Control of seed-to-seedling transition by an upstream open reading frame in ABSCISIC ACID DEFICIENT2

Significance

Seed dormancy and germination are critical phases during plant development. The hormone abscisic acid (ABA) has important roles in seed dormancy, and ABA DEFICIENT2 (ABA2) acts in ABA biosynthesis. Here, we revealed a translational regulation of ABA during seed-to-seedling transition, we found that an upstream open reading frame (uORF) in the ABA2 messenger RNA (mRNA) affects its translation in rice and Arabidopsis. Notably, we found that nucleotide variations in the uORF sequence of OsABA2 among rice cultivars lead to distinct preharvest sprouting phenotypes. This work underscores the significant role of ABA2 uORF in seed biology and offers promising strategies for crop improvement.

Abstract

The start of seed germination is a major decision point in plant life cycle, which relies on seed stored mRNA. However, the underlying translational mechanism remains less illustrated. Here, we demonstrate that inhibiting translation using translation inhibitors and ribosome-defective mutants delays germination in Arabidopsis. Through comprehensive transcriptome deep sequencing (RNA-seq) and polysome profiling analyses, we elucidated the dynamic interplay of regulation at the transcriptional and translational levels during germination. We show that delayed germination in some ribosome-defective mutants is partially regulated by the gene ABSCISIC ACID DEFICIENT2 (ABA2), with an upstream open reading frame (uORF) in the 5′ untranslated region that represses translation of the downstream ORF encoding ABA2. In addition, disrupting rice OsABA2 uORF inhibited preharvest sprouting (PHS). Furthermore, we found two main haplotypes for the uORF among rice cultivars that result in different OsABA2 expression levels, thus contributing to diverse PHS phenotypes. This work highlights the critical role of translational control and genetic variation in seed dormancy and germination, with implications for crop improvement.

The transition from dormant dry seeds to active seedling growth is an irreversible process that undergoes complete reprogramming (1). Adequate dormancy and timely germination are crucial for plant growth and crop production. For example, preharvest sprouting (PHS) is a phenomenon whereby seeds of grain crops germinate prematurely on the panicles, owing to insufficient dormancy (2). Seed germination is an important agronomic trait controlled by multiple genes, such as MOTHER OF FT AND TFL 1/2 (OsMFT1/2) (3, 4), SEED DORMANCY 6 (SD6) (5), and SEED DORMANCY 4 (Sdr4) (6, 7). PHS poses a significant threat to agricultural production and yield. Therefore, understanding how seed dormancy and germination are regulated is essential for ensuring high yields and achieving food security worldwide.

Seed dormancy, which is established during seed maturation, is a state of metabolic inactivity that enables seeds to withstand adverse conditions until the environment becomes suitable for germination (8). The dormancy status reduces during subsequent dry storage of the seeds (afterripening) (9). Seed dormancy and germination are tightly regulated by a combination of environmental cues and internal signals, including the action of phytohormones (10). The two most prominent plant hormones involved in this regulation are abscisic acid (ABA) and gibberellins (GAs), which exert opposite effects on seed germination: ABA suppresses germination and maintains seed dormancy, whereas GAs promotes seed germination (11–13). ABA biosynthesis is mediated by several key enzymes, such as the zeaxanthin epoxidase ABA DEFICIENT1 (ABA1), the dehydrogenase/reductase ABA2, and a family of NINE-CIS-EPOXYCAROTENOