|Home | About | Journals | Submit | Contact Us | Français|
The self-incompatibility response of crucifers is a barrier to fertilization in which arrest of pollen tube development is mediated by allele-specific interactions between polymorphic receptors and ligands encoded by the S-locus haplotype. The binding of the stigma-expressed S-locus receptor kinase (SRK)  to the pollen coat-localized S-locus cysteine-rich (SCR) ligand [2-5], activation of the receptor, and pollen rejection occurs only if receptor and ligand are encoded by the same S haplotype [4, 6-8]. To identify residues within the extracellular domain of SRK (eSRK) that are required for its ligand-selective activation, we assayed chimeric receptors and receptor variants containing substitutions at polymorphic sites in transgenic Arabidopsis thaliana [9, 10]. We show that only a small number of the ~100 polymorphic residues in eSRK are required for ligand-specific activation of self-incompatibility in vivo. These essential residues occur in two non-contiguous clusters located at equivalent positions in the two variants tested. They also correspond to sites showing elevated levels of substitutions in other SRKs, suggesting that these amino acids could define SI specificity in most SRKs. The results demonstrate that the majority of eSRK residues that show signals of positive selection and previously surmised to function as specificity determinants are not essential for specificity in the SRK-SCR interaction.
To understand the basis of ligand-selective activation of SRK, we focused on polymorphic residues in the eSRK, a region subject to strong diversifying selection, with extraordinarily high levels of polymorphisms that have persisted for long periods of time [11-16]. Comparison of SRK alleles, which can number over 50 in one species, has shown that their amino-acid sequences can diverge by as much as 35% in Brassica species [1, 17] and 51% in A. lyrata . Sequence alignments have demonstrated that polymorphic residues, although scattered over the length of the eSRK, are particularly prevalent in several “hypervariable regions” [17, 19], in which non-synonymous to synonymous substitution (Ka/Ks) ratios are significantly greater than 1  or elevated relative to the rest of the protein . These “hypervariable regions” also contain many of the residues having a high posterior probability of being under selective pressure to change in physicochemical property . These features have suggested that hypervariability in these regions is not due to relaxed constraint but rather to diversifying selection, and that the variable residues within these regions function as SI specificity determinants.
To test the hypothesis that SI specificity resides in the hypervariable (hv) regions, specifically hvI, hvII, and hvIII (Figure 1A), we assayed eSRK chimeras in which a segment encompassing these regions was swapped between different SRK variants in transgenic plants of the Arabidopsis thaliana C24 accession . We first generated AtS1preSRKx:AlSRKb fusions (Figure 1A; Boggs et al. submitted; Supplemental Data) in which the stigma-specific AtS1 promoter  drives expression of an SRK consisting of the eSRK (minus the last 23 amino acids) from one SRK variant (eSRKx) fused to the last 23 amino acids of AleSRKb, followed by the AlSRKb transmembrane and kinase domains. The hvI-hvIII region of an AtS1preSRKx:AlSRKb fusion was then replaced with the corresponding region from other SRK variants to generate eSRKx(y)x chimeras (Figure 1B; Supplemental Data), where “x” and “y” correspond to the number or letter of the SRK allele from which the various regions were derived, and the central swapped hvI-hvIII region is indicated in parentheses (Figure 1B). For each chimera, several independent transgenic plants were assayed by pollinating stigmas with transgenic A. thaliana pollen expressing the SCRs that correspond to the parental SRKs used in chimera construction (Supplemental Data; Figure S1). In this expression system, AtS1prAleSRKa:AlSRKb and AtS1prAleSRK16:AlSRKb (Figure 1A) confer Sa and S16 specificity, respectively: transgenic stigmas expressing these chimeras inhibit transgenic A. thaliana AlSCRa- and AlSCR16-expressing pollen, respectively, but not wild type or AlSCRb-expressing pollen (Boggs et al., submitted).
We constructed 11 AtS1preSRKx(y)x:AlSRKb chimeras (hereafter “eSRK chimeras”) (Table 1; Figure S2) using AleSRKa, AleSRKb, and AleSRK16 (which are ~62% similar), AleSRK25 (which is 82% similar to AleSRK16; Boggs et al. submitted), and Capsella grandiflora CgeSRK7 (which is 77% similar to AleSRKa ). Pollination assays of stigmas from several independent primary (T1) transformants using pollen that expresses SCR corresponding to the hvI-hvIII region of eSRK chimeras revealed that seven chimeras (Table 1; Figure S2B) failed to confer an incompatibility response, possibly due to the disruptive steric effect of combining diverged eSRK segments. However, the eSRK16(b)16:SRKb, eSRK16(25)16:SRKb, eSRKa(7)a:SRKb, and eSRKa(b)b:SRKb chimeras (Figure 1B; Figure S2A) were functional and each conferred the SI specificity of the SRK allele from which the hvI-hvIII region was derived (Table 1), as determined by pollination assays in T1 and T2 transgenic progenies (Figure 1C). Thus, residues within the ~160-aa segment spanning the hvI-hvIII region of eSRK are sufficient for determining SI specificity in transgenic Arabidopsis.
To identify residues within the hvI-hvIII region that determine SI specificity, we focused on the S7- and S25-determining regions of eSRKa(7)a and eSRK16(25)16 (Table 1). The eSRKa(7)a and eSRK16(25)16 chimeras were modified by site-directed mutagenesis (Supplemental Data) to generate a series of mutants containing single-site substitutions at each of the 42 polymorphic sites that differ between the hvI-hvIII regions of eSRKa and eSRK7, and 44 out of the 46 polymorphic sites that differ between the hvI-hvIII regions of eSRK16 and eSRK25 (Figure 2A). Each of the resulting eSRKa(7)a and eSRK16(25)16 mutants had the chimeric eSRK sequence except for one amino-acid residue within the hvI-hvIII region that was replaced with the residue found at the corresponding position in eSRKa or eSRK16, respectively (Figure 2A). The stigmas of T1 plants transformed with each of the 42 eSRKa(7)a and 44 eSRK16(25)16 mutants were pollinated with A. thaliana SCR7- and SCR25-expressing pollen (Figure S3), respectively. In all, 35 eSRKa(7)a mutants and 38 eSRK16(25)16 mutants (Figure 2A) conferred a strong incompatibility response in at least some of the T1 transformants analyzed, indicating that the mutant chimeras retained the specificity of the “wild type” eSRka(7)a or eSRK16(25)16 chimeras, and that residues that were replaced in these mutants are not required for S7 or S25 specificity.
For the remaining mutants (Figure 2A), none of the 10-18 independent T1 transformants analyzed per construct conferred a strong incompatibility response towards pollen expressing the SCR corresponding to the swapped specificity-determining region, indicating that the substituted residues are important for the S7 or S25 specificity to be manifested. The majority of these mutant eSRKa(7)a and eSRK16(25)16 chimeras produced only plants whose stigmas were fully compatible with SCR7- and SCR25-expressing pollen, respectively. However, chimeras carrying the L218V substitution in eSRKa(7)a and the Q285E substitution in eSRK16(25)16 conferred a weakened incompatibility phenotype (Figure S3), characterized by variable expressivity of the response or a degree of leakiness that allowed the germination of some SCR7- or SCR25-expressing pollen grains. This partial disruption of SRK function might be due to the partial insensitivity of the sites to volume- (L218V) or charge- (Q285E) changing substitutions.
To determine if the identified essential sites are not only necessary for eSRKa(7)a:SRKb and eSRK16(25)16:SRKb function, but are sufficient for S7 or S25 specificity, we generated a construct that combined substitutions at the seven sites essential for eSRKa(7)a function into the eSRKa backbone, and another construct that combined substitutions at the six sites essential for eSRK16(25)16 function into the eSRK16 backbone. However, neither of these two multiple-substitution mutants conferred an incompatibility response in transgenic stigmas pollinated with SCR7- or SCR25-expressing pollen.
To exclude the possibility that the non-functionality of mutant eSRK chimeras is due to their sub-optimal accumulation in stigmas, the original “wild type” eSRKa(7)a and eSRK16(25)16 chimeras, as well as their mutant derivatives, were tagged by inserting a hemagglutinin (HA) tag at the N-terminus of mature eSRK (Supplemental Data). The HA-eSRK16(25)16 chimeras, including the “wild type” chimera, were not functional, likely due to disruptive effects of the HA tag, and were not investigated further. In contrast, the HA-eSRKa(7)a chimeras recapitulated the pollination phenotypes observed with their non-tagged counterparts. Importantly, for all non-functional HA-eSRKa(7)a chimeras, T1 transformants were obtained whose stigmas accumulated SRK to levels equivalent to, or higher than, those of “wild type” HA-eSRKa(7)a in stigmas that expressed S7 specificity (Figure 2B). Thus, the non-functionality of chimeras containing substitutions at essential sites cannot be explained by reduced SRK levels in transgenic stigmas. Therefore, amino acids at essential sites most likely function as specificity-determining residues. However, the failure of multiple-substitution eSRKa(7)a and eSRK16(25)16 mutants to confer an incompatibility response suggests that additional residues in the hI-hvIII region might also be required for SRK specificity. Such residues might have escaped detection because a single amino-acid substitution at these sites might not produce a detectable effect on SRK function.
Nevertheless, the observation that the majority of substitutions (even those that changed charge, volume, or polarity) were not disruptive, indicates that the specific amino acids occupying most polymorphic sites in the hvI-hvIII region are not critical for function. Interestingly, the few sites found to be essential for eSRKa(7)a and eSRK16(25)16 function are clustered in two non-contiguous regions located at equivalent positions in the hvI and hvII regions of the two chimeras, and two of these sites were found to be essential for the function of both chimeras.
To determine if the amino-acid sites we identified as likely specificity determinants in CgSRK7 or AlSRK25 might also determine specificity in other SRK variants, we performed pairwise alignments of the hvI-hvIII region for 34 intra- or inter-specific pairs of the most closely-related eSRK variants that are known or assumed to encode different SI specificities (Figure S4). We reasoned that this strategy is preferable to comparing highly-diverged SRK sequences, which can impede the prediction of specificity-determining residues due to overall high variability in eSRKs . Alignment of pairwise consensus sequences (Figure 3, Figure S4) showed that none of the 11 essential sites identified in planta was polymorphic in all sequences. However, seven of these sites were polymorphic in >50% of the comparisons, and the two sites found to be essential for both CgSRK7 and AlSRK25 were polymorphic in >60% of the comparisons (Figure 3). Additionally, several of the 15 sites in hvI and hvII that exhibited polymorphisms in >50% of pairwise comparisons coincided with, or clustered near, the functionally-important residues identified in planta (Figure 3).
Our results provide the first empirical support for the hypothesis suggested by previous comparisons of SRK sequences [24, 27], that SCR-specific activation of SRK is a function of hvI and hvII (which are conserved in functionally-equivalent SRKs) rather than hvIII (which differs by many substitutions between functionally-equivalent SRKs). Although theory predicts high diversity at amino acids that are closely linked to sites subject to balancing selection , the finding that so very few of the residues previously shown to bear signals of positive selection (and therefore presumed to function as specificity determinants) are essential for SRK function (Figure 3) was unexpected. While some “positively-selected” sites that lie outside the hvI, hvII, and hvIII regions [22, 24] might be important for receptor functions unrelated to ligand recognition [e.g. see 29], polymorphisms at many “positively-selected” sites within the hvI-hvIII region of eSRK may have little functional importance. Rather, they might exhibit false signals of positive selection, as shown for “rapidly-evolving” regions in the human genome, in which “positively-selected” polymorphisms were found to result, not from selection, but from biased gene conversion [30-32]. Our finding that 17 out of 24 “positively-selected” sites in eSRK (Figure S4) can be replaced with other amino acids with no consequence for receptor function and ligand selectivity, together with the documented occurrence of gene conversion at the SRK gene [20, 33], suggests that a non-selective process might similarly drive accumulation of polymorphisms in eSRK.
We propose that for many if not all SRK variants, SI specificity is determined primarily by small amino-acid regions located toward the C-terminal end of hvI and C-terminal half of hvII. We further suggest that it is the overall sequence or three-dimensional conformation of these small segments, rather than individual residues within them, which determines SI specificity. The fact that residues in two discrete regions underlie SRK specificity is not surprising, as similar results were obtained for other recognition molecules [34-38]. It is tempting to speculate that the essential eSRK residues identified in this study, or the two clusters that encompass them, might be surface-exposed regions that are brought into close proximity in a three-dimensional structure to form part of an SCR-binding pocket. A high-resolution three-dimensional structure of the eSRK in its ligand-bound and unbound forms is required to address this issue. Nevertheless, three-dimensional models of eSRK sub-domains predict with confidence that hvI is a solvent-exposed segment of the LLD2 domain . Accordingly, it is not surprising that the charge-, polarity-, or volume-changing substitutions (Figure S3) that were introduced at essential residues in this region would be disruptive. In contrast, the resolution of the hvII region, which maps to the structurally distinct LLD2 and EGF-like domains, is less clear , and it is difficult to surmise how substitutions in this region might impact SRK structure and function. Future in vivo structure-function analyses of other SRK variants, together with high-resolution structural studies, will undoubtedly elucidate the contribution of essential eSRK residues to ligand binding, and help explain how large numbers of SRK and SCR variants have coevolved to maintain their highly-specific interaction.
We thank Tiffany Crispell and Lei Zhu for technical help. A. thaliana C24 seed was obtained from the Arabidopsis Biological Resource Center in Columbus, Ohio. This work was supported by US National Institutes of Health Grant GM057527 and National Science Foundation Grant IOS-0744579.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.