Generation of Transgenic Self-Incompatible Arabidopsis thaliana Shows a Genus-Specific Preference for Self-Incompatibility Genes

Brassicaceae species employ both self-compatibility and self-incompatibility systems to regulate post-pollination events. Arabidopsis halleri is strictly self-incompatible, while the closely related Arabidopsis thaliana has transitioned to self-compatibility with the loss of functional S-locus genes during evolution. The downstream signaling protein, ARC1, is also required for the self-incompatibility response in some Arabidopsis and Brassica species, and its gene is deleted in the A. thaliana genome. In this study, we attempted to reconstitute the SCR-SRK-ARC1 signaling pathway to restore self-incompatibility in A. thaliana using genes from A. halleri and B. napus, respectively. Several of the transgenic A. thaliana lines expressing the A. halleri SCR13-SRK13-ARC1 transgenes displayed self-incompatibility, while all the transgenic A. thaliana lines expressing the B. napus SCR1-SRK1-ARC1 transgenes failed to show any self-pollen rejection. Furthermore, our results showed that the intensity of the self-incompatibility response in transgenic A. thaliana plants was not associated with the expression levels of the transgenes. Thus, this suggests that there are differences between the Arabidopsis and Brassica self-incompatibility signaling pathways, which perhaps points to the existence of other factors downstream of B. napus SRK that are absent in Arabidopsis species.


Introduction
Self-incompatibility is one of the most important breeding systems present in many hermaphroditic flowering plants as it causes self-pollen rejection to promote out-crossing. Self-incompatibility systems under investigation include species in Brassicaceae, Solanaceae, Rosaceae, Scrophulariaceae and Papaveraceae [1][2][3][4]. In the Brassicaceae (crucifer family), self-incompatibility is controlled by tightly linked multi-allelic genes at the S-locus region, which rarely recombine and so have collectively been named S-haplotypes [4]. Three S-locus-linked genes have been identified in Brassica species. The S-locus glycoprotein (SLG) gene encodes an abundant, secreted glycoprotein that is located in the cell wall of stigma papillae, and SLG has been reported to be connected to pollen adhesion on stigmatic surfaces [5,6]. The S-locus receptor kinase (SRK) is the female determinant in stigma, and mediates the self-incompatibility signaling pathway [7][8][9]. Finally, the S-locus Protein 11/S-locus Cysteine-Rich (SP11/SCR) gene encodes the male determinant and is expressed in the anther tapetum and pollen [10][11][12]. SP11/SCR is recognized by SRK as a ligand in an S-haplotype-specific manner, and SRK is then autophosphorylated to induce various signaling cascades [13,14].
Models of the downstream self-incompatibility signaling pathways have been proposed in Brassica species [15,16]. The M locus protein kinase (MLPK) was identified as a positive regulator of self-incompatible response, since the B. rapa mlpk mutant lost its self-incompatibility phenotype and knocking out the B. napus MLPK genes by CRISPR/Cas9 completely knocked out the self-incompatibility response [17,18]. MLPK is tethered on the plasma membrane and can be phosphorylated by SRK by forming a complex with SRK to mediate signal transduction in self-incompatibility response [19,20]. ARC1 (Armed Repeat Containing 1) and THL1/2 (thioredoxin-h like 1/2) were isolated for interacting with the kinase domain of SRK in yeast two-hybrid analysis [21,22]. THL1/2 were believed to regulate self-incompatibility response negatively [21,23]. Down-regulation of ARC1 expression in the stigma caused a breakdown of self-incompatibility in the B. napus line 'W1 [22,24]. ARC1 is an E3 ubiquitin ligase that can direct the degradation of target proteins required for compatible pollen responses [25]. One target is EXO70A1, which directly interacts with ARC1 and can be ubiquitinated by ARC1 in vitro. This action is proposed to block the transfer of secretory vesicles to the pollen-stigma contact site resulting in pollen rejection [26]. GLO1 (glyoxalase I), a key enzyme involved in detoxification of methylglyoxal, and has been identified as another ARC1 target. ARC1 was proposed to negatively regulate GLO1 in self-pollen rejection, leading to the accumulation of the methylglyoxal toxin in papillar cells [27].
In addition to Brassica species, Arabidopsis species have been used for self-incompatibility studies in Brassicaceae. By analyzing the sequences of Arabidopsis S-locus genes, naturally self-incompatible species, such as Arabidopsis lyrata and Arabidopsis halleri, were shown to have intact SRK and SCR/SP11 genes. Arabidopsis thaliana ecotypes are self-compatible and do not carry functional SRK or SCR/SP11 genes [28][29][30][31][32]. Efforts have been made to reconstruct the self-incompatibility response in A. thaliana by introducing functional SCR and SRK genes, and significant differences were observed in the stability and strength of the self-incompatibility traits among various ecotypes tested [31,[33][34][35][36][37]]. When B. napus self-incompatibility stigma genes (SLG, SRK and ARC1 genes) were transformed into A. thaliana, the transgenic plants still produced seeds, and their stigmas failed to reject the corresponding self-incompatible B. napus W1 pollen [38]. Similarly, transgenic A. thaliana ecotype Col-0 plants expressing the A. lyrata SCRb-SRKb genes were fully self-fertile [33,35,39]. However, when the same experiments were conducted with the A. thaliana C24 ecotype, the transformation of the AlSCRb-AlSRKb genes resulted in transgenic plants showing a strong self-incompatibility response [33,35,40]. Self-incompatible A. thaliana C24 lines were also created using the A. halleri AhSCR 1 and AhSRK 1 genes of the most recessive S 1 haplotype [41]. Finally, ARC1 was found to be a pseudogene in A. thaliana, and a homolog of MLPK, AtAPK1b, did not appear to be involved in regulating self-incompatibility in transgenic A. thaliana, suggesting that SRK-mediated signaling in transgenic A. thaliana might be different from the Brassica models [37,42]. This is also supported by differences in the post-pollination cellular responses of Brassica and Arabidopsis stigmas [15,16]. Nevertheless, stable self-incompatible A. thaliana were finally generated in the Col-0 ecotype when AlARC1 or BnARC1 were expressed along with the AlSCRb-AlSRKb transgenes [43].
Here, we further examine the question of whether Brassica and Arabidopsis self-incompatibility genes can share a common self-incompatibility signaling pathway in transgenic A. thaliana using a new set of self-incompatibility genes from B. napus and A. halleri. The B. napus S 1 -haplotype SCR/SP11 and SRK genes [44][45][46] along with BnARC1 were transformed into two A. thaliana accessions, Col-0 and C24. As both the pollen (BnSCR/SP11 1 ) and stigma (BnSRK 1 , BnARC1) factors were included for this study, this differs from a previous study [38] where only stigma B. napus factors were transformed. A new set of Arabidopsis self-incompatibility genes from A. halleri S 13 -haplotype SCR/SP11 and SRK genes [47], along with AhARC1, were also tested into Col-0 as a comparison. The S 1and S 13 -haplotypes were previously found to be the most dominant S-haplotypes in B. napus and A. halleri, respectively [41,44,45]. Our results show that self-incompatibility genes from B. napus were not sufficient to restore the self-incompatibility in both the Col-0 and C24 ecotypes, while self-incompatibility was successfully transmitted into Col-0 with the A. halleri self-incompatibility genes.

Self-Incompatibility Genes
To examine if the SCR-SRK-ARC1 model is conserved using genes from another naturally self-incompatible Arabidopsis species, A. halleri SCR 13 -SRK 13 genes were introduced into A. thaliana Col-0 plants in the presence or absence of AhARC1, with all three transgenes in a single construct. Thirty and 34 independent T 1 Col-0 lines were generated with the addition of AhSCR 13 -AhSRK 13 and AhSCR 13 -AhSRK 13 -AhARC1, respectively ( Table 1). The self-incompatibility phenotype of the transgenic plants was examined by staining their stigmas with aniline blue to visualize the pollen grains and the pollen tubes growing into papillar cells. Pollinated pistils from the different transgenic lines showed variations in the levels of rejection/acceptance of self-pollen, which was also reported by Indriolo et al. [43]. Accordingly, the phenotypes were defined as self-compatible (similar to Col-0), moderately self-incompatible (>5 pollen tubes/pistil, but visibly reduced compared to Col-0) or strongly self-incompatible (<5 pollen tubes/pistil). All 30 independent transgenic T 1 plants with AhSCR 13 -AhSRK 13 were found to be fully self-compatible, as they could accept self-pollen similarly to that observed for wild-type self-pollinated Col-0 plants in stage-14 stigmas (Table 1 and Figure 1A-D). In contrast, four out of 34 transgenic T 1 plants carrying AhARC1 along with AhSCR 13 -AhSRK 13 showed strong self-incompatibility (Table 1 and Figure 1E,F). To ensure that the strong self-incompatibility phenotype was not caused by other reproduction factors, reciprocal crosses were performed between the transgenic lines and wild-type Col-0 plants, leading to complete self-compatibility (Supplementary Figure S1). In addition, 16 of the 34 AhSCR 13 -AhSRK 13 -AhARC1 transgenic T 1 plants displayed moderate self-incompatibility, and the final 14 AhSCR 13 -AhSRK 13 -AhARC1 transgenic T 1 plants were fully self-compatible (Table 1 and Figure 1G-J). The fact that only the combination of AhSCR 13 -AhSRK 13 -AhARC1 transgenes could reconstruct self-incompatibility in the A. thaliana Col-0 ecotype confirms a role for ARC1 in this self-pollen rejection signaling pathway, as previously reported by Indriolo et al. [43].  To test whether the B. napus SCR 1 -SRK 1 genes were sufficient to restore self-incompatibility in A. thaliana plants, and to test if BnARC1 was also required, the BnSCR 1 -BnSRK 1 genes were transformed with or without BnARC1 into the A. thaliana Col-0 and C24 ecotypes. Independent transgenic plants were obtained for each transformation event, and the pollination phenotypes were noted (Tables 1 and 2). Twenty-one independent T 1 Col-0 plants were obtained with the transformation of BnSCR 1 -BnSRK 1 and all of them exhibited a self-compatible phenotype in stage-14 self-pollinated pistils stained with aniline blue (Table 1 and Figure 2A-D). Similarly, 13 independent T 1 C24 lines were created with the BnSCR 1 -BnSRK 1 transgenes and showed a similar level of self-pollen acceptance to wild-type C24 plants at developmental stage 14 (Table 2 and Figure 2G-J). Previous studies have reported that A. thaliana Col-0 plants carrying the A. lyrata SCRb-SRKb transgenes were fully self-compatible, while for A. thaliana C24 plants transformed with the SCRb-SRKb transgenes, self-incompatibility was successfully reinstated [33,35,40]. However, the introduction of B. napus BnSCR 1 -BnSRK 1 into C24 plants failed to restore self-incompatibility, indicating that BnSCR 1 -BnSRK 1 may need other unknown factors in the self-incompatibility pathway that do not exist in the C24 ecotype.  To examine if the transformation of BnARC1 along with BnSCR 1 -BnSRK 1 could produce a functional self-incompatible pathway in Col-0 and C24 plants, 12 and 9 independent T 1 Col-0 and C24 plants were generated, respectively. The phenotypes of these transgenic lines were again scored for self-pollen rejection by aniline blue staining. Unexpectedly and in contrast to the AhSCR 13 -AhSRK 13 -AhARC1 transgenic Col-0 plants, all Col-0 lines carrying BnSCR 1 -BnSRK 1 -BnARC1 genes showed self-compatible phenotypes (Table 1 and Figure 2E,F). In addition, no self-incompatible or moderately self-incompatible phenotypes were observed in the BnSCR 1 -BnSRK 1 -BnARC1 transgenic C24 lines (Table 2 and Figure 2K,L). These results indicate that the addition of BnARC1 with BnSCR 1 -BnSRK 1 also does not work to restore the self-incompatibility signaling pathway in the A. thaliana Col-0 and C24 ecotypes.

Seed Production is Significantly Reduced in Col-0 Lines Expressing AhSCR 13 -AhSRK 13 -AhARC1
As the transformation of SCRb-SRKb-ARC1 was previously found to lead to an approach herkogamous phenotype in A. thaliana Col-0 and Sha plants [43], we examined whether the transgenic SCR-SRK-ARC1 Col-0 and C24 lines displayed any changes in their floral morphology. The approach herkogamous phenotype was only exhibited in the strongly self-incompatible Col-0 plants carrying the AhSCR 13 -AhSRK 13 -AhARC1 transgenes ( Figure 3J,K and Supplementary Figure S2). The physical stigma-anther separation (SAS) was measured in flowers from wild-type and representative self-incompatible Col-0 plants. A positive SAS value was indicative of approach herkogamy which is consistent with the self-incompatible phenotype in the AhSCR 13 -AhSRK 13 -AhARC1 plants ( Figure 3L). However, all the self-compatible transgenic lines showed a normal flower trait with the anthers positioned above the stigmas, similar to wild-type Col-0 and C24 plants, which ensures that the released pollen falls onto the stigmas (Supplementary Figure S2). Therefore, the approach herkogamous phenotype correlates with the degree of self-incompatibility in the transgenic A. thaliana plants. Since self-incompatibility has been defined as a reproductive mechanism that prevents self-fertilization, seed production of representative transgenic Col-0 and C24 lines was examined. The plants were allowed to self-fertilize naturally, and photographs of branches with siliques from plants with each combination of transgenes are shown in Figure 3. The wild-type A. thaliana Col-0 and C24 plants are self-compatible and should produce well-developed siliques full of seeds with self-pollination ( Figure 3A,G). The transgenic Col-0 plants with AhSCR 13 -AhSRK 13 , AhSCR 13 -AhSRK 13 -AhARC1, BnSCR 1 -BnSRK 1 or BnSCR 1 -BnSRK 1 -BnARC1 that did not display self-incompatibility produced siliques with similar sizes to the wild-type Col-0 plants ( Figure 3B-D,F). Regular-sized siliques were also observed in the self-compatible transgenic C24 plants carrying the BnSCR 1 -BnSRK 1 or BnSCR 1 -BnSRK 1 -BnARC1 transgenes ( Figure 3H,I). However, the strongly self-incompatible transgenic Col-0 plants with the AhSCR 13 -AhSRK 13 -AhARC1 transgenes displayed much smaller silique sizes, as a result of very reduced seed production with self-pollination ( Figure 3E).
Fully developed siliques from each of these lines were harvested and dissected to record the number of seeds per silique (Figure 4). Siliques from wild-type A. thaliana Col-0 plants produced an average of 44.7 seeds/silique via self-pollination ( Figure 4A). Similarly, the siliques from the three self-compatible AhSCR 13 -AhSRK 13 transgenic Col-0 plants contained 40.4, 42.5 and 46.1 seeds/silique on average, respectively, showing no significant difference with wild-type Col-0 ( Figure 4A). The siliques of the three self-compatible BnSCR 1 -BnSRK 1 and BnSCR 1 -BnSRK 1 -BnARC1 transgenic Col-0 plants generated similar averages of 48.5, 47, 47.3, and 47.2, 47.3, 48.5 seeds/silique, respectively ( Figure 4B). Thus, these self-compatible transgenic Col-0 lines displayed an equivalent ability to accept self-pollen compared to the wild-type Col-0 plants. The addition of BnARC1 with BnSCR 1 -BnSRK 1 did not promote the self-pollen rejection in A. thaliana Col-0. In contrast, some reduction in the number of seeds/silique was observed for the three moderately self-incompatible AhSCR 13 -AhSRK 13 -AhARC1 transgenic Col-0 lines, with averages of 28.6, 23.8, and 26.5 seeds/silique, respectively ( Figure 4A). In addition, significant reduction in seed production was observed for the three strong self-incompatible AhSCR 13 -AhSRK 13 -AhARC1 transgenic Col-0 lines following self-pollination, with the lowest value of 1.5 seeds/silique scored for AhSCR 13 -AhSRK 13 -AhARC1 line 2 ( Figure 4A). Therefore, the expression of AhARC1 along with AhSCR 13 -AhSRK 13 in A. thaliana Col-0 induced a strong self-incompatibility response resulting in very reduced seed production. Mature siliques of natural self-pollination transgenic A. thaliana C24 lines were also examined for seed set and compared to wild-type A. thaliana C24 plants, which produced an average of 48.9 seeds/silique ( Figure 4B). All the self-compatible BnSCR 1 -BnSRK 1 and BnSCR 1 -BnSRK 1 -BnARC1 transgenic C24 lines showed similarly high numbers of seeds/silique following self-pollination, which ranged from 44.3 to 47.5 seeds/silique ( Figure 4B). Accordingly, the B. napus SCR 1 -SRK 1 transgenes in the A. thaliana C24 ecotype could not elicit a self-incompatibility response, and the addition of BnARC1 with BnSCR 1 -BnSRK 1 did not result in any reduction of seed set in the C24 ecotype.
2.4. The Expression Levels of the SCR-SRK-ARC1 Transgenes are Not Associated with the Intensity of Self-Incompatibility Response in A. thaliana Given the differences observed in the ability of the AhSCR 13 -AhSRK 13 -AhARC1 and BnSCR 1 -BnSRK 1 -BnARC1 transgenes to confer self-incompatibility in A. thaliana, the transgene expression levels were investigated to reveal if they correlated with the observed phenotypes. Mature buds from representative transgenic lines were harvested to test the relative expression levels of each transgene using quantitative RT-PCR (qRT-PCR). Overall, the transgene expression levels were quite variable, and there was no correlation between a particular transgene expression level and the self-incompatibility trait ( Figure 5). For example, the strongly self-incompatible  Although the addition of the BnSCR 1 -BnSRK 1 and BnSCR 1 -BnSRK 1 -BnARC1 transgenes did not restore the self-incompatibility in the A. thaliana Col-0 and C24 ecotypes, the expression levels of these transgenes were also quantified to confirm that they were successfully being expressed in the transgenic lines ( Figure 5D-F). The BnSCR 1 transgene tended to show quite high relative expression levels, particularly in the C24 ecotype ( Figure 5D), and the relative levels in both Col-0 and C24 showed similar ranges to that observed for AhSCR 13 ( Figure 5A). Conversely, the BnSRK 1 relative expression levels were generally lower than those observed for several of the transgenic lines carrying the AhSRK 13 transgene ( Figure 5B,E). However, the BnSCR 1 -BnSRK 1 -BnARC1 transgenic lines did have BnSRK 1 expression levels that were comparable to the AhSRK 13 transgene in the strongly self-incompatible AhSCR 13 -AhSRK 13 -AhARC1 transgenic lines ( Figure 5B,E). Finally, the BnARC1 expression levels were generally higher than the observed relative expression values for the AhARC1 (Figure 5C,F). Thus, no trend was seen in the various expression levels for the BnSCR 1 -BnSRK 1 -BnARC1 transgenes that could account for lack of self-pollen rejection in both the A. thaliana Col-0 and C24 transgenic plants. This suggests that the B. napus self-incompatibility signaling pathway cannot be reconstructed in the A. thaliana Col-0 and C24 ecotypes.

Discussion
The transition to selfing has made the model plant A. thaliana an ideal system for the reconstruction of self-incompatibility. It is believed that the loss-of-function mutations in SRK and SCR genes has led to self-fertility in all A. thaliana ecotypes [28,[30][31][32]35,[49][50][51]. Self-incompatibility has been restored in some A. thaliana ecotypes by introducing functional A. lyrata SCR and SRK genes, while several other ecotypes have remained self-fertile [31,33,35,40,43]. Notably, transgenic A. thaliana expressing A. lyrata SCRb-SRKb in the C24 ecotype were self-incompatible and produced very few seeds, while the same combination in the Col-0 ecotype resulted in full seed production with selfing [33,35,40]. On the other hand, the addition of ARC1 with A. lyrata SCRb-SRKb in the Col-0 ecotype resulted in self-incompatible plants with reduced seed set [43].
In this study, we tested the reconstitution of the SCR-SRK-ARC1 signaling pathway in self-compatible A. thaliana using new sets of self-incompatibility genes from two other species, self-incompatible B. napus and A. halleri. Our results showed that self-pollen rejection was achieved only in the A. thaliana AhSCR 13 -AhSRK 13 -AhARC1 Col-0 lines. The other combination of the AhSCR 13 -AhSRK 13 transgenes was not sufficient to restore self-incompatibility in A. thaliana Col-0 ecotypes, similar to that previously observed [43]. Interestingly, only flowers from the AhSCR 13 -AhSRK 13 -AhARC1 strongly self-incompatible Col-0 plants displayed an approach herkogamous phenotype, which had been previously detected with the SCRb-SRKb-ARC1 transgene combination [43]. The self-compatible A. thaliana Col-0 AhSCR 13 -AhSRK 13 -AhARC1 plants maintained the same floral phenotype as wild-type Col-0 plants, indicating that the approach herkogamous phenotype is associated with the intensity of self-incompatibility of the transgenic plants. The expression of SRKb was previously found to enhance pistil elongation and stigma exertion in the rdr6 mutant background [52]. Thus, the self-incompatibility signaling transduction mediated by AhSCR 13 -AhSRK 13 -AhARC1 might also activate downstream components that are regulating pistil development.
As nearly half of the A. thaliana Col-0 AhSCR 13 -AhSRK 13 -AhARC1 plants still exhibited a self-compatible phenotype, quantitative RT-PCR was conducted to test if the expression levels of the transgenes were correlated with the ability of self-pollen rejection. However, there was no clear relationship between the relative expression levels of the three transgenes and the intensity of self-incompatibility response. Thus, the trigger for self-pollen rejection in the transgenic A. thaliana Col-0 AhSCR 13 -AhSRK 13 -AhARC1 lines did not appear to be dependent on the expression levels of the transgenes. Unexpectedly, no self-incompatible lines were generated in both the A. thaliana Col-0 and C24 ecotypes with the transformation of BnSCR 1 -BnSRK 1 or BnSCR 1 -BnSRK 1 -BnARC1 genes. These B. napus transgenes displayed similar expression patterns to the A. halleri transgenes, and so their relative expression levels did not appear to account for the lack of a self-incompatibility phenotype. A potential reason may be the design of these transgenes, where only coding sequences were cloned and driven by promoters from other genes, potentially impacting their expression. As the BnARC1 transgene was previously shown to be functional in the A. thaliana Col-0 background [43], it is more likely that BnSCR1 or BnSRK1 are not functional in A. thaliana. Similar, the BnSCR1 construct was found to be functional in B. napus [46], and so BnSRK1 would be the only previously untested construct. One other possible reason is that Arabidopsis and Brassica species may not completely share the same SRK-mediated self-incompatibility signaling pathway. These differences may potentially be due to B. napus SRK 1 activity, as B. napus ARC1 was previously shown to interact with A. lyrata SRK, and the A. lyrata SCRb-SRKb and B. napus ARC1 transgenes produced self-incompatible Col-0 transgenic plants [43,53]. Ideally, the potential activity of the B. napus transgenes could be tested by reciprocal pollinations between the A. thaliana Col-0 BnSCR1-BnSRK1-BnARC1 lines and the B. napus S1 haplotype transgenic line [46]. However, large physical differences in pollen and pistil sizes between these species (Supplementary Figure S2, [38]) as well as potential complications from interspecies crosses may obscure any self-incompatibility reactions.
