In flowering plants, prezygotic reproductive barriers may occur at one or more check points before fertilization to prevent production of unfavourable offspring1. In nature, the stigma of an open flower is exposed to pollen from its own species, closely and distantly related interspecies, therefore must respond accordingly. Within a species, SI is widely utilized as a mechanism to avoid inbreeding depression and promote hybrid vigour by rejecting self-pollen and accepting intraspecific cross-compatible (CP) pollen2,3. Between species, interspecific incompatibility maintains species integrity and often follows the SI × SC rule, that is, interspecific pollen is UI on SI pistils but UC on SC pistils1,4,5,6. So far, little is known about the molecular mechanisms regulating these compatibility systems and their interconnections in the Brassicaceae.

In Brassicaceae SI, self-pollen is recognized by the ligand–receptor interaction between the pollen-expressed SCR/SP11 and its receptor, the stigma-expressed, plasma membrane-localized SRK2,3. This activates two downstream positive regulators of SI, the M-locus protein kinase (MLPK)15 and ARM-repeat containing 1 (ARC1) E3 ubiquitin ligase16. How MLPK functions remains unclear2,15,17. ARC1 targets compatible factors such as the exocyst component EXO70A1 for degradation, blocking pollen hydration2,16,18. We discovered that the female fertility regulator FER receptor kinase7,8,9 maintains a stigmatic gate in Arabidopsis thaliana14 and has a dual role in rejecting SI pollen and facilitating SC pollen germination in Brassica rapa10,11 by signalling a rapid elevation of stigmatic reactive oxygen species (ROS) or its decline, respectively.

SRK controls rejection of UI pollen

We first characterized pollen-induced responses during SI, SC, UI or UC pollination in B. rapa and A. thaliana stigmas by depositing pollen from B. rapa, closely related interspecific Brassica oleracea, more distant intergeneric Barbarea vulgaris and A. thaliana19,20 (Fig. 1a). Of particular interest is crosses involving B. vulgaris owing to its high resistance to fungal and insect pathogens21,22, but intergeneric barriers prevent introgression of its desirable traits into Brassica crops.

Fig. 1: SRK controls stigmatic ROS in SI to reject intraspecific and interspecific pollen.
figure 1

a, Open flowers and phylogenetic tree for Brassicaceae species from maximum likelihood analysis using internal transcribed spacer sequences. Distantly related Cochlearia danica served as an outgroup. Scale bar, mean number of nucleotide substitutions per site. b, Aniline blue staining showing intraspecific and interspecific pollen growth in mature and bud-stage B. rapa stigmas. c, H2DCFDA staining of ROS in unpollinated (UP) or pollinated B. rapa stigmas. d,e, Relaxed SI and UI in AS-BrSRK46-treated B. rapa S46 stigmas (d) and stigmas from SRK-defective B. rapa, BrSRK∆TM (e). f,g, Intraspecific and interspecific pollen (f) and ROS responses (g) in stigmas from SC A. thaliana, that is, wild type (WT), and stigmas from SI A. thaliana expressing A. halleri S13 genes (A. tha-S13). The values in the bg images, shown as average ± s.d., indicate average number of pollen tubes in the stigma (b,d,e,f) and average ROS intensity (c,g). The same data are also presented in box plots with all data points (Extended Data Figs. 1, 2). Scale bars, 0.5 cm (a), 500 μm (be) and 200 μm (f,g). Each experiment was repeated at least three times with consistent results.

We used B. rapa var. pekinensis, an economically important vegetable crop also called heading Chinese cabbage, as a representative SI species for pollen-induced responses on the stigma. A hallmark of SI in Brassicaceae is that its strength is developmentally regulated, due to the progressive increase in the amount of SRK accumulation from bud stage to maturity23 (Extended Data Fig. 1a). Compared with extensive CP pollen tube penetration into mature B. rapa stigmas, SI and interspecific pollen were severely inhibited (Fig. 1b and Extended Data Fig. 1b). Bud-stage B. rapa stigmas were comparably well penetrated by both SI and B. oleracea pollen, reflecting the involvement of an SRK-mediated mechanism in UI, mimicking SI. However, they remained impenetrable by B. vulgaris and A. thaliana pollen (Fig. 1b and Extended Data Fig. 1b), and it is likely that even the low level of SRK in bud-stage B. rapa stigmas is enough to reject intergeneric pollen.

Given that the profile of SRK expression aligns with that of ROS accumulation10, we examined whether stigmatic redox status has a crucial role in the UI response. Like the SI-induced stigmatic ROS increase10,11 but contrary to CP-induced ROS decline, UI pollen stimulated notable ROS increase in mature B. rapa stigmas. Bud-stage stigma ROS was reduced by B. oleracea pollen but was increased by B. vulgaris and A. thaliana pollen by 10 min after pollination (Fig. 1c and Extended Data Fig. 1c,d). Furthermore, sequestrating ROS by sodium salicylate (Na-SA)8,24,25 not only broke the barrier for B. oleracea pollen on mature B. rapa stigmas but also the barrier for B. vulgaris pollen on bud-stage stigmas (Extended Data Fig. 1e–g). Together, these results support UI pollen-induced ROS being essential for their rejection on SI stigmas.

Next, we tested whether SRK, in addition to rejecting SI pollen, also underlies the UI response. Treating S46 B. rapa with AS-BrSRK46 (antisense oligodeoxyribonucleotide (AS-ODN) that specifically suppresses BrSRK46)10, allowed the penetration of B. oleracea pollen tubes into mature stigmas and B. vulgaris pollen tubes into bud-stage stigmas, consistent with ROS reduction (Fig. 1d and Extended Data Fig. 2a–d). Moreover, ‘B. rapa fast plant self-compatible (FPsc)’ expresses a transmembrane domain-deleted SRK, hereafter named BrSRKΔTM, failed to respond with SI and UI pollen-triggered ROS increase and showed compromised rejection of SI and UI pollen (Fig. 1e and Extended Data Fig. 2e–j). Together, these results clearly demonstrate that SRK not only controls SI but also the rejection of interspecific and intergeneric pollen.

