Multi‐layered RNA decay mechanisms underlie subgenome‐biased homoeologous mRNA stability in salt‐stressed allopolyploid plants

SUMMARY

Many crop species are polyploid plants, yet how post‐transcriptional regulation influences the RNA abundance of homoeologous genes from different subgenomes under stress conditions remains poorly understood. In this study, we systematically investigated transcriptional and post‐transcriptional regulation in the allopolyploid crop Brassica napus under salinity stress. Integrated transcriptome and degradome profiling on hypocotyl and cotyledon revealed tissue‐specific gene expression and mRNA stability associated with their developmental changes under salt stress. Notably, mRNAs undergoing exon junction complex (EJC)‐marked RNA decay (EMRD) and co‐translational RNA decay (CTRD) exhibited higher degradation propensity. Transcriptome‐wide profiling of mRNA abundance and mRNA decay in homoeologous genes revealed subgenome‐specific patterns in both gene expression and mRNA stability. Dual‐luciferase assays confirmed that the 3′ UTR mediates the divergent mRNA stability observed between homoeologous gene pairs. Further analysis revealed that EMRD, CTRD, and miRNA cleavage are associated with asymmetric RNA degradation across subgenomes. These findings provide comprehensive insights into the multi‐layered regulation of transcript stability under salt stress and reveal a critical role for RNA degradation in shaping tissue‐specific and subgenome mRNA asymmetry in polyploid crops.

Significance Statement

This study reveals that in allopolyploid crops under salt stress, a multi‐layered RNA degradation network, consisting of exon junction complex (EJC)‐marked decay, co‐translational decay, miRNA cleavage, and 3′ UTR‐mediated regulation, serves as a crucial post‐transcriptional mechanism governing subgenome‐biased mRNA stability.

INTRODUCTION

Soil salinization is an increasingly critical environmental factor that limits crop productivity worldwide (Song & Liu, 2017). It adversely affects seed germination, seedling establishment, and overall plant growth by disrupting ionic balance, inducing osmotic stress, and generating oxidative damage (Kumar et al., 2013; van Zelm et al., 2020; Zhou et al., 2024). Brassica napus (Rapeseed) is a globally important oilseed crop, and salinity stress has seriously affected rapeseed growth, photosynthesis, and nutrient uptake through osmotic stress and ionic imbalance, particularly during the early seedling stage, ultimately reducing its productivity (Gyawali et al., 2019). Improving salt tolerance in B. napus is therefore a major objective for breeders and plant scientists seeking to enhance crop resilience and ensure food security.

Plants have evolved complex regulatory networks to cope with salt stress, involving transcriptional reprogramming, epigenetic regulation, hormonal signaling, and post‐transcriptional mechanisms, such as mRNA degradation and miRNA‐mediated gene silencing (Barrero et al., 2006; Chen et al., 2025; Yan et al., 2023; Yin et al., 2019; Zhao, Yang, et al., 2023). In recent years, high‐throughput RNA sequencing (RNA‐seq) has revealed transcriptome‐wide salt‐responsive genes across multiple plant species (Liang et al., 2024; Luo et al., 2016; Wang et al., 2021; Wang et al., 2022; Yang et al., 2021). However, while transcriptional responses to salt stress have been well studied, the contribution of RNA turnover mechanisms, particularly mRNA degradation pathways to salt stress adaptation remains poorly understood in B. napus.

mRNA degradation is a crucial layer of gene expression regulation that determines transcript stability and turnover rates, especially under stress conditions (Kawa & Testerink, 2017; Yamada & Akimitsu, 2019). GMUCT‐seq (Genome‐wide Mapping of Uncapped and Cleaved Transcripts sequencing) has emerged as a powerful tool to quantify transcriptome‐wide degradation and to detect post‐transcriptional features such as 5′‐phosphate‐containing (5′P) decay intermediates, including miRNA‐directed cleavage sites, exon junction complex (EJC)‐marked RNA decay (EMRD) and co‐translational RNA decay (CTRD) (Ganguly & Gregory, 2026; Yu et al., 2016; Zhang et al., 2023). Up to now, miRNA‐mediated cleavage represents a major post‐transcriptional regulatory mechanism during stress responses, with cleavage efficiency modulated by environmental signals (Nayak et al., 2025). CTRD has been first discovered in yeast (Hu et al., 2009). XRN4‐mediated CTRD has been identified in multiple species, including Arabidopsis, tomato, soybean, rice, and maize (Carpentier et al., 2020; Carpentier et al., 2024; Guo et al., 2023; Merret et al., 2015). Notably, CTRD involves the degradation of transcripts while they remain associated with translating ribosomes and has been shown to dynamically regulate mRNA abundance in response to various stress conditions across plant species (Carpentier et al., 2020; Crisp et al., 2017; Merret et al., 2015). Many eukaryotic pre‐mRNAs undergo splicing, in which intervening sequences (introns) are excised and the exon junction complex (EJC) is deposited in a region 20–24 nucleotides upstream of the exon‐exon junction (Le Hir et al., 2000). During the pioneer round of translation, EJCs residing upstream of the termination codon are disassembled and removed through the ribosome‐associated protein Partner of Y14 and Mago (Gehring et al., 2009). If mRNAs are degraded by exonucleases without being fully translated, the EJCs will remain on the mRNA strands. In this case, EJC‐marked RNA decay (EMRD) in the RNA degradome can serve as signatures of mRNA degradation prior to steady‐state translation (Lee et al., 2020).

B. napus (AACC) is an allopolyploid plant generated by natural hybridization between two diploid progenitors, B. rapa (AA) and B. oleracea (CC) (Chalhoub et al., 2014). While a pan‐genome of B. napus has been assembled from thousands of accessions (Song et al., 2020a, 2020b), the fundamental question of how allopolyploidization drives substantial genomic alterations via the merger of divergent genomes remains unresolved. A common outcome of these changes is subgenome‐biased gene expression and alternative splicing, which have been subject to dynamic selection during B. napus domestication and adaptation to diverse environments (Chalhoub et al., 2014; Zhang et al., 2021). The asymmetry is likely attributable to differential epigenetic modifications and differences in the abundance of transposable elements between the subgenomes (Zhang et al., 2021). Since steady‐state RNA abundance is determined by the balance between transcription and decay rates (Timmers & Tora, 2018), RNA decay likely contributes to the observed asymmetries in steady‐state RNA levels. In diverse species, several mRNA features have been shown to influence RNA decay, including overall transcript length, 3′ UTR length, the presence of AU‐rich elements within the 3′ UTR, codon optimality, and RNA secondary structure (Chen & Shyu, 1995; Feng & Niu, 2007; Hanson & Coller, 2018; Mishima & Tomari, 2016; Zhang et al., 2024). Despite the diversity of associative measurements in plants, characterizing mRNA stability‐associated factors in B. napus remains poorly understood.

In this study, we integrated RNA‐seq and GMUCT‐seq analysis to investigate the transcriptomic and post‐transcriptional landscapes of B. napus seedlings under salt stress. By comparing mRNA expression and degradation patterns in hypocotyl and cotyledon tissues, we identified salt‐responsive genes and examined how mRNA decay contributes to transcript regulation under salt stress. We further compared expression levels and RNA stability between AA and CC subgenomes, focusing on homoeologous gene pairs. We then validated the role of 3′ UTR sequences in mediating differential RNA decay through structural predictions and luciferase reporter assays. Moreover, our study reveals that different modes of RNA decay, including EMRD, CTRD, and miRNA cleavage–directed RNA decay, are associated with the asymmetric RNA degradation of homoeologous gene pairs among subgenomes. Our findings provide insights into the molecular mechanisms associated with salt stress responses in B. napus, highlighting the importance of post‐transcriptional regulation in subgenome asymmetry of mRNA abundance.

RESULTS

Overview of phenotypes and omics profiling design for rapeseed seedlings response to salt stress

To uncover subgenome‐ and tissue‐specific differences in gene expression and mRNA stability in the allopolyploid rapeseed (Brassica napus) upon salt stress, we profiled the transcriptome and degradome of seven‐day‐old seedlings subjected to salt stress (Figure 1a). To determine the optimum salt concentration for screening the salt‐responsive genes of B. napus, B. napus seeds were sown on MS medium containing NaCl at a range of concentrations from 0 to 200 mM to impose salt stress (Figure 1b). After 7 days culture, the phenotypic variations were determined. Compared with the 0 mM (Figure 1b), the moderate stress (80 and 120 mM) concentrations were well tolerated by the seedlings, while the seedlings were severely inhibited under 200 mM NaCl.

Figure 1.

Experimental workflow and phenotypic responses of B. napus seedlings under salt treatment.

(a) Schematic representation of RNA‐seq and GMUCT‐seq using seedlings subjected to mock or salt treatment for 7 days.

(b) Salt stress inhibited hypocotyl elongation but promoted cotyledon expansion.

(c) Represented phenotypes of rapeseeds under different concentrations of NaCl, and statistical analysis of hypocotyl length and cotyledon fresh weight with or without salt treatment (120 mM). Statistical significance was assessed by the Wilcoxon rank‐sum test.

The phenotypic analysis of seedlings under salt treatment indicated that B. napus seedlings may be capable of tolerating moderate salt stress. Therefore, we selected the 120 mM NaCl treatment to further investigate the mechanism of salt tolerance at the seedling stage in B. napus. Under 120 mM NaCl stress, the hypocotyls were significantly shortened, while the cotyledons were heavier than those in the mock treatment (Figure 1b,c). These contrasting morphological changes indicate that B. napus seedlings may adopt a trade‐off strategy between hypocotyl shortening and cotyledon expansion as an adaptive mechanism under salt stress.

Identification of tissue‐specific salt‐responsive genes associated with their developmental changes in rapeseeds

We carried out RNA‐seq using cotyledons (cot) and hypocotyls (hyp) from seedlings cultured on MS medium without (Control) or with 120 mM NaCl (Treatment) after 7 days culture. A total of 12 RNA‐seq libraries were generated from four groups, namely C‐cot, T‐cot, C‐hyp, and T‐hyp. “C” represents tissues harvested from seedlings without salt treatment, whereas “T” represents tissues harvested from seedlings subjected to salt treatment. Totally, we generated 19–24 million 150 bp paired‐end reads. The raw reads were processed to remove adapters and then mapped to the B. napus genome reference. Overall, 83%–90% of clean reads were uniquely mapped to the B. napus genome reference (Table S1). Principal component analysis (PCA) displayed that the 3 biological replicates for each sample clustered together (Figure S1A), with the hypocotyl and cotyledon tissues separated from each other. This result indicates that the biological reproducibility of the sequenced data was well presented.

In total, 12 949 significant DEGs were identified, including 8539 salt‐responsive genes (3765 upregulated; 4774 downregulated) in rapeseed hypocotyls (Table S2) and 6180 genes (2976 upregulated; 3204 downregulated) in rapeseed cotyledons (Figure 2a and Table S3). Among them, the expression level of HKT1, NHX2, and NCED3, which have been previously reported to regulate plant salt tolerance, was specifically induced by salt stress in hypocotyls, while the mRNA abundance of CBL10 and P5CSA was specifically increased by salt stress in cotyledons (Figure S2A,B) (Barragán et al., 2012; Barrero et al., 2006; Hua et al., 2024; Kubala et al., 2015; Yin et al., 2019). To validate these findings, quantitative PCR (qPCR) was performed to confirm the expression changes of HKT1, NHX2, CBL10, P5CSA, and NCED3 under salt stress (Figure 2b). Overall, the expression changes of these genes were consistent with known mechanisms of salt response.

Figure 2.

Identification of salt stress–responsive genes in B. napus seedlings after 7 days of 120 mM NaCl treatment.

(a) Volcano plot showing DEGs between mock and salt treatments in hypocotyls (left) and cotyledons (right).

(b) Relative fold changes of representative salt‐responsive genes. Student's t‐test, *P < 0.05; **P < 0.01. C‐cot represents cotyledon without salt treatment; T‐cot represents cotyledon with salt treatment; C‐hyp represents hypocotyl without salt treatment; and T‐hyp represents hypocotyl with salt treatment.

(c) Venn diagram showing commonly upregulated and downregulated genes in hypocotyls (Hyp) and cotyledons (Cot).

(d) Heatmap of salt‐responsive genes in 3 replicates of hypocotyls (hyp) and cotyledons (cot) under control (C) and salt treatment (T) condition.

The Gene Ontology (GO) enrichment analysis of those DEGs indicated their potential roles in responses to multiple biological process, as well as many stress conditions. This analysis revealed that DEGs in hypocotyls were significantly enriched in pathways related to the electron transport chain, photosynthetic electron transport chain, response to red light, response to blue light and cellular response to light stimulus (Figure S2C). Notably, several DEGs (e.g., LHB1B1, LHCB3, LHB1B2, LHCB2.4) involved in light‐responsive and photosynthetic electron transport processes were differentially expressed, implying that salt stress may interfere with the photomorphogenic development of seedlings (Figure 2c). Compared with hypocotyls, GO enrichment analysis of DEGs in cotyledons revealed significant enrichment in biological processes such as organic acid catabolic process, carboxylic acid catabolic process, cell wall modification, detoxification, and cellular modified amino acid metabolism (Figure S2D). These findings indicate that catabolic and metabolic processes are more active in cotyledons under salt stress. We further compared the DEGs identified in hypocotyls with those in cotyledons and found that 951 genes were commonly upregulated, while 716 were commonly downregulated. GO enrichment analysis of the commonly upregulated genes revealed significant enrichment in biological processes such as cell wall modification, organic acid catabolic process, carboxylic acid catabolic process, and regulation of stomatal movement (Figure S3A). In contrast, the commonly downregulated genes were significantly enriched in functions associated with the responses to red light, hydrogen peroxide, and cytokinin (Figure S3B). Additionally, 58 DEGs were downregulated in the hypocotyl but upregulated in the cotyledon, whereas 45 DEGs showed the opposite pattern: upregulated in the hypocotyl but downregulated in the cotyledon. Notably, many of these genes were associated with photosynthesis, including BCA1, LHCA6, LHCB1, GAPA1, and PHYB (Figure S3C). These opposing expression patterns suggest tissue‐specific regulatory mechanisms in response to salt stress.

To further identify the genes directly involved in salt resistance, we extracted the DEGs associated with the GO term GO:0042538 (hyperosmotic salinity response). In total, 30 DEGs were identified in the hypocotyl, and 26 DEGs were found in the cotyledon (Figure 2d and Table S4). Taken together, these findings allowed us to identify a set of candidate genes potentially contributing to B. napus resistance to salt stress.

GMUCT‐seq reveals post‐transcriptional regulation in salt signaling pathways of rapeseeds

GO enrichment analysis revealed that the DEGs were also enriched in multiple RNA processes and RNA degradation, such as the production of small RNA involved in gene silencing (Figure 3a). We further analyzed the mRNA abundance of several small RNA processing genes, including BnC05.DCL2, BnC01.HEN1, BnC04.SE, BnA05.SMD1A, BnC01.RDR1, and BnC03.AGO7; these genes have been reported to serve vital roles in small RNA biosynthesis (Li et al., 2024). The mRNA abundances of these genes were significantly affected by salt stress (Figure 3b), suggesting that RNA metabolism, especially for sRNA‐mediated RNA degradation, played a vital role in plant response to salt stress.

Figure 3.

Effects of salt stress on mRNA stability.

(a) GO enrichment analysis showing that salt stress‐responsive genes are enriched in multiple RNA metabolic processes.

(b) Relative fold changes in mRNA abundance of salt‐responsive genes associated with RNA processes. Statistical significance was assessed by the student's t‐test.

(c) Volcano plot of differentially degraded genes between control and salt treatments.

(d) Salt stress induced mRNA degradation in both hypocotyl and cotyledon tissues. Statistical significance was tested by the Wilcoxon test.

(e) Venn diagram analysis of the overlap between genes with significant differential mRNA degradation and those with significant differential mRNA expression across groups. mRNA degradation contributes to maintaining the dynamic changes in mRNA abundance under salt stress.

(b–d) C‐cot represents cotyledon without salt treatment; T‐cot represents cotyledon with salt treatment; C‐hyp represents hypocotyl without salt treatment; and T‐hyp represents hypocotyl with salt treatment.

To comprehensively investigate how salt affects RNA degradation, we conducted GMUCT‐seq (degradome‐seq) on the same samples, which were also used for RNA‐seq (Figure 1a). Totally, the GMUCT‐seq produced 4–24 million 150 bp reads per sample. The raw reads were processed to remove adapters and subsequently mapped to the Brassica napus reference genome (Darmor‐bzh V10). Overall, 75% to 89% of the clean reads were uniquely mapped to the reference genome (Table S1). The PCA analysis showed that the 3 biological replicates for each sample clustered closely together (Figure S1B), whereas the hypocotyl and cotyledon tissues were clearly distinct from each other. These findings indicate a high degree of biological reproducibility in the degradome data. To assess the degradation level of each gene, we calculated the proportion uncapped (PU), defined as the ratio of normalized GMUCT‐seq reads to normalized RNA‐seq reads (Figure 1a) (Yu et al., 2021; Zhang et al., 2023). To evaluate differences in PU between the control and treatment groups, we performed a differential PU analysis, with parameter |log2FoldChange| ≥ 0.58 and P value <0.05. The results showed that 2221 genes (1473 upregulated, 748 downregulated) exhibited significantly altered mRNA stability under salt stress in hypocotyl. Similarly, 3896 genes (3413 upregulated, 483 downregulated) were significantly differentially degraded after salt treatment in cotyledon (Figure 3c). Overall, PU values were higher in the treatment group than in the control group, suggesting that salt stress tends to induce mRNA degradation in both hypocotyl and cotyledon tissues (Figure 3d).

Next, we performed an intersection analysis between genes showing significant differential mRNA degradation and those with significant differential mRNA expression. In the hypocotyl, 112 genes showed upregulated mRNA expression under salt stress, accompanied by decreased mRNA degradation, while 195 genes exhibited downregulated expression along with increased degradation. In the cotyledon, 61 genes were upregulated in expression with reduced mRNA instability, and 454 genes showed downregulated expression coupled with enhanced degradation (higher PU). For instance, in the hypocotyl, the mRNA levels of BnC02.NAC5 and BnA03.GT‐3B were upregulated, while their degradation levels were downregulated. In contrast, BnA05.HY5 and BnA01.SAUR50 exhibited the opposite pattern, with downregulated expression and increased mRNA degradation (Figure S4). In the cotyledon, BnA05.ERF3 and BnA01.ERF6 showed elevated mRNA expression accompanied by reduced degradation levels. Conversely, BnA01.HAT2 and BnA05.REV demonstrated decreased expression and increased degradation under salt stress (Figure S4). These results suggest that mRNA degradation and transcription may act in a coordinated manner to regulate and maintain appropriate mRNA levels in plant tissues under salt stress.

Additionally, we identified 3160 upregulated and 4450 downregulated genes in the hypocotyl, and 2836 upregulated and 2740 downregulated genes in the cotyledon that did not overlap with any differentially degraded mRNAs (Figure 3e). This phenomenon may reflect transcriptional adjustments that operate independently of mRNA degradation, representing an autonomous transcriptional response to salt stress in B. napus.

Differential gene expression of homoeologous gene pairs from two subgenomes in Brassica napus

We compared the expression levels of genes (Reads per million >1) between the AA and CC subgenomes in both the hypocotyl and cotyledon. We found that gene expression in the AA subgenome was significantly higher than in the CC subgenome (Figure 4a). The asymmetrical transcription of the AA and CC subgenomes was primarily due to the differential expression of subgenome‐unique genes (Figure 4b). Previous studies have shown that orthologous gene pairs in the respective progenitor genomes (B. rapa and B. oleracea) were retained as homoeologous gene pairs in B. napus (Chalhoub et al., 2014). Totally, we found 31 766 homoeologous gene pairs between AA and CC subgenomes (Table S5). We further investigated the relationship between the expression levels of homoeologous gene pairs in the two subgenomes. Overall, there were no significant differences in the expression levels of homoeologous gene pairs between the AA and CC subgenomes in either the hypocotyl or cotyledon, both before and after salt treatment (Figure 4c).

Figure 4.

Expression bias of homoeologous gene pairs in B. napus seedlings.

(a–c) Box plot showing mRNA abundance for all genes from the AA or CC subgenomes (a), subgenome‐unique genes (b), and homoeologous gene pairs (c) across four groups of rapeseed samples. Statistical significance was tested by Wilcoxon test.

(d–g) Volcano plots of differentially expressed genes between homoeologous gene pairs in each of the four comparisons.

(h, i) Numbers of overlapping gene pairs exhibiting CC‐biased expression (h) and AA‐biased expression (i) across the four comparisons.

(j) Bar plot showing subgenome‐specific salt responses of homoeologous gene pairs. C‐cot represents cotyledon without salt treatment; T‐cot represents cotyledon with salt treatment; C‐hyp represents hypocotyl without salt treatment; and T‐hyp represents hypocotyl with salt treatment.

To explore whether some homoeologous gene pairs exhibit significant differences in expression levels, we performed the difference expression analysis on homoeologous gene pairs, with parameter |log2FoldChange| ≥ 0.58 and P value <0.05. We found that 3810 gene pairs exhibited AA‐biased expression in hypocotyl tissue under control condition, while 4696 gene pairs showed CC‐biased expression. Similarly, in hypocotyl tissue under salt treatment, 4126 gene pairs were AA‐biased and 5303 were CC‐biased (Figure 4d,e). In cotyledon tissue under control condition, 4016 AA‐biased and 5151 CC‐biased gene pairs were observed (Figure 4f). Whereas under salt treatment, 4029 AA‐biased and 5058 CC‐biased gene pairs were identified (Figure 4g). Overall, more than 70% of homoeologous gene pairs showed no significant difference in subgenome‐biased expression. A total of 2333 CC‐biased and 1907 AA‐biased expression gene pairs were commonly identified across the four groups, respectively (Figure 4h,i). We also found that salt stress perturbs the expression of homoeologous gene pairs. For instance, in the hypocotyl, no differential expression was detected between the homoeologous pair A04p03920.1_BnaDAR and C04p36280.1_BnaDAR under control condition, whereas A04p03920.1_BnaDAR exhibited significantly higher mRNA abundance than C04p36280.1_BnaDAR upon salt stress treatment. By contrast, an opposite expression pattern was observed for the homoeologous pair A03p35230.1_BnaDAR and C03p42830.1_BnaDAR in cotyledons (Figure 4j). Notably, the number of common gene pairs was larger than that of other intersections, suggesting that these biased expression gene pairs may be consistently regulated across different tissues and environmental conditions. For instance, A07p14690.1_BnaDAR (corresponding to BnaA07G0124900ZS in the ZS11.v0 genome reference) is homoeologous to C07p21580.1_BnaDAR (corresponding to BnaC07G0183400ZS). Previous studies reported that BnaA07G0124900ZS exhibits higher levels of active epigenetic marks (H3K27ac and H3K4me3) and TPM, whereas BnaC07G0183400ZS shows elevated levels of the repressive mark H3K27me3 and lower TPM (Figure S5A–D). Consistent with these findings, our study revealed that A07p14690.1_BnaDAR was expressed at significantly higher levels than C07p21580.1_BnaDAR (Figure S5E,F). These results suggest that homoeologous genes displaying subgenome‐biased expression are a conserved mechanism that is tightly regulated by epigenetic modifications.

Subgenome‐biased mRNA stability of homoeologous gene pairs and the regulatory role of 3′ UTRs in rapeseed

The subgenomes of rapeseed exhibit distinct preferences for RNA abundance. We then sought to determine whether RNA degradation also displays subgenome preferences. To reveal the effect of subgenome on RNA decay efficiency, we compared the PU values between the AA and CC subgenomes in both hypocotyl and cotyledon tissues. We found that PU values in the CC subgenome were significantly higher than those in the AA subgenome across both tissues, indicating that mRNAs derived from the CC subgenome are more prone to degradation (Figure 5a). To further examine the subgenome‐biased RNA degradation efficiency, we utilized a publicly available dataset from rapeseed (Jian et al., 2018). We also found that the CC subgenome exhibited higher degradation rates than the AA subgenome (Figure S6A). Similarly to gene expression, we found that the subgenome‐unique genes were responsible for the CC‐biased degradation (Figure 5b,c), while the PU values of homoeologous gene pairs showed no significant difference across tissues.

Figure 5.

RNA decay bias of homoeologous gene pairs in B. napus seedlings.

(a) Box plot showing Proportion uncapped (PU) indicating mRNA instability for all genes from the AA or CC subgenomes (a), subgenome‐unique genes (b), and homoeologous gene pairs (c) across four groups of rapeseed samples. Statistical significance was evaluated using the Wilcoxon test.

(d–g) Volcano plots showing homoeologous gene pairs with significant differential PU in each of the four comparisons.

(h) Heatmap showing two representative homoeologous gene pairs with differential PU.

