Evolution of the Self-Incompatibility System in the Brassicaceae: Identiﬁcation of S-Locus Receptor Kinase (SRK ) in Self-Incompatible Capsella Grandiﬂora

Home , Arabidopsis, Arabidopsis lyrata, Capsella grandiflora

Heredity (2006) 97, 283–290 & 2006 Nature Publishing Group All rights reserved 0018-067X/06 $30.00 www.nature.com/hdy Evolution of the self-incompatibility system in the Brassicaceae: identiﬁcation of S-locus receptor kinase (SRK ) in self-incompatible Capsella grandiﬂora

M Paetsch, S Mayland-Quellhorst and B Neuffer Department of Systematic Botany, University of Osnabrueck, Barbarastrae 11, Osnabrueck 49076, Germany

Self-incompatibility (SI) has been well studied in the genera sequence and structure. Our phylogenetic analysis supports Brassica and Arabidopsis, which have become models for the scenario of differing SI evolution along the two lineages investigation into the SI system. To understand the evolution (The Brassica lineage and Arabidopsis/Capsella lineage). of the SI system in the Brassicaceae, comparative analyses Our results also argue that the ancestral S-locus lacked the of the S-locus in genera other than Brassica and Arabidopsis SLG gene (S-locus glycoprotein) and that the diversification are necessary. We report the identification of six putative of S-alleles predates the separation of Arabidopsis and S-locus receptor kinase genes (SRK) in natural populations Capsella. of Capsella grandiflora, an SI species from a genus which is Heredity (2006) 97, 283–290. doi:10.1038/sj.hdy.6800854; closely related to Arabidopsis. These S-alleles display published online 14 June 2006 striking similarities to the Arabidopsis lyrata SRK alleles in

Keywords: self-incompatibility; Capsella; S-locus receptor kinase; S-alleles; Brassicaceae; natural populations