There are several pieces of evidence that would support the idea of differences in the Brassica and Arabidopsis SRK-mediated signaling pathways. A. thaliana diverged from A. lyrata about 10 MYA and diverged from Brassica approximately 24 MYA [54][55][56][57][58]. As would be expected within the Arabidopsis genus, the inactivated SCR and SRK alleles in A. thaliana share high sequence similarity with the functional S-haplotype genes in A. lyrata and A. halleri [28,30,31,34,35,50,59,60]. In contrast, there is high variation in their sequences when compared to Brassica SP11/SCR and SRK alleles, and so this variation could lead to differences in SRK interactions with other downstream components (the ARC1 interaction is conserved). The ability of the SCRb-SRKb transgenes to restore self-incompatibility in the C24 ecotype (i.e., no ARC1) supports the idea that C24 plants might employ other unknown self-incompatibility signaling players; for example, a strong calcium flux mediated by glutamate receptor-like channels has been shown to occur in self-incompatible C24 stigmas [33,35,40]. In addition, while MLPK is required for self-incompatibility both in B. rapa and B. napus [17,18], an A. thaliana homolog of MLPK (APK1b) was found to be unfunctional in regulating the transient self-incompatibility response observed in SCRb-SRKb Col-0 lines [42]. Finally, there are differences in the cellular responses that take place in the stigmatic papillae of Brassica and Arabidopsis species, such as autophagy and the trafficking of vesicles and multivesicular bodies [16,61]. The potential redundancy shaped by genome triplication and the formation of distinct subgenomes in Brassica species may have increased robustness in the self-incompatibility system, allowing further evolution of the underlying signaling pathways and distinction from Arabidopsis species [62,63].
In conclusion, we have demonstrated that the previously untested A. halleri S13-haplotype genes, along with AhARC1, can restore the self-incompatibility phenotype in A. thaliana Col-0 plants, and that AhARC1 is required along with AhSCR 13 -AhSRK 13 for self-pollen rejection in the Col-0 ecotype. The generation of these self-incompatible A. thaliana AhSCR 13 -AhSRK 13 -AhARC1 Col-0 plants will be a very useful resource for further studies on sporophytic self-incompatibility in A. thaliana. Our results also show that the equivalent set of self-incompatibility genes from Brassica (BnSRK 1 , BnSCR 1 and BnARC1) was not sufficient to establish self-incompatibility in both the A. thaliana Col-0 and C24 ecotypes. These observations indicate that the transfer of self-incompatibility into A. thaliana is based on the phylogenetic relationships with the transgene donor, and it would be good to verify this by testing other combinations of SP11/SCR and SRK alleles from different Brassicaceae species. Since the Brassica-specific MLPK gene has been proven to be necessary for self-incompatibility in Brassica species [17,18], it would be worth transforming B. napus SCR-SRK-MLPK-ARC1 all together into A. thaliana to determine if MLPK is the missing component for reconstructing the Brassica self-incompatibility pathway. As well, the reciprocal experiment of transferring Arabidopsis SCR and SRK genes into self-compatible B. napus would be quite interesting in order to see if the Arabidopsis SCR and SRK genes can restore self-incompatibility in a Brassica species, or if the diversification of the self-incompatibility pathway extends in both directions.

Plant Materials
The wild-type A. thaliana Col-0 and C24 plants and the transgenic lines were grown under long-day conditions in a greenhouse with a 16-h light/8-h dark photoperiod at 22 • C. The transgenic T1 plants were selected from seeds collected following floral dipping by being plated on 1/2 MS medium containing kanamycin.

Vector Construction and Plant Transformation
The functional promoter and CDS of BrSP11-47 (the CDS sequence is the same as BnSCR 1 ) was amplified from the pCAMBIA2301-1+4 vector [46]. And the Nopaline synthase polyadenylation signal (NosT) was amplified from the pCAMBIA2301 vector used as the terminator of BnSCR 1 [64]. These two fragments were subcloned into the Bsp120I and Psp1406I sites of the pORE_O4 binary vector, respectively [65]. The SLR1 promoter is expressed in particular in stigma of Arabidopsis, and was amplified from B. napus 'Westar' genome DNA using forward and reverse primers with KpnI and XbaI sites [66,67]. The full-length CDS of BnSRK 1 was cloned from 'Westar' cDNA with primers carrying XbaI and BstEII sites. The SLR1 promoter and BnSRK 1 CDS were joined using the XbaI site by being cloned into the pCAMBIA2301 vector, then the SLR1pro-BnSRK 1 fragment was amplified from the medium vector with primers with Cfr9I and NotI sites for directional cloning into the pORE_O4 vector. A 989 bp-terminator of BnSRK 1 was obtained from 'Westar' genome DNA and was cloned into the pORE_O4 vector at the NotI site. For the reconstitution of the BnSCR 1 -BnSRK 1 -BnARC1 construct, the full-length CDS of BnARC1 was amplified from 'Westar' cDNA with primers with SalI and KpnI restriction sites. BnARC1 was also driven by the SLR1 promoter, which was amplified using primers with HindIII and SalI sites that were then cloned into the pCAMBIA2301 vector to be joined. The SLR1pro-BnARC1 fragment was further cloned by primers carrying the KpnI site and ligated into the pORE_O4 vector.
All promoters and genes from the A. halleri S 13 haplotype [47] and AhARC1 together with its promoter were synthesized by Invitrogen GeneArt (Thermo Fisher Scientific, Waltham, MA, USA). It should be noted that AhSRK 13 was driven by AhSRK 20 promoter, since AhSRK 13 and AhSRK 20 were codominant in A. halleri [41]. AhSRK 20 promoter, AhARC1pro-AhARC1, AhSCR 13 pro-AhSCR 13 and AhSRK 13 were ligated into the pORE_O4 vector using the Cfr9I-NotI, XbaI, Bsp120I and NotI-Acc65I sites or pairs of sites in the order they were mentioned (see Supplementary File S1 for further details and sequences). AhARC1 was not cloned into the pORE_O4 vector for the combination of AhSCR 13 -AhSRK 13 transgenes. All PCR products were amplified with PrimeSTAR HS DNA polymerase (Takara, Kusatsu, Shiga, Japan), and the PCR products were then ligated to pGEM-T Easy (Promega, Madison, WI, USA) and sequenced. These constructs were introduced into Agrobacterium tumefaciens GV3101 host cells. PCR primers used in vector construction steps are listed in Supplementary Table S1. The plant transformation was performed following the floral-dip method [49,68].

Pollination Assays
All the floral buds of the A. thaliana Col-0 and C24 transgenic plants were emasculated and covered with a paper bag one day before anthesis to avoid pollen contamination. Pollinations were performed the next day (day of anthesis). The pistil was cut at the peduncle 4 h after pollination, then fixed for 2 h in ethanol-acetic acid (3:1), softened in 1 N NaOH at 60 • C for 1 h and stained with 0.01% (w/v) decolorized aniline blue for 2.5 h in 2% (w/v) K 3 PO 4 . Pistils were gently squashed onto a microscopic slide glass by placing the cover glass over the pistils. Samples were examined under a fluorescence microscope (Eclipse 80i, Nikon, Minato, Tokyo, Japan).

Quantitative RT-PCR Assays
Total RNAs were extracted using the SV Total RNA Isolation System (Promega Madison, WI, USA). The RNA samples were quantified using a NanoDrop Spectrophotometer (Nanodrop Technologies, Wilmington, DE, USA), and 1 µg RNA of each sample was used to synthesize the first-strand cDNA with a Thermo RT kit (Thermo Fisher, Waltham, MA, USA). The cDNA samples were used as templates of the qRT-PCR assays. The qRT-PCRs on the transgenic lines were performed using cDNA of the mature buds and 2× Power SYBR green (Toyobo, Osaka, Osaka Prefecture, Japan). qRT-PCRs were then conducted on the CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA), and the conditions used were a pre-denaturation at 95 • C for 3 min, followed by a three-step cycle of 10 s denaturation at 95 • C, 10 s annealing at 65 • C and a 30 s extension at 72 • C for 47 cycles with a melt curve. The results were calculated with CFX Manager Software (Bio-Rad Hercules, CA, USA). Primers used in qRT-PCR analysis are shown in Supplementary Table S2.

Accession Numbers
BnSCR 1 gene (Genbank accession AB270773), BnSRK 1 gene (Genbank accession AB270771), BnARC1 gene (Genbank accession AF024625). Sequence data and accession numbers of B. napus can be found in http://www.genoscope.cns.fr/brassicanapus/. Sequence and accession numbers information for A. thaliana and A. halleri genes used in this study were obtained from TAIR and Phytozome.   Table S1. Primers used in vector construction. Table S2. Primers used in qRT-PCR assay. Supplementary File S1. Map and sequences for AhSCR13-AhSRK13-AhARC1.

Conflicts of Interest:
The authors declare that they have no competing interests.