A Stigmatic Gene Confers Interspecies Incompatibility in the Brassicaceae

Articles https://doi.org/10.1038/s41477-019-0444-6

A stigmatic gene confers interspecies incompatibility in the Brassicaceae

Sota Fujii 1,2,3*, Takashi Tsuchimatsu4, Yuka Kimura1, Shota Ishida1, Surachat Tangpranomkorn1, Hiroko Shimosato-Asano2, Megumi Iwano2,7, Shoko Furukawa2, Wakana Itoyama2, Yuko Wada2, Kentaro K. Shimizu 5,6 and Seiji Takayama 1,2*

Pre-zygotic interspecies incompatibility in angiosperms is a male–female relationship that inhibits the formation of hybrids between two species. Here, we report on the identification of STIGMATIC PRIVACY 1 (SPRI1), an interspecies barrier gene in Arabidopsis thaliana. We show that the rejection activity of this stigma-specific plasma membrane protein is effective against distantly related Brassicaceae pollen tubes and is independent of self-incompatibility. Point-mutation experiments and functional tests of synthesized hypothetical ancestral forms of SPRI1 suggest evolutionary decay of SPRI1-controlled interspecies incompatibility in self-compatible A. thaliana. Hetero-pollination experiments indicate that SPRI1 ensures intraspecific fertilization in the pistil when pollen from other species are present. Our study supports the idea that SPRI1 functions as a barrier mechanism that permits entrance of pollen with an intrinsic signal from self species.

he pre-zygotic reproductive barrier is a biological mechanism pistil-side S-locus receptor kinase (SRK)10. The lock-and-key-like that has an important role in speciation. Most eukaryotes pos- specific interaction between these two molecules induces incom- Tsess such a mechanism preventing fertilization before it can patibility, resulting in the rejection of self pollen grains. Similar form a costly unfavourable hybrid. This mechanism can be split into but distinct molecules have been adopted in the SI system of the incongruity and incompatibility functions1,2. Papaveraceae, in which the receptor-like protein PrpS is the pollen An incongruity is caused by passive loss of gene functions determinant of SI11 and directly binds to the ligand PrsS, which is involved in the fertilization process, and may be a driver of repro- secreted from the pistil side. The interaction between PrpS and PrsS ductive isolation during speciation1,2. The Dobzhansky–Muller triggers a series of signals in the pollen that lead to cell death12,13. model suggests that an incongruity may result from evolution- The unilateral incompatibility relationship between species is ary divergence of two species in any of the genes involved in the known as the ‘SI × SC rule’: self-compatible (SC) species are prone relationship between male and female. In angiosperms, the rela- to accept the pollen tubes of self-incompatible species, whereas pol- tionship between the pistil-side factor LURE13, a pollen attrac- len is rejected in reciprocal crosses. This rule is widely distributed tant secreted from the synergid cells, and its pollen-side receptor among angiosperm families such as Solanaceae14,15 or Brassicaceae16. PRK64 is a prominent example. Efficient interactions between these Some factors involved in SI might also be involved in the expression two molecules occur in a species-specific manner4, and divergence of interspecies incompatibility. For example, the absence of S-RNase of one or both molecules could have resulted in interspecific breed- or the SLF complex could explain unilateral incompatibility between ing barriers. some Solanaceae species17–20. While this type of interspecies incom- By contrast, incompatibility is an active reproductive barrier patibility may evolve as a byproduct of the SI system, most occur- that rejects an undesired partner1. For instance, numerous studies rences of the SI × SC rule, other than in Solanaceae, cannot be have been conducted to date on another active reproductive barrier, explained by known SI-related molecules. Ecological contexts, such self-incompatibility (SI). SI is a mechanism that prevents inbreed- as reproductive interference, an interspecific reproductive interac- ing by discrimination of self and non-self, and is widely distrib- tion that may reduce the fitness of one or both species21, could be uted in eukaryotes5–7. In fungi, mating types are determined by the an important driver for the evolution of pre-zygotic interspecies interaction of pheromones with specific receptors5. In a tunicate, incompatibility. However, how such an active interspecific incom- a tightly linked polycystin 1-related male receptor and fibrinogen- patibility system that does not result as a consequence of SI might like ligand female genes comprise the self-recognition pair that have evolved remains unclear. induces SI8. In some angiosperm species such the Solanaceae, a In this study, we investigated a pistil factor involved in the pre- cytotoxic pistil (female)-side protein S-RNase and a pollen (male)- vention of pollen-tube entrance from different species in the model side detoxification protein S-locus F-box (SLF) protein govern the plant Arabidopsis thaliana. We found an interspecies incompatibility non-self-recognition-based SI mechanism9. The determinants of SI factor specifically localized in the plasma membrane of stigma, and in the Brassicaceae are a pollen ligand factor, the S-locus protein 11 named the protein STIGMATIC PRIVACY 1 (SPRI1). SPRI1 was (SP11, also known as the S-locus cysteine-rich protein), and the required to efficiently reject pollen from many other Brassicaceae

1Graduate School of Agricultural and Life Sciences, The University of Tokyo, Tokyo, Japan. 2Graduate School of Biological Sciences, Nara Institute of Science and Technology, Nara, Japan. 3Japan Science and Technology Agency, Precursory Research for Embryonic Science and Technology, Saitama, Japan. 4Department of Biology, Graduate School of Science, Chiba University, Chiba, Japan. 5Department of Evolutionary Biology and Environmental Studies, University of Zurich, Zurich, Switzerland. 6Kihara Institute for Biological Research, Yokohama City University, Yokohama, Japan. 7Present address: Graduate School of Biostudies, Kyoto University, Kyoto, Japan. *e-mail: [email protected]; [email protected]

Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants 731 Articles NatuRe PlantS species, and was functionally unrelated to SI. Function of SPRI1 has To understand the ancestral function of this gene, we searched been lost multiple times in A. thaliana, in line with the SI × SC rule. for and found a single phylogenetic orthologue of SPRI1 (AlSPRI1) SPRI1 function is required to ensure full ovule fertilization when in Arabidopsis lyrata, a self-incompatible relative of A. thaliana pollen from another species is simultaneously present. that is estimated to have diverged at least 6 Myr ago27. Although genes encoding a PGG-domain protein with size similar to SPRI1 Results (200 to 300 amino acids) were commonly found in angiosperms, Identification of an interspecific incompatibility related gene clear phylogenetic orthologues were only found in the Brassicaceae. locus via genome-wide association study. SC evolved in A. thaliana Therefore, we reconstructed the ancestral coding sequences of via disruption of the S-locus and the loss of SI22. In accordance SPRI1 in the genus Arabidopsis using SPRI1 homologues from with the SI × SC rule, we hypothesized that interspecies incompat- other Brassicaceae species as outgroups (Fig. 2c and Supplementary ibility could also occur during the process of decay in this species, Fig. 4b). These simulated sequences (Anc_N4 and Anc_N5) were which should be apparent as within-species genetic variation in each fused to the SPRI1 promoter and introduced into the spri1-1 incompatibility phenotypes. We used a single nucleotide polymor- mutant. Transgenic mutants carrying either of the ancestral coding phism (SNP) dataset for A. thaliana23 in a genome-wide associa- sequences or AlSPRI1 rejected M. littorea pollen grains (Fig. 2d). tion study (GWAS) to find relevant genes. To initiate the study, we This experiment confirmed that pollen rejection is the ancestral pollinated pistils of eight strains roughly representing the genetic state of SPRI1 in A. thaliana. diversity of A. thaliana with pollen from six different Brassicaceae SPRI1A and SPRI1B differ by two amino acids, and SPRI1B has species. One of the clearest phenotypic variations was found when an additional extended C terminus generated by a frameshift at the Malcolmia littorea was used as the pollen donor in crosses with stop codon (Fig. 2e). To test whether these residues contribute to A. thaliana (Fig. 1a and Supplementary Fig. 1). Only a few M. littorea the functional divergence of the two haplotypes, we created a series pollen grains germinated on stigmas of A. thaliana Col-0, whereas of SPRI1 variants by introducing point mutations into SPRI1A and many germinated on stigmas of the Cvi-0 strain (Fig. 1b). Therefore, SPRI1B (Fig. 2f) and introduced them into spri1-1. Pollination we performed pollination assays using M. littorea pollen and stig- assays indicated that the extension of the C terminus or the simulta- mas from 338 A. thaliana strains. The GWAS identified a locus neous mutation of the two amino acids in SPRI1A (E2V and N92H) (P < 10−15; using the accelerated mixed mode23) on chromosome 4 compromised its function in rejecting M. littorea pollen (Fig. 2g). It (Fig. 1c). The SNP with the strongest association was located in should be noted that the 92nd amino acid in the ancestral A. thaliana At4g13266 (Fig. 1d), a previously uncharacterized gene encoding a protein is histidine, suggesting that the E2V mutation is the sub- 221-amino-acid protein. stitution event that caused the loss-of-function during evolution (Fig. 2e). We also found at least five other frameshift events caus- Stigmatic Privacy 1 is required to reject pollen from another spe- ing loss-of-function of SPRI1 that are phylogenetically independent cies. We expressed a fusion protein comprising At4g13266 with (Fig. 2a) and geographically diverse (Supplementary Fig. 5). In sum, the fluorescent protein Venus at its C-terminus in leaf epidermal these results suggest that function of SPRI1 has been lost at least cells or stigma. In line with the subcellular localization predicted seven times independently during the evolution of A. thaliana. using SUBAcon24, Venus fluorescence was observed in the plasma membrane and cytosolic regions (Supplementary Fig. 2a, b). The SPRI1 is functionally independent from SI. We investigated results of our promoter-activity study (Supplementary Fig. 2c) whether SPRI1 is engaged to the SI mechanism. In a previous study and transcriptome analysis of laser-microdissected stigma sam- we reconstructed SI in the A. thaliana C24 strain by introducing ples25 (Supplementary Fig. 2d), as well as the TRAVA database26 SP11 and SRK25. We used the genome-editing system to manipulate (Supplementary Table 1) all suggest that this gene is specifically the line expressing SRK, inserting a premature stop codon at the expressed in the stigma. To test the role of this gene in interspecies 64th codon of SPRI1 by creating a two-base-pair deletion (Fig. 3a). incompatibility, we used two independent transfer DNA insertion As with the spri1 mutants, the genome-edited mutant was unable to lines of At4g13266 (Supplementary Fig. 3a, b). Although wild-type reject M. littorea pollen grains (Fig. 3b,c) but was still able to reject Col-0 stigmas largely rejected M. littorea pollen grains, stigmas of pollen from the line expressing SP11 (Fig. 3d,e). These results indi- both mutant lines permitted pollen-tube entrance (Fig. 1e). A simi- cate that the SPRI1 pollen-rejection mechanism is independent of SI. lar level of M. littorea pollen-tube entrance was observed in the two alleles, and both were considered to be null alleles. We named the Col-0 pollen can mentor the penetration of M. littorea pollen gene STIGMATIC PRIVACY 1 (SPRI1) for its function in prohib- tubes into A. thaliana stigma. SPRI1 was required for specific iting entrance of foreign pollen tubes into the stigma. SPRI1 was rejection of M. littorea pollen, and had no effect on the entrance of predicted to contain four transmembrane domains (Supplementary A. thaliana pollen tubes (Fig. 1e). We therefore hypothesized that Fig. 3c) and a PGG domain, according to the pfam database (pfam SPRI1 rejects a specific type of pollen. To examine the nature of ID: PF13962). The PGG domain is present in 70 angiosperm spe- such a pollen factor involved in this recognition process, we per- cies, and is considered to be ubiquitous, although its biochemical formed pollination assays using a micromanipulator. Single polli- function is unknown. SPRI1 has a predicted molecular mass of nation of M. littorea pollen grains showed that SPRI1 reduces the 25.1 kDa and a predicted isoelectric point (pI) of 9.24. rate of pollen germination and prevents pollen-tube penetration (Fig. 4a,b and Supplementary Video 1). However, when the M. lit- SPRI1 function was lost multiple times in the self-compatible torea pollen grains were placed adjacent to the Col-0 grains, 41.0% A. thaliana. To investigate the evolutionary trajectory of SPRI1, we of their pollen tubes penetrated into the stigma cells (Fig. 4c,d and sequenced its coding regions from 270 A. thaliana strains and con- Supplementary Video 2) without affecting the penetration of adja- structed a haplotype network based on amino acid polymorphisms cent pollinated Col-0. (Fig. 2a, Supplementary Fig. 4a and Supplementary Tables 2 and 3). We named the two major haplotypes that exhibited different phe- SPRI1 rejects pollen from distantly related Brassicaceae spe- notypes as SPRI1A (rejects M. littorea pollen) and SPRI1B (does not cies. We then pollinated pollen from 24 species broadly covering reject M. littorea pollen) (Fig. 2a), and introduced the genomic frag- the genetic diversity of Brassicaceae, from a common ancestor esti- ments containing these genes into the spri1-2 mutant. As expected, mated date from approximately 27 Myr ago28 to the spri1-1 mutant, the mutants containing SPRI1A rejected M. littorea pollen grains, to understand the relationship between phylogeny and SPRI1 func- but those with SPRI1B did not (Fig. 2b). tion. In general, species that are relatively distant to A. thaliana were

732 Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants NatuRe PlantS Articles

a c M. littorea Chr. 1 Chr. 2Chr. 3Chr. 4Chr. 5 15

) 10

A. thaliana –log( P 5

0 01020 010010 20 0100 10 20 Locus (Mb)

b d e ♂M. littorea ♂A. thaliana 15 ♂M. littorea Chr. 4 Stigmatic papillae ] 10

Style ♀ Col- 0

–log[ P 5 ♀ Col- 0 Ovary 0 7,670,000 7,690,000 7,710,000 7,730,000 Locus (bp) AT4G13210 AT4G13240 AT4G13266 (SPRI1) ♀ spri1- 1 Pollen tubes ♀ Cvi- 0

AT4G13250 AT4G13260 AT4G13220 AT4G13261 AT4G13230 AT4G13263

AT4G13235 ♀ spri1- 2

Fig. 1 | Identification of an interspecific incompatibility-related gene locus via GWAS. a, Photographs of A. thaliana and M. littorea open flowers. Scale bars, 2 mm. b, Representative fluorescent images of aniline-blue-stained A. thaliana Col-0 and Cvi-0 pistils 6 h after pollination with M. littorea pollen grains. Numbers of pistils pollinated: Col-0, n = 6; Cvi-0, n = 5. All replicates showed similar results. Scale bars, 200 µm. c,d, Manhattan plot of the GWAS. Pollination assays performed using M. littorea pollen and stigmas from 338 A. thaliana strains. The GWAS was performed using the platform on the GWA-Portal website. SNPs with minor allele frequencies >0.15 are shown. The horizontal line indicates the nominal P < 0.05 threshold after Bonferroni correction. d, Magnified view of the peak region on chromosome 4. e, Fluorescent images of aniline-blue-stained pistils after pollination. Numbers of pistils pollinated: M. littorea pollen, n = 12 for Col-0, n = 12 for spri1-1 and n = 5 for spri1-2; A. thaliana pollen, n = 12 for Col-0, n = 12 for spri1-1 and n = 5 for spri1-2. All replicates showed similar results. Scale bars, 200 µm. Chr., chromosome. rejected by SPRI1 (Fig. 5a,b). This was consistent with the identifica- (Supplementary Fig. 7a, b), and competition between M. littorea tion of SPRI1 orthologues only in species from lineage I that are rela- and Col-0 pollen tubes at the stage of polyspermy block appeared tively close to A. thaliana, and their absence from the species outside unlikely to be the cause of seed-set reduction. lineage I (Supplementary Fig. 6). To determine whether SPRI1 can We further performed a hetero-pollination experiment with flu- distinguish the species origin of pollen, we introduced the SPRI1 orescently labelled A. thaliana pollen and M. littorea pollen. In the orthologue from Capsella rubella (CrSPRI1) into spri1-1. In contrast control without pre-pollination by M. littorea pollen, we observed to A. thaliana SPRI1A, which rejected pollen from both M. littorea fluorescent pollen tubes penetrating through the spri1-1-trans- and C. rubella, stigmas expressing CrSPRI1 rejected pollen of mitting racts after tissue clearing (Figs. 6b,c). The penetration of M. littorea but accepted that of C. rubella (Fig. 5c,d). This indicates fluorescent pollen tubes into the spri1-1 transmitting tracts was functional diversification between the two genes and supports the markedly reduced when M. littorea was pre-pollinated (Fig. 6b,c). idea that SPRI1 does not reject pollen from the same species. These results suggest that spatial occupation of the style by M. littorea pollen tubes in the spri1-1 mutant may interfere with SPRI1 ensures efficient fertilization under heterologous fertilization. Competition for nutrients required for fertilization pollination conditions. Pre-zygotic interspecies incompatibility (such as arabinogalactans29) may also have caused the reduction of may protect resources of female plants by eliminating competition Col-0 pollen-tube entrance through the transmitting tract in the between pollen tubes from different species inside the pistils1,16. hetero-pollinated samples. We performed a competitive hetero-pollination experiment with pollen grains from M. littorea and A. thaliana on spri1-1 and wild- Discussion type pistils, and counted the numbers of seeds obtained. Compared SPRI1 encodes a previously unidentified transmembrane with the wild type, seed set was reduced in the spri1-1 mutant when protein. SPRI1 was predicted to encode a previously unidentified M. littorea pollen was pre-pollinated 6 h before A. thaliana pollen, 221-amino-acid protein with four predicted transmembrane and only a few seeds were produced in spri1-1 when M. littorea domains. This transmembrane domain organization is similar to pollen was pre-pollinated 24 h before A. thaliana pollen (Fig. 6a). that of the eukaryotic tetraspanin proteins, which are 221 to 327 M. littorea pollen tubes were rarely guided to Arabidopsis ovules amino acids in length in A. thaliana, and are known to participate in

Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants 733 Articles NatuRe PlantS

a b c 6 Cagra.0736s0033 N2 5 AT3G13950 4 N1 CrSPRI1 B 3 *** N3 AlSPRI1 A 2 N4 SPRI1B Compatibility score 1 N5 SPRI1_Col-0 0 N6 + SPRI1A + SPRI1B SPRI1A spri1-2 0.2

d 6 e 1 50 5 SPRI1A *** SPRI1B 4 ** 51 100 SPRI1A *** *** SPRI1B 3 101 150 SPRI1A 2 SPRI1B

Compatibility score 151 200 1 *** SPRI1A SPRI1B 201 0 SPRI1A 221 +AlSPRI1 +Anc_N4 +Anc_N5 +NL SPRI1B 233

spri1-1 A. lyrata

f 2 92 221 g * *

SPRI1A EN 6 *** ** SPRI1B VH 5 SPRI1_m1 V N 4 *** SPRI1_m2 E H 3 *** SPRI1_m3 EN 2 SPRI1_m4 E H Compatibility scor e 1 SPRI1_m5 V N SPRI1_m6 VH 0 +SPRI1 +SPRI1 +SPRI1 +SPRI1 +SPRI1 +SPRI1 spri1-1 _m1 _m2 _m3 _m4 _m5 _m6

Fig. 2 | SPRI1 function has been lost multiple times in the self-compatible A. thaliana. a, A haplotype network of the SPRI1 gene. The colours of the circles indicate the mean compatibility scores of the haplotypes (blue, <3.0; red, ≥3.0) and their sizes reflect the frequencies of the haplotypes. Arrowheads indicate frameshift events. b, Compatibility scores of spri1-2 transformants containing SPRI1A and SPRI1B. Numbers of pistils pollinated (left to right, as plotted): spri1-2, n = 5; + SPRI1A, n = 3, 4, 5, 3 and 5; + SPRI1B, n = 4, 3, 3, 3 and 3. Dunnett’s two-sided test versus spri1-2 (left to right, as plotted): + SPRI1A, P = 1.58 × 10−8, 3.88 × 10−10, 5.27 × 10−11, 3.95 × 10−8 and 4.72 × 10−8; +SPRI1B, P = 6.23 × 10−2, 2.26 × 10−1, 7.10 × 10−1, 9.97 × 10−1 and 1.00. Data are mean ± s.d. c, Phylogenetic relationships of the SPRI1 homologues used for the inference of the ancestral sequences. N1–N6 indicate positions of the ancestral nodes. d, Compatibility scores of spri1-1 transformants containing the SPRI1 promoter-driven AlSPR1, Anc_N4 and Anc_N5 coding sequences. NL, line carrying the SPRI1-Nanolantern construct used as the vector control. Numbers of pistils pollinated (left to right, as plotted): spri1-1, n = 6; + AlSPR1, n = 8, 4 and 5; Anc_N4, n = 5, 5 and 4; Anc_N5, n = 4, 5 and 6; + NL, n = 4; A. lyrata, n = 4. Dunnett’s two-sided test against spri1-1 (left to right, as plotted): + AlSPR1, 5.38 × 10−6, 5.81 × 10−4 and 1.82 × 10−3; + Anc_N4, 2.40 × 10−6, 6.12 × 10−6 and 2.46 × 10−5; + Anc_N5, 1.78 × 10−5, 5.29 × 10−6 and 1.57 × 10−6; + NL, 1; A. lyrata, 5.66 × 10−9. Data are mean ± s.d. e, Amino acid alignment of SPRI1A and SPRI1B. Transmembrane regions predicted by TMHMM are highlighted in yellow. f, Illustration of point-mutated SPRI1 variants. g, Compatibility scores of spri1-1 transformants containing the point-mutated variants shown in f. Numbers of pistils pollinated (left to right, as plotted): spri1-1, n = 9; + SPRI1_m1, n = 7, 7, 6, 11 and 8; + SPRI1_m2, n = 6, 7, 8, 10 and 8; + SPRI1_m3, n = 8, 9, 7, 8 and 11; + SPRI1_m4, n = 11, 12, 11, 7 and 11; + SPRI1_m5, n = 11, 11, 12, 13 and 6; + SPRI1_m6, n = 7, 9, 14, 8 and 8. Dunnett’s two-sided test against spri1-1 (left to right, as plotted): SPRI1_m1, P = 2.22 × 10−16, 5.22 × 10−15, 3.92 × 10−14, 5.55 × 10−16 and 3.28 × 10−11; SPRI1_m2, P = 0.00, 0.00, 0.00, 0.00 and 1.14 × 10−14; for SPRI1_m1, P = 2.10 × 10−2, 2.40 × 10−1, 9.75 × 10−1, 1.00 and 1.00; for SPRI1_m4, P = 1.19 × 10−1, 4.89 × 10−1, 5.34 × 10−1, 1.00 and 1.00; SPRI1_m5, P = 6.14 × 10−2, 9.87 × 10−1, 6.79 × 10−1, 1.00 and 1.00; for SPRI1_m6, P = 3.15 × 10−4, 1.10 × 10−3, 3.56 × 10−2, 1.26 × 10−1 and 4.64 × 10−1. In b,d,g, compatibility scores are shown for independent T1 transgenic lines pollinated with M. littorea pollen. Dunnett’s two-sided test compared with spri1-2 or spri1-1; *P < 0.05, **P < 0.01 and ***P < 0.005. cell–cell adhesion and signalling, and in cell motility in mammalian plants32. The SI determinant PrpS in Papaver rhoeas is an approxi- systems30. CD9 is one of the best-studied tetraspanin proteins—it is mately 190-amino-acid protein with four predicted transmem- involved in many biological processes, and knockout of this protein brane domains. The pistil factor of the Ipomoea trifida SI system causes female sterility in mouse31. Their expression patterns suggest is most probably a protein with a membrane-spanning structure33. that the tetraspanins may have a role in the fertilization process of Although the predicted secondary structure and size of SPRI1 are

734 Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants NatuRe PlantS Articles

a reminiscent of these proteins, there are some notable differences. ATG Exon1 Exon2 Stop For example, tetraspanins have one short (about 10 amino acids) SPRI1 and one long (about 100 amino acids) extracellular loop, whereas SPRI1 has two extracellular domains with 39 and 14 amino acids, according to the protein topology prediction method TMHMM (Supplementary Fig. 8). 55 SRK-C24; WT 60 65 SPRI1 is predicted to contain a PGG domain—a domain com- G V M V N P P G G V W Q S E monly found in angiosperms. Accelerated Cell Death 6, which is known to be involved in disease resistance, is the only gene encoding a PGG protein that has been characterized in A. thaliana34. G V M V N S R R C L A E * In total, there are 40 genes that are predicted to encode a protein 55 SRK-C24; spri1-ge 60 containing the PGG domain in A. thaliana. Further studies are b c ♂M. littorea 6 needed to understand the biochemical function of SPRI1 and the *** 11 PGG domain proteins. In analogy with PrpS , the rejection system 5 SRK-C24; SRK-C24; controlled by SPRI1 may involve a specific pollen–pistil biomolec- ♀ ♀ C24 WT spri1-ge ular interaction; this includes the possibility that SPRI1 functions ♀ 4 as a receptor for an as-yet-unidentified pollen-derived ligand. 3 Ligand-like cysteine-rich proteins abundantly found in the pollen coat fraction of the Brassicaceae species35 are possible candidate 2 Compatibility scor e SPR1 ligands. 1 ♂ M. littorea SPRI1 has evolved independently from SI and other known 0 C24 -C24; incompatibility mechanisms. Knockout of SPRI1 did not affect -C24;WT the SI mechanism based on SP11–SRK interactions (Fig. 3). SPRI1 SRK SRKspri1-ge pollen-grain-rejection activity is also independent of SRK activity, d e SP11 as Col-0 lacks SRK function22. It is possible that downstream mol- 6 ♂ ♀SRK-C24; ♀SRK-C24; ecules of the SP11–SRK pathways are shared with the pollen rejec- ♀C24 WT spri1-ge 5 tion mechanism of SPRI1; however, our pollination data from the micromanipulator experiments suggests that the mechanisms of 4 pollen rejection by the two systems are different, since SI inhibits 3 pollen hydration25, whereas SPRI1 inhibits germination and pen- 2 etration (Fig. 4a,b). Our results suggest that this pre-zygotic inter- ♂ SP11 Compatibility scor e specific incompatibility system evolved independently from the 1 S-locus. This is in contrast with previous reports that identified 0 SI-related molecules as the causes of interspecific incompatibility in C24 17–20 -C24; the S-RNase system . Factors involved in the intraspecific incom- -C24;WT patibility systems have been identified recently. A duplicated S-locus SRK SRKspri1-ge was found to be responsible for the interpopulation incompatibility 36 in Brassica rapa . A pectin methylesterase gene was identified as the Fig. 3 | SPRI1 is functionally independent of SI. a, The SPRI1 genome-edited male factor in the unilateral intraspecific cross-incompatibility Ga1 line in the SRK-C24 background (SRK-C24; spri1-ge) carries a 2-bp deletion 37 locus . The identification of these molecules and SPRI1 provide a in the coding sequence. b, Representative fluorescent images of pistils basis for future studies into the molecular mechanisms of pollen– pollinated with M. littorea pollen. c, Compatibility scores for pollination pistil interaction. with M. littorea pollen. Number of pistils pollinated: C24, n = 8; SRK-C24; WT, n = 11; SRK-C24; spri1-ge, n = 13. SRK-C24; WT versus SRK-C24; SPRI1 rejects pollen from distantly related species. SPRI1 is spri1-ge, P = 9.68 × 10−13; two-tailed Student’s t-test. Data are mean ± s.d. involved in the rejection of distantly related Brassicaceae species. d, Representative fluorescent images of pistils pollinated with SP11- The results of the pollination assay using micromanipulators sug- expressing pollen. e, Bar plots of compatibility scores for pollination with gested that a factor in Col-0 pollen may locally suppress the func- SP11-expressing pollen. Number of pistils pollinated: C24, n = 6; SRK-C24; tion of SPRI1, thereby allowing penetration by adjacently attached WT, n = 13; SRK-C24; spri1-ge, n = 9. SRK-C24; WT versus SRK-C24; spri1-ge, M. littorea pollen (Fig. 4c,d). This is reminiscent of the mentor P = 1.00; two-tailed Student’s t-test. Data are mean ± s.d. In b,d, scale bars, 1 effects of pollen , in which mixed pollination of pollen grains from 200 µm. Two-tailed Student’s t-test; ***P < 0.005. WT, wild type. the same and different species increases the frequency of interspecies hybrids. It is possible that SPRI1 is involved in this phenom- enon. The exceptional rejection by SPRI1 of pollen from C. rubella Evolutionary decay of SPRI1 function in A. thaliana and possible that belong to lineage I (Fig. 5a,b) may indicate diversification of role of SPRI1. SPRI1 function has been lost at least seven times such a factor in this species. independently, including six frameshift events and the missense Alternatively, it is possible that a factor in M. littorea pollen E2V mutation in A. thaliana. The number of the loss-of-function caused the incompatibility response by activating SPRI1. However, events may be an underestimation because the haplotype network penetration rates of Col-0 pollen grains attached adjacent to analysis (Fig. 2a, Supplementary Fig. 4a and Supplementary Table 3) M. littorea were unaffected (Fig. 4c,d). Therefore it is unlikely that a implies that there may be missense mutations that compromised diffusible incompatibility signal operates to prevent the penetration the function of SPRI1. Conservatively, approximately 24% of strains of M. littore pollen grains, as in the case of the calcium-ion-mediated investigated in this study were considered to have lost SPRI1 func- SI response in the Brassicaceae25. For some species, such as Diplotaxis tion in some way. muralis or Iberis gibraltarica, one or more factors other than SPRI1 A central question is the selective pressure that has driven the may be involved, because their pollen tubes could not germinate evolution of the SPRI1 gene. In contrast with self-compatible species, irrespective of the presence or absence of SPRI1 (Fig. 5a,b). self-incompatible species tend to reject pollen from other species16.

Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants 735 Articles NatuRe PlantS

a b Hydrated Germinated Penetrated 0 min 30 min 60 min 100

60 ♀ Col- 0

20 Percentage of attached pollen grains

0 ♀ spri1-1

♀Col-0 ♀Col-0spri1-1spri1-2 ♀Col-0spri1-1 ♀spri1-1♀spri1-2 ♀ ♀ ♀ ♀spri1-2

spri1-2 + SPRI1A spri1-2 + SPRI1A ♀ ♀ ♀spri1-2 + SPRI1A

At Ml c d ♂ ♂ Hydrated Germinated Penetrated *** *** 100

0 min 30 min 60 min 80 Ml ♂At ♂Ml ♂Ml ♂ ♂At 60 At ♀ Col- 0 ♂ 40

20 Percentage of attached pollen grains 0 Single Dual SingleDual SingleDual

Fig. 4 | Col-0 pollen can mentor the penetration of M. littorea pollen tubes into A. thaliana stigma. a, Representative image of the single-pollination assays. The arrowhead indicates pollen-tube penetration. Scale bars, 50 µm. b, M. littorea pollen grains were attached to stigmatic papilla cells of Col-0, spri1-1, spri1-2 and spri1-2 + SPRI1A. Two independent T3 transgenic lines were used for spri1-2 + SPRI1A. Time-lapse images were taken at 1 min intervals after pollination. Pollen grains that hydrated, germinated and/or penetrated after 1 h were counted. Numbers of pollen grains pollinated: Col-0, n = 89; spri1-1, n = 69; spri1-2, n = 68; spri1-2 + SPRI1A, n = 52 and n = 56. c, Representative image of dual pollination on spri1-2 + SPRI1A pistils. ♂At, Col-0 pollen grain; ♂Ml, M. littorea pollen grain. Adjacent (that is, in physical contact) pollen grains are almost simultaneously pollinated. The green arrowhead indicates Col-0 pollen-tube penetration. The magenta arrowhead indicates M. littorea pollen-tube penetration. Scale bars, 20 µm. d, Time-lapse images were taken at 1 min intervals after pollination. Pollen grains that hydrated, germinated and/or penetrated after 1 h were counted. Numbers of papilla cells pollinated by designated pollen grains: n = 50 for dual pollination; n = 38 for single-pollination ♂At. Data for single ♂Ml is identical to the first spri1-2 + SPRI1A data shown in b. The P values determined by two-sided Fisher’s exact test for the comparison between single and dual pollinations were 0.00 for germinated ♂Ml and 0.00 for penetrated ♂Ml. Fisher’s exact test; ***P < 0.005.

A protection mechanism against germplasms from other species A. thaliana is about 5–10% (refs. 39,40), there may be a selective pres- may be more critical for self-incompatible species because SI pro- sure to maintain the SPRI1 function of rejecting foreign pollen in motes outcrossing. A recent report suggested that selfing evolved some A. thaliana populations. in A. thaliana before the last glacial period38, and all known strains are self-compatible. It is possible that SPRI1 was released from Conclusion the selective pressure to maintain its interspecific pollen-rejection In this study, we used a GWAS to identify a stigma-specific plasma function once SC evolved in A. thaliana. Although this decay path- membrane-localized protein, SPRI1, in A. thaliana, which func- way may be strongly correlated with the SI-to-SC transition, the tions to reject pollen grains from other Brassicaceae species. The mechanisms of foreign pollen rejection and SI are independent in SI is a mechanism that has been acquired repeatedly in plants and this case (Fig. 3). several of the molecular factors required for SI have been identified In our experimental conditions, absence of SPRI1 resulted in in various plant species7. However, far less is understood about the inefficient fertilization during hetero-pollination (Fig. 6). SPRI1 incompatibility factor that selectively rejects gametes from foreign may have evolved in an outcrossing species in which pistils were species. The pre-zygotic interspecies incompatibility conferred by more likely than those of an inbreeding species to be exposed to SPRI1 is the first SI-independent example of molecular rejection of heterogeneous pollen. Alternatively, since the outcrossing rate of foreign male by female in plants. Future goals include identifying

736 Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants NatuRe PlantS Articles

a 200 *** ♀Col-0 ♀spri1-1 ***

150 *** * *** *** *** ** *** * ** *** 100 *** ** *

50 Number of pollen tubes

0 A. lyrata I. amara L. annua T. glabra C. annua E. allionii E. lyratus O. pumila C. rubella I. tinctoria M. littorea T. parvula D. muralis A. saxatile A. thaliana E. gallicum L. maritima E. hybridum O. violaceus I. gibraltarica E. pieninicum A. androsacea I. sempervirens T. aphanoneurum

c d 6 ♂M. littorea 6 *** ♂C. rubella 5 5

4 4

3 3 *** 2 2 Compatibility scor e Compatibility scor e 1 1

0 0

+ CrSPRI1 + CrSPRI1 spri1-1spri1-1 spri1-1 spri1-1 C. rubella C. rubella + SRPI1A + SRPI1A

