Identification of the pollen self-incompatibility determinant in Papaver rhoeas.

Higher plants produce seed through pollination, using specific interactions between pollen and pistil. Self-incompatibility is an important mechanism used in many species to prevent inbreeding; it is controlled by a multi-allelic S locus. 'Self' (incompatible) pollen is discriminated from 'non-self' (compatible) pollen by interaction of pollen and pistil S locus components, and is subsequently inhibited. In Papaver rhoeas, the pistil S locus product is a small protein that interacts with incompatible pollen, triggering a Ca(2+)-dependent signalling network, resulting in pollen inhibition and programmed cell death. Here we have cloned three alleles of a highly polymorphic pollen-expressed gene, PrpS (Papaver rhoeas pollen S), from Papaver and provide evidence that this encodes the pollen S locus determinant. PrpS is a single-copy gene linked to the pistil S gene (currently called S, but referred to hereafter as PrsS for Papaver rhoeas stigma S determinant). Sequence analysis indicates that PrsS and PrpS are equally ancient and probably co-evolved. PrpS encodes a novel approximately 20-kDa protein. Consistent with predictions that it is a transmembrane protein, PrpS is associated with the plasma membrane. We show that a predicted extracellular loop segment of PrpS interacts with PrsS and, using PrpS antisense oligonucleotides, we demonstrate that PrpS is involved in S-specific inhibition of incompatible pollen. Identification of PrpS represents a major advance in our understanding of the Papaver self-incompatibility system. As a novel cell-cell recognition determinant it contributes to the available information concerning the origins and evolution of cell-cell recognition systems involved in discrimination between self and non-self, which also include histocompatibility systems in primitive chordates and vertebrates.

It has been established that self-incompatibility (SI) has evolved independently several times. Three SI systems have been well characterised at a molecular level1,2. Both pollen and pistil S determinants are expected to have co-evolved and be physically linked to the S locus in order to maintain a functional SI system. Other characteristics expected of them are high levels of allelic polymorphism and tissue specific expression. Most importantly, they should function in mediating the SI response. To fully understand how different SI systems operate, identification of both the pistil and pollen S locus components, together with establishing mechanisms involved in pollen inhibition is crucial. Previously, we identified the pistil S determinant for Papaver rhoeas8-10 and established several components involved in pollen inhibition3-7,11. Although we identified a glycoprotein in pollen that bound to the pistil S protein, studies indicated that it was not the pollen S determinant, though it might modulate the SI response12. Recent analysis of the S locus enabled identification of the pollen component of the S locus on a cosmid clone comprising a 42 kb region at the S locus.

Nucleotide sequencing and analysis identified a novel putative open reading frame (ORF) 457 bp from the S pistil gene (Fig 1a). Expression analysis using RT-PCR revealed that the ORF was specifically transcribed in pollen (Fig 1b), appearing during anther development (Fig 1c). The temporal expression pattern is very similar to that of the pistil S gene8. These data suggested that this ORF was a candidate for the Papaver pollen S gene (designated PrpS: ). We propose renaming the gene that determines SI specificity in pistils (currently designated simply S) to provide a clearer nomenclature; we suggest PrsS (for Papaver rhoeas stigma S determinant).

Fig 1

Organization and expression of PrpS

(a) Organization of the S locus. Arrows indicate pistil PrsS and pollen PrpS coding sequences and their orientation; transcription start site (+1). An intron is located 84 bp from the 3′ end.

(b) RT-PCR shows PrpS is expressed in pollen (P) but not in leaf (L), ovary (O) or stigma (S) (top panel). (c) RT-PCR showing PrpS expression increases during anther development. Immature anthers (IA); anthers 3 (-3A), 2 (-2A), 1 (-1A) days pre-anthesis, and at anthesis (+1A). Glyceraldehyde-3-phosphate dehydrogenase (GAPD) shows equal loading (bottom panels).

(d) Alignment of PrpS, PrpS and PrpS deduced amino acid sequences. The predicted extracellular loop segment (TMHMM) is indicated for PrpS1 (grey box).

(e) Linkage of PrpS and PrsS to the S locus. Full-sib families segregating for haplotypes (i) S and S and (ii) S and S were used for PCR. Pistil PrsS-, PrsS- and PrsS- and pollen PrpS, PrpS and PrpS sequences were amplified only if plants carried the corresponding PrpS allele. S haplotypes are indicated: S (1,3), S (3,8), S (1,8).

The cDNA of PrpS comprises 1206 bp containing a coding region of 579 bp encoding a 192 amino acid polypeptide with a predicted Mr of 20.5 kDa, pI 7.55. We subsequently cloned PrpS and PrpS from S pollen RNA. The PrpS and PrpS coding sequences are 576 bp and 582 bp (191 and 193 amino acids) respectively (Fig 1d); PrpS and PrpS encode proteins of predicted Mr of 21.1 kDa (pI 6.57) and 20.9 kDa (pI 8.51) respectively. Southern blotting revealed that PrpS is single copy (Supplemental Fig 1), so the related sequences identified as PrpS and PrpS are clearly allelic to PrpS, rather than being related/paralogous genes.

Segregation analyses were conducted to obtain evidence of genetic linkage at the S locus between PrsS and PrpS and between PrsS and PrpS. Specific primers were used to amplify regions of the pistil PrsS, PrsS and PrsS and pollen PrpS, PrpS and PrpS genes from genomic DNA from two full-sib families segregating for these haplotypes (45 and 25 individuals, a total of 140 PrpS/PrsS pairs). PrpS was amplified only from plants carrying S; PrpS was amplified only from plants carrying S and PrpS was amplified only from plants carrying the S haplotype. The pistil PrsS, PrsS and PrsS sequences were also amplified only from plants carrying the respective S haplotypes (Fig 1e), as expected. This demonstrates co-segregation and linkage of PrpS, PrpS and PrpS and their cognate PrsS genes, as no recombination was detected (recombination frequency <0.021). Thus, at the 95% rejection level we can be confident that there is no recombination.

A strikingly high level of allelic sequence polymorphism is a well-documented feature of S locus proteins; S alleles have unusually high amino acid sequence divergence within species13-15. Papaver is no exception; the pistil proteins PrsS1 and PrsS3 exhibit 46% sequence divergence; PrsS1 and PrsS8 have 40%, and PrsS3 and PrsS8 46% divergence. PrpS proteins exhibit a similar level of polymorphism (Fig 1d); the PrpS1 and PrpS3 predicted amino acid sequences are 50% divergent; PrpS1 and PrpS8 exhibit 40% divergence; PrpS3 and PrpS8 are 47% divergent16.

The pollen and pistil S determinants should exhibit evidence of co-evolution. Examination of the PrpS sequences for non-synonymous to synonymous (Ka/Ks) substitutions reveal that the PrsS alleles have a mean Ka/Ks ratio of 0.234, and the PrpS alleles have a mean Ka/Ks ratio of 0.368 (Supplemental Table 1). A two-tailed t-test showed no significant difference between substitution rates in PrpS and PrsS genes. These data suggest that the pollen and pistil S alleles co-evolved and are likely to be similarly ancient.

PrpS has no significant sequence homology to any protein in existing databases. Sequence analysis, using a range of prediction programmes, indicated that PrpS has 3-5 predicted transmembrane helices and alignment of the three PrpS alleles indicates that they share a similar topology (Fig 1d; Supplemental Fig 2). In support of predictions that PrpS is a transmembrane protein, western analysis using antisera raised against PrpS1, revealed that PrpS1 was detected as a ~20 kDa protein specifically in S pollen membrane-enriched extracts (Fig 2a-c). Moreover, immunolocalization studies revealed that PrpS1 is associated with the pollen tube plasma membrane (Fig 2d). Although the PrpS sequences do not show any particular bias for the “positive inside rule”17, structural predictions suggest an extracellular loop segment, comprising amino acids ~60-100 (63-97 using TMHMM, Fig 1d; Supplemental Fig 2). We hypothesized that this region might be involved in the interaction with PrsS and show that a peptide corresponding to part of the predicted PrpS extracellular loop interacted with the PrsS protein, while the corresponding randomized peptide did not (Fig 2e).

Fig 2

PrpS is pollen membrane-associated

(a) Western blotting detects PrpS1 at ~20 kDa (arrow) in pollen (P), but not stigma (S), leaf (L) or root (R) membrane-enriched protein extracts (left-hand panel). Coomassie staining shows equal loading (right-hand panel).

(b) PrpS is expressed in pollen samples carrying S, but not other alleles (left-hand panel). Coomassie shows equal loading (right-hand panel).

(c) Western blot of fractionated pollen extracts. PrpS is not present in cytosolic extracts, but is present in a Triton-X-100 enriched fraction.

(d) Immunolocalization shows PrpS1 localization to the pollen tube plasma membrane. Scale bar, 10 μm.

(e) PrpS1 binds PrsS1. A 15-mer peptide corresponding to part of the PrpS1 predicted 35 amino acid extracellular loop region (DQKWVVAFGTAAICD) binds recombinant PrsS1 in a concentration-dependent manner (top panel). A corresponding randomized peptide (FTVDVKDCAAAWGQI) did not bind PrsS1 (bottom panel). Concentrations are as indicated; see Methods for details; n=8.

To determine whether PrpS is functionally involved in the SI response, we investigated whether it mediates S specific pollen inhibition, using in vitro SI bioassays8. Peptides based on extracellular domains of receptors have been used to identify ligand-binding epitopes via their ability to block the receptor-ligand interaction18. Preliminary experiments with the peptide used in the binding assay tested if it could block SI-mediated inhibition. Pollen from plants with haplotypes S, when challenged with incompatible recombinant PrsS, were rescued from inhibition by PrpS1 peptides (n = 6; P < 0.001, ***), while randomized peptides based on the same amino acids had no effect (n=3; N.S.; Supplemental Fig 3). This is consistent with the hypothesis that this region is involved in recognition and indicated that PrpS might mediate pollen inhibition. To confirm this possibility, we used an antisense oligonucleotide approach3,19. We hypothesized that if PrpS functions as the pollen S determinant, knockdown of its expression should result in alleviation of pollen tube inhibition in an S-specific manner. We induced SI in vitro in the presence of either antisense- (as-ODNs) or sense- (s-ODNs) oligonucleotides to test this hypothesis. As our plants are heterozygous for S haplotypes, the pollen phenotype of plants with S haplotypes should theoretically be 50% S and 50% S. Thus, if the interaction is S-specific, as-ODNs specific for PrpS should only affect the 50% of pollen (carrying S).

SI induced strong inhibition of pollen tube growth (a 79% reduction in length compared to the controls) and we observed a significant alleviation of this inhibition in an incompatible combination in the presence of as-ODNs and not with corresponding s-ODNs (Fig 3). With pollen from plants with S haplotypes, SI induced strong inhibition of pollen tube length (22.1%, P<0.001 ***, n=300) and addition of as-PrpS-ODNs gave a highly significant recovery of SI-treated tubes (58.3% increase in length compared to SI-treated; P<0.001, ***, n=150), and incompatible pollen responded in a bimodal manner, consistent with only S pollen being affected (Supplemental Fig 4). When as-PrpS-ODNs were added to the same pollen from plants with S and S haplotypes, they did not alleviate SI-induced inhibition (P=0.604, N.S., n=150). This demonstrates that the PrpS and PrpS as-ODNs had an S-specific effect. As expected, s-PrpS-ODNs did not affect the SI response (P=0.591, N.S., n=150).

Fig 3

PrpS determines S-specific pollen inhibition

PrpS and PrpS antisense oligonucleotides (as-ODNs: as-PrpS, as-PrpS) “rescue” pollen from plants with S or S haplotypes, from SI-induced inhibition in an S-specific manner, while PrpS and PrpS sense oligonucleotides (s-ODNs: s-PrpS, s-PrpS) do not. Controls: untreated pollen and as-ODNs without SI induction controls (white bars); SI-induced pollen (black bars); SI-induced in presence of as-ODN (crosshatched bars); SI-induced in presence of s-ODNs (dotted bars). 50 pollen tubes were measured in three independent experiments (150 in total); error bars indicate s.e.m.