Furthermore, A. thaliana, a typical SC Brassicaceae with loss-of-function mutations or complete deletions of the SRK and/or SCR genes23,26,27, allowed pollen penetration from A. thaliana, B. rapa, B. oleracea and B. vulgaris, and showed stigmatic ROS reduction after pollination. However, transgenic A. thaliana expressing SP11/SCR, SRK and ARC1 from the A. halleri of S13 (referred to as A. tha-S13 hereafter), which recapitulates SI28, strongly inhibited interspecific pollen and rapidly increased stigmatic ROS (Fig. 1f,g and Extended Data Fig. 2k–m). In addition, B. oleracea S36 pollen was similarly rejected by B. rapa stigmas of several different S-haplotypes (Extended Data Fig. 2n), suggesting that SRK variations do not differ in their function in UI. Together, these results demonstrated unambiguously that SRK and SRK-dependent high stigmatic ROS levels positively correlate with the rejection of UI pollen, similar to the SI response.

UI and SI activate SRK–FER-regulated ROS

Given the role of FER in regulating ROS8,29, we suppressed BrFER1 in B. rapa stigmas by AS-ODN or crossed fer-4 mutant into SI Arabidopsis, and observed effective inhibition of SI and UI pollen-triggered stigmatic ROS increase, and marked SI and UI breakdown (Fig. 2a,b and Extended Data Fig. 3a–g). Suppressing BrANJEA1 (BrANJ1), which complexes with FER14, or BrRBOHF, which produces ROS10, also severely inhibited ROS increase and compromised SI and UI pollen rejection (Extended Data Figs. 3h–l and 4). Furthermore, yeast two-hybrid, bimolecular fluorescent complementation (Extended Data Fig. 5a,b) and protein pull-down assays demonstrated that BrSRK46 interacted with BrFER1 (Fig. 2c,d). Furthermore, protein extracts from SI and UI pollen, and SI determinant BrSCR46, markedly augmented BrSRK46–HA pulled down by MBP–BrFER1 (kinase domain (KD)), whereas CP pollen extracts and BrSCR12 had no effect (Fig. 2c,d and Extended Data Fig. 5c–f). Moreover, when co-expressed in tobacco leaves, BrSCR46 enhanced BrFER1–MYC co-immunoprecipitation with BrSRK46–HA and ROS production in tobacco leaves (Fig. 2e and Extended Data Fig. 5g–i). Together, these results suggest that SCR from SI pollen2,3 and an unknown signal from UI pollen trigger SRK-dependent activation of FER-regulated ROS production by boosting SRK–FER interaction to reject the incompatible pollen.

Fig. 2: UI and SI converge on stigmatic ROS activation for pollen rejection via the SRKFER interaction.
figure 2

a,b, Treating B. rapa stigmas with AS-BrFER1 or crossing fer-4 into SI A. thaliana stigmas, S13/fer-4, relaxed SI and UI. The values in the a,b images, shown as average ± s.d., indicate average number of pollen tubes in the stigma. The same data are also presented in box plots with all data points (Extended Data Fig. 3). ce, Pull-down (pd) (c,d) and co-immunoprecipitation (co-IP) (e) assays showing protein extracts from SI and UI pollen. BrSCR46 augmented the BrSRK46–BrFER1 interaction. The protein samples were derived from the same experiment and the blots were processed in parallel (ce). For gel source data, see Supplementary Fig. 1. Scale bars, 500 μm (a) and 200 μm (b). Each experiment was repeated at least three times with consistent results.

We then explored how the FER–ROS signalling pathway might connect with known SRK signalling15,16,17,18. Suppressing BrMLPK and BrARC1 by AS-ODN allowed the penetration of SI as well as B. oleracea pollen tubes in mature B. rapa stigmas (Extended Data Fig. 6a–d). However, although AS-BrMLPK inhibited SI and UI-triggered ROS increase, AS-BrARC1 did not (Extended Data Fig. 6e,f). Together, these results suggest that after SRK senses SI and UI pollen, it activates two parallel intracellular signalling pathways: the FER to ROS pathway that mediates pollen rejection, possibly involving MLPK, and the ARC1-mediated pathway for the degradation of compatible factors required for pollen growth.

FER and PCP-Bs favour SC over UC pollen

We used A. thaliana for pollen-induced responses on SC stigmas. Although pollen tubes from B. rapa, B. oleracea or B. vulgaris all penetrated A. thaliana stigmas by 6 h after pollination (HAP) (Fig. 1f), UC pollen tubes were much shorter than SC pollen tubes when examined earlier (1 HAP) (Fig. 3a), suggesting a more stringent barrier for UC pollen. We determined that FER has an important role in suppressing UC pollen in A. thaliana stigma as B. rapa pollen hydrated considerably faster and their tubes were much longer in fer-4 stigmas than those in wild type by 1.5 HAP (Fig. 3b). Moreover, simultaneous deposition of A. thaliana with B. rapa pollen, or A. thaliana with B. oleracea pollen on the same A. thaliana stigma showed a notable reduction in stigmatic barrier strength in fer and rbohd stigmas (Fig. 3c and Extended Data Fig. 7a–c). UC pollen was also slower than SC pollen in inducing ROS decline (Fig. 3d), consistent with FER-mediated ROS functioning as a barrier in SC stigmas to discriminate against interspecific pollen for species integrity.

Fig. 3: Species-preferential interaction between PCP-Bs and FER maintains interspecies barrier.
figure 3

a, In A. thaliana stigmas, A. thaliana pollen tubes were longer than those of B. rapa, B. oleracea and B. vulgaris at 1 HAP. b, Faster hydration and growth of B. rapa pollen on A. thaliana fer-4 stigmas. The orange plots indicate relative pollen width and the blue plots indicate pollen tube length. The equatorial diameter of a pollen grain, indicated by white dashed lines, was measured in ImageJ for pollen width. c, Relaxed interspecies barrier in fer-4 stigmas. Ratios of pollen tube length in intraspecies and interspecific crosses are used as a measure for barrier strength. In ac, the arrows indicate pollen tube front. d,e, Species-preferential ROS reduction in A. thaliana stigmas by pollen (d) and PCP-Bs (e) from intraspecies and interspecies. f, Pull-down assay showing AtPCP-Bγ competed dose-dependently with GST–BrPCP-B3 for interaction with AtFER (ED)–FLAG. The protein samples were derived from the same experiment and the blots were processed in parallel (f). For gel source data, see Supplementary Fig. 1. Scale bars, 200 μm (ae) and 20 μm (pollen in b). For box plots (ae), the centre line indicates the median, the box limits denote the lower and upper quartiles, the dots indicate individual data points, and the whiskers denote the highest and lowest data points. P values were determined by two-tailed Student t-tests. n (in blue) indicates the number of stigmas or pollen grains. In f, for the data bar, average ± s.d. is shown; average intensities from three biological replicates of the blot are represented on the left (two-tailed t-test, n = 3). Each experiment was repeated at least three times with consistent results.

Source data

Pollen coat proteins B-class (PCP-Bs) are highly polymorphic cysteine-rich peptides that are widespread in the Brassicaceae (Extended Data Fig. 7d,e) and are important for pollen hydration12,13,14. In A. thaliana, PCP-Bγ binds to FER and reduces stigmatic ROS to allow pollen growth14. We therefore examined the efficacy of PCP-Bs from different species in suppressing ROS. On A. thaliana stigmas, AtPCP-Bγ induced a marked reduction of ROS as early as 5 min after treatment, but BrPCP-B3 (ref. 12) reduced ROS to comparable levels only by 60 min after treatment. Reciprocally, B. rapa stigma ROS responded to BrPCP-B3 earlier than A. thaliana stigmas (Fig. 3e and Extended Data Fig. 7f,g). Moreover, species-specific PCP-B-induced ROS suppression was recapitulated in Arabidopsis roots, which also express FER29 (Extended Data Fig. 7h,i), thus providing further support. Pull-down assays showed that AtPCP-Bγ competed dose-dependently with BrPCP-B3, but not vice versa, for interaction with the AtFER extracellular domain (ED) (Fig. 3f and Extended Data Fig. 7j). Together, these results suggest that the species-specific match between PCP-Bs and FER is important for a rapid compatibility response and serves as an interspecific barrier to favour intraspecific pollen for fertilization.

Compatible pollen induced-NO reduces ROS

As signalling of ROS and nitric oxide (NO) is intimately engaged30,31, we investigated whether NO regulates stigmatic ROS during SC and UC responses. In A. thaliana stigmas, NO levels increased rapidly, as early as 2 min after pollination with SC pollen, reaching a maximum by 5 min after pollination, and declined sharply to pre-pollination level, paralleling that of pollen hydration (Fig. 4a, upper panel, and Extended Data Fig. 8a,b). However, B. rapa pollen and BrPCP-B3 were notably slower than A. thaliana pollen and AtPCP-Bγ, respectively, in stimulating NO in A. thaliana stigmas (Fig. 4b,c). Furthermore, unlike wild type, fer-4 stigma NO was non-responsive to SC pollination and AtPCP-Bγ treatment (Fig. 4d,e and Extended Data Fig. 8c,d). Species-preferential PCP-B-induced NO increase was also recapitulated in roots (Extended Data Fig. 8e–g). The intraspecific and interspecific pollen and respective PCP-B-induced differential NO increases (Fig. 4b,c) were opposite to their induced changes in ROS (Fig. 3d,e), revealing an inverse functional relationship between NO and ROS in discriminating against interspecific compatible pollen.

Fig. 4: Compatible pollination induces NO to suppress FER-mediated ROS production.
figure 4

a, DAF-FM DA staining of pollination-induced NO responses and pollen hydration in A. thaliana stigmas. The equatorial diameter of a pollen grain, indicated by white dashed lines, was measured in ImageJ for pollen width. b,c, Species-preferential elevation of NO in A. thaliana stigmas. d,e, FER-dependent NO elevation in A. thaliana stigmas induced by pollen (d) and AtPCP-Bγ (e). The values in the ae images, shown as average ± s.d., indicate average NO intensity (ae) and equatorial diameter of pollen grains (a). The same data are also presented in box plots with all data points (Extended Data Fig. 8). f,g, Nitrosation of FER. BrFER1 was nitrosated in vitro by the NO donor GSNO (f), and AtFER–GFP from transformed A. thaliana stigmas was nitrosated in vivo by SC pollination (g). h,i, Pull-down assays showing that nitrosation of FER in vitro (h) and by pollination (i) reduced its interaction with the downstream ROP2 signalling module. The protein samples were derived from the same experiment and the blots were processed in parallel (fi). For gel source data, see Supplementary Fig. 1. Scale bars, 200 μm (ae) and 50 μm (pollen in a). For box plots (b,c), the centre line indicates the median, the box limits denote the lower and upper quartiles, the dots indicate individual data points, and the whiskers denote the highest and lowest data points. n (in blue) indicates the number of stigmas or pollen grains. P values were determined by two-tailed Student t-tests. Each experiment was repeated at least three times with consistent results.

Source data

Additional observations further supported the role of NO in regulating stigmatic ROS and pollen growth. Scavenging A. thaliana stigma NO by cPTIO31 suppressed SC-induced NO increase and ROS reduction and prevented SC pollen growth (Extended Data Fig. 8h,i). Pollination experiments carried out in NO-deficient32,33,34 and NO-overaccumulating35 mutants similarly supported the importance of NO involvement (Extended Data Fig. 8j–p). Furthermore, in B. rapa, stigmatic NO responded similarly, stimulated by SC but not by SI or UI pollen (Extended Data Fig. 9a,b). Scavenging or increasing NO by cPTIO or S-nitrosoglutathione (GSNO), respectively, also induced opposite changes in ROS and pollen growth in B. rapa stigmas (Extended Data Fig. 9c–h). These results established firmly that NO is specifically induced in stigmas by compatible pollen for ROS reduction and pollen growth.

NO nitrosates FER in compatible response

A major bioactivity of NO is S-nitrosation of specific cysteines in many proteins9,36,37. Given the role of FER and RBOHs examined here (Extended Data Figs. 3 and 4), they could be targets for nitrosation. A tandem mass tag-switch analysis9,38 of GSNO-treated MBP–BrFER1 showed nitrosation of Cys730, and Cys752 in the KD, which are conserved among FERs from many Brassicaceae species (Fig. 4f and Extended Data Fig. 9i–l). AtFER–GFP in transformed A. thaliana stigmas8,29 was also nitrosated after SC pollination (Fig. 4g), suggesting that nitrosation of FER has important roles during the compatible response.

Nitrosation of proteins may affect their stability, biochemical properties and interaction with other proteins9,39. GSNO-treated BrFER1 (KD) and a Cys730Trp (BrFERC730W) conversion, which mimics nitrosation40, both showed compromised interaction with its downstream BrROP2 signalling complex (Fig. 4h and Extended Data Fig. 9m,n). We also observed substantial reduction in AtFER–GFP pulled down by AtROP2 in A. thaliana stigmas with SC pollen (Fig. 4i). Moreover, stigma-expressed RBOHs were also nitrosated in vitro and by pollination (Extended Data Fig. 9o–t), similar to immunity signalling-induced nitrosation36. Together, these results indicate that stigmatic ROS decline is a consequence of compatible pollination-stimulated NO, nitrosating FER to inactivate downstream RAC/ROP-regulated RBOH-dependent ROS production, and the already produced RBOHs to rapidly quell ROS-producing activity in stigmas.

Breaking the barrier for distant breeding

Having established the mechanisms underlying the interspecific barrier at the stigma, we next explored to what extent breaking the barrier might promote distant breeding. We treated B. rapa pistils with Na-SA to reduce the levels of ROS and GSNO to increase NO levels, or AS-ODNs to disrupt the BrSRK–BrFER interaction and BrFER-to-BrRBOH signalling, then pollinated them with SI and UI pollen. By 12 days after pollination, SI B. rapa pollen, B. oleracea pollen and B. vulgaris pollen all resulted in enlarged ovules with developing embryos in these treated pistils (Fig. 5a,b and Extended Data Fig. 10). Additional barriers beyond the stigma must have precluded robust cross-fertilization and hampered hybrid embryo development (Fig. 5a,b and Extended Data Fig. 10). Combining the strategies used here with embryo rescue, an in vitro culture technique widely utilized in distant breeding41, should allow successful development of interspecific hybrid embryos into viable plants.

Fig. 5: Breaking the stigmatic barrier for distant breeding of Brassicaceae crops.
figure 5

a,b, Reducing stigmatic ROS by a ROS scavenger, increasing NO by a NO generator (a), and disrupting the BrSRK–BrFER interaction and BrFER1-to-BrRBOH signalling by AS-ODNs (b) alleviated the interspecific and intergeneric reproductive barrier. The arrowheads indicate enlarged ovules. The dashed lines denote the outline of hybrid embryos. The values in the a,b images, shown as average ± s.d., indicate average number of enlarged ovules in the pod. The same data are also presented in box plots with all data points (Extended Data Fig. 10). Scale bars, 0.5 cm (siliques) and 100 μm (emb ryos). Each experiment was repeated at least three times with consistent results. c, Model of FER-regulated ROS as a shared signalling node in SI, UI, SC and UC responses. In stigmas from SI species, SI pollen and UI pollen activate ROS via the SRK–FER–ROP2–RBOHs pathway, functioning along with the ARC1-mediated processes. Stigmas from SC species are compatible to interspecific pollen, but species-preferential interaction between PCP-Bs and FER initiates a faster compatible response via NO-mediated nitrosation of FER and RBOHs to suppress ROS production, promoting intraspecies precedence and protecting species integrity. The dashed lines and ‘?’ indicate ‘to be determined’. P, phosphate; P-some, proteosome; Ub, ubiquitin.


Upon pollination, pollen and stigma engage in a series of communications to facilitate growth of desirable pollen and discourage invaders and less desirable pollen1. Exploring SI, UI, SC and UC together in one study, we demonstrate how the interplay between two receptor kinases, SRK and FER, provides the capacity to perceive different pollen signals to fine tune the stigmatic redox conditions and determine acceptance or rejection of pollen from intraspecies or sympatric interspecies with similar flowering time. As summarized in our working model (Fig. 5c), we demonstrate how incompatible signals from SI and UI pollen activate FER-mediated ROS production for pollen rejection in SI stigmas, functioning in parallel with ARC1-mediated processes2. We also demonstrate how the species-preferential interaction of PCP-Bs and FER leads to interspecific barriers in SC stigmas.

The Brassicaceae includes many important vegetable and oil crops. Breeding within a species is far from maximizing hybrid vigour owing to the relatively narrow genetic diversity. Breeding between species enriches germplasm resources but is restricted by interspecific barriers, rendering distant breeding with slim chances of success. Achieving interspecific and intergeneric fertilization here by breaking the stigmatic barrier in B. rapa, in particular the production of hybrid embryos with the fungal-resistant and insect-resistant B. vulgaris21,22 (Fig. 5a,b), represents a remarkable feat and major breakthrough, enabling introgression of desirable traits into crops from distant species.


Plant materials and growth conditions

Stigmas from B. rapa var. pekinensis, also known as heading Chinese cabbage, and A. thaliana were mostly used for pollination responses. For interactions on SI stigmas, B. rapa stigmas from a double haploid line of S46 were pollinated with pollen from B. rapa S46 or S12 as SI or CP pollinations, or pollen from B. oleracea (S36), B. vulgaris (SC) and A. thaliana (SC) as UI pollinations. For interactions on SC stigmas, a variety FPsc with a defective allele of SRK that lacks the coding sequence for the transmembrane domain and SC A. thaliana were used. B. rapa stigmas from S46, S12, S9, S40 and S38 were pollinated with B. oleracea (S36) for haplotype-dependent analysis. Transgenic SI A. thaliana28, pAtFER::AtFER–GFP29 plants, homozygous fer-4 (GK-106A06, GABI-Kat)29, srn (the γ-ray mutant)42, hot5-4 (FLAG_298F11, Versailles Genomic Resource Centre)35, noa1 (CS6511)9, rbohd-3 (Salk_070610C)10 and rbohd-4 (CS9555)10 mutants and their corresponding wild-type plants Col-0, C24 or WS have been previously described. Transgenic plants pAtFER::GFP–AtRBOHD were generated by the Agrobacterium tumefaciens-mediated floral dip method43. S13/fer-4 was generated by crossing fer-4 mutant into SI Arabidopsis (S13 of A. halleri); / indicates that these two genetic modifications are in the same Arabidopsis plant.

Seeds of B. rapa, B. oleracea and B. vulgaris were germinated in potted soil (Pindstrup substrate, Denmark). Vernalization was performed in a growth chamber with 10 °C–5 °C, 14 h–10 h light–dark cycles, and light intensity of 100 mmol m2 s−1. After 1 month of cold treatment for 1-week-old B. rapa, and 3 months for 7–8 leaf stage B. oleracea and B. vulgaris plants, these plants were planted in soil under greenhouse conditions with 25 °C–15 °C, 16 h–8 h light–dark cycles, and light intensity of 300 mmol m2 s−1. Seeds of A. thaliana and Nicotiana benthamiana were germinated and grew in potted soil in a greenhouse at 22 °C, 16–8 h light–dark cycle with a relative humidity of 60%.

Statistical analysis

Data involving ROS, NO, pollen hydration and pollen tube length were presented as box plots generated in GraphPad Prism v8.0.1. Unless otherwise indicated, the centre line of box plots denotes the median, the box limits denote the lower and upper quartiles, and the whiskers denote the lowest and highest data points. Protein plots and quantitative PCR with reverse transcription (qRT–PCR) related data were presented as data bars (average ± s.d., n = 3). Data involving changes of stigma NO over time after pollination were presented as a line chart (average ± s.d.; n indicates the number of stigmas). All statistics data were labelled with the exact P value, and dots in data bars and box plots denote individual data points. Each experiment was repeated at least three times with consistent results.

Stigma treatment and pollen growth observation

Compatibility was demonstrated by the number of pollen tubes that had penetrated the stigma papilla cells. Stigma feeding assays followed that of refs. 10,44. B. rapa flowers that have just opened but before anther dehiscence, or bud-stage flowers were emasculated and cut at 3 mm away from the stigmatic surface. Excised stigmas were inserted into basic PGM (5 mM CaCl2, 5 mM KCl, 0.01% H3BO3, 1 mM MgSO4•7H2O, 10% sucrose and 0.8% agarose, pH 7.5) or the treatment medium and kept in a chamber with constant temperature (22.5 °C) and humidity (45%) for the indicated period of time (S-ODNs or AS-ODNs for 1 h; Na-SA, GSNO and cPTIO for 6 h). Treated stigmas were transferred to basic PGM medium and manually pollinated with similar amount of SI, CP or various UI pollen and maintained in the same condition for 6 h or as indicated. Stigmas were then fixed in Canoy’s fixative (ethanol to acetic acid 3:1), softened in 10 M NaOH, and stained in 0.1% aniline blue. Pollen tubes were visualized by epifluorescence (Ex375-328, DM415 and BA351p) on a Nikon Eclipse Ni. Images were captured by a DS-Ri2 digital camera.

For in planta treatment, just open flowers, or bud-stage flowers, on inflorescences of B. rapa plants were treated twice at a 30-min interval, with S-ODN and AS-ODN, Na-SA, GSNO or the corresponding mock solution, supplemented with 0.0125% Tween, then pollinated with a similar amount of SI, CP, B. oleracea or B. vulgaris pollen. The number of enlarged ovules was counted at 12 days after pollination. Embryo clearing and observation followed that of ref. 45 with modifications. Enlarged ovules were cleared in Hoyer’s medium (7.5 g gum arabic, 100 g chloral hydrate, 5 ml glycerol and 60 ml H2O) for 5 days, then observed under differential interference contrast on a Nikon Eclipse Ni microscope equipped with a DS-Ri2 digital camera.

The effect of chemicals or the mutation of stigma-expressed genes on the growth of intraspecific or interspecific pollen on SC A. thaliana stigmas was demonstrated by the rate of pollen hydration or pollen tube length. SC A. thaliana flowers were emasculated at stage 12 (ref. 46), cultured in PGM for 14 h, then pollinated with A. thaliana pollen or B. rapa pollen. For pollen hydration, images were taken at each time point after pollination under differential interference contrast on a Nikon Eclipse Ni microscope equipped with a DS-Ri2 digital camera. The equatorial diameters of pollen grains at various time points were measured in ImageJ v1.53c. For pollen tube length, A. thaliana stigmas with A. thaliana pollen, B. rapa pollen or B. oleracea pollen were processed for aniline blue staining at 1 or 1.5 HAP. In dual-pollination assays, A. thaliana WT and fer-mutant pistils (or other mutant pistils) were simultaneously pollinated with A. thaliana pollen on half and either B. rapa or B. oleracea pollen on the other half of the same stigma, with a clear boundary in between. Pollen growth from each half of the stigma was clearly confined to the corresponding half of the pistils and readily distinguishable.

ODN design and treatment

ODN design and treatment of stigmas followed that of ref. 10. S-ODNs and AS-ODNs were used to target the following genes (accession numbers shown in Supplementary Table 1): BrSRK46, BrFER1 and BrRBOHF. S-ODNs or AS-ODNs were designed based on Sfold ( The BLAST program ( was used to assess potential off-target effect. The ODNs were synthesized in the Beijing Genomics Institution (BGI). Three bases at both 5′ and 3′ end of S-ODN and AS-ODN were phosphorothioate-modified to maintain stability. The sequences of S-ODNs and AS-ODNs were listed in Supplementary Table 2. Stigmas were excised at the style 1 mm away from the top, inserted in PGM containing the S-ODN or AS-ODN and treated for 1 h. Stigmas were subjected to aniline blue assay at 6 HAP to observe pollen growth.

Staining of ROS and NO

For stigmatic ROS staining, B. rapa stigmas or A. thaliana stigmas, unpollinated or at 10 min after pollination with SI, CP or UI pollen, were pretreated in MES buffer (10 mM MES, 5 μM KCl and 50 μM CaCl2, pH 6.15) for 30 min, stained with 50 μM H2DCF-DA (2′,7′-dichlorofluorescein diacetate; Sigma-Aldrich) for 30 min, respectively, then washed at least three times in buffer before observation. For stigmatic NO staining, B. rapa stigmas or A. thaliana stigmas before or after pollination were soaked in Tris buffer (10 mM Tris-HCl and 10 mM KCl, pH 7.5) for 30 min, stained with 20 μM DAF-FM DA (3-amino,4-aminomethyl-2′,7′-difluorescein diacetate; Thermo Scientific) for 1 h, then washed at least three times in buffer before observation.

For treatments, stigmas were first soaked in MES buffer for 30 min, then treated with S-ODNs or AS-ODNs, chemicals, peptides or pollen extracts at indicated concentration and duration in MES buffer supplemented with 0.0125% Tween 20. After washing three times in MES buffer, treated stigmas were stained for ROS or NO as above described.

For root ROS or NO staining, 3-day-old A. thaliana seedlings or 2-day-old B. rapa seedlings were soaked in the corresponding buffer for 30 min, treated with 0.1 μM of AtPCP-Bγ or BrPCP-B3 for the indicated period of time, then stained with 50 μM H2DCF-DA or 20 μM DAF-FM DA for 1 h.

Imaging was carried out under eGFP epifluorescence (Ex470-440, DM4951p and BA525/550), using a Nikon Eclipse Ni and equipped with a DS-Ri2 digital camera. The exposure time for all comparative samples were exactly the same within one experiment. Average ROS or NO signals outlined in the dotted area were measured in Image J v1.53c; ROS or NO in control stigmas were set at 1 for comparative analyses.

For comparison, stigmas were directly stained with the same concentrations of H2DCF-DA or DAF-FM DA for 10 min without buffer pretreatment, and imaged under a confocal microscope (Zeiss LSM880). Changes of ROS and NO were consistent with those imaged under wide-field fluorescence microscope. Wide-field observation, due to its considerably high expedition to sample entire stigma specimens, was used for the massive amount of data gathering required for this study.

For ROS staining of infiltrated tobacco leaves, agrobacterium cells containing BrFER1–MYC, GFP–BrRBOHD2 and BrSRK46–HA were mixed and infiltrated into tobacco leaves. Two days after infiltration, 10 μM GST–BrSCR12 or GST–BrSCR46 were injected into the same leaf and treated for 10 min. Nitro blue tetrazolium (NBT) staining47 was used to detect ROS of the infiltrated leaves. The leaves were vacuum infiltrated with 5 mg ml−1 NBT (in 10 mM sodium citrate, pH 7.0) for 40 min at room temperature. Leaves were then decolourized three times by boiling in decolourized solution (95% ethanol to glycerine 3:1) for 10 min. The images were captured by a digital camera and measured for ROS intensity in Image J v1.53c. For ROS staining of protoplasts, protoplasts were isolated from the above infiltrated tobacco leaves co-expressing BrFER1–MYC, GFP–BrRBOHD2 and BrSRK46–HA, following that of ref. 48. Protoplasts were then treated with buffer, BrSCR12 and BrSCR46, respectively, for 15 min, before staining with 10 μM H2DCF-DA for 10 min. Protoplasts from tobacco leaves infiltrated with buffer were stained for ROS as a control.

Enzymatic activity of RBOHs

Stigmas before or after pollination or treatment were washed three times with the MES buffer and then freezed in liquid N2. Enzymatic activity of RBOHs was measured following the manufacturer’s instructions (Nanjing JC bio). In brief, approximately 0.05 g stigma tissue (approximately 100 stigmas) was ground in liquid N2, extracted in buffer (0.2 M NaH2PO4 and 0.2 M Na2HPO4, pH 7.2) and centrifuged at 4,000g for 20 min at 4 °C. The resulting supernatant was used to measure RBOH activity spectrophotometrically at 340 nm using FAD and NADPH as substrates. The result was shown as the average of three technical replicates of one sample.

RNA isolation and qRT–PCR

Total RNA was extracted via the SteadyPure Universal RNA Extraction Kit (AG21017, Accurate Biotechnology) and reverse transcribed with HiScript III 1st Strand cDNA Synthesis Kit (Vazyme). qRT–PCR was performed on a QuantStudio 3 system (Applied Biosystems) with ChamQ SYBR qPCR Master Mix (Vazyme), using BrACTIN2 as internal controls. A list of gene-specific primers used for qRT–PCR is included in Supplementary Table 3.

Molecular cloning

For the BrSRK46–HA or BrRBOHD2–GFP construct, sequences encoding the full length of BrSRK46 or BrRBOHD2 were amplified from B. rapa stigma cDNA using 2× Phanta Max Master Mix (Vazyme). The fragment was cloned into the modified pCAMBIA1300 vector with a HA or GFP tag using pEASY-Basic Seamless Cloning and Assembly Kit (TransGen Biotech). For the GST fusion protein constructs, sequences encoding BrPCP-B3 (residues 1–76), BrSCR46 (residues 1–78), BrSCR12 (residues 1–76), full length of BrROP2, C-terminal cytoplasmic region of BrRBOHD1 (CT; residues 758–923), BrRBOHD2 (CT; residues 614–904) and BrRBOHF (CT; residues 767–949) were amplified and cloned into a pGEX-4T-1 vector. For the MBP fusion protein constructs, sequences encoding ED of BrFER1 (residues 14–420), KD of BrFER1 (residues 486–783), mature AtPCP-Bγ (residues 23–76) and AtROP2 were amplified and cloned into a pMAL-p2X vector. For the FLAG fusion protein construct, the ED of AtFER (residues 29–446) and the KD of BrSRK46 (residues 448–860) were amplified and cloned into the pFLAG-CTS vector. For the MYC fusion protein construct, BrFER1 (residues 1–788) was amplified and cloned into pCXSNF. For the bimolecular fluorescent complementation constructs, the KD of BrSRK46 (residues 448–860) and BrFER1 (residues 486–783) were amplified and cloned into pDONR207 by the Gateway BP reaction via Gateway BP Clonase Enzyme Mix (Invitrogen Life Technologies) and cloned into the pEARLY GATE 201 vector or pEARLY GATE 202 by the LR reaction via Gateway LR Clonase Enzyme Mix (Invitrogen Life Technologies). For the yeast two-hybrid constructs, the KD of BrSRK46 (residues 448–860) and the KD of BrFER1 (residues 486–783) were amplified and cloned into the pGADT7 or pGBKT7 vectors. For MBP–BrFER1C730W (KD), the mutant fragment was amplified from the MBP–BrFER1 (KD) vector with appropriate PCR primers using the Fast Mutagenesis System (TransGen Biotech) according to the manufacturer’s instructions. A list of gene-specific primers used for the above constructs is included in Supplementary Table 4.

Protein expression and purification

The constructs of GST, MBP and FLAG fusion protein were transformed into Escherichia coli BL21 (DE3) for protein expression. After induction by 0.5 mM IPTG at 37 °C for 4–6 h, the cells were spun down and resuspended in 5 ml PBS (140 mM NaCl, 2 mM KCl, 2 mM KH2PO4 and 10 mM Na2HPO4.7H2O). Sonicated (SCIENTZ) protein was purified by magnetic GSH beads (BEAVER), amylose magnetic beads (PuriMag Pro) or anti-FLAG magnetic beads (BeyoMag), respectively. The eluted protein was separated by SDS–PAGE and detected by the corresponding antibody after western blot.

For peptide preparation, MBP–AtPCP-Bγ and GST–BrPCP-B3 were expressed in E. coli BL21 cells and purified by corresponding magnetic beads. The AtPCP-Bγ (residues 23–76) peptides were also synthesized by Scilight Biotechnology with more than 95% of purity. The peptides were diluted to 1 mM in sterile ddH2O as stock solution.

For proteins expressed in tobacco leaves, Agrobacterium tumefaciens GV3101 (Weidi) cells containing corresponding constructs were infiltrated into N. benthamiana leaves following standard procedure48. In brief, Agrobacterium cells were spun down at 5,000g for 10 min at 4 °C and resuspended to an optical density at 600 nm of 0.6 in the infiltration buffer (10 mM MES, 10 mM MgCl2 and 0.5% glucose, pH 6.5), with 100 μM acetosyringone added before infiltration. Leaves from 5–6-week-old tobacco plants were infiltrated using a 1-ml syringe. Two days after infiltration, leaves were homogenized in liquid N2, mixed in 1 ml plant protein extraction buffer (75 mM KAc, 300 mM NaCl, 100 mM arginine, 10 mM EDTA, 0.25% Triton X-100, pH 7.4, 1 mM PMSF and 1 mM cocktail protease inhibitor) in a 1.5-ml centrifuge tube and stayed on ice for 20 min. After centrifuge at 13,800g for 10 min at 4 °C, the supernatant was transferred into a new tube for pull-down or co-IP assays.

Protein interaction assays

For the yeast two-hybrid assay of BrFER1 (KD) and BrSRK46 (KD), vector construction, yeast transformation, growth on drop out medium and X-Gal staining followed standard procedure29.

For the bimolecular fluorescent complimentary assay, the pEARLY GATE 201 containing BrFER1 (KD)–nYFP and the pEARLY GATE 202 containing BrSRK46 (KD)–cYFP were transformed into Agrobacterium tumefaciens strain GV3101. Equal volumes of two cultures were mixed and infiltrated into N. benthamiana leaves as described above. Two days after infiltration, imaging was carried out under eGFP epifluorescence (Ex470-440, DM4951p and BA525/550), using a Nikon Eclipse Ni and equipped with a DS-Ri2 digital camera.

For the pull-down assay of BrSRK46–HA by MBP–BrFER1 (KD), BrSRK46–HA protein was extracted from infiltrated tobacco leaves, mixed with amylose magnetic beads (PuriMag Pro)-bound MBP–BrFER1 (KD) bait protein for 6 h at 4 °C. The beads were washed three times in pull-down buffer (50 mM Tris-HCl, pH 7.0, 100 mM NaCl and 0.1% Triton X-100 (v/v)) and boiled in SDS–PAGE loading buffer for 5 min. The proteins were then processed for SDS–PAGE, western blot and immunodetection by anti-MBP antibody (1:15,000; Abmart) or anti-HA antibody (1:5,000; Abmart), and with anti-mouse horseradish peroxidase (HRP)-conjugated secondary antibody (1:15,000; Abmart) followed by the HRP detection kit (Vazyme) analysis with a chemiluminescence imaging system (TIAN NENG). For the effect of peptides, GST–BrSCR46 or GST–BrSCR12, or protein extracts from SI (S46), CP (S12) pollen or interspecific (B. oleracea) pollen on the interaction of BrSRK46–BrFER1 (KD), 0.5 μM peptides or 0.65 mg ml−1 protein extracts were incubated with BrSRK46–HA protein for 1 h before the pull-down assay.

For the co-IP assay of BrFER1–MYC by BrSRK46–HA, GV3101 Agrobacterium cells containing 35S::BrSRK46–HA or 35S::BrFER1–MYC were mixed and infiltrated together into N. benthamiana leaves. Two days after infiltration, 10 μM GST–BrSCR12 or GST–BrSCR46 were infiltrated into the same leaf and treated for 10 min. The leaves were then used for protein extraction. Aliquots of 1 ml supernatant were incubated with 100 μl anti-HA magnetic beads (PuriMag Pro) at 4 °C overnight with rotation. After washing three times, the beads were boiled and the proteins were processed for SDS–PAGE, western bolt and immunodetection by anti-MYC antibody (1:5,000; Abmart) and anti-HA antibody (1:5,000; Abmart) and anti-mouse HRP-conjugated secondary antibody (1:10,000; Abmart).

For the pull-down assay of MBP–BrFER1 (KD) by GST–BrROP2, MBP–BrFER1 (KD) protein was treated with 1, 2 and 5 mM GSNO for 3 h at room temperature, then mixed with GSH bead (BEAVER)-bound, GST–BrROP2 bait protein for 6 h at 4 °C. The beads were boiled and processed for SDS–PAGE, western bolt and immunodetection by anti-MBP antibody (1:15,000; Abmart) or anti-GST antibody (1:5,000; Abmart), and with anti-mouse HRP-conjugated secondary antibody (1:15,000; Abmart). MBP–BrFER1C730W (KD), without GSNO treatment, was processed similarly for the pull-down assay with GST–BrROP2.