Fig. 5 | SPRI1 rejects pollen from distantly related species. a, Box plots of pollen-tube numbers in Col-0 and spri1-1 pistils after pollination with 24 species of Brassicaceae. Species tested, number of pistils pollinated, P value by two-tailed Student’s t-test, respectively, are: A. thaliana, n = 12, 12, 1.99 × 10−1; A. lyrata, n = 12, 11, 6.28 × 10−1; Olimarabidopsis pumila, n = 4, 4, 9.38 × 10−1; Turritis glabra, n = 5, 3, 3.70 × 10−1; C. rubella, n = 11, 9, 1.03 × 10−4; Erysimum allionii, n = 7, 5, 9.79 × 10−1; Erysimum pieninicum, n = 6, 6, 7.41 × 10−1; Erysimum hybridum, n = 3, 3, 3.38 × 10−1; D. muralis, n = 7, 6, 4.52 × 10−1; Trachystoma aphanoneurum, n = 10, 10, 5.99 × 10−4; Erucastrum gallicum, n = 5, 5, 4.00 × 10−2; Enarthrocarpus lyratus, n = 10, 9, 3.27 × 10−3; Carrichtera annua, n = 4, 4, 3.93 × 10−2; Orychophragmus violaceus, n = 13, 14, 2.14 × 10−2; Isatis tinctoria, n = 5, 5, 5.03 × 10−5; Thellungiella parvula, n = 6, 4, 1.08 × 10−2; Arabis androsacea, n = 11, 11, 1.66 × 10−9; Iberis amara, n = 7, 7, 7.37 × 10−3; Iberis sempervirens, n = 8, 8, 9.30 × 10−3; M. littorea, n = 12, 12, 4.25 × 10−9; Lobularia maritima, n = 5, 4, 8.26 × 10−3; Alyssum saxatile, n = 3, 3, 2.02 × 10−3; Lunaria annua, n = 9, 7, 1.89 × 10−2. Centre line shows the median, box limits indicate the 25th and 75th percentiles, whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles and data points are plotted as open circles. Significant differences between Col-0 and spri1-1 are indicated: *P < 0.05, **P < 0.01 and ***P < 0.005. b, Phylogenetic relationships of the species used for the pollination test. Green branches indicate lineage I species and magenta branches indicate species belong to other lineages, according to a previous 67 study . c,d, Bar plots of the compatibility scores. Result of three independent T1 lines are shown for spri1-1 + CrSPRI1. Numbers of pistils pollinated for c: spri1-1, n = 7; spri1-1 + SRPI1A, n = 6; +CrSPRI1, n = 5, 5, 5; C. rubella, n = 9. Dunnett’s test versus spri1-1: spri1-1 + SRPI1A, P = 0.00; +CrSPRI1, P = 0.00, 0.00, 1.11 × 10−16; C. rubella, P = 0.00. Numbers of pistils pollinated for d: spri1-1, n = 8; spri1-1 + SRPI1A, n = 6; +CrSPRI1, n = 6, 5, 10; C. rubella, n = 9. Dunnett’s test against spri1-1: spri1-1 + SRPI1A, P = 6.51 × 10−10; +CrSPRI1, P = 2.83 × 10−1, 5.55 × 10−1, 5.46 × 10−1; C. rubella, P = 8.21 × 10−1. Values indicate means of replicates and whiskers indicate standard deviations. Dunnett’s test against spri1-1; ***P < 0.005. the cognate pollen factor(s) of SPRI1, which should further clarify obtained from the Brassicaceae seed bank of Tohoku University. Arabidopsis halleri the ecological role of this system. (Mino) seeds were sampled from a feld site in Minoo, Japan. Cardamine scutata seeds were sampled from a feld site in Nara, Japan. A. halleri (Fum), C. rubella, O. pumila, T. parvula and T. glabra seeds were a gif from A. Kawabe (Kyoto Sangyo Methods University). A. lyrata SaSb seeds were a gif from Y. Takada (Tohoku University). Plant materials. spri1-1 (SALK_047439C), spri1-2 (SALK_085995C) and A. I. amara and L. maritima seeds were gifs from Y. Shiroto (Yamato Noen). E. allionii lyrata MN47 (N22696) seeds and the seed set for the GWAS (stock ID: N78942) and O. violaceus seeds were purchased from Takii Seeds. L. annua seeds were were obtained from the Arabidopsis Biological Resource Center. Te C. annua, purchased from Kaneko Seeds. I. tinctoria and E. hybridum seeds were purchased D. muralis, E. lyratus, E. gallicum, M. littorea and T. aphanoneurum seeds were from Mikasa Seeds. A. saxatile, A. androsacea, Barbarea vulgaris, E. pieninicum,

Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants 737 Articles NatuRe PlantS

a a ab 20 a b ab b a

c 10

5 Number of developed seeds at 4 DAP

Pistil Col-0 Spri1-1 Col-0 Spri1-1 Col-0 Spri1-1 Col-0 Spri1-1 ♂M. littorea – – + + + + + + Interval – – 0 h 0 h 6 h 6 h 24 h 24 h ♂Col-0 + + + + + + + +

b c ♂M. littorea > 24 h > ♂FP 24 h > ♂FP *** 700 600 500 400 300

♀ spri1-1 200 Distance from the top

of the style to tip 100 the longest FP tube ( µ m)

200 µm200 µm ♀spri1-1 ♀spri1-1 ♂M. littorea 24 h > ♂FP > 24 h > ♂FP

Fig. 6 | SPRI1 ensures efficient fertilization under heterologous pollination conditions. a, Box plots of the numbers of developed seeds counted 4 d after pollination (DAP) of Col-0 or spri1-1 pistils with pollen from Col-0 or M. littorea. Pistils were pollinated with M. littorea pollen and then with Col-0 pollen after the indicated interval. Numbers of pistils pollinated (left to right, as plotted): n = 11, 10, 11, 15, 26, 19, 21 and 21. Tukey multiple comparisons of means with a 95% family-wise confidence level. b, Representative maximum intensity projection images of fluorescent pollen (FP) tubes in the pistils. The pistils from the spri1-1 mutant were collected and kept on agar plates for 24 h with (left) or without (right) pollination by M. littorea pollen. Pistils were then pollinated by fluorescent pollen and incubated for a further 6 h. Green, fluorescence of the Venus moiety of YC3.6; magenta, pistil autofluorescence. c, Box plots of the distance from the top of the style to the tip of the longest fluorescent pollen tube. Numbers of pistils pollinated (left; right, as plotted): n = 10; n = 10. In a and c, centre line shows the median, box limits indicate the 25th and 75th percentiles, whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles and data points are plotted as open circles. Two-tailed Student’s t-test; ***P < 0.005.

I. sempervirens and I. gibraltarica seeds were purchased from Plant World. For SPRI1 coding sequence was fused with the Venus fragment by PCR. The clustered all experiments using M. littorea, fowers from clonally propagated progenies of a regularly interspaced short palindromic repeats (CRISPR)–CRISPR-associated 9 single individual were used. Te mating systems (SI or SC) of the species used for (Cas9)-mediated genome editing of SPRI1 was done using the pHEE401 system, the pollination tests are listed in Supplementary Table 4. which expresses the Cas9 gene specifically in the egg cell43. pHEE401E was a gift from Q.-J. Chen (Addgene plasmid 71287). The target sequence starting from base Growth conditions. All plant materials were grown in mixed soil in a growth 154 of the SPRI1 coding region (5′- GGGTAATGGTGAACCCTCC-3′) was fused chamber under controlled conditions (light intensity 120–150 µmol m−2 s−1, 14 h to the guide RNA following the described protocol43. All transgenic plants were light:10 h dark cycle at 22 ± 2 °C). generated using the Agrobacterium infiltration procedure44.