(i) Schematic representation of vector construction with 3′ UTRs of homoeologous gene pairs fused downstream of the luciferase gene. The green rectangle represents 3′ UTR from the AA subgenome homoeolog, while purple rectangle represents 3′ UTR from the CC subgenome homoeolog.

(j) The bar plot showing the relative foldchange of mRNA abundance of homoeologous gene pairs detected by qPCR.

(k) The bar plot showing the relative luciferase activity in homoeologous gene pairs.

(j, k) The Error bars represent ± SD (n = 3 biological replicates). Statistical significance was determined using a Student's t‐test (*P < 0.05, **P < 0.01). C‐cot represents cotyledon without salt treatment; T‐cot represents cotyledon with salt treatment; C‐hyp represents hypocotyl without salt treatment; and T‐hyp represents hypocotyl with salt treatment.

To explore subgenome‐wide homoeologous gene pairs that exhibit significant differences in RNA decay efficiency, we performed the differential analysis based on PU values, with parameter |log₂FoldChange| ≥ 0.58 and P value <0.05. In hypocotyl tissue under control condition (G1 group), 2657 gene pairs exhibited AA‐biased degradation, while 2740 gene pairs showed CC‐biased degradation; under salt treatment (G2 group), 2707 gene pairs were AA‐biased and 2854 were CC‐biased; in cotyledon tissue (G3 group), 1827 AA‐biased and 1834 CC‐biased gene pairs were observed, whereas under salt treatment (G4 group), 3123 AA‐biased and 3139 CC‐biased gene pairs were identified (Figure 5d–g). Venn diagram analysis revealed that 1257 gene pairs were commonly identified across all four groups (Figure S6B). For instance, in the G1 group, the PU levels of BnC07.PAA2 and BnA04.PAL1 were upregulated compared to their homoeologous gene pairs. In G2, BnC01.PPR and BnA02.TCP22 were upregulated. Similarly, BnC03.WRKY33 and BnA03.TRE1 were upregulated in G3, while BnC01.WRKY7 and BnA04.DOF2.1 were upregulated in G4 (Figure S6C).

We highlighted two homoeologous gene pairs from the shared group; we found that A09p13000.1_BnaDAR/C09p15400.1_BnaDAR showed CC‐biased degradation in all four groups, while the pair A003p33340.1_BnaDAR/C03p40580.1_BnaDAR displayed AA‐biased degradation (Figure 5h). Previous studies in other species have reported that mRNA structures in the 3′ UTR tend to be associated with mRNA stability (Wu et al., 2024). An 18 bp deletion was found in the 3′ UTR of A09p13000.1_BnaDAR compared to C09p15400.1_BnaDAR (Figure S7A). Using the RNAFold web server, we predicted the RNA structures in the 3′ UTR region. Notably, the predicted RNA secondary structures of their 3′ UTR showed substantial differences (Figure S7B), suggesting a potential structural basis for the observed variation in RNA decay efficiency. A similar result was also observed in the gene pair A03p33340.1_BnaDAR/C03p40580.1_BnaDAR, where several nucleotide differences were found in the 3′ UTR between the two genes (Figure S7C,D). Interestingly, comparative sequence alignments between allotetraploid B. napus and its diploid progenitors revealed an 18‐bp deletion in the 3′ UTR of A09p13000.1_BnaDAR arose post‐allopolyploidization, whereas nucleotide differences in the 3′ UTRs of A03p33340.1_BnaDAR and C03p40580.1_BnaDAR predate allopolyploidization (Figure S8).

By fusing the 3′ UTRs of this homoeologous gene pair to the coding region of the Firefly luciferase (LUC) reporter gene, along with an identical Renilla luciferase as an internal control. Using qPCR, we found that the LUC mRNA abundance associated with the 3′ UTR of A09p13000.1_BnaDAR, which exhibits a smaller PU value, was significantly higher mRNA abundance than that fusing with the 3' UTR of  C09p15400.1_BnaDAR (Figure 5i,j). In another homoeologous gene pair, the 3′ UTR of A03p33340.1_BnaDAR exhibited a higher PU value compared with that of C03p40580.1_BnaDAR (Figure 5h). Using qPCR, we found that the LUC mRNA abundance associated with the 3′ UTR of A03p33340.1_BnaDAR was significantly lower than that fusing with the 3' UTR of C03p40580.1_BnaDAR (Figure 5i,j). Furthermore, we quantified the relative Firefly luciferase activity of two homoeologous gene pairs. We found that the Firefly luciferase activity associated with the 3′ UTR of A09p13000.1_BnaDAR was significantly higher than that of C09p15400.1_BnaDAR (Figure 5k). In another gene pair, A03p33340.1_BnaDAR and C03p40580.1_BnaDAR, the Firefly luciferase activity linked to the 3′ UTR of A03p33340.1_BnaDAR was significantly lower than that of C03p40580.1_BnaDAR (Figure 5k). These results support the functional significance of 3′ UTRs in regulating RNA stability and indicate that 3′ UTR is an important factor contributing to the differences in RNA decay between subgenomic homoeologous genes, highlighting the role of RNA decay in shaping the imbalance of transcript abundance between the two subgenomes.

Incomplete translation mediated by EJC‐marked RNA decay (EMRD) drives subgenomic mRNA imbalance

Previous studies suggest that, across Arabidopsis and human HEK293 cells, 5′P ends in degradome displayed a clear enrichment in the canonical EJC region (25–30 nt upstream of the exon‐exon junction) (Ganguly & Gregory, 2026; Lee et al., 2020), suggesting that a subset of transcripts underwent incomplete translation. Thus, transcripts from genes with EJC‐marked RNA degradation (EMRD) were degraded prior to the first round of translation. However, whether genes undergo subgenome‐specific EMRD remains unclear. In this study, we profiled 5′P ends in four groups and compared their abundances in the canonical EJC regions. Meta‐analysis showed an accumulation of 5′P‐end reads at 26–28 nt upstream of exon 3′ ends, corresponding to the 5′ boundary of the EJCs deposited at the canonical position (Figure 6a), consistent with previous findings in Arabidopsis and human (Lee et al., 2020). In total, we identified 510, 1476, 574, and 1383 exons with EMRD events (enrichment score ≥3) in the C‐hyp, T‐hyp, C‐cot, and T‐cot groups, respectively, indicating that salt stress could induce incomplete translation of transcripts. Furthermore, these mRNAs exhibited higher PU values than the genome‐wide average, suggesting that they are generally less stable (Figure 6b). We next investigated whether EJC footprints display subgenome preferences. By comparing the 5′P ends between the AA and CC subgenomes in both hypocotyl and cotyledon tissues, we observed a greater number of 5′P ends within the canonical EJC region in the CC subgenome than in the AA subgenome across both tissues (Figure 6c). This indicates that mRNAs derived from the CC subgenome are more prone to undergoing incomplete translation. Previous studies have shown that EJC footprints promote nonsense‐mediated mRNA decay (NMD) (Lee et al., 2020). In Arabidopsis, the gene RS2Z33 (AT2G37340), which produces premature termination codon (PTC)‐containing transcripts through alternative splicing, displays strong 5′P peaks within the EJC regions of exons downstream of the PTC (Gloggnitzer et al., 2014). We identified a homoeologous gene pair (A03p20610.1_BnaDAR and C03p20610.1_BnaDAR) that is homologous to RS2Z33. Notably, C03p20610.1_BnaDAR consistently produced more 5′P ends than A03p20610.1_BnaDAR across all tissues (Figure 6d; Figure S9). Taken together, our observations reveal transcriptome‐wide EMRD responsive to salt stress and subgenome‐specific EMRD in rapeseeds.

Figure 6.

Widespread EJC‐marked RNA decay displaying subgenome bias in B. napus seedlings under salt stress.

(a) The distribution of 5′P end reads around the upstream 51 nt of 3′ end of exons. Average relative abundance of 5′P end reads detected by GMUCT within a 51‐nucleotide window upstream of exon 3′ ends. The y‐axis represents the mean relative frequency (%) across all exons, while the x‐axis denotes the nucleotide position within the −51 nt upstream window of the exon 3′ ends. A sharp enrichment peak was observed approximately 27 nt upstream of the 3′ end of exons, a pattern consistent across all four libraries.

(b) Genes undergoing EJC‐marked RNA decay exhibited higher PU than the whole transcriptome. The statistical significance was tested by the Wilcoxon test.

(c) EJC footprints display subgenome preferences. Specifically, across both tissues, a greater number of 5′P ends were detected within the canonical EJC region in the CC subgenome than in the AA subgenome.

(d) Subgenome‐biased accumulation of 5′P ends in the RS2Z33 homoeologous gene pair in hypocotyl. C03p20610.1_BnaDAR consistently produced more 5′P ends than A03p20610.1_BnaDAR under both control and salt stress conditions in hypocotyl. C‐cot represents cotyledon without salt treatment; T‐cot represents cotyledon with salt treatment; C‐hyp represents hypocotyl without salt treatment; and T‐hyp represents hypocotyl with salt treatment.

The impact of miRNA cleavage efficiency on asymmetric degradation of homeologs

GMUCT‐seq has been used to identify miRNA targets and quantify miRNA cleavage efficiency in plants (Yu et al., 2021; Zhang et al., 2023). In this study, we used the online tool psRNAtarget to predict the miRNA targets in B. napus. The conserved Brassica mature miRNAs were downloaded from miRBase (Kozomara et al., 2018), and potential cleavage sites were detected using GMUCT‐seq. As expected, we observed that 5′P read ends were significantly enriched in predicted miRNA cleavage sites compared to their flanking regions (Figure 7a). In total, we identified 208 miRNA target sites in the C‐hyp group; 224 in the T‐hyp group; 184 in the C‐cot group, and 217 in the T‐cot group (Figure 7b–e). These targets included known miRNA‐regulated genes such as SPL3, SPL10, MYB65, SPL11, SCL27, AGL16, SPL13. The abundance of cleavage reads at these target sites was significantly higher than in their flanking regions, supporting the accuracy of GMUCT‐based detection (Figure 7b–e). To assess how salt affects the miRNA cleavage efficiency, we defined the miRNA cleavage efficiency as the ratio of 5′P read ends at miRNA cleavage sites to the mean 5′P read end coverage in 100 nt flanking regions (Yu et al., 2016; Zhang et al., 2023). In the hypocotyl under salt stress, cleavage efficiency increased for 144 targets and decreased for 104 targets (Figure 7f). Similarly, in the cotyledon, 144 targets showed increased cleavage efficiency, 119 showed decreased efficiency, and 77 were unaffected by salt stress (Figure 7f). For instance, the cleavage efficiency of miR1885a‐BnaCO3.TNL, miR156‐BnaC04.SPL9, and miR396‐BnaA02.GRF7 was elevated under salt stress in the hypocotyl. In contrast, the efficiency of miR171‐BnaC09.SCL6, miR161‐BnaC04.PPR, and miR172‐BnaTOE1 was inhibited (Figure 7g). In the cotyledon, similar increases were observed for miR164‐BnaA05.NAC1‐L, miR396‐BnaA02.GRF7, miR161‐BnaA04.PPR, while decreases were found for miR319‐BnaC03.TCP4, miR161‐BnaA09.PPR, and miR159‐BnaC03.MYB101 (Figure 7h). These results indicated that salt stress broadly affects the miRNA cleavage efficiency in B. napus, potentially modulating the stability of transcripts.

Figure 7.

Identification of salt‐induced miRNA cleavage at target mRNA sites, revealing subgenome‐bias miRNA cleavage efficiency at homoeologous gene pairs from B. napus seedlings.

(a) 5′P read ends are significantly enriched at miRNA cleavage sites compared to their flanking regions. The statistical significance was tested by the Wilcoxon test.

(b–e) Scatterplots showing miRNA cleavage sites with significantly enriched 5′P read ends in each of the four comparisons. The statistical significance was tested by the Wilcoxon test.

(f) Effects of salt stress on miRNA cleavage efficiency in B. napus seedlings.

(g, h) Representative miRNA‐target genes showing altered cleavage efficiency under salt stress in hypocotyls (g) and cotyledons (h).

(i) The number of homoeologous gene pairs exhibiting subgenome‐biased miRNA cleavage efficiency.

(j) Subgenome‐biased miRNA cleavage efficiency and PU pattern in a representative homoeologous gene pair. Statistical significance was determined using a student's t‐test (*P < 0.05, **P < 0.01). C‐cot represents cotyledon without salt treatment; T‐cot represents cotyledon with salt treatment; C‐hyp represents hypocotyl without salt treatment; and T‐hyp represents hypocotyl with salt treatment.

To investigate the contribution of miRNA cleavage efficiency to subgenome‐biased RNA decay, we analyzed homoeologous gene pairs among miRNA targets. The targets were found to be nearly equally distributed between subgenome‐unique genes and homoeologous pairs (Figure S10). Notably, most homoeologous pairs exhibited subgenome‐biased miRNA cleavage efficiency, either AA‐biased or CC‐biased, suggesting its role in driving subgenome‐biased RNA decay (Figure 7i). For example, in the pair A04p01530.1_BnaDAR/C04p32030.1_BnaDAR, the C04p32030.1_BnaDAR showed higher miRNA cleavage efficiency in both control and treatment groups and a corresponding CC‐subgenome‐biased PU across groups (Figure 7j). Collectively, these results suggest that miRNA cleavage may act as a key factor associated with subgenome‐biased degradation patterns.

Co‐translational RNA decay associates with asymmetric degradation of Homeologous gene pairs

Co‐translational RNA decay, a process in which 5′‐3′ exonuclease‐degraded mRNA is still associated with translating ribosome, is widespread in plants (Guo et al., 2023; Yu et al., 2016; Zhang et al., 2023). CTRD contributed to rapid RNA turnover of genes in plant (Yu et al., 2016). To analyze transcriptome‐wide CTRD in B. napus, GMUCT‐seq reads were mapped to the B. napus cDNA reference. The 5′P read ends were then extracted for downstream analysis. Each codon within the coding sequence (CDS) was divided into 3 potential reading frames, labeled 0, 1, and 2, with frame 0 corresponding to the translation frame (Yu et al., 2016; Zhang et al., 2023). We observed that 5′P read ends accumulated at significantly higher levels in frames 1 and 2 compared to frame 0 (Figure 8a). This pattern indicates the presence of ribosome‐associated RNA degradation with a characteristic 3 nucleotide periodicity across the CDS regions in all GMUCT‐seq libraries. In our previous studies, we found that GMUCT‐seq data revealed prolonged ribosome stalling at translational termination sites compared to other regions within the CDS region (Yu et al., 2016; Zhang et al., 2023). The translational termination stalling index (TSI) was defined as the ratio of the number of 5′P read ends at the ribosome boundary (16–17 nt upstream from stop codon) to the median number of 5′P read ends within the flanking 100 nt (Guo et al., 2023; Zhang et al., 2023). In total, we identified 1092 CTRD genes using a TSI threshold of >3, including 667, 208, 467, and 388 genes in the C‐hyp, T‐hyp, C‐cot, and T‐cot libraries, respectively (Figure 8b; Figure S11A,B). Notably, only 84 CTRD genes were commonly identified across all four libraries (Figure 8c), indicating that CTRD events occur recurrently in a manner influenced by environmental conditions and tissue‐specific contexts. Gene Ontology enrichment analysis revealed that these 1092 CTRD genes are primarily involved in biological processes such as negative regulation of signal transduction, negative regulation of cell communication, negative regulation of signaling, as well as peptidyl‐serine modification and peptidyl‐serine phosphorylation (Figure S11C). Furthermore, under both control and salt stress conditions, CTRD mRNAs displayed higher PU values than the genome‐wide average, indicating that they are generally less stable (Figure 8d).

Figure 8.

Identification of mRNAs undergoing co‐translational RNA decay and the impact of CTRD on subgenome‐biased mRNA stability in B. napus seedlings.

(a) GMUCT revealed a 3‐nt periodic distribution of 5′P read ends in frames 0, 1, and 2 across all protein‐coding sequences in each of the four libraries. Statistical significance was determined using a chi square test (*P < 0.05, **P < 0.01).

(b) Heatmap showing mRNAs with enriched 5′P read ends at the 5′ boundary of ribosome stalling sites near translational termination in each of the four libraries.

(c) Venn diagram showing the overlap of CTRD genes among the four libraries.

(d) Genes undergoing CTRD exhibited higher PU than the whole transcriptome.

(e–h) Four representative homoeologous pairs exhibited high TSI values accompanied by significantly high PU values and decreased mRNA abundance. C‐cot represents cotyledon without salt treatment; T‐cot represents cotyledon with salt treatment; C‐hyp represents hypocotyl without salt treatment; and T‐hyp represents hypocotyl with salt treatment.

We next investigated the association between CTRD and subgenome‐biased RNA decay across tissues. Through intersection analysis between CTRD genes and genes exhibiting subgenome‐biased degradation in the four groups, we identified 156, 166, 130, and 176 overlapping genes, respectively (Figure S11D). The association analysis showed that CTRD genes were significantly enriched among genes showing subgenome‐biased RNA decay, including both AA‐ and CC‐biased decay, while being markedly underrepresented among subgenome‐unique genes, suggesting that CTRD may be associated with subgenome‐biased degradation (Figure S12). Furthermore, several homoeologous pairs exhibited highly divergent TSI values alongside significantly different PU values and total mRNA abundance. For instance, compared with C03p16220.1_BnaDAR, A03p13790.1_BnaDAR exhibited higher TSI and PU values, which correlated with reduced mRNA levels (Figure 8e). By contrast, some homoeologous pairs showed opposite patterns (Figure 8f–h). Demonstrating that CRTD contributes to RNA instability and consequently impacts mRNA abundance. In summary, these results reveal that CTRD signatures are crucial regulators of mRNA stability under both normal and stress conditions, shaping subgenome‐divergent RNA decay patterns through the modulation of RNA decay dynamics.

DISCUSSION

Salt stress is a major constraint on crop productivity, and understanding how plants coordinate transcriptional and post‐transcriptional regulation is critical for improving stress tolerance. Previous studies have primarily focused on transcriptional reprogramming under salt stress, while the contribution of mRNA decay and RNA stability remains less explored, particularly in polyploid crops such as B. napus. In this study, we integrate transcriptome and degradome analyses to dissect both transcriptional and post‐transcriptional regulatory responses to salt stress in B. napus seedlings. Our findings reveal that salt stress not only reprograms gene expression in a tissue‐specific manner, but also extensively impacts mRNA degradation, miRNA activity, and subgenome‐biased RNA decay.

In our study, RNA‐seq analysis identified 12 949 significantly DEGs, including 8539 genes in the C‐hyp vs T‐hyp comparison and 6180 in the C‐cot vs T‐cot comparison (Figure 2a). The enrichment of light signaling pathways in hypocotyls is consistent with previous reports showing that photoreceptors such as PHYB and PIFs modulate elongation growth under salt stress. However, our results further suggest that this regulation is tightly coupled with tissue‐specific metabolic reprogramming, as cotyledons preferentially activate catabolic and detoxification pathways. This division of labor between tissues highlights a coordinated strategy to balance growth and survival, extending previous findings that largely focused on single tissues or pathways (Qi et al., 2025; Zhao, Bo, et al., 2023). Those identified DEGs could provide valuable insights and potential targets for rapeseed breeding programs.

Using GMUCT‐seq, we quantified mRNA degradation dynamics and observed a general increase in degradation rates under salt stress (Figure 3d), as evidenced by higher PU values. This is consistent with previous observations that stress conditions often accelerate RNA turnover as a means to rapidly reprogram gene expression (Smith et al., 2024; Yu et al., 2021; Zhang et al., 2023). Integration with transcriptome data revealed a subset of genes whose expression was inversely correlated with degradation, suggesting transcription and mRNA stability act in concert to fine‐tune gene expression (Figure 3e; Figure S4). However, many DEGs did not overlap with differentially degraded mRNAs (Figure 3e), implying independent transcriptional responses. This partial uncoupling highlights the complexity of gene regulation under stress conditions, where distinct regulatory layers may be selectively engaged depending on gene function or cellular context. Collectively, these findings underscore the critical role of RNA turnover as an additional regulatory layer shaping transcriptome dynamics under salt stress.

Consistent with previous findings, B. napus exhibits subgenome‐biased expression, with the AA subgenome generally showing higher mRNA abundance (Figure 4a) (Chen et al., 2025; Zhang et al., 2021). Based on previous studies, we concluded that this asymmetry is likely attributable to differential epigenetic modifications and differences in the abundance of transposable elements between the subgenomes (Chen et al., 2025; Zhang et al., 2021). Specifically, active histone marks (H3K4me3, H3K27ac, and H3K4me1) and RNAPII occupancy are more enriched on AA subgenome genes compared to those in the CC subgenome. In contrast, the AA subgenome showed lower average levels of DNA methylation and repressive histone marks (H3K27me3 and H3K9me2) than the CC subgenome.

Although several studies have focused on subgenome‐biased expression, no attention has been paid to subgenome‐biased RNA decay. In this study, we found that the decay efficiency in the CC subgenome were significantly higher than those in the AA subgenome (Figure 5a). Combined with the observation that AA subgenome generally showing higher mRNA abundance, these findings suggest that not only transcriptional regulation but also RNAs degradation correlates with subgenome‐biased expression in B. napus. Genetic variations in promoters regulate gene transcription, thereby modulating agronomic traits (Lin et al., 2012; Zhu et al., 2023). More recently, natural variation in regulatory elements within 5′ UTRs or 3′ UTRs has been shown to significantly affect mRNA abundance through post‐transcriptional regulation (Srivastava et al., 2018; Wang et al., 2024; Zhang et al., 2024). Our study demonstrates that differences in 3′ UTR sequences between homoeologous genes affect their mRNA abundance at post‐transcriptional level, providing valuable cues for selecting which homoeologous gene to target for effective gene editing at their UTR region.

Subgenome expression asymmetry in allopolyploids can arise from both inherited regulatory differences from diploid progenitors and evolutionary changes following polyploidization. Previous studies have shown that part of the expression bias reflects parental legacy, while additional divergence may occur due to epigenetic reprogramming and regulatory interactions between subgenomes after genome merger (Li et al., 2021; Wei et al., 2021; Wu et al., 2018). Although our study did not include the diploid progenitors of B. napus for direct transcriptome comparison, the observed subgenome‐biased RNA degradation patterns may partially reflect parental legacy. We found that an 18‐bp deletion in the 3′ UTR of A09p13000.1_BnaDAR resulted in asymmetry mRNA stability emerging after allopolyploidization, while nucleotide variation between A03p33340.1_BnaDAR and C03p40580.1_BnaDAR produced unbalanced mRNA degradation that originated prior to allopolyploidization (Figures S7, S8; Figure 5), This finding indicates that some homoeologous gene pairs in B. rapa (AA) and B. oleracea (CC) already exhibit unbalanced mRNA degradation, whereas this imbalance may have arisen in others only after the formation of the allotetraploid. Future studies integrating diploid progenitor data will be necessary to disentangle their relative contributions.

In this study, functional assays indicate that 3′ UTR sequences play a critical role in determining mRNA stability. Differences in UTR length and secondary structure between homoeologous genes may influence RNA‐protein interactions or accessibility to decay machinery. It is possible that these structural features modulate the recruitment of RNA‐binding proteins or miRNA complexes, thereby leading to differential degradation. Future studies combining structural probing and protein interaction assays will be valuable to dissect the underlying mechanisms.

Our findings highlight a pervasive layer of post‐transcriptional regulation mediated by incomplete translation and subsequent EJC‐marked RNA decay. The sharp enrichment of 5′P ends at canonical EJC sites indicates that a substantial subset of transcripts fails to complete the pioneer round, thereby entering degradation pathways before productive translation. Notably, CC‐derived transcripts showed consistently higher levels of EJC footprints than those from the AA subgenome, suggesting intrinsic differences in translational efficiency or decay susceptibility (Figure 6). Such subgenome biased incomplete translation may shape overall transcriptome balance and could represent an underappreciated mechanism contributing to subgenome dominance.

GMUCT‐seq has been used to identify miRNA targets and estimate miRNA cleavage efficiency in plants (Yu et al., 2021; Zhang et al., 2023). In this study, we found that salt stress reprogrammed miRNA activity. In the hypocotyl, salt stress increased the cleavage efficiency of 144 targets and decreased it for 104 targets; in the cotyledon, 144 targets showed increased cleavage efficiency, and 119 showed decreased efficiency (Figure 7f). GMUCT‐based identification of miRNA cleavage sites revealed that cleavage efficiency was differentially affected among targets. Increased cleavage of known salt‐responsive targets, such as miR396‐GRF7, suggests that salt stress alters miRNA‐mediated gene silencing (Figure 7g,h). Conversely, reduced cleavage of other targets may stabilize transcripts critical for stress adaptation. These results highlight a dynamic regulation of miRNA–target interactions during salt stress. Importantly, many homoeologous gene pairs exhibited asymmetric cleavage efficiency. This indicates that miRNA‐mediated decay substantially contributes to subgenome biased RNA turnover. This differential targeting also resulted in unequal mRNA abundance highlighting miRNA cleavage as a key regulatory layer.

Our results further confirm that CTRD significantly affects mRNA stability; they all exhibited higher PU values compared to the genome‐wide average across groups (Figure 8d). Moreover, the divergent TSI and PU values observed between homoeologous gene pairs further reveal that CTRD associates with asymmetric mRNA decay. In summary, CTRD provides a new perspective for investigating post‐transcriptional regulation in polyploid plants.

In summary, our study reveals a multi‐layered regulatory framework in which transcriptional control, mRNA decay, miRNA activity, and 3′ UTR‐mediated mechanisms collectively shape the response of B. napus to salt stress. Importantly, we demonstrate that subgenome asymmetry is influenced not only by epigenetic regulation but also by differential RNA stability, providing new insights into the regulatory complexity of polyploid genomes. From an applied perspective, the identification of key regulators of RNA stability and miRNA–target interactions offers potential targets for improving stress tolerance in rapeseed. Future efforts integrating genome editing and synthetic biology approaches could enable precise manipulation of RNA decay pathways to enhance crop resilience under adverse environmental conditions.

METHODS AND MATERIALS

Plant materials

The wild‐type Brassica napus variety ‘K407’, a semi‐winter ecotype inbred line, was grown on solid Murashige and Skoog (MS) medium supplemented with 1% sucrose and 0–200 mM NaCl under a 16‐h light/8‐h dark photoperiod at 22°C. The light intensity was maintained at 250 μmol m⁻² s⁻¹ using warm white fluorescent tubes. After 1 week of growth, hypocotyls and cotyledons were harvested from the seedlings, immediately frozen in liquid nitrogen, and stored at −80 C until total RNA extraction for RNA‐seq and GMUCT library construction.

Library construction

For RNA‐seq analysis, three independent biological replicates were collected per sample. Total RNA was extracted using the RNA Isolation Kit (Bioer, Cat. BSC65S1, China) according to the manufacturer's instructions. RNA quality and quantity were assessed using a NanoDrop 2000 spectrophotometer and agarose gel electrophoresis. RNA‐seq libraries were prepared using the RNA Library Preparation Kit (NEB, Cat. #E7775S and #E7770S, USA) following the manufacturer's protocol. The size and quality of the libraries were evaluated using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). Final libraries were constructed with an Illumina kit following the manufacturer's instructions and sequenced using 150 bp paired‐end reads on an Illumina platform.

For GMUCT‐seq, libraries were generated from total RNA extracted from three biological replicates per sample, based on previously published methods with minor modifications (Yu et al., 2016). Briefly, total RNA was extracted using the RNA Extraction Kit (Bioer, Cat. BSC65S1, China), and poly(A)⁺ mRNA was enriched using mRNA capture beads (Vazyme, Cat. N401‐01, China). Degraded 5′‐phosphate‐containing mRNAs (5′P mRNAs) were ligated with 5′ RNA adaptors, and excess adaptors were removed using RNA Clean Beads (Vazyme, China, Cat. No. N412‐01). Reverse transcription was performed using primers containing 3′ DNA adaptors fused to random hexamers. Residual RT primers were removed using DNA Clean Beads (Vazyme, Cat. N411). GMUCT libraries were then amplified with indexed primers using a small RNA Illumina kit. The size and quality of the libraries were evaluated using an Agilent 2100 Bioanalyzer.

Bioinformatic analysis

The adaptor sequences from both RNA‐seq and GMUCT‐seq data were trimmed prior to alignment, and then, the RNA‐seq clean data were aligned to the B. napus the reference genome (Darmor‐bzh V10 version: https://yanglab.hzau.edu.cn/BnIR/download?module=genomics) using STAR tool (version: 2.7.1a), with the default parameter (github.com/alexdobin/STAR). The GMUCT‐seq clean reads was aligned to the B. napus cDNA reference (Darmor‐bzh V10) using STAR tool (version: 2.7.1a). The Gene‐level read counts were calculated using HTSeq 2.0 (version: 0.13.5) (Putri et al., 2022), with key parameter: ‐q ‐t exon ‐f sam ‐m intersection‐nonempty ‐‐stranded = no, and RPM (Reads of exon model per Million mapped reads) values were employed to quantity the expression level. Differentially expressed genes (DEGs) were analyzed using the DESeq2 R package (version: 1.42.1). DEGs were determined based on thresholds of |log2FoldChange| ≥ 0.58 and adjusted P value <0.05 compared to the control group. The PU metric was calculated by RPM_GMUCT‐seq/RPM_RNA‐seq for each replicate, the average PU ratio across 3 replicates was used to define the proportion uncapped of each gene.

Identification of genes undergoing Co‐translational RNA decay

The features of CTRD were analyzed as previously described with minor modifications (Guo et al., 2023; Zhang et al., 2023). Briefly, adapter‐trimmed reads were mapped to the reference cDNA using STAR (version: 2.7.1a) with default parameters (github.com/alexdobin/STAR). To identify co‐translational decayed mRNAs exhibiting ribosome stalling at stop codon, the translational termination stalling index (TSI) was defined as the ratio of the number of 5′P read ends at the ribosome boundary (16–17 nt upstream from stop codon) to the median number of 5′P read ends within the flanking 100 nt (Guo et al., 2023; Zhang et al., 2023). Genes with a TSI value greater than 3 and more than 2 reads in 16–17 site were defined as genes undergoing CTRD.

Identification of miRNA targets and measurement of miRNA cleavage efficiency

The online tool psRNAtarget was used to predict the miRNA targets in B. napus cDNA reference (https://www.zhaolab.org/psRNATarget/), with default parameters and a maximum expectation value of 4.0. The conserved Brassica mature miRNAs were downloaded from miRbase (https://www.mirbase.org/). Putative miRNA targets were defined as genes with 5′P read ends from GMUCT data located within predicted miRNA cleavage sites. Additionally, miRNA cleavage efficiency was calculated as the ratio of 5′P read ends at the miRNA cleavage site to the mean number of 5′P read ends within a 100‐nucleotide flanking region (Yu et al., 2016; Zhang et al., 2023).

Gene Ontology (GO) enrichment analysis

Genes were submitted to BnIR (https://yanglab.hzau.edu.cn/BnIR/GO) for GO enrichment analysis, and the top enriched items (based on gene count) were selected for bar plot or heatmap visualization using a customized R package.

Identification of homoeologous genes between AA and CC subgenomes

The protein sequence was downloaded from BnIR website (https://yanglab.hzau.edu.cn/BnIR/download?module=genomics). The protein sequences were then divided into AA and CC groups with corresponding gene ID. We first used the protein sequences from the AA group to perform a BLAST search against the CC group and then used the protein sequences from the CC group to perform a BLAST search against the AA group. We then compared the results from both searches to extract the best reciprocal matches.

Real time PCR

A total of 2 μg of high‐quality total RNA (A260/280 ≈2.00), treated with DNase I, was reverse‐transcribed into cDNA using the PrimeScript RT Reagent Kit (Takara, Japan, Cat. No. RR037Q). The ACT7 gene was used as an internal control. qPCR was performed using SYBR Green Master Mix (YEASEN, China, Cat. No. 11201ES) on a Bio‐Rad CFX96 instrument. Three biological replicates and three technical replicates were included in the analysis. Gene‐specific primer pairs (listed in Table S6) were used for amplification. Statistical significance was evaluated using Student's t‐test.

Plasmid construction and dual‐luciferase reporter assay

The 3′ UTRs of the pairs of homoeologous genes were synthesized at BGI (https://en.genomics.cn/). Briefly, Rluc is driven by the CaMV 35S promoter and serves as an internal control for normalization. Based on the sequence information of the vector used in this study (pC1300‐35S‐Rluc‐35S‐Fluc), the construct contains two independent expression cassettes: (1) a Renilla luciferase (Rluc) gene driven by the 35S promoter, and (2) a firefly luciferase (Fluc) gene also driven by the 35S promoter. The 3′ UTR sequences of interest were cloned downstream of the Fluc coding region to evaluate their regulatory effects. The 3′ UTR fragments were inserted into the vector using ClonExpress Ultra One Step Cloning Kit (Vazyme, China, Cat. No. C115‐01), ensuring precise and seamless integration into the designated site. All recombinant constructs were verified by Sanger sequencing before being introduced into Agrobacterium tumefaciens strain (GV3101(pSoup‐p19)) for transient expression assays in Nicotiana benthamiana (tobacco) leaves via agroinfiltration, and co‐infiltrated leaves were collected after 48 h (Sparkes et al., 2006). Relative gene expression was quantified using a dual‐luciferase reporter system. Firefly luciferase activity was measured with the Dual‐Lumi™ Luciferase Reporter Gene Assay Kit and Dual‐Lumi™ Luciferase Assay Kit (Beyotime Biotechnology, China, Cat. No. RG089S). Detailed sequences of primers, 3′ UTRs, and plasmid constructs are provided in Table S6.

Metagene analysis of 5′P ends

For the metagene analysis, we extracted normalized GMUCT‐seq reads (RPM) corresponding to 5′P ends mapped to the 50‐nt region upstream of exon–exon junctions. To minimize background noise, we retained only exons with total normalized 5′P counts ≥1. For each qualifying exon, we calculated the mean 5′P density across the coding sequence (CDS) as the background level. The signal within the canonical EJC window (26–29 nt upstream of the junction) was then quantified, and its ratio to the exon‐wide mean background was defined as the EJC enrichment score. Exons with an enrichment score ≥3 and a maximum 5′P peak located within the EJC window were identified as transcripts containing an EJC footprint.

Author Contributions

PX, YBW, SH, and XY designed the study. PX prepared the rapeseed material for RNA‐seq and GMUCT. PX and YBW performed the GMUCT experiments with input from XY. PX, YW, and SH compiled all sequencing data and carried out the computational analysis with input from XY. YBW performed experiments with input from PX and XY. PX, YBW, and XY wrote the manuscript. All authors read and approved the manuscript.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Figure S1. Principal component analysis (PCA) of RNA‐seq (A) and GMUCT‐seq (B) libraries based on RPM‐normalized gene expression values. Each point represents an individual sample, colored by experimental group (C‐hyp, T‐hyp, C‐cot, and T‐cot). C‐cot represents cotyledon without salt treatment; T‐cot represents cotyledon with salt treatment; C‐hyp represents hypocotyl without salt treatment; and T‐hyp represents hypocotyl with salt treatment.

Figure S2. Bar plot of salt‐responsive genes and GO enrichment analysis of DEGs. (A) Bar plot showing the expression levels (RPM values) of BnA09.HKT1, BnC05.NHX2, BnC05.NCED3 and BnA02.ACT7 in hypocotyls under normal and salt treatment. (B) Bar plot showing the expression levels (RPM values) of BnC03.CBL10 and BnC05.P5CSA in cotyledons under normal and salt treatment. (C, D) Bar plots showing the top 20 significantly enriched Gene Ontology (GO) terms (based on gene counts) for salt‐responsive DEGs identified in hypocotyls (C) and cotyledons (D).

Figure S3. Functional enrichment analysis of common DEGs and heatmap visualization of photosynthesis‐related genes. (A, B) Bar plots showing the top 10 significantly enriched GO terms (based on gene counts) of common upregulated DEGs (A) and common downregulated DEGs (B) in both tissues under salt treatment. (C) Heatmap visualization of DEGs exhibiting opposite salt response between hypocotyl and cotyledon tissues. Highlighted photosynthesis‐related genes, including BCA1, LHCA6, LHCB1, GAPA1, and PHYB.

Figure S4. Point plot of eight represented genes displaying coordinated salt‐responsive patterns between mRNA expression and mRNA instability (PU) in hypocotyl and cotyledon. Red dots represent the changes of mRNA abundance, black dots represent the change of proportion uncapped (PU), indicating mRNA instability, and ∩ indicates the intersection.

Figure S5. Asymmetrical distribution of epigenomic marks in BnaA07G0124900ZS and BnaC07G0183400ZS. (A–D) BnaA07G0124900ZS exhibits higher levels of active epigenetic marks (H3K27ac and H3K4me3) (A) and higher mRNA abundance (C), whereas its homeolog BnaC07G0183400ZS shows elevated levels of the repressive mark H3K27me3 (B) and lower mRNA abundance (D). The genome browser screenshots (A, B) were generated from https://yanglab.hzau.edu.cn/BnIR. (E, F) The A‐subgenome BnaDAR homoeolog (A07p14690.1 in Darmor‐bzh, syntenic to BnaA07G0124900ZS in ZS11) (E) exhibited significantly higher expression than the C‐subgenome homoeolog (C07p21580.1 in Darmor‐bzh, syntenic to BnaC07G0183400ZS in ZS11) (F).

Figure S6. Homoeologous gene pairs in B. napus seedlings exhibit degradation bias. (A) Genes in the CC subgenome showed significantly higher degradation rates (PU) than their AA subgenome counterparts under both control (CK) and Sclerotinia sclerotiorum‐infected conditions at 48 hpi (T48) in a public dataset. The p value was tested by the Wilcoxon test. (B) Venn diagram showing the overlap of homoeologous genes with subgenome‐biased degradation among the four groups (G1: hypocotyl under normal condition; G2: hypocotyl under salt treatment; G3: cotyledon under normal condition; G4: cotyledon under salt treatment.). (C) Bar plot showing representative homoeologous genes that display subgenome‐biased degradation in each sample group, respectively. Statistical significance was assessed by the student's t‐test.

Figure S7. Representative examples of homoeologous gene pairs showing subgenome‐biased mRNA degradation, with distinct 3′ UTR sequence divergence and altered RNA secondary structures. (A) In one homoeologous gene pair, an 18 bp deletion is present in the 3′ UTR of A09p13000.1_BnaDAR compared with C09p15400.1_BnaDAR. (B) This deletion alters the secondary structure of the RNA 3′ UTR in A09p13000.1_BnaDAR compared with C09p15400.1_BnaDAR. (C) In another homoeologous gene pair, mismatches occur between the 3′ UTRs of A03p33340.1_BnaDAR and C03p40580.1_BnaDAR. (D) These mismatches alter the secondary structure of the RNA 3′ UTR in A03p33340.1_BnaDAR compared with C03p40580.1_BnaDAR.

Figure S8. Sequence alignments of genes between the allopolyploid and its diploid progenitors. (A) Alignment of the 3′ UTR of A09p13000.1_BnaDAR with its B. rapa homolog, and of C09p15400.1_BnaDAR with its B. oleracea homolog, reveals that deletions in A09p13000.1_BnaDAR arise post‐allopolyploidization. (B) Alignment of the 3′ UTR of A03p33340.1_BnaDAR with its B. rapa homolog, and of C03p40580.1_BnaDAR with its B. oleracea homolog, indicate that mutations and indels between homoeologs predate allopolyploidization.

Figure S9. Overview of 5′ P reads distribution across an homoeologous gene pair in cotyledon. C03p20610.1_BnaDAR consistently produced more 5′P ends than A03p20610.1_BnaDAR under both control and salt stress conditions in cotyledon.

Figure S10. Number of homoeologous gene pairs and subgenome‐specific genes among predicted miRNA targets in B. napus.

Figure S11. Genes undergoing co‐translational RNA decay (CTRD) under salt stress. (A) Bar plot showing the number of genes undergoing CTRD across four libraries. (B) Metagene analysis displaying the distribution of 5′P reads around stop codons. (C) GO enrichment analysis of genes undergoing CTRD. (D) Intersection analysis between CTRD genes and genes exhibiting subgenome‐biased degradation across the four libraries.

Figure S12. CTRD genes are significantly enriched among subgenome‐biased degradation events across the four libraries. Expected and observed numbers of genes with CTRD in each library of subgenome‐biased genes. Significantly over‐ or underrepresented groups are shown (Chi‐squared test, *P < 0.05, **P < 0.01).

Table S1. Summary of the RNA‐seq and GMUCT‐seq data.

Table S2. Differentially expressed genes in rapeseed hypocotyl between control and salt treatments.

Table S3. Differentially expressed genes in rapeseed cotyledon between control and salt treatments.

Table S4. The genes involved in the GO item of hyperosmotic salinity response.

Table S5. The homoeologous gene pairs between AA and CC subgenomes.

Table S6. List of primers used in this study.

Data Availability Statement

12 transcriptome data (RNA‐seq) and 12 degradome data (GMUCT) reported in this paper have been deposited in the Genome Sequence Archive in National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA029705) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa.