2.1 Plant materials

This study used hardy water lily (Nymphaea. ‘Peter Slocum’) and tropical water lily (Nymphaea. ‘NangKwaug Fah’) as experimental materials. The water lily plants were preserved in the Bai Ma Education Demonstration Base of Nanjing Agricultural University, China. There are two main reasons for choosing these two cultivars for this study. Firstly, both cultivars have high fruit set rates and good pollen vitality, eliminating the influence of fruit set rates and pollen vitality in the experiment. Secondly, both cultivars exhibit obvious pre-fertilization hybridization barriers. N. ‘Peter Slocum’ and N. ‘NangKwaug Fah’ were used as the female and male parents, respectively. The hybridization was conducted following the methods described by Sun et al. (2018, 2019).

2.2 Phytohormone detection by LC–MS/MS

The stigma of water lily was collected 6 h after pollination, including self-pollination (SP), cross-pollinated in the same subgenus (CP), and inter-subgenus cross-pollination (ISCP) treatments. An unpollinated sample (Mock) was used as control. Each sample had 4 biological replicates. After being flash-frozen in liquid nitrogen, the samples were stored at −80°C. The levels of ABA, IAA, GA, JA, SA, and BR, which belong to the six major classes of hormones, were detected in the stigmas of water lily.

The stigmas were frozen in liquid nitrogen, ground into fine powder, and 50 mg of powder was taken for extraction in a mixture of methanol:water:formic acid (15:4:1, v/v/v). The extraction solution was filtered through a 0.22 μm PTFE membrane (Anpel) and then analyzed using LC-ESI-MS/MS (Hui et al. 2018). The obtained mass spectrometry data were qualitatively analyzed by establishing the MWDB plant hormone database based on relevant standards (Šimura et al. 2018). The hormone content detection experiment was conducted at Suzhou Panomic Bio-medical Technology Co., Ltd..

2.3 Hormone treatments and pollen tube visualization

The stamens of the first day blooming waterlily flower were removed, and then 30 μM ABA, 0.3 mM GA or 0.5 μM BR were added into the pistil fluid of the waterlily flower. The hormone concentrations used in the experiment were modified by reference to the hormone treatment concentrations in previous articles (Kovaleva et al. 2005, Kovaleva et al. 2017, Breygina et al. 2023, Wang et al. 2023). After about 10 minutes of exogenous hormone application, pollination was performed.

Six hours after pollination, the stigmas were fixed in FAA fixative for 6 hours, then soften overnight in 1 mM NaOH. The softened stigmas were stained with aniline blue using a previously detailed experimental protocol (Duan et al. 2014, Su et al. 2020, Zhang et al. 2021) and observed under a fluorescence microscope at a wavelength of 405 nm. Images were captured using a DS-Ri2 digital camera.

For the calculation of fruiting percentage, we did 3 independent experiments, each consisting of 6 independent pollination events. The ratio of the number of water lily fruits obtained in each experiment to the total number of flowers in the treatment was calculated as the fruition rate. For the calculation of the number of seeds, we had six replications for each treatment and went through three experiments. The average number of mature seeds harvested from each water lily fruit in each trial was calculated.

2.4 Antioxidant enzyme activity assay

For the determination of antioxidant enzyme activity, 1 mL of ice-cold 25 mM HEPES buffer (pH = 7.8) containing 0.2 mM EDTA, 2 mM ascorbic acid and 2% polyvinylpyrrolidone was used to extract total protein from 0.1 g of waterlily stigmas. The homogenate was centrifuged at 12,000 g for 20 min at 4°C, and then the supernatant was carefully collected for enzyme activity assays. The enzyme activities of superoxide dismutase (SOD), peroxidase (POD) and catalase (CAT) were measured. The measurement of enzyme activity follows the method described by Wang et al. (2011).

2.5 RNA extraction and sequencing

The samples used in this experiment were the same as the samples described in the hormone detection. The FastPure Universal Plant Total RNA Isolation Kit (Novazyme Biotechnology) was used to extract total RNA from plants, following the instructions of the kit. The integrity of the total RNA was determined by 1% agarose gel electrophoresis. The integrity and concentration of the RNA were accurately measured using the Agilent 2100 RNA Nano 6000 Assay Kit (Agilent Technologies). The detection indicators included RIN value, 28S/18S ratio, presence of baseline uplift in the profile and 5S peak. RNA-seq was performed by Suzhou Panomic Bio-medical Technology Co., Ltd (Suzhou, China) on the Illumina Novaseq6000 platform.

2.6 RNA sequencing data processing

After filtering adapter sequences and low-quality regions, the clean reads were aligned to the N. colorata genome using HISAT2 (Kim et al. 2015, Pertea et al. 2016). The gene expression levels for each sample were calculated using featureCounts (Liao et al. 2014) as FPKM (fragments per kilobase of exon per million mapped fragments). An expression matrix was generated using Trinity to assess differential gene expression among different samples (Grabherr et al. 2011). To identify enriched differentially expressed genes (DEGs) in terms of GO terms and metabolic pathways, functional enrichment analyses, including GO and KEGG, were performed relative to the whole transcriptome background using a bonferroni-corrected P-value threshold of <0.05. For GO functional enrichment and KEGG pathway analysis, the R package ClusterProfiler was utilized (Wu et al. 2021).

2.7 Co-expression network analysis

We performed co-expression network analysis of hormones and gene expression using the weighted gene co-expression network analysis (WGCNA) R package (Langfelder and Horvath 2008). To determine the relationship between gene expression and hormone levels, we analyzed the normalized gene expression data along with the levels of ABA, GA3, Me-JA, SAG, and BR. We visualized module correlations and co-expression networks using the LinkET R package and Cytoscape software (v3.9.1) (Smoot et al. 2011, Huang 2021). The obtained genes were further analyzed for correlation with genes involved in hormone synthesis, and annotation of the final genes was performed using hmmscan. Based on the results of qRT-PCR, we selected the most feasible genes for follow experiments (Johnson et al. 2010).

2.8 Quantitative real-time PCR

HiScript III RT SuperMix for qPCR (gDNA wiper) kit (Novazyme Biotechnology) was used to synthesize the first strand of cDNA. 1 μg total RNA was used in each 20-μL reaction. The cDNA products were diluted 10 times with deionized water before use. qRT-PCR experiments were conducted on a CFX96 Touch™ Real-Time PCR detection system (Bio-Rad) using the ChamQ SYBR qPCR Master Mix kit (Novazyme Biotechnology). Each reaction used 2 μL of diluted cDNA, and other reaction components and conditions were carried out according to the manufacturer's instructions. Specific primers were designed for qRT-PCR (Table S4). The ACTIN11 gene of waterlily was used as the internal reference gene (Luo et al. 2010), and three biological and three technical replicates were performed for each treatment. The relative expression levels were calculated using the 2^−ΔΔCt method (Livak and Schmittgen 2001, Xu et al. 2012).

2.9 Oligonucleotide design and treatment

Design positive (sense phosphorothioate oligodeoxynucleotides, S-ODN) and antisense (antisense phosphorothioate oligodeoxynucleotides, AS-ODN) oligonucleotides based on the cDNA sequence of NCED2 using the Sfold database (http://sfold.wadsworth.org/cgi-bin/soligo.pl) (Su et al. 2020, Zhao et al. 2020, Zhang et al. 2021, Li et al. 2022)(Table S5). Assess potential off-target effects using the BLAST program (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The 5′ and 3′ ends of S-ODN and AS-ODN were modified with thiol groups to maintain stability. S-ODN and AS-ODN were synthesized by Beijing Tsingke Biotech Co., Ltd.

Before ODN treatment, the stamens were removed from selected water lily flowers first. Then, the designed ODN (the secretions from water lily stigmas were diluted to a concentration of 20 μM) was added to the stigmas of water lily flowers. After 1 h of ODN treatment, the pollination procedure was performed.

2.10 Statistics

For the analysis of intergroup significant differences, a one-way analysis of variance (ANOVA) was used. When P < 0.05, Tukey's test was employed for post-hoc comparisons. Two-group comparisons were conducted using a two-tailed Student's t-test. All analyses were performed using GraphPad Prism 8 software.