Gene cloning and transformation. All oligonucleotides used in this study are Gene-expression studies. The microarray analysis was done in our previous listed in Supplementary Table 5. Genomic fragments of SPRI1A and SPRIB study25. The Transcriptome Variation Analysis database26 was searched to obtain (approximately 3,500 bp) were amplified by PCR from the strains Ale-Stenar-44-4 the tissue expression pattern of SPRI1. Total RNA was extracted from stigmas (stock ID: N76652) and Ste 2 (stock ID: N77274), respectively, using the KOD FX of Col-0 and the spri1 mutants using the RNeasy Plant Mini Kit (Qiagen). The Neo polymerase (Toyobo). The fragments were introduced into the pCambia1300 real-time PCR reaction was performed with the QuantiTect SYBRGreen PCR vectors using the In-Fusion HD Cloning Kit (Takara Bio). To generate the SPRI1 kit (Qiagen) using the LightCycler 96 system (Roche). cDNA synthesis promoter-Nanolantern construct, the SPRI1 promoter sequence was amplified for the real-time PCR analysis was done using the SuperScript IV Reverse from Col-0 genomic DNA and then fused upstream of Nanolantern, which Transcriptase kit (Thermo Fisher Scientific). All kits were used according encodes a chimeric Renilla luciferase and Venus fluorescent protein41. Nanolantern- to the manufacturer’s protocols. pcDNA3 was a gift from T. Nagai (Addgene plasmid 51970). Further constructs were generated by replacing the Nanolantern cassette of this construct with the Pollination experiments. Flowers were emasculated before anthesis. At anthesis, coding sequences of SPRI1, the A. lyrata orthologue (AlSPRI1: AL9U10580 from pistils were collected, placed on 1% agar plates, pollinated and incubated for 6 h the A. lyrata genome data v.2.1 (ref. 42)) or the two simulated ancestral forms of at 22 ± 2 °C and humidity 50 ± 5%. Pistils were pollinated in a manner such that SPRI1 (Anc_N4, Anc_N5). The DNA fragments for AlSPRI1, Anc_N4, and Anc_ entire stigmatic surface was covered with pollen grains. Pollinated pistils were fixed N5 were chemically synthesized by FASMAC. The genomic fragment of CrSPRI1 overnight at room temperature in ethanol:acetate 3:1 (v/v), then at 60 °C for 30 min was cloned from C. rubella. To construct the SPRI1–Venus fusion protein, the in 1 M sodium hydroxide. The pistils were stained in 2% tripotassium phosphate,

738 Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants NatuRe PlantS Articles

0.01% aniline blue for 3 h at room temperature. For the hetero-pollination assays, the cutadapt program60. The short reads were mapped against SPRI1 sequences approximately half of the stigmatic surface was first pollinated with M. littorea from A. lyrata or A. halleri obtained in this study using the bowtie2 program61. pollen grains and the pistils were incubated for 24 h as described above. The Consensus SPRI1 sequences for each run were generated by the bcftools programs remaining stigma surface was then covered with Col-0 pollen grains. Pistils were implemented in the samtools suite62. Sequences with average coverage depth ≥5 placed on half-strength Murashige–Skoog medium (Fujifilm Wako Pure Chemical were used for further phylogenetic analysis. Corporation) and incubated as above for 4 d. Cloning of SPRI1 fragments from different species. We amplified SPRI1 Microscopic observation of pollen tubes. Pollen tubes on pistils stained with fragments from A. halleri (Fum and Mino) and A. lyrata (SaSb) by PCR using the aniline blue were observed and counted under an epifluorescence microscope as Tks Gflex DNA Polymerase (Takara). We used degenerate primers to amplify the previously described45. To facilitate large-scale phenotyping during the GWAS, fragments from Barbarea vulgaris, Cardamine scutata, E. allionii, E. hybridum and we defined arbitrary compatibility scores based on the numbers of pollen tubes in O. pumila. These fragments were cloned into pGEM T-Vector systems (Promega) the styles: 1, no tubes observed; 2, 1–19 tubes; 3, 20–39 tubes; 4, 40–59 tubes; 5, after A-tailing using 10× A-attachment mix (Toyobo) and sequenced. ≥60 tubes. To count over 60 pollen tubes, an LSM880 confocal laser-scanning microscope (Carl Zeiss) was used with 440 nm excitation from a diode laser. Ancestral sequence reconstruction. Ancestral sequence reconstruction of SPRI1 in the genus Arabidopsis was done using the FASTML program63. Three SPRI1 Dual pollination of fluorescent pollen with M. littorea or non-fluorescent Col-0 sequences from A. thaliana (SPRI1Col-0, SPRI1A and SPRI1B) and AlSPRI1 were pollen. Transgenic A. thaliana plants carrying the YellowCameleon3.6 (YC3.6) used to infer ancestral forms. The SPRI1 orthologue from C. rubella (renamed gene under the control of the Act1 promoter44 was used as the donor of fluorescent CrSPRI1, Carubv10007342m from the C. rubella genome v.1.064), a SPRI1-like pollen. To observe fluorescent pollen tubes in pistils, we used the tissue-clearing paralogue in A. thaliana (AT3G13950) and another SPRI1-like sequence in C. reagent TOMEI (Tokyo Chemical Industry). We mostly followed the protocol grandiflora (Cagra.0736s0033 from the C. grandiflora genome data v1.164) were provided by the manufacturer46. In brief, pistils were vacuum-treated in fixative used as the outgroups. Their phylogenetic relationships are schematized in Fig. solution (4% paraformaldehyde in PBS, pH 7.0). Pistils were incubated for 1 h 2c. A codon-wise alignment of the SPRI1 homologues was performed using the at room temperature and washed three times in PBS. Pistils then were treated tranalign function in the EMBOSS package65, guided by the protein alignment sequentially with 10, 30, 50, 70 and 100% TOMEI solution in PBS for 10 min at created with muscle66 and manual correction. The phylogenetic topology and room temperature. Finally, the pistils were incubated in 100% TOMEI solution for aligned SPRI1-homologous sequences were then used as the input for the FASTML at least 1 h at room temperature. run. The most probable ancestral sequences in N4 and N5 were obtained and used for further functional tests. Microscopic observation of fluorescent proteins. For the subcellular localization experiment in leaf epidermal cells (Supplementary Fig. 2a), the gene encoding Phylogenetic analysis. We used the phylogenetic relationships of the Brassicaceae the SPRI1–Venus fusion protein under the control of the CAMV 35S promoter species from a previous study67 for Fig. 5b. Protein sequences of the SPRI1 was stably introduced into Col-0. For the SPRI1 translational Venus-fusion homologues were aligned by muscle66 followed by manual correction. A codon- experiment (Supplementary Fig. 2b) and the SPRI1-promoter activity experiment wise alignment of the SPRI1 homologues was performed using the tranalign (Supplementary Fig. 2c), pistils at anthesis were collected from a transgenic function in the EMBOSS package65. The skipredundant function in the EMBOSS line expressing the SPRI1 promoter:Nanolantern construct. Fluorescence in package65 was used to remove identical sequences. This resulted in the multiple the emission range 520–555 nm was observed using an LSM880 confocal laser- alignment of 75 SPRI1 homologues. Poorly aligned regions were removed by the scanning microscope, with 514nm excitation from an argon laser. Gblocks program68. The phylogenetic tree was constructed on the basis of Bayesian To visualize fluorescent-pollen tubes, pistils were mounted on glass slides inference using the MrBayes v.3.2.2 program69. Four chains of the Metropolis- after tissue-clearing treatment and were observed using an LSM880 confocal coupled Markov Chain Monte Carlo processes were run for 1,000,000 generations, laser-scanning microscope. Fluorescence in the emission range 514–544 nm was with trees sampled every 1,000 generations. The first 25% of trees were discarded, observed, with 514 nm excitation from an argon laser. Fluorescence in the and the remaining trees were used to support the majority-rule consensus tree emission range 624–729 nm was simultaneously obtained to detect the topology with posterior probabilities. The figtree program (http://tree.bio.ed.ac.uk/ autofluorescence of the pistils. For each pistil, 25 to 38 z-stack images were taken software/figtree/) was used to visualize the phylogenetic tree. with 1.79 µm intervals. Reporting Summary. Further information on research design is available in the GWAS. The GWAS was performed using the platform in the GWA-Portal website47. Nature Research Reporting Summary linked to this article. We used the 1001 imputed full-genome SNP set48 and the accelerated mixed model. We removed SNPs with minor allele frequencies of <0.15 for the Manhattan plot. Data availability Sequence data can be found at The Arabidopsis Information Resource database Haplotype network analysis. We sequenced SPRI1 coding sequence in 270 strains (https://www.arabidopsis.org/) or in the 1001 genomes website (1001genomes.org). by the Sanger method. The SPRI1 region was amplified using the primer pair Raw phenotype data used for the GWAS has been deposited in the GWA-Portal listed in Supplementary Table 4. The network based on the deduced amino acid (https://gwas.gmi.oeaw.ac.at/#/study/4205/overview) and Arapheno (https://doi. sequences was visualized using the median-joining algorithm in the 49 org/10.21958/study:37). Raw data for pollen tube count was deposited to Mendeley popart program . (https://doi.org/10.17632/yzy85dtwk3.1). All other data are available in the article or in the Supplementary information. Geographic-distribution analysis. The rworldmap package50 was used to visualize the geographic distribution of the SPRI1 haplotypes. SNPs for most strains were retrieved from the 1001 genomes website (1001genomes.org). SNP information of Received: 27 December 2018; Accepted: 9 May 2019; African and relict strains are from a previous study51. Published online: 1 July 2019

Protein secondary structure prediction. TMHMM server v2.052, TMPred server53, References Phobius server54, TOPCONS server v2.055, SOSUI56 and SPLIT server v4.057 were 1. de Nettancourt, D. Incompatibility and Incongruity in Wild and Cultivated used to predict the transmembrane region of SPRI1. Plants (Springer, 2001). 2. Hogenboom, N. G. & Mather, K. Incompatibility and incongruity: two Sequence retrieval from the database. Genome sequences of Aethionema diferent mechanisms for the non-functioning of intimate partner arabicum, Leavenworthia alabamica and Sisymbrium irio58 were retrieved from the relationships. Proc. Roy. Soc. B 188, 361–375 (1975). NCBI (https://www.ncbi.nlm.nih.gov), A. halleri was from the Phytozome (https:// 3. Okuda, S. et al. Defensin-like polypeptide LUREs are pollen tube attractants phytozome.jgi.doe.gov/pz/ahalleri), Brassica napus was from the Geoscope (http:// secreted from synergid cells. Nature 458, 357–361 (2009). www.genoscope.cns.fr/brassicanapus/), B. rapa and oleracea were from the Brassica 4. Takeuchi, H. & Higashiyama, T. Tip-localized receptors control pollen tube Database (http://brassicadb.org/brad/downloadOverview.php), Thellungiella growth and LURE sensing in Arabidopsis. Nature 531, 245–248 (2016). parvulum was from thellungiella.org (http://thellungiella.org/index.php) and 5. Boehm, T. Quality control in self/nonself discrimination. Cell 125, Raphanus raphanistrum was from the Radish Database (http://radish.plantbiology. 845–858 (2006). msu.edu/index.php?title=RadishDB). SPRI1 homologues were identified using the 6. Igic, B., Lande, R. & Kohn, J. R. Loss of self-incompatibility and its blastp program59. evolutionary consequences. Int. J. Plant Sci. 169, 93–104 (2008). 7. Fujii, S., Kubo, K. & Takayama, S. Non-self- and self-recognition models in Retrieval of SPRI1 sequences in the genus Arabidopsis. We used the resequencing plant self-incompatibility. Nat. Plants 2, 16130 (2016). dataset of the genus Arabidopsis27 to find polymorphisms of SPRI1. Datasets for 8. Harada, Y. et al. Mechanism of self-sterility in a hermaphroditic chordate. 49 sequence runs generated in the study were retrieved from the Sequence Read Science 320, 548–550 (2008). Archive at the National Center for Biotechnology Information (https://www. 9. Kubo, K. et al. Collaborative non-self recognition system in S-RNase-based ncbi.nlm.nih.gov/sra). The sequences were subjected to adaptor-trimming by self-incompatibility. Science 330, 796–799 (2010).

Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants 739 Articles NatuRe PlantS

10. Takayama, S. et al. Direct ligand-receptor complex interaction controls 41. Saito, K. et al. Luminescent proteins for high-speed single-cell and Brassica self-incompatibility. Nature 413, 534–538 (2001). whole-body imaging. Nat. Commun. 3, 1262 (2012). 11. Wheeler, M. J. et al. Identifcation of the pollen self-incompatibility 42. Rawat, V. et al. Improving the annotation of Arabidopsis lyrata using determinant in Papaver rhoeas. Nature 459, 992–995 (2009). RNA-seq data. PLoS ONE 10, e0137391 (2015). 12. Wilkins, K. A. et al. Self-incompatibility-induced programmed cell death in 43. Wang, Z.-P. et al. Egg cell-specifc promoter-controlled CRISPR/Cas9 feld poppy pollen involves dramatic acidifcation of the incompatible pollen efciently generates homozygous mutants for multiple target genes in tube cytosol. Plant Physiol. 167, 766–779 (2015). Arabidopsis in a single generation. Genome Biol. 16, 144 (2015). 13. Tomas, S. G. & Franklin-Tong, V. E. Self-incompatibility triggers 44. Iwano, M. et al. Fine-tuning of the cytoplasmic Ca2+ concentration is essential programmed cell death in Papaver pollen. Nature 429, 305–309 (2004). for pollen tube growth. Plant Physiol. 150, 1322–1334 (2009). 14. Lewis, D. & Crowe, L. K. Unilateral interspecifc incompatibility in fowering 45. Shiba, H. et al. Alteration of the self-incompatibility phenotype in Brassica by plants. Heredity 12, 233–256 (1958). transformation of the antisense SLG gene. Biosci. Biotechnol. Biochem. 64, 15. Lewis, D. Incompatibility in fowering plants. Biol. Rev. 24, 472–496 (1949). 1016–1024 (2000). 16. Hiscock, S. J. & Dickinson, H. G. Unilateral incompatibility within the 46. Hasegawa, J. et al. Tree-dimensional imaging of plant organs using a simple Brassicaceae: further evidence for the involvement of the self-incompatibility and rapid transparency technique. Plant Cell Physiol. 57, 462–472 (2016). (S)-locus. Teor. Appl. Genet. 86, 744–753 (1993). 47. Seren, Ü. GWA-Portal: Genome-wide association studies made easy. Methods 17. Li, W. & Chetelat, R. T. A pollen factor linking inter- and intraspecifc pollen Mol. Biol. 1761, 303–319 (2018). rejection in tomato. Science 330, 1827–1830 (2010). 48. Te 1001 Genomes Consortium 1,135 genomes reveal the global pattern of 18. Murfett, J. et al. S RNase and interspecifc pollen rejection in the genus polymorphism in Arabidopsis thaliana. Cell 166, 481–491 (2016). Nicotiana: multiple pollen-rejection pathways contribute to unilateral 49. Bandelt, H.-J., Forster, P. & Röhl, A. Median-joining networks for inferring incompatibility between self-incompatible and self-compatible species. Plant intraspecifc phylogenies. Mol. Biol. Evol. 16, 37–48 (1999). Cell 8, 943–958 (1996). 50. South, A. rworldmap: A new R package for mapping global data. R Journal 3, 19. Li, W. & Chetelat, R. T. Unilateral incompatibility gene ui1.1 encodes an 35–43 (2011). S-locus F-box protein expressed in pollen of Solanum species. Proc. Natl 51. Durvasula, A. et al. African genomes illuminate the early history and Acad. Sci. USA 112, 4417–4422 (2015). transition to selfng in Arabidopsis thaliana. Proc. Natl Acad. Sci. USA 114, 20. Tovar-Méndez, A. et al. Restoring pistil-side self-incompatibility factors 5213–5218 (2017). recapitulates an interspecifc reproductive barrier between tomato species. 52. Krogh, A., Larsson, B., von Heijne, G. & Sonnhammer, E. L. Predicting Plant J. 77, 727–736 (2014). transmembrane protein topology with a hidden Markov model: application to 21. Burdfeld-Steel, E. R. & Shuker, D. M. Reproductive interference. Curr. Biol. complete genomes. J. Mol. Biol. 305, 567–580 (2001). 21, R450–R451 (2011). 53. Hofmann, K. & Stofel, W. TMBASE-A database of membrane spanning 22. Tsuchimatsu, T. et al. Evolution of self-compatibility in Arabidopsis by a protein segments. Biol. Chem. 374, 166 (1993). mutation in the male specifcity gene. Nature 464, 1342–1346 (2010). 54. Käll, L., Krogh, A. & Sonnhammer, E. L. L. Advantages of combined 23. Seren, U. et al. GWAPP: A web application for genome-wide association transmembrane topology and signal peptide prediction-the Phobius web mapping in Arabidopsis. Plant Cell 24, 4793–4805 (2012). server. Nucleic Acids Res. 35, W429–W432 (2007). 24. Hooper, C. M. et al. SUBAcon: a consensus algorithm for unifying the 55. Tsirigos, K. D., Peters, C., Shu, N., Käll, L. & Elofsson, A. Te TOPCONS subcellular localization data of the Arabidopsis proteome. Bioinformatics 30, web server for consensus prediction of membrane protein topology and 3356–3364 (2014). signal peptides. Nucleic Acids Res. 43, W401–W407 (2015). 25. Iwano, M. et al. Calcium signalling mediates self-incompatibility response in 56. Hirokawa, T., Boon-Chieng, S. & Mitaku, S. SOSUI: classifcation and the Brassicaceae. Nat. Plants 1, 15128 (2015). secondary structure prediction system for membrane proteins. Bioinformatics 26. Klepikova, A. V., Kasianov, A. S., Gerasimov, E. S., Logacheva, M. D. & Penin, 14, 378–379 (1998). A. A. A high resolution map of the Arabidopsis thaliana developmental 57. Juretić, D., Zoranić, L. & Zucić, D. Basic charge clusters and predictions of transcriptome based on RNA-seq profling. Plant J. 88, 1058–1070 (2016). membrane protein topology. J. Chem. Inf. Comput. Sci. 42, 620–632 (2002). 27. Novikova, P. Y. et al. Sequencing of the genus Arabidopsis identifes a complex 58. Haudry, A. et al. An atlas of over 90,000 conserved noncoding sequences history of nonbifurcating speciation and abundant trans-specifc provides insight into crucifer regulatory regions. Nat. Genet. 45, 891–898 (2013). polymorphism. Nat. Genet. 48, 1077–1082 (2016). 59. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local 28. Murat, F. et al. Understanding Brassicaceae evolution through ancestral alignment search tool. J. Mol. Biol. 215, 403–410 (1990). genome reconstruction. Genome Biol. 16, 1–17 (2015). 60. Martin, M. Cutadapt removes adapter sequences from high-throughput 29. Mizukami, A. G. et al. Te AMOR arabinogalactan sugar chain induces sequencing reads. EMBnet. J. 17, 10 (2011). pollen-tube competency to respond to ovular guidance. Curr. Biol. 26, 61. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. 1091–1097 (2016). Nat. Methods 9, 357–359 (2012). 30. Hemler, M. E. Tetraspanin functions and associated microdomains. Nat. Rev. 62. Li, H. et al. Te sequence alignment/map format and SAMtools. Mol. Cell Biol. 6, 801–811 (2005). Bioinformatics 25, 2078–2079 (2009). 31. Miyado, K. et al. Requirement of CD9 on the egg plasma membrane for 63. Ashkenazy, H. et al. FastML: a web server for probabilistic reconstruction of fertilization. Science 287, 321–324 (2000). ancestral sequences. Nucleic Acids Res. 40, W580–W584 (2012). 32. Boavida, L. C., Qin, P., Broz, M., Becker, J. D. & McCormick, S. Arabidopsis 64. Slotte, T. et al. Te Capsella rubella genome and the genomic consequences of tetraspanins are confned to discrete expression domains and cell types in rapid mating system evolution. Nat. Genet. 45, 831–835 (2013). reproductive tissues and form homo- and heterodimers when expressed in 65. Rice, P., Longden, I. & Bleasby, A. EMBOSS: the European Molecular Biology yeast. Plant Physiol. 163, 696–712 (2013). Open Sofware Suite. Trends Genet. 16, 276–277 (2000). 33. Tsuchiya, T. in Sexual Reproduction in Animals and Plants (eds Sawada, H. 66. Edgar, R. C. MUSCLE: multiple sequence alignment with high accuracy and et al.) 305–325 (Springer, 2014). high throughput. Nucleic Acids Res. 32, 1792–1797 (2004). 34. Vanacker, H., Lu, H., Rate, D. N. & Greenberg, J. T. A role for salicylic 67. Couvreur, T. L. P. et al. Molecular phylogenetics, temporal diversifcation, and acid and NPR1 in regulating cell growth in Arabidopsis. Plant J. 28, principles of evolution in the mustard family (Brassicaceae). Mol. Biol. Evol. 209–216 (2001). 27, 55–71 (2010). 35. Doughty, J., Wong, H. Y. & Dickinson, H. G. Cysteine-rich pollen coat 68. Talavera, G. & Castresana, J. Improvement of phylogenies afer removing proteins (PCPs) and their Interactions with stigmatic S (incompatibility) and divergent and ambiguously aligned blocks from protein sequence alignments. S-related proteins in Brassica: putative roles in SI and pollination. Ann. Bot. Syst. Biol. 56, 564–577 (2007). 85, 161–169 (2000). 69. Ronquist, F. et al. MrBayes 3.2: efcient Bayesian phylogenetic inference and 36. Takada, Y. et al. Duplicated pollen–pistil recognition loci control intraspecifc model choice across a large model space. Syst. Biol. 61, 539–542 (2012). unilateral incompatibility in Brassica rapa. Nat. Plants 3, 17096 (2017). 37. Zhang, Z. et al. A pectin methylesterase gene at the maize Ga1 locus Acknowledgements confers male function in unilateral cross-incompatibility. Nat. Commun. 9, We thank M. Okamura, M. Nara, T. Manabe, Y. Yamamoto, M. Niidome and M. Ishii 3678 (2018). for their technical assistance and A. Kawabe and Y. Kato for helpful discussions. This 38. Tsuchimatsu, T. et al. Patterns of polymorphism at the self-incompatibility work was supported in part by Grants-in-Aid for Scientific Research on Innovative Locus in 1,083 Arabidopsis thaliana genomes. Mol. Biol. Evol. 34, 1878–1889 Areas (23113002, 16H06467 and 16H06464 to S. Takayama; 16H01467 and 18H04776 (2017). to S. Fujii; 17H05833 and 18H04813 to T.T. and 16H06469 to K.K.S.), Grants-in-Aid for 39. Platt, A. et al. Te scale of population structure in Arabidopsis thaliana. PLoS Scientific Research (25252021 and 16H06380 to S.Takayama and 18H02456 to S. Fujii), Genet. 6, e1000843 (2010). Grant-in-Aid for Challenging Exploratory Research (15K14626 to S. Fujii) from the 40. Shimizu, K. K., Kudoh, H. & Kobayashi, M. J. Plant sexual reproduction Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT), Swiss during climate change: gene function in natura studied by ecological and National Science Foundation to K.K.S. and Japan Science and Technology Agency (JST) evolutionary systems biology. Ann. Bot. 108, 777–787 (2011). PRESTO programme (JPMJPR16Q8) to S. Fujii.