For the pull-down of AtFER–GFP by MBP–AtROP2, MBP or MBP–AtROP2 bait proteins were bound with amylose magnetic beads (PuriMag Pro) for 6 h and washed three times with pull-down buffer. A. thaliana stigmas from pAtFER::AtFER–GFP transgenic plants, unpollinated or 10 min after pollination, were used for total protein extraction. AtFER–GFP protein was incubated with corresponding bait protein for 8 h at 4 °C. After washing three times, the beads were boiled and processed for SDS–PAGE, western bolt and immunodetection by anti-GFP antibody (1:10,000; Abmart) or anti-MBP antibody (1:15,000; Abmart), and with anti-mouse HRP-conjugated secondary antibody (1:10,000; Abmart).

For the pull-down assay to examine peptide competition, anti-FLAG magnetic bead (BeyoMag)-bound AtFER (ED)–FLAG bait protein was incubated with GST–BrPCP-B3 with rotation at 4 °C for 2 h, then increasing concentrations of AtPCP-Bγ peptides were added for competition at 4 °C for 3 h. The competition by GST–BrPCP-B3 peptides with MBP–AtPCP-Bγ in interaction with AtFER (ED)–FLAG bait protein was performed similarly. The beads were boiled and processed for SDS–PAGE, western bolt and immunodetection by anti-FLAG antibody (1:5,000; Abmart), anti-GST antibody (1:5,000; Abmart) or anti-MBP antibody (1:15,000; Abmart), and anti-mouse HRP-conjugated secondary antibody (1:5,000; Abmart).

Protein nitrosation assay

In vitro S-nitrosation assay followed that of ref. 35 with modifications, using Pierce S-Nitrosylation Western Blot Kit (90105, Thermo Scientific). Purified proteins of GST–BrRBOHD1 (CT), GST–BrRBOHD2 (CT), GST–BrRBOHF (CT) and MBP–BrFER1 (KD) were desalted by acetone precipitation and resuspended in HENS buffer (100 mM HEPES, pH 7.0, 1 mM EDTA, 0.1 mM neocuproine and 2.5% SDS). Approximately 150 μg proteins per sample were incubated with 1 mM GSNO in a reaction volume of 100 μl HENS buffer for 2 h at room temperature in the dark. The sample was precipitated with cold acetone and resuspended in 100 μl HENS buffer. After incubation at 50 °C for 1 h with 200 mM N-ethylmaleimide (Solarbio) and at room temperature for 30 min, the sample was precipitated with cold acetone and resuspended in HENS buffer. The sample was treated with 60 mM sodium ascorbate and 0.4 mM iodoTMTzero label reagent for 2 h. All the above steps were carried out in the dark. The proteins were finally precipitated by cold acetone and resuspended in 2 M urea buffer. Aliquots of each protein were separated by SDS–PAGE and then analysed by immunoblotting with anti-TMT antibody (1:5,000; Thermo Scientific) and goat anti-mouse IgG (H+L)–HRP (1:10,000; Thermo Scientific) were used for detection of nitrosated protein. In parallel, the proteins were detected for loading with anti-GST antibody (1:5,000; Abmart) or anti-MBP antibody (1:15,000; Abmart), and goat anti-mouse IgG–HRP (1:10,000; Abmart).

For the analysis of S-nitrosation in stigmas before and after pollination, 200 stigmas (approximately 0.1 g) from A. thaliana plants expressing pAtFER::AtFER–GFP or pAtFER::GFP–AtRBOHD, unpollinated or at 10 min after pollination with intraspecific compatible pollen (WT Arabidopsis pollen), were homogenized in liquid N2 and resuspended in 0.7 ml of the plant protein extraction buffer. Protein extracts, 1.5 mg in 150 μl, were precipitated with cold acetone and resuspended in HENS buffer for the detection of nitrosated proteins similar to in vitro nitrosation detection.

Mass spectrometry analysis of nitrosated residue

Mass spectrometric identification of S-nitrosated cysteine residues was carried out by Shanghai Bioprofile Biotechnology (China). The protein pellets were resolved with HENS buffer and N-ethylmaleimide (Sigma) was used to block the free cysteine. The S-nitrosation sites of protein were reduced by the sodium ascorbate (Sigma) specifically and then labelled with iodoTMT zero reagent (Thermo Scientific). The processed proteins were digested with trypsin in 50 mM NH4HCO3 overnight at 37 °C. The peptides were then desalted with C18 cartridge (Thermo Scientific). The iodoTMT-labelled peptides were enriched by anti-TMT resin as instructed by the manufacturer (Thermo Scientific). Then, the enriched peptides were loaded into liquid chromatography–mass spectrometry for analysis. The Q Exactive HF-X mass spectrometer coupled to Easy nLC1200 (Thermo Scientific) were performed on a 2-h time gradient for peptide mass spectrometry detection. The mass spectrometry raw data were imported into MaxQuant software v1.6.0.16 for data interpretation and protein identification against the B. rapa genome database49 ( The search results were filtered and exported with a less than 1% false discovery rate at the site level, peptide-spectrum-matched level and protein level. MaxQuant analysis was filtered only for those nitrosated sites (iodoTMT labelled) that were confidently localized (class I, localization probability of more than 0.75) and the score of the modified peptide was more than 40.

Bioinformatic analysis

All sequences analysed were retrieved from the B. rapa genome database49 (, NCBI GenBank50 ( or EnsemblPlants51 ( or Phytozome52 ( C. danica was selected as the outgroup to infer the species relationships among B. rapa, B. oleracea, B. vulgaris, A. thaliana, Raphanus sativus, Arabidopsis lyrata and A. halleri. The maximum likelihood tree based on the alignment of nuclear ribosomal internal transcribed spacers from the above-mentioned species was inferred under a Tamura–Nei nucleotide substitution model with 1,000 bootstraps. For the phylogenetic tree of RBOHs, PCPs and SRKs, corresponding sequences were aligned using the MUSCLE algorithm implemented in MEGA X53, and constructed using the neighbour-joining method in MEGA X with 1,000 bootstraps. The amino acid sequence of PCP-Bs and FER alignment was performed by the online software Clustal Omega54 ( with default parameters then import into ESPript 3.0 (ref. 55) ( to generate pictures.

Figure preparation

Average ROS and NO signals, pollen grain width and pollen tube length were measured by Image J v1.53c ( Histograms were prepared by GraphPad Prism v8.0.1 ( The main figures were assembled in Adobe Illustrator; all other figures were assembled in Adobe Photoshop. Some cartoon components were from for model drawing.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.