressed specifically in the stigma epidermal cells. Different variants of theS locus, designated S haplotypes, are characterized by highly polymorphic alleles for SLG andSRK (1, 3, 4). Although one function of SLG is to stabilize SRK, SRK itself is viewed as a ligand-activated receptor kinase and the primary female determinant of SI (1, 5). Selective binding of a pollen-borne ligand is thought to initiate a signal transduction cascade resulting in the SI pollen rejection reaction. This view requires that the S locus contain at least one more SI gene, an as-yet unidentified gene that would encode the male determinant and provide directly or indirectly the ligand for the SRK receptor (6–8).

Recombination analysis of the B. campestris (synonymB. rapa) S8 haplotype has shown that the male and female SI determinants are contained in a 65-kb chromosomal segment encompassing SLG8 and SRK8 (2,7). During sequence analysis of the 13-kb region betweenSLG8 and SRK8 , we discovered that the 400–base pair (bp) Hind III–Xba I restriction fragment indicated in Fig. 1A contained a small segment with an unusually high frequency of cysteine residues in one of the deducible reading frames. RNA gel blot analysis with the 400-bp Hind III–Xba I fragment as probe revealed that this segment is part of a transcribed gene that is expressed specifically in anthers (see below). Indeed, screening of an S8 anther cDNA library allowed the isolation of a 450-bp full-length cDNA clone. Sequence comparison of this clone and theSLG8 -SRK8 intergenic segment determined that the gene consists of two exons (110 and 300 bp) separated by an unusually large intron of 4.1 kb (Fig. 1A). In DNA gel blot analysis, the 400-bp Hind III–Xba I probe detected restriction fragments with sizes predicted by the restriction map (Fig. 1B). Thus, we had identified a single-copy, S locus–encoded, anther-expressed gene, which we named S locus cysteine-rich protein (SCR) gene.

Figure 1

Genomic organization of theSCR8 gene. (A) Map of theSLG8 -SRK8 intergenic region and of the SCR8 cDNA. Restriction sites: B, Bam HI; E, Eco RI; H, Hind III; S, Sac I; and X, Xba I. The translation start sites of SLG8 ,SRK8 , and SCR8 are marked by arrows. Exons are indicated by filled boxes. The bar labeled “400-bp probe” indicates the Hind III–Xba I fragment used for hybridization experiments. The positions of cysteine codons in theSCR8 cDNA are marked by arrowheads, and the positions of stop codons flanking the SCR open reading frame are marked by asterisks. (B) DNA gel blot analysis ofSCR8 . Genomic DNA fromS8S8 homozygotes, digested with Xba I (X), Sac I (S), Hind III (H), or Bam HI (B), was hybridized with the 400-bp probe (A).

S Locus–Associated Polymorphism of SCR

We isolated SCR cDNA clones from the B. oleracea strains S6 andS13 (9). These two cDNAs, designatedSCR6 and SCR13 , represent true SCR alleles because they detected the same size bands as the SCR8 probe on Brassica DNA gel blots. Furthermore, SCR6 andSCR13 cosegregated in F2populations, respectively, with theSLG6/SRK6 andSLG13/SRK13 gene pairs as well as with S6 and S13 SI specificities, demonstrating their genetic linkage to theS locus (10). The S haplotype association of SCR was also demonstrated by DNA gel blot analysis of other S homozygotes. Under low-stringency conditions, which were used in order to maximize the detection of possibly highly divergent alleles, most of the seven Shomozygotes we analyzed exhibited a single band hybridizing with theSCR probe (Fig. 2A). However, the length of the restriction fragments varied between strains, consistent with the notion that the SCR gene exists as a single-copy gene localized at the polymorphic S locus. These results confirm that the SCR gene is a consistent feature of the S haplotype (11).

Figure 2

DNA gel blot analysis of SCR in plants homozygous for the S2 ,S6 , S13 ,S14 , S22 , andS29 haplotypes of B. oleracea and theS8 haplotype of B. campestris. Gel blots of DNA digested with Hind III or Eco RI were sequentially hybridized with the SCR13 probe at 65°C (A and C) and a combinedSCR6/SCR8 probe at 55°C (B) and washed under low (A and B) or high stringency (C). Final wash solutions were (A) 2× SET, 0.5% SDS at 65°C, (B) 0.2× SSC, 0.2% SDS at 55°C, and (C) 0.2× SET, 0.1% SDS at 65°C.

Examination of the DNA gel blots also revealed that the intensity of the hybridization signal, a qualitative indication of nucleic acid sequence similarity, showed S haplotype–associated variation. Under low-stringency conditions, anSCR13 -derived probe hybridized with DNA from theS6 , S13 ,S14 , S22 ,S29 , and S8 homozygotes, albeit with varying degrees of intensity, but not with DNA from theS2 homozygote (Fig. 2A). Only simultaneous hybridization with SCR6 andSCR8 allowed the detection of a weak hybridization signal in S2 DNA (Fig. 2B). This pattern of S haplotype–associated variation mirrors that observed for the SLG/SRK gene pair: Alleles of this gene pair isolated from the S6 ,S13 , S14 ,S22 , S29 , andS8 haplotypes share >85% DNA sequence similarity with each other but only <70% similarity with alleles isolated from the S2 haplotype (3, 4). The parallelism betweenSCR and SLG/SRK was also apparent in the hybridization patterns obtained with the SCR13 probe under high-stringency conditions, which resulted in the loss of hybridizing bands in all S homozygotes except for theB. oleracea S13 andB. campestris S8 homozygotes (Fig. 2C). This relatively high interspecific sequence similarity betweenSCR13 and SCR8 , verified by sequence analysis (see below), coincides with the observation that the B. oleracea SLG13/SRK13 gene pair exhibits higher amino acid sequence similarity to B. campestris SLG8/SRK8 than to otherB. oleracea SLG/SRKs such asSLG6/SRK6 (4). In conclusion, SCR appears to have coevolved with the SLG/SRK gene pair.

Function as the Male Determinant of SI

To test whether SCR is indeed the male determinant of SI, we initiated a detailed analysis of both its expression and function. First, we analyzed the expression of SCR in wild-type self-incompatible plants by RNA gel blot analysis (12). SCR shows an anther-specific, developmentally regulated expression profile (Fig. 3A, lanes 1 to 6). SCRtranscripts can be detected in anthers only after the generation of haploid microspores, with transcripts accumulating in the microspores (Fig. 3A, lane 7). Therefore, the SCR gene is active postmeiotically and gametophytically. However, our data do not exclude expression of SCR in sporophytically derived cells of the anther such as the tapetum, a secretory cell layer that lines the anther locule in which microspores develop and that is a source for several pollen coat components.

Figure 3

Expression and functional analysis ofSCR. (A) Gel blots of polyadenylated RNA were sequentially hybridized with an SCR probe (SCR8 for lanes 1 to 7, SCR13 for lanes 8 to 10, SCR6 for lanes 11 to 14) and a Brassica actin probe. Lanes 1 to 7 contain RNA isolated from B. campestris S8S8 plants: stigmas (lane 1), anthers of 2.5- to 3.9-mm buds containing uni- and binucleate microspores (lane 2), anthers of 4.0- to 5.0-mm buds containing bi- and trinuclear microspores (lane 3), anthers of 5.1- to 6.0-mm buds containing trinuclear microspores (lane 4), anthers of open flowers containing mature pollen grains (lane 5), leaves (lane 6), and microspores collected from 4.0- to 6.0-mm buds (lane 7). Lanes 8 to 10 contain anther RNA isolated from a self-incompatible B. oleracea S13S13 homozygote (lane 8), a self-incompatible S13Sf1 heterozygote (lane 9), and the self-compatible mutant m1600(lane 10). Lanes 11 to 14 contain anther RNA from four representative S2S2 plants transformed with SCR6 , three of which express SCR6 transcripts (lanes 11, 13, and 14) and one of which lacks SCR6 transcripts (lane 12). (B to F) Pollination response of pollen from SCR6 transformants onS6S6 stigmas. (B) Pollen from a transformant expressing SCR6 [same plant as in (A), lane 11]. (C) Pollen from a transformant without detectable levels of SCR6 transcript [same plant as in (A), lane 12]. Representative details of pollinatedS6S6 stigmas display the incompatible reaction exhibited by pollen fromS6S6 plants (D) andS2S2/SCR 6 +plants (E), and the compatible reaction exhibited by pollen fromS2S2/SCR 6 plants (F). Absent or aborted short pollen tubes are indicative of an incompatible reaction and are clearly distinguished from the dense array of pollen tubes penetrating the stigma in a compatible reaction.

We also analyzed the accumulation of SCR transcripts in anthers isolated from a self-compatible mutant B. oleraceastrain, designated m1600, which was generated by γ-irradiation of an S13S13 homozygote (13). Reciprocal pollinations betweenm1600 and self-incompatibleS13S13 plants had shown that the lesion in m1600 resulted in the loss ofS13 specificity in pollen but not in stigma. Gel blot analysis of anther RNA demonstrated that, in contrast to plants carrying the wild-type S13 haplotype,m1600 anthers lacked detectable SCR13 transcripts (Fig. 3A, lanes 8 to 10). The correlation between loss of pollen S13 specificity and absence of detectableSCR13 transcripts in the m1600 mutant strain provides strong evidence that SCR is necessary for pollen SI specificity.

To obtain definitive proof that SCR functions as the pollen determinant of SI, we transformed a B. oleracea S2S2 homozygous strain with a chimeric gene consisting of the SCR8 promoter fused to the SCR6 cDNA (14). Among 14 independent hygromycin-resistant plants, 12 plants expressed theSCR6 transgene and were designatedS2S2/SCR 6 +, whereas two plants failed to produce detectable levels ofSCR6 transcript and were designatedS2S2/SCR 6 (Fig. 3A, lanes 11 to 14). Pollination assays showed that the stigmas of theS2S2/SCR 6 +andS2S2/SCR 6 plants were compatible with pollen fromS6S6 plants, a response identical to that of the S2S2 transformation host strain. Pollen from the twoS2S2/SCR 6 plants germinated and produced pollen tubes onS6S6 stigmas (15) (Table 1 and Fig. 3, C and F). In contrast, pollen from each of the 12S2S2/SCR 6 +plants was inhibited by S6S6 stigmas (Fig. 3, B and E), even though pollen was viable as demonstrated by its ability to elaborate pollen tubes on stigmas homozygous for an unrelated S haplotype such asS22 (Table 1). Thus, pollen ofS2S2/SCR 6 +transformants has acquired S6 specificity. The correlation of SCR6 expression with gain ofS6 specificity not only proves the functional involvement of SCR in SI, but also demonstrates that this gene is sufficient for determining male SI specificity. With the identification of SCR as the male SI determinant, we can now define the S locus as a tightly linked genetic unit encompassing the array of SLG/SRK and SCR genes.

Table 1

Pollen phenotype ofS2S2/SCR 6 +andS2S2/SCR 6 plants assayed on stigmas of S6 andS22 homozygotes. The data represent the number of pollen tubes observed onS6S6 stigmas in at least three experiments performed on different dates, with each experiment consisting of two to three pollinated stigmas. Control pollinations with S22S22 stigmas were performed on two different dates.

View this table:

SCR as a Potential Ligand for the S-Locus Receptor Kinase

To analyze the structure and polymorphism of the SCRgene products, we compared the deduced amino acid sequences of the three cloned SCR sequences (Fig. 4A). The SCR alleles encode polypeptides of 74 to 77 amino acids, which are hydrophilic except for an NH2-terminal hydrophobic stretch of 19 amino acids (16). Based on the compliance of amino acid residues 24 and 26 in the SCR sequences with the “(-3,-1)-rule” for cleavage of signal peptides (17), we predict that the mature SCR proteins are small (8.4 to 8.6 kD), basic (isoelectric point 8.1 to 8.4), secreted proteins. Sequence comparison of the three SCR preproteins reveals that although conservation is high in the NH2-terminal region extending two amino acids beyond the putative signal peptide cleavage site, conservation in the remaining sequence is limited to 11 amino acids, 8 of which are cysteines. Indeed, overall sequence identities are only 42, 30, and 37% for the pairwise comparisons SCR8/SCR13, SCR8/SCR6, and SCR6/SCR13, respectively. This extensive divergence of the SCR proteins is consistent with their role as male SI specificity determinants.

Figure 4

(A) Amino acid sequence alignment of the predicted SCR8 ,SCR13 , and SCR6 proteins (GenBank accession numbers AF195627, AF195626, and AF195625, respectively). Gaps, represented by hyphens, were introduced to optimize the alignment. The arrow marks the conserved site of the intron, which has a length of 4.1 kb in SCR8 , but only 0.75 kb in SCR13 and 0.88 kb inSCR6 . The putative signal peptides are underlined. The two positions complying with the “(-3,-1)-rule” for signal peptides (17) are indicated. Bold letters mark amino acids that are identical in at least two sequences. A consensus sequence is shown with numbered cysteine residues. (B) Cysteine pattern in mature proteins of the PCP family, represented by PCP-A1 (19), and the plant defensins, represented by Rs-AFP1 (21). Lines indicate the disulfide bridges determined for Rs-AFP by x-ray crystallography. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

Database searches did not identify any significant homologies. However, small cysteine-rich proteins, the pollen coat proteins (PCPs), have been discussed previously as candidates for the male SI determinant (18–20). PCPs are basic 6 to 8-kD proteins with a cysteine pattern suggesting a cysteine-stabilized tertiary structure related to that of plant defensins (21). PCPs constitute the major component in a pollen coat fraction that was reported to modify pollen SI phenotype (18), but they lack both S-locus linkage and S haplotype–associated polymorphism. Therefore, the PCPs analyzed to date are unlikely to function in SI specificity. The predicted SCR proteins, even though resembling PCPs in charge and molecular size, have a distinct cysteine pattern that is most apparent when the spacing of cysteine residues Cys3, Cys4, and Cys5 in SCR and the PCPs/defensins is compared (Fig. 4, A and B). Thus, SCRs represent a new class of small, secreted, cysteine-rich proteins, distinguishable from members of the PCP/defensin family. Nevertheless, we speculate that the even-numbered cysteine residues in mature SCR proteins are engaged in disulfide bridges, as has been shown for the cysteine residues of defensins and inferred for the ones of PCPs (19, 21). Although a cysteine-stabilized three-dimensional structure may be common to SCR proteins, we expect the amino acid stretches between the cysteine residues, varying in length and composition between SCR alleles (Fig. 4A), to form loops at the surface of the folded protein. By imparting extensive structural diversity on the small SCR polypeptide molecules, such loops could mediate specificity in the SI recognition reaction.

Having established that the SCR gene determines pollen SI specificity, we suggest that the SCR gene product represents the pollen-borne ligand postulated to activate the stigmatic SRK receptor. The small hydrophilic polypeptide predicted by theSCR sequence is expected to localize to the pollen coat after its secretion from developing microspores [similar to the secretion of other gametophytically expressed components of the pollen coat (19)] and also possibly from cells of the tapetum. In either case, SCR molecules would mix readily within the anther locule and consequently, the pollen coat of all pollen grains in anS-locus heterozygote would incorporate SCR proteins encoded by each of the two parental S haplotypes, as predicted by sporophytic control of SI in Brassica (22). SCR would translocate into the cell walls of the stigma epidermal cell through the pollen coat–stigma contact zone. In the case of self-pollination, SCR would bind to a structurally complementary SRK receptor, resulting in activation of the receptor and initiation of a signal transduction cascade that ultimately leads to pollen inhibition.

  • * To whom correspondence should be addressed. E-mail: jbn2{at}


View Abstract

Navigate This Article