740 Nature Plants | VOL 5 | JULY 2019 | 731–741 | www.nature.com/natureplants NatuRe PlantS Articles

Author contributions Additional information S. Fujii, K.K.S. and S. Takayama conceived the study. S. Fujii, T.T., K.K.S. and S. Takayama Supplementary information is available for this paper at https://doi.org/10.1038/ wrote the manuscript. S. Fujii conducted most of the experiments and data analysis. s41477-019-0444-6. T.T. conducted the GWAS and the geographical analysis. H.S.-A., S. Furukawa, W.I. Reprints and permissions information is available at www.nature.com/reprints. and Y.W. contributed to phenotyping in the GWAS. Y.K. performed the fluorescent pollen experiment. S.I. contributed to the transgenic experiments. S. Tangpranomkorn Correspondence and requests for materials should be addressed to S.F. or S.T. contributed to the sequence analysis. M.I. performed the microarray analysis and W.I. Peer review information: Nature Plants thanks Daphne Goring and the other, performed the real-time PCR. anonymous, reviewer(s) for their contribution to the peer review of this work. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in Competing interests published maps and institutional affiliations. The authors declare no competing interests. © The Author(s), under exclusive licence to Springer Nature Limited 2019

Corresponding author(s): Sota Fujii; Seiji Takayama

Last updated by author(s): May 3, 2019 Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.

Statistics For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section. n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated

Our web collection on statistics for biologists contains articles on many of the points above. Software and code Policy information about availability of computer code Data collection GWA-Portal website

Data analysis R (3.5.1) popart cutadapt (1.3) samtools (0.1.19) bowtie2 (2.1.0) EMBOSS (6.6.0.0) MrBayes (3.2.2) figtree (1.4.2) muscle (3.8.31) Gblocks (0.91) FASTML (3.1) blastp (2.2.29) TMHMM server (2.0) TMPred server Phobius server October 2018 TOPCONS server (2.0) SOSUI SPLIT server (4.0)

For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

1 Data nature research | reporting summary Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability

Sequence data can be found at The Arabidopsis Information Resource database (https://www.arabidopsis.org/) or in the 1001 genomes website (1001genomes.org). Raw phenotype data used for the GWAS is deposited in the GWA-portal (https://gwas.gmi.oeaw.ac.at/#/home). Raw data for pollen tube count was deposited to Mendeley (DOI: http://dx.doi.org/10.17632/yzy85dtwk3.1). All other data are available in the manuscript or in the supplementary materials.

Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Life sciences study design All studies must disclose on these points even when the disclosure is negative. Sample size No sample size calculation was performed. In this study, we used at least three independent transgenic lines to determine the function of each introduced genes. This number is sufficient considering most studies use two independent transgenic lines for such experiment.

Data exclusions No data was excluded from the analysis.

Replication For each pollination pairs, at least three pistils were pollinated and evaluated. All experiments were reproducible.

Randomization In this study, inbred or clonal propagated plants were used for the analysis. Therefore, for each experiment, samples carry the exact same genotypes. We consider randomization is not required in such case.

Blinding This study does not involve clinical experiments. We consider blinding is not required in such case.

Reporting for specific materials, systems and methods We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response. Materials & experimental systems Methods n/a Involved in the study n/a Involved in the study Antibodies ChIP-seq Eukaryotic cell lines Flow cytometry Palaeontology MRI-based neuroimaging Animals and other organisms Human research participants Clinical data

Animals and other organisms Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research Laboratory animals The study did not involve laboratory animals. October 2018

Wild animals The study did not involve wild animals.

Field-collected samples The study did not involve samples collected from the field.

Ethics oversight The study was restricted to using plants and is not relevant to ethics oversight. Note that full information on the approval of the study protocol must also be provided in the manuscript.