Introduction S-locus is a complex locus spanning about 20 kb up to 200 kb and containing several physically linked tran- In more than 50% of angiosperms species, outbreeding scriptional units that cosegregate with SI phenotype. A is promoted by the action of a self-incompatibility (SI) subset of genes within the S-locus-complex is highly system (Brewbaker, 1959), which allows the stigma to polymorphic as would be expected for genes involved in reject self-pollen. The Brassicaceae are characterised by a recognition. To reflect this complexity, the ‘S-alleles’ of sporophytic SI system; the SI phenotype is determined classical SI genetics are sometimes referred to as ‘S by the diploid genotype of the parent plant. The genetics haplotypes’ (eg Nasrallah and Nasrallah, 1993). Com- of SI in the Brassicaceae were deciphered in the early parative mapping of different S haplotypes has demon- 1950s by Bateman (1955) who described control by a strated that they vary not only in the sequence of their SI single Mendelian locus, the S-locus, which exists as genes but also in their order, relative orientation, and multiple alleles, each of which encodes a distinct mating spacing between the genes (Boyes and Nasrallah, 1993; specificity. Starting in the 1980s, molecular methods were Cui et al, 1999; Suzuki et al, 1999). Three highly applied to the study of SI, focussing on the genera polymorphic genes are involved in the SI response: Brassica (reviewed by Kemp and Doughty, 2003) and SLG (S-locus glycoprotein), SCR/SP11 (S-locus cystein Arabidopsis (Kusaba et al, 2001; Schierup et al, 2001; rich), and SRK (S-locus receptor kinase). The genes are Charlesworth et al, 2003a, b; Mable et al, 2003) as model tightly linked and behave as a single Mendelian locus plants for sporophytic SI. Recent molecular knowledge (S-locus). SRK is a plasmamembrane-spanning protein strongly argues for an ancient and monophyletic origin localised in stigmas and has been found to be the female of SI in the Brassicaceae (reviewed by Fobis-Loisy et al, determinant (reviewed by Kemp and Doughty, 2003). Its 2004). Thus it appears, that SI is the ancestral system in extracellular domain (Exon 1: ‘S-domain’) is highly the Brassicaceae, that self-fertility is a derived condition, polymorphic between S-alleles. SCR/SP11 has been and that self-compatible species (SC) have arisen via identified as the male determinant of the SI response mutation of the genes important for SI. (Schopfer et al, 1999; Suzuki et al, 1999). It encodes a small cystein-rich protein which is secreted to the surface of the pollen grain and shows an even higher degree of The S-locus in Brassica and Arabidopsis polymorphism than SRK.InBrassica, the majority of Molecular analyses of the S-locus region in the Brassica- polymorphisms in the S-domain are concentrated in ceae, namely in the genus Brassica, have shown that the three hypervariable (HV-) regions which have been suggested to play a role in determining allele-specificity Correspondence: M Paetsch, Department of Systematic Botany, University and thus might be an important factor in the generation of Osnabrueck, Barbarastrae 11, Osnabrueck 49076, Germany. E-mail: [email protected] of new S-haplotypes (Miege et al, 2001). HV-regions in Received 12 July 2005; accepted 3 May 2006; published online 14 Arabidopsis are not so clearly distinctive; polymorphic June 2006 sites are distributed more evenly over the complete Identification of S-alleles in Capsella M Paetsch et al 284 sequence (Schierup et al, 2001). SLG is similar to the Materials and methods extracellular domain of SRK and probably originated by a duplication of the SRK S-domain (eg Nasrallah, 1994). Plant material and DNA preparation It is found in Brassica and Raphanus (Sakamoto et al, C. grandiflora seed material was individually collected on 1998), but not in Arabidopsis lyrata (Kusaba et al, 2001; the mainland of Greece and the Greek island of Corfu, C. Schierup et al, 2001). It appears not to be essential for the rubella from different Mediterranean countries (Table 1). SI response but may have an accessory function in Progeny of 5–10 individuals of each population were stabilising the active S-receptor complex (reviewed in grown in the greenhouse. Voucher specimens are Kemp and Doughty, 2003). The current model of SI deposited in the Herbarium OSBU. DNA was extracted response is that SCR/SP11 interacts with the extracellular from young leaves (0.5–1.0 g) using the following CTAB domain of the SRK protein, and the binding of the two method (Doyle and Doyle, 1987; modified). Leaves were proteins induces phosphorylation of SRK which initiates homogenised in liquid nitrogen, mixed with 2 Â CTAB a signalling cascade leading to inhibition of self-pollen. and b-mercaptoethanol and incubated for 1 h at 601C. DNA was extracted with chloroform in two steps, and precipitated through incubation with ice-cold The genus Capsella isopropanol for 30 min. After centrifugation, pellets were The genus Capsella comprises three species, two diploids washed twice with 70% ethanol, dried, resuspended in and one tetraploid. Distribution of Capsella grandiflora 100 ml water (autoclaved HPLC-water), and stored (FAUCHEZ &CHAUB)BOISS.(2Â ) is restricted to Western at 41C. Greece, Albania and northern Italy. C. rubella REUTER. (2 Â ) is distributed in the Mediterranean, the Middle East and followed European settlers occasionally into the New World and Australia (Figure 1) (Hurka and Neuffer, Table 1 Origin of Capsella seed material 1997; H Hurka, personal communication). C. grandiflora Pop. no. Provenance Elevation represents the ancestral obligate outbreeding species (SI) a.s.l. from which two self-compatible species, C. rubella and Capsella bursa-pastoris (L.) MED.(4Â ), originated (Hurka C. grandiflora and Neuffer, 1997). The breakdown of the SI system in 900 Greece: Corfu: Acharavi 3 m 391470N, 191480E Capsella SC-species coincides with reduction in flower 923 Greece: W-Peloponnes: Vrosina 90 m size (Figure 1) (Hurka and Neuffer, 1997; Neuffer and 391390N, 201310E Hurka, 1999). Capsella appears to be closely related to 924 Greece: W-Peloponnes: Botzaras 250 m Arabidopsis (Koch et al, 2000, 2001) on the basis of 391400N, 201350E sequence data from various nuclear genes and cpDNA. 926 Greece: W-Peloponnes: Votonosi 750 m 0 0 Comparative genome analysis has revealed extensive 39146 N, 21107 E 927 Greece: W-Peloponnes: Metsovo 1150 m conservation of genome organisation among Arabidopsis 391460N, 211100E thaliana and C. rubella (Acarkan et al, 2000; Koch and 934 Greece: W-Peloponnes: Metsovo 1350 m Kiefer, 2005). Owing to this close relationship it should 391460N, 211100E be possible to make use of the present knowledge of Arabidopsis SI in research on Capsella. It is therefore a C. rubella promising new object for further investigations on the 774 Italy: Mt. Gargano; 800 m 411500N, 161000E Brassicaceae S-locus in natural populations. 1241 Spain: Sierra de Guadalupe: 511 m The overall objective of our research is to better monastery; understand the evolution of the SI system within the 391270N, 51190W Brassicaceae. There is urgent need to analyse more wild 1249 Portugal: Sintra near Lisbon; 143 m species for comparative analyses. The aim of the present 381240N, 71230W 1504 Spain: Canary Islands: La Palma; 750 m study was to identify S-locus genes in the genus Capsella, 1 0 1 0 beginning with the self-incompatible C. grandiflora, and 28 40 N, 17 52 W to relate the SI systems of Capsella to Arabidopsis and Pop. no. ¼ Population numbers in collection of Capsella Seed Bank Brassica. (University of Osnabrueck).

Capsella grandiflora 2x; self-incompatible

0.5 cm

Capsella rubella 0.5 cm 2x; self-compatible

Figure 1 Distribution and flower morphology of diploid Capsella species. The outcrossing C. grandiflora presents fragrant and attractive flowers due to dependence on pollinators and is restricted to a small area (black asterisk). In the selfing C. rubella breakdown of SI system coincides with reduction of flower size, loss of fragrance and colonising ability (hatched dots).

Heredity Identification of S-alleles in Capsella M Paetsch et al 285 Amplification, cloning and sequencing Capsella sequences were aligned with the most similar For PCR amplification primers SLGf(50-AGAACCTATG sequences indicated by BLAST (for accession numbers CATGGGTTGC-30) and SLGr(50-ATCTGACATAAA see Table 2). The degree of similarity revealed by BLAST GATCTTGACC-30) (Charlesworth et al, 2000) were used searches was on average 86% over a mean length of which were based on Brassica S-domain gene sequences 588 bp, ranging from 79% of 323 bp to 91% of 829 bp. The and are known to amplify, among others, the S-domain alignment corresponded to positions 244–1234 of 1305 bp of SRK in A. lyrata. Plant genomic DNA was mixed with of A. lyrata SRK b exon 1. Assuming that the S-domain of a pair of primers, 10 pmol each, 5 mlof10Â buffer SRK in the genus Capsella is comparable in length and (160 mM (NH4)2SO4, 670 mM TriS-HCL pH 8.8, 15 mM structure to known Brassicaceae S-domains, we appear MgCl2, 0.1% Tween-20), 3 ml of dNTP mixtures, 0.5 mlTaq to have amplified about two-thirds of exon 1 of SRK in C. DNA Polymerase (BioThermt; Genecraft), and sterile grandiflora. By setting the start codon by analogy with water to make a final volume of 50 ml. PCR conditions published SRK, the amino-acid sequences of Capsella were: denaturation at 941C for 2 min followed by 40 species were predicted (Figure 2). cycles of 941C for 1 min, 52–561C for 45 s, and 721C for 2 min, followed by final extension at 721C for 7 min. An annealing temperature of 541C appeared to be optimal. Tree topology and sequence divergence PCR was performed using a PTC-200 thermocycler (MJ The neighbour joining tree (Figure 3) assigned Capsella Research). PCR products were subjected to electro- sequences to two different groups (II, III), which are phoresis in a 1% agarose gel containing 1 Â TBE buffer. clearly separated with 100% bootstrap support from After electrophoresis at 120 V bands were stained with group I, containing the S-locus sequences of Brassica ethidiumbromide and examined under UV light. species and R. sativus. Most of the amplified PCR PCR fragments of the expected size (approximately fragments of C. grandiflora and all of C. rubella appear 1 kb) were cloned in pCR 2.1 vector using the TA cloning to be highly similar to A. thaliana ARK 3 (group III), kit (Invitrogen). Separate clones (10–20) were picked per which is expressed in roots and the root-hypocotyl individual and sequenced in both directions using transition zone (Dwyer et al, 1994). The region containing universal M13 primers. To avoid errors that may have ARK 3 is located on A. thaliana chromosome IV, contig occurred during the PCR process and sequencing, three fragment No. 55 (Genbank accession Nr. AL161555) and independent clones obtained from the same individual has been identified as a homeolog of the A. lyrata S-locus that appeared to contain SRK PCR fragments were region (Kusaba et al, 2001). The S-domain gene ARK 3 sequenced. directly flanks the S-locus in both Arabidopsis species, as DNA sequencing was done on an ABI 377 automatic was shown by Kusaba et al (2001) by comparative sequencer using Big Dye (Applied Biosystems). Se- mapping of A. lyrata and A. thaliana. These sequences quences were aligned using BioEdit 5.0.9 (Hall, 1999). are ruled out as candidates for SRK in the Capsella species due to the high degree of sequence identity (88%; Table 3) Sequence analyses between A. thaliana ARK 3 and Capsella sequences of To get an initial hint about the character of the sequences, group III, and the low degree of sequence divergence BLAST searches were performed. Capsella sequences within Capsella group III (2%; Table 3). It is more likely were aligned to published SRK/SLG sequences of that they represent an orthologue of ARK 3. Brassica species, Raphanus sativus, A. lyrata and A. thaliana In group II, six C. grandiflora sequences (pop. 900.5.1, SRK pseudogene. A second member of the S-gene family, 934.3a.2., 923.1.1, 924.6.4, 926.5.5, 926.5.3) cluster within ARK 3ofArabidopsis species, which revealed similarities to Capsella sequences in BLAST searches was included in the alignment. A neighbour joining tree was constructed Table 2 Accession numbers of Brassicaceae taxa used in analyses using the program MEGA version 2.1 (Kumar et al, 2001) using the Kimura 2-parameter distance measure. Nu- Species Gene Genbank accession no. cleotide sequence divergence (P distance; MEGA version A. lyrata Aly 13 set A AF328990–AF328993; 2.1) was calculated between alleles and within each AF328995; AF328997; species group. Support for nodes in the tree was assessed AF328998; AF329000 by bootstrapping (1000 replicates). Initial insight into the A. lyrata Aly 13 set B AY186763–AY186766; structure of sequences was gained through translation of AY186770 DNA- to amino-acid sequences. The start codon of A. lyrata SRK a/b AB052755; AB052756 A. thaliana ARK 3 NM_118258 Capsella sequences was set by analogy with the known A. thaliana SRK pseudogene NM 105252 SRK sequences used in the alignment, while recognising B. oleracea/B. SLG AB054060; D30050; that this reading frame – although likely to be correct – rapa/B. AB054815; X79431; remains tentative as long as the complete sequence of the campestris AB024417; D84468 Capsella SRK gene is unknown. B. oleracea/B. SRK X79432; D38563; rapa/B. AB054698; AB054704; campestris AB054710; AB054718; Results AB054725; AB054726 R. sativus SRK AB114846–AB114850; The primers SLGf and SLGr amplified fragments of AB114852–AB114854 B1 kb length in both Capsella species. BLAST searches R. sativus SLG AB009677; AB009681; revealed high similarities to exon 1 (or S-domain) of the AB009684 Brassicaceae S-locus receptor kinase SRK as well as to C. grandiflora Putative SRK DQ530637–DQ530642 another gene of the S-domain gene family, ARK 3 C. grandiflora Putative ARK 3 DQ534555–DQ534562 C. rubella Putative ARK 3 DQ534551–DQ534554 (Arabidopsis receptor kinase 3) of Arabidopsis species. All

Heredity Identiﬁcation of S-alleles in Capsella M Paetsch et al 286

Figure 2 Structure of putative Cgr SRK alleles. Alignment of the predicted amino-acid sequences of C. grandiﬂora putative SRK alleles (Cgr SRK 1–6). Dots pointed out the 12 conserved cystein residues of all members of the S-domain gene family. The conserved motif WQSFDXPTDT characteristic for receptor-like protein kinases (Walker, 1994) is framed by a dotted box. Asterisks mark identical amino acids, broad lines indicate regions with more than two variable sites. It becomes clear that variable regions are not concentrated in three distinct hypervariable regions reported to be exhibited in the lineage of Brassica S-locus genes (eg Nasrallah, 1994).

the A. lyrata alleles of SRK, termed Aly 13 by Charles- follow their division into sets A and B in our gene tree worth et al (2000) and SRK a and SRK b (Kusaba et al (Figure 3). Based on the high similarities of the C. (2001), respectively. Charlesworth et al (2003a) divided grandiflora sequences with Aly 13 alleles we appear to Aly 13 sequences into subsets A and B based on have identified six alleles of SRK in the self-incompatible investigations of the S-domain as well as the kinase C. grandiflora, designated Cgr SRK 1–6 in the following. domain. Their findings were combined with the results These Capsella sequences show nearly the same level of of linkage analyses gained from extensive crossing polymorphism as the A. lyrata SRK; the mean sequence experiments. As four from six of our C. grandiflora group divergence within each group is 27% for Cgr SRK and II sequences cluster in Aly 13 set B with 93% bootstrap 29% for our selection of Aly 13 sequences (Table 3). Cgr support we have to take the interpretations of Charles- SRK 1, 4, 5, 6 which cluster in Aly 13 set B exhibit a lower worth et al (2003a) into consideration and therefore degree of divergence (19%) than Cgr SRK 2 and Cgr SRK

Figure 3 Neighbour-joining tree of Brassicaceae S-domain genes. S-domain genes of four genera of the Brassicaceae are clearly separated into groups I–III. To simplify the differentiation between the genera they are marked by dots in scales of grey. In group I SRK and SLG alleles of Brassica species and R. sativus are clustered, exclusively (dark grey/light grey dots). Putative SRK alleles of C. grandiﬂora (black dots) cluster within A. lyrata SRK alleles (circles) in group II. The latter were subdivided into set A and set B by Charlesworth et al (2003a) due to striking separation of subsets in all of their gene trees, which is reported recently to be correlated with different dominance classes (Prigoda et al, 2005). Interestingly, Capsella putative S-alleles appear to show the same subdivision, suggesting that this subdivision predates separation of the genera Arabidopsis and Capsella. Note, that Cgr SRK 3 is closer to Aly 13-1 than to any other Cgr SRK allele, indicating that the two genera might share an ancient allele. Group III contains the ARK 3 gene of A. thaliana and its putative orthologue in C. grandiﬂora and C. rubella. Phylogeny was constructed using the Kimura 2-parameter distance measure and the neighbour-joining procedure. Numbers on branches represent bootstrap values 465% and are based on 1000 replicates.

Heredity Identiﬁcation of S-alleles in Capsella M Paetsch et al 287 3 (divergence 31%). The latter two sequences represent those in A. lyrata (0.29% in set A, 0.19% in set B). alleles of one individual plant and cluster within Aly 13 Calculating pairwise distances between Aly 13 alleles set A. Thus, divergences in C. grandiﬂora correlate with and the putative Capsella S-alleles of set A revealed

Heredity Identification of S-alleles in Capsella M Paetsch et al 288 striking low sequence divergence between Aly 13-1 and hypothesise that none of the Capsella sequences repre- Cgr SRK 3 of only 9% (Table 4). Such a very low value of sents an orthologue of the SLG gene. divergence is neither observed between Aly 13 alleles nor between putative Cgr SRK alleles. As SRK alleles in Structure of the sequences Brassica and Arabidopsis are characterised by a high level Twelve conserved cystein residues characteristic for of sequence polymorphism these analyses serve as a members of the S-domain gene family could be found further indication that the sequences represent SRK in in Capsella sequences of all groups (Figure 2). A Capsella. conserved block of amino acids – WQSFDXPTDT – SRK in C. rubella remains elusive. All PCR products of which is reported to be characteristic for S-domain C. rubella showed considerably higher similarities to ARK kinases (Walker, 1994) is found in all the putative Cgr 3 than to SRK and exhibit no significant sequence SRK alleles. These findings support the results of the diversity. Decreasing PCR stringency did not lead to sequence alignment and allow the identification of the amplification of products other than the putative ARK 3 sequences as S-domain kinases. Concentration of poly- orthologues in C. rubella. morphism in three distinct hypervariable regions as is known for Brassica cannot yet be demonstrated. Variable sites are distributed more evenly in the alignment as has SLG in Capsella species been described for Arabidopsis (Schierup et al, 2001). An important aspect for insight into the evolution of the Brassicaceae S-locus is the question whether the Capsella S-locus has an SLG gene or not. SLG has not been found Discussion in A. thaliana and A. lyrata so far (Kusaba et al, 2001; By applying primers from Brassica SRK to genomic DNA Schierup et al, 2001). To reveal possible similarities of of two taxa of the genus Capsella we succeeded in Capsella sequences to SLG, a selection of Brassiceae SLG amplifying six putative SRK alleles in the self-incompa- alleles were added into the alignment. All S-locus tible C. grandiflora. Evidence for this identification is sequences (SLG and SRK)ofBrassica species and their provided by the high sequence similarities of putative close relative R. sativus cluster together in group I and are Cgr SRK alleles with A. lyrata Aly 13 alleles, the high level clearly separated from the Capsella-containing groups, of polymorphism between the sequences and the shared supported by a bootstrap value of 100% (Figure 3). We structure of the putative SRK, particularly twelve conserved cystein residues and the conserved amino- acid motif WQSFDXPTDT that are located identically in Table 3 Comparison of mean pairwise distances of taxa groups Brassicaceae SRK proteins. The fact that in one individual (926.5) the primers amplified two different putative Species S-domain gene Classification Overall in the NJ gene average alleles (Cgr SRK 2, Cgr SRK 3), which cluster together tree with Aly 13 set A alleles supports this interpretation. The division of Aly 13 sequences into subsets A and B B. species SRK Group I 0.12 was introduced by Charlesworth et al (2003a) and was B. species SLG Group I 0.14 recently corroborated by Prigoda et al (2005), who succeeded in correlating this classification with different R. sativus SRK Group I 0.14 R. sativus SLG Group I 0.14 dominant classes (Class A 1–3 and Class B). Set A ( ¼ Class A) consists of SRK alleles of A. lyrata which are A. lyrata SRK (set A and B) Group II 0.29 all linked to the S-locus (Charlesworth et al, 2003b). They Aly 13 SRK set A 0.29 are expressed exclusively in flower buds and appear to Aly 13 SRK set B 0.19 be dominance to set B ( ¼ Class B) alleles (Prigoda et al, 2005). Set B contains two sequences (Aly 13-2, Aly 13-7) C. grandiflora putative SRK Group II 0.27 Cgr 2/3 (assigned to set A) 0.31 which are not linked to the S-locus, presumably due to Cgr 1/4/5/6 (assigned to 0.19 sequence exchange between the S-locus and one or more set B) unlinked loci. They are not, therefore, candidates for being SRK alleles. The other sequences of set B are linked A. thaliana ARK 3/Capsella Group III 0.12 to the S-locus and also represent SRK alleles, but they are ARK 3 type more similar to the unlinked sequences Aly 13-2/Aly 13-7 Capsella ARK 3 type 0.02 and show well-supported separation from alleles from

Table 4 Pairwise nucleotide distances between set A S-alleles of Arabidopsis and Capsella Aly 13–20 Aly 13–13 Aly 13–22 Aly 13–19 Aly 13–5 Aly 13–23 Aly 13–1 Cgr SRK 3 Cgr SRK 2

Aly 13–20 Aly 13–13 0.29 Aly 13–22 0.27 0.32 Aly 13–19 0.28 0.30 0.28 Aly 13–5 0.27 0.20 0.31 0.29 Aly 13–23 0.27 0.32 0.26 0.30 0.31 Aly 13–1 0.27 0.30 0.27 0.32 0.31 0.28 Cgr SRK 3 0.30 0.31 0.28 0.34 0.33 0.29 0.09 Cgr SRK 2 0.29 0.30 0.31 0.30 0.31 0.31 0.31 0.33

Heredity Identification of S-alleles in Capsella M Paetsch et al 289 set A in all gene trees (Charlesworth et al, 2003a). Prigoda that has been reported for A. lyrata (Charlesworth et al, et al, (2005) found that Class B sequences are recessive to 2003a; Prigoda et al, 2005) and, recently, for A. halleri all except one Class A haplotype. The majority of (Castric and Vekemans, 2004). One putative allele of C. putative Cgr SRK alleles (1, 4, 5, 6) cluster within Aly grandiflora assigned to Aly set A, Cgr SRK 3, differs by 13 set B sequences. It is possible that some of our only 9% from A. lyrata S-allele Aly 13-1 (Table 4), which sequences represent orthologues of an unlinked locus, reflects a remarkably high degree of sequence identity, similar to the findings in A. lyrata (Charlesworth et al, especially between S-alleles of two different genera. So it 2003a). It remains open whether the division of Aly 13 becomes conceivable that both genera share ancient into sets A and B is transferable to our data. alleles. The existence of common alleles and the division Final evidence for identification of our sequences as in set A and set B would, then, appear to have occurred Capsella SRK as well as establishing dominance relation- not only before speciation within the genus Arabidopsis, ships between S-alleles will be gained through linkage but even before separation of the two genera Arabidopsis analyses. Extensive crossing experiments addressing and Capsella. It is clear that the understanding of the these questions are currently being performed. evolution of the SI system in Brassicaceae would be enhanced by further studies of S-loci in natural popula- SRK in self-compatible C. rubella tions of other Brassicaceae genera from these two lineages. Comparative analyses may well reveal intra- Primers failed to amplify any candidate for SRK in C. and interspecific differences in sequence and structure of rubella. Instead, another member of the large S-domain presently known Brassicaceae S-locus genes. gene family was amplified, which was identified as an orthologue of ARK 3(Arabidopsis receptor kinase 3). All sequenced PCR products show considerably more similarities to the A. thaliana ARK 3 gene than to any Acknowledgements other S-domain gene of the Brassicaceae. The most We thank D Charlesworth for helpful discussion of our obvious explanation of these results is that the primer results, W Bleeker and H Hurka for comments on the set was not suitable for C. rubella SRK. The primer set manuscript, U Coja for technical assistance and the DFG has, however, proved capable of amplifying several (German research foundation) for financial support. S-domain genes in one Capsella species. It could be suggested then, that it should amplify more than one member of the S-domain gene family in C. rubella. Thus, it remains an open question, whether C. rubella has an References intact S-domain or not. Acarkan A, Rossberg M, Koch M, Schmidt R (2000). Compara- tive genome analysis reveals extensive conservation of Evolution of the SI system in Capsella species genome organisation for Arabidopsis thaliana and Capsella By adding information from Capsella, we have strength- rubella. Plant J 23: 55–62. ened the hypothesis that SI in the Brassicaceae has Bateman AJ (1955). Self-incompatibility systems in angiosperms evolved differently in two lineages (reviewed by Fobis- III. Cruciferae. Heredity 9: 53–69. Loisy et al, 2004): one including Brassica and Raphanus, Boyes DC, Nasrallah JB (1993). Physical linkage of the SLG and the other Arabidopsis and Capsella. Evidence is provided SRK genes at the self-incompatibility locus of Brassica by the clear and well-supported separation of group I, oleracea. Mol Gen Genet 236: 369–373. S- Brassica Brewbaker JL (1959). Biology of the angiosperm pollen grain. containing locus genes of species and their Indian J Genet Plant Breed 19: 121–124. close relative R. sativus, from group II, in which putative Charlesworth D, Awadalla P, Mable BK, Schierup MH (2000). S-alleles of C. grandiflora cluster among S-alleles of A. Population-level studies of multiallelic self-incompatibility lyrata. Like Arabidopsis, Capsella appears to have no SLG loci, with particular reference to Brassicaceae. Ann Bot gene, and variable sites are not concentrated in three 85(Suppl. A): 227–239. distinct hypervariable regions known of Brassica and Charlesworth D, Bartolome C, Schierup MH, Mable BK (2003a). Raphanus species. Capsella therefore appears to have an S- Haplotype structure of the stigmatic self-incompatibility locus similarly structured to Arabidopsis (Schierup et al, gene in natural populations of Arabidopsis lyrata. Mol Biol 2001), which supports the model of an ancient S-locus Evol 20: 1741–1753. complex lacking the SLG gene (reviewed by Fobis-Loisy Charlesworth D, Mable BK, Schierup MH, Bartolome C, Awadalla P (2003b). Diversity and linkage of genes in the et al, 2004). The high level of allelic variation in self-incompatibility gene family in Arabidopsis lyrata. Genetics Arabidopsis Aly 13 alleles and Capsella putative SRK 164: 1519–1535. alleles (425% sequence divergence in the Arabidopsis/ Castric V, Vekemans X (2004). Plant self-incompatibility in Capsella lineage, compared to o15% sequence diver- natural populations: a critical assessment of recent theore- gence in the Brassica/Raphanus lineage; see Table 3) tical and empirical advances. Mol Ecol 13: 2873–2889. implies their antiquity. Cui Y, Brugiere N, Jackman L, Bi YM, Rothstein SJ (1999). Our present results support the suggestion that S- Structural and transcriptional comparative analysis of the alleles of the Arabidopsis/Capsella lineage diverged earlier S-regions in two self-incompatible Brassica napus lines. Plant than those of Brassica (Fobis-Loisy et al, 2004). The Cell 11: 2217–2231. Brassica Arabidopsis Capsella Doyle JJ, Doyle JL (1987). A rapid DNA isolation procedure for lineages of and / were esti- small quantities of fresh leaf tissue. Phytochem Bull Bot Soc mated to have separated approximately 12.2–19.5 million Am 19: 11–15. years ago, whereas for Arabidopsis and Capsella, the Dwyer KG, Kandasamy MK, Mahosky DI, Acciai J, Kudish BI, estimate is of 6.2–9.8 million years (Acarkan et al, 2000). Miller JE et al (1994). A superfamily of S-locus-related Putative Capsella S-alleles seem to exhibit the same sequences in Arabidopsis: diverse structures and expression division into at least two sets or classes (set A and set B) patterns. Plant Cell 6: 1829–1843.

Heredity Identification of S-alleles in Capsella M Paetsch et al 290 Fobis-Loisy I, Miege C, Goude T (2004). Molecular evolution of Miege C, Ruffio-Chable V, Schierup ME, Cabrillac D, Dumas C, the S-locus controlling mating in the Brassicaceae. Plant Biol Gaude T et al (2001). Intrahaplotype polymorphism at the 6: 109–118. Brassica S-locus. Genetics 159: 811–822. Hall TA (1999). BioEdit: a user-friendly biological sequence Nasrallah JB (1994). Evolution of the Brassica self-incompat- alignment editor and analysis program for Windows 95/98/ ibility locus: A look into S-locus gene polymorphism. Proc NT. Nucl Acids Symp Ser 41: 95–98. Natl Acad Sci USA 94: 9516–9519. Hurka H, Neuffer B (1997). Evolutionary processes in the genus Nasrallah JB, Nasrallah ME (1993). Pollen-stigma signaling in Capsella (Brassicaceae). Plant Syst Evol 206: 295–316. the sporophytic self-incompatibility response. Plant Cell 5: Kemp BP, Doughty J (2003). Just how complex is the Brassica 1325–1335. S-receptor complex? J Exp Bot 54(Plant reproductive biology Neuffer B, Hurka H (1999). Colonization history and introduc- special issue): 157–168. tion dynamics of Capsella bursa-pastoris (Brassicaceae) in Koch M, Haubold B, Mitchell-Olds T (2000). Comparative North America: isozymes and quantitative traits. Mol Ecol 8: evolutionary analysis of chalcone synthase and alcohol 1667–1681. dehydrogenase loci in Arabidopsis, Arabis, and related genera Prigoda NL, Nassuth A, Mable BK (2005). Phenotypic and (Brassicaceae). Mol Biol Evol 17: 1483–1498. genotypic expression of self-incompatibility haplotypes in Koch M, Haubold B, Mitchell-Olds T (2001). Molecular Arabidopsis lyrata suggests unique origin of alleles in different systematics of the Brassicaceae: evidence from coding dominance classes. Mol Biol Evol 22: 1609–1620. plastidic matK and nuclear chs sequences. Amer J Bot 88: Sakamoto K, Kusaba M, Nishio T (1998). Polymorphism of 534–544. the S-locus glycoprotein gene (SLG) and the S-locus related Koch MA, Kiefer M (2005). Genome evolution among crucifer- gene (SLR1)inRaphanus sativus L. and self-incompatible ous plants: a lecture from the comparison of the genetic maps ornamental plants in the Brassicaceae. Mol Gen Genet 258: of three diploid species: Capsella rubella, Arabidopsis lyrata 397–403. subsp. petraea, and Arabidopsis thaliana. Amer J Bot 92: Schierup MH, Mable BK, Awadalla P, Charlesworth D (2001). 761–767. Identification and characterization of a polymorphic receptor Kumar S, Tamura K, Jakobsen IB, Nei M (2001). MEGA2: kinase gene linked to the self-incompatibility locus of Molecular Evolutionary Genetics Analysis Software, Version 2.1, Arabidopsis lyrata. Genetics 158: 387–399. Arizona State University, Tempe: Arizona, USA. Schopfer CR, Nasrallah ME, Nasrallah JB (1999). The male Kusaba M, Dwyer K, Hendershot J, Vrebalov J, Nasrallah JB, determinant of self-incompatibility in Brassica. Science 286: Nasrallah ME (2001). Self-incompatibility in the genus 1697–1700. Arabidopsis: characterization of the S-locus in the outcrossing Suzuki G, Kai N, Hirose T, Fukui K, Nishio T, Takayama S et al A. lyrata and its autogamous relative A. thaliana. Plant Cell 13: (1999). Genomic organization of the S locus: Identification 627–643. and characterization of genes in SLG/SRK region of an Mable BK, Schierup MH, Charlesworth D (2003). Estimating the S9 haplotype of Brassica campestris (syn. rapa). Genetics 153: number, frequency, and dominance of S-alleles in a natural 391–400. population of Arabidopsis lyrata (Brassicaceae) with sporo- Walker JC (1994). Structure and function of the receptor-like phytic control of self-incompatibility. Heredity 90: 422–431. protein-kinases of higher-plants. Plant Mol Biol 26: 1599–1609.

Heredity