3.1 Hormones play an important role in pollen and stigma recognition in water lily

To understand the regulation of plant hormones on pollination in water lily, we analyzed the hormone levels in the stigma (Table S1). The results revealed a significant accumulation of ABA, GA3, jasmonic acid-isoleucine (JA-lle) and BR in stigmas from self-pollination (SP) and cross-pollination with the same subgenus (CP). In contrast, the levels of these four hormones were lower in the unpollinated (Mock) stigma and stigma from inter-subgenus cross-pollination (ISCP) (Figure 1). Interestingly, the content of GA8 in the stigma exhibited an opposing trend to that of GA3. This could be attributed to the conversion of active gibberellins to inactive gibberellins in the stigma after Mock and ISCP pollination. Methyl jasmonate (MeJA) and glycosylated salicylic acid (SAG), which are closely associated with biotic stress, were only detected in ISCP-pollinated stigmas. This suggests that certain regulatory networks associated with disease resistance in water lily stigmas were activated after ISCP treatment (Figure 1). In addition, other hormones such as IAA, SA and JA appeared to have no direct correlation with the different pollination systems (Figures 1 and S1; Table S1).

3.2 ABA and BR treatments increase fruit set after hybridization between water lily subgenera

To further investigate the role of hormones in the pollination hybridization process of water lily, we applied ABA, GA and BR onto the ISCP-pollinated stigmas. Six hours after pollination, we collected the stigma and observed the pollen tubes' germination. The results showed a significant germination of pollen tubes on stigmas treated with SP and CP, while no pollen tube germination was observed on stigmas treated with ISCP (Figure 2A,B). When exogenous ABA and BR were applied, we observed that they promoted the germination of pollen tubes on ISCP-pollinated stigmas, while GA treatment did not (Figure 2A,B).

To further understand the relationship between ABA, BR, GA and the fruit set rate in water lily hybridization, we collected the hybrid fruits of water lily 25 days after pollination. The results revealed that water lily treated with SP and CP successfully produced fruits with a high fruit set rate, and the seeds developed normally. However, water lily treated with ISCP did not show fruit enlargement and experienced rotting. Interestingly, when ABA and BR were applied to water lily's ISCP-pollinated stigmas, it partially prevented the rotting of the ovary, significantly improving the fruit set rate and yielding a certain number of hybrid seeds. On the other hand, GA treatment, despite not promoting pollen tube germination on ISCP-pollinated stigmas, stimulated the enlargement of water lily ovaries, resulting in an increased fruit set rate. However, the majority of seeds in the fruits showed abortion (Figure 2C-E).

The germination experiment was conducted on the obtained seeds, it was found that the SP- and CP-derived seeds exhibited a higher germination rate. Additionally, the hybrid seeds obtained through ISCP with exogenously applied ABA and BR were also able to germinate successfully (Figure 2F). However, compared to the seeds obtained through SP and CP, the hybrid seeds obtained through ISCP required a longer germination time and exhibited a significantly slower plant growth rate (Figure 2F).

3.3 ABA and BR synergistically promote antioxidant enzyme activities in water lily stigmas

Reactive oxygen species (ROS) play a significant role in regulating the interaction between water lily pollen and the stigma as also previously demonstrated (Sun et al. 2019, Sun et al. 2023). SOD, POD and CAT are crucial antioxidant enzymes that directly regulate ROS accumulation (Kao et al. 2018). To gain further insights into how ABA and BR promote pollen germination on ISCP water lily, we assessed the activity of antioxidant enzymes in the stigma following pollination. Our results showed that higher SOD, POD and CAT activity were observed in the SP and CP stigmas than in the in the Mock and ISCP stigmas. Furthermore, exogenous application of ABA and BR on the ISCP stigma promoted the enzyme activity of SOD, POD and CAT (Figure 3C-E). As studies have shown that there is a synergistic regulation between ABA and BR (Li et al. 2021, An et al. 2023), we further measured their levels in the stigma. The results revealed that BR application on the ISCP stigma promoted the accumulation of endogenous ABA content, while ABA application on the ISCP stigma increased the endogenous BR content (Figure 3A,B). This suggests a potential synergistic regulation of ABA and BR in the process of hybridization in water lily.

3.4 Transcription profiles of water lily stigmas from different pollination combinations

In order to gain a deeper understanding of the specific mechanisms underlying the inter-subgeneric hybridization barriers in water lily, we utilized RNA-seq technology to investigate the gene transcription levels in Mock, SP, CP and ISCP stigmas. The experimental design consisted of four sample groups, with four independent biological replicates for each sample, resulting in a total of 16 RNA-seq libraries. On average, each library generated 45.81 million single-end reads, ranging from 41.76 million to 50.44 million. The average number of clean reads was 43.92 million, ranging from 40.06 million to 49.56 million. Among these reads, 97.81% had a base quality score of Q20, and 93.65% had a base quality score of Q30 (Table S2). Principal component analysis (PCA) based on FPKM values revealed that the first principal component (PC1) explained 40.8% of the variance in the dataset, while the second principal component (PC2) explained 25.3% of the variance. The Mock, SP, CP, and ISCP treatments formed distinct clusters, with the SP and CP groups exhibiting relatively closer clustering, and the clustering of all biological replicates was well-differentiated (Figure S2). These data indicate the reproducibility and reliability of the RNA-seq data.

Venn plot analysis revealed that in the ISCP_vs_Mock comparison, there were 1262 upregulated differentially expressed genes (DEGs) and 610 downregulated DEGs that were unique to this group (Figure 4A,B). These DEGs may play an important role in the inter-subgeneric hybridization barriers in water lily. To gain further insights into the functions of these 1872 DEGs, we conducted GO and KEGG analyses. The KEGG results showed that the 1262 upregulated genes were mainly involved in signal transduction, membrane transport, and cellular cytoskeleton (Figure 4C). Additionally, the GO analysis indicated that the upregulated genes were mainly enriched in kinase activity and signal transduction, with Ca²⁺ transport and Ca²⁺ signaling exhibiting significant enrichment (Figure S3A). On the other hand, the 610 downregulated genes were primarily enriched in secondary metabolites, such as phenylpropanoid biosynthesis and flavonoid biosynthesis (Figure 4D). The GO analysis revealed that the downregulated genes were mainly enriched in processes such as abscisic acid metabolism, carotenoid metabolism, and protein phosphorylation (Figure S3B). DEGs in SP_vs_Mock and CP_vs_Mock cross groups are considered to be keys in promoting pollen germination on the water lily's stigma. Through GO and KEGG enrichment analysis, we found that the 819 DEGs upregulated were mainly related to carbohydrate synthesis and metabolic process, hormone synthesis and the synthesis of some secondary metabolites related to antioxidants (e.g., flavonoids, carotenoids, etc.) (Figure S4A,B). The 749 down-regulated DEGs are mainly associated with signaling, cell wall formation, and biological stress (Figure S4C,D).

3.5 Co-expression network analysis and key gene identification

The above results indicated that the changes in hormone levels play a significant role in regulating the formation of inter-subgeneric hybridization barriers in water lily. To understand the regulatory mechanisms behind the variations in hormone levels, we performed weighted gene co-expression network analysis (WGCNA) using the changes in ABA, GA3, MeJA, SAG and BR contents as phenotype data. The genes were divided into 26 co-expression modules using the WGCNA package (Figure S5 A,B,C). Genes within the same module exhibited highly similar expression patterns and were considered to be tightly co-regulated. The correlation between the phenotype data and the modules was visualized using the LinkET R package (Figure 5A). The results showed that the magenta, turquoise, blue and yellow modules were positively correlated with MeJA and SAG contents, while the lightcyan module was negatively correlated with MeJA and SAG contents. Modules associated with ABA, GA3 and BR contents exhibited high similarity, such as the yellow, tan, black, brown and darkturquoise modules, which were positively correlated with these three hormones. On the other hand, the green, turquoise, blue and orange modules showed negative correlations with ABA and BR contents (Figures 5A and S5D).

It is interesting that two modules, blue and turquoise, showed correlations with all five hormones (Figure 5A). Therefore, we selected the blue and turquoise modules for further analysis. GO and KEGG enrichment analysis revealed that genes in the blue module were mainly enriched in cellular cytoskeleton movement, signal transduction, kinase activity, and phosphorylation (Figure S6A,C). On the other hand, genes in the turquoise module were primarily enriched in regulating pollen tube growth, organ development, cell growth and kinase activity (Figure S6B,D). We constructed a gene regulatory network using WGCNA and visualized the network using Cytoscape software (Figure 5B,C). We extracted the genes with the most connections, referred to as hub genes, and performed GO and KEGG analysis on them. The GO enrichment analysis showed that these hub genes were primarily enriched in regulating pollen tube growth, cell development, kinase activity, and phosphorylation (Figure 5D). KEGG enrichment analysis revealed enrichment in protein kinases, carbohydrate metabolism, and chromatin-related proteins (Figure 5E). In addition, we performed GO and KEGG enrichment analysis for DEGs in the module with significant positive correlation between MeJA and SAG content. The results showed that these DEGs were mainly related to flavonoid synthesis, pathogen defense and hormone signaling (Figure S7A,B). These clues suggested that when stigmata is exposed to incompatible pollen, stigmata misinterprets the pollen as a foreign invading microorganism, and then activates the expression of genes related to biological stress response, and rejects the incompatible pollen.

3.6 Identification of genes involved in hormone biosynthesis

To further understand the relationship between the obtained hub genes and the levels of these five hormones (ABA, GA3, BR, MeJA, and SAG), we analysed the expression of structural genes involved in their biosynthesis (Figure 6A) and performed a correlation analysis with the hub genes. The results revealed that key genes involved in ABA biosynthesis, named ZEP, NCEDs, ABA2.1, and AAO3, exhibited high expression levels in the stigmas of SP and CP, which was consistent with the observed changes in ABA content. KAO, GA3ox and GA20X1 were identified as potential important factors contributing to the accumulation of GA3 in the stigmas of SP and CP. Within the BR synthesis pathway, most structural genes, such as CYP90A1 and CYP85A1/2, were directly associated with the accumulation of BR in the stigmas of SP and CP. The specific expression of LOX and OPCL in ISCP stigma is the main reason for the specific presence of MeJA in ISCP-pollinated stigmas. In the SA synthesis pathway, both the branched-chain pathway and the phenylalanine pathway appeared to participate in regulating the synthesis of SAG in ISCP stigmas. Genes such as ICS in the branched-chain pathway and PAL1.4 in the phenylalanine pathway were identified as potentially important factors leading to the specific accumulation of SAG in the transmitting tract of the pistil.

The top 13 genes with the highest connectivity in the blue and turquoise modules were subjected to correlation analysis with the structural genes involved in the synthesis of the five hormones (Figure 6B). The results revealed that the expression of nine genes was highly correlated with the expression of the structural genes involved in the synthesis of the five hormones. Additionally, we observed that NCED2 had the highest number of connections among all the structural genes, suggesting that NCED2 may be a key node gene for the regulation of these nine candidate genes. Visualization of the expression of these nine genes showed that their expression was low in the stigmas of SP and CP, but high in the stigmas of Mock and ISCP (Figure 6C).

To further validate the response of candidate genes to different pollination in water lily, qRT-PCR was used to confirm the expression of the candidate genes. The results showed that the expression of most genes was consistent with the transcriptome results (Figure 7). However, there were still some differences between the expression levels of some genes and the transcriptome data, such as NCED1, CIPK6, KIN10, and GRF10. Interestingly, CIPK6 and KIN10 are specifically expressed in the stigma of ISCP, which seems to suggest that these two genes play an important role in promoting the formation of intergeneric hybridization disorder of waterlily.

3.7 Silencing of NCED2 on SP and CP stigmas inhibits pollen tube germination

The high connectivity of NCED2 among all the structural genes in the network caught our interest. We further investigated the expression of NCED2 in the stigmas of SP and CP. The antisense oligonucleotide (ASO)-mediated was used to transiently silence the expression of NCED2 (Figure 8C). When NCED2 was silenced, the abundant pollen tube germination on the stigmas of SP and CP disappeared (Figure 8A,B), accompanied by a decrease in ABA content (Figure 8D). However, when exogenous ABA was applied, pollen tube germination resumed on the stigmas (Figure 8A,B). These results indicated that NCED2 is an important rate-limiting enzyme in ABA synthesis in the water lily pistil and that NCED2 promotes pollen germination by increasing ABA accumulation on the stigmas. In addition, the BR content on the stigma of SP and CP water lily was also downregulated after NCED2 silencing. Interestingly, the BR content could be restored by exogenous ABA treatment (Figure 8E).