To further confirm the S-specific effect of the as-ODNs, we also tested their effect on pollen from plants with haplotypes S. SI resulted in inhibited pollen tubes (19.8% of the control, n=300), and addition of as-PrpS-ODNs alleviated the SI-induced inhibition, giving a highly significant 100.3% increase in pollen tube length (P=<0.001, ***, n=150), whereas there was no effect using as-PrpS-ODNs (P=0.336, N.S., n=150) or s-PrpS-ODNs (P=0.565, N.S., n=150) (Fig 3). Together these data demonstrate that PrpS plays a crucial role in SI-induced S haplotype-specific pollen tube inhibition.

In summary, we have cloned a polymorphic pollen-expressed gene, PrpS. Together our data are consistent with the hypothesis that PrpS is the Papaver pollen S determinant as it mediates S-specific recognition and inhibition. SI in Papaver is distinct from the other well characterized SI systems (i.e. from both the Brassica pollen S determinant SCR/SP11 and the pollen F-box protein SLF/SFB from the S-RNase based SI system)1,20-23. As PrpS has no homologues, its nature is intriguing. Self-nonself discrimination and other recognition systems which are controlled by a highly polymorphic locus are not limited to SI; other systems include disease resistance in plants24, and histocompatibility systems in animals25-27. These parallels between non-analogous recognition systems were recognized, and their importance appreciated, long before the molecular basis of these systems were elucidated25 and the nature of their polymorphism has intrigued population and evolutionary biologists for decades. The identification of PrpS as a novel cell-cell recognition determinant thus contributes to the available information regarding the evolution of self-nonself recognition systems.

METHODS SUMMARY

Cloning of PrpS, PrpS and PrpS

A genomic clone of PrpS was identified by nucleotide sequence analysis of a 42 kb clone carrying the S locus, obtained by screening a Papaver rhoeas S cosmid genomic DNA library (SuperCos1, Stratagene) with PrsS cDNA. The DNA upstream and downstream of the PrsS gene was sequenced and analyzed using BLAST (http://ncbi.nlm.nih.gov/BLAST) and ORF Finder (http://searchlauncher.bcm.tmc.edu)28. The organisation of PrpS and PrsS genes was confirmed using PCR on genomic DNA of S- and non-S- containing plants. PrpS and PrpS cDNAs were obtained using RT-PCR, 3′ and 5′-RACE PCR (see On-line Methods for primer details) on pollen cDNA from suitable S-haplotypes, using low annealing temperatures (48 °C).

Ka/Ks calculations

DNAsP29 was used to estimate the Ka (the number of non-synonymous substitutions per non-synonymous site), and Ks (the number of synonymous substitutions per synonymous site) for pairs of PrsS and PrpS nucleotide sequences.

Peptide binding assay

A 15 aa peptide (DQKWVVAFGTAAICD) corresponding to part of the predicted extracellular loop segment of PrpS1 (TMHMM; http://www.cbs.dtu.dk/services/TMHMM​30) and a randomized version (FTVDVKDCAAAWGQI) were synthesized (Alta Bioscience, University of Birmingham). The peptides (10 μg, 1 μg, 0.1 μg) were bound to PVDF membrane. This was incubated with recombinant PrsS1 and then probed for binding using α-PrsS1 antisera and alkaline phosphatase detection.

Antisense oligo silencing of PrpS expression

Phosphorothioated gene-specific antisense oligodeoxynucleotides (as-ODN) and their sense controls (s-ODN) were designed (PrpS as-ODN: gtccTCCCAGTATTAttga, PrpS s-ODN: tcaaTAATACTGGGAggac, PrpS as-ODN: ttccCACCAGCACAGCaatt, PrpS s-ODN: aattGCTGTGCTGGTGggaa; lower case letters indicate bases linked by phosphorothioate bonds). Pollen was grown in vitro and pre-treated with as-ODNs and s-ODNs3,19 for 1 h prior to induction of SI with recombinant PrsS1, PrsS3 and PrsS8​8. After 2 hours, pollen tubes were fixed in 2% formaldehyde and 150 pollen tube lengths were measured in three independent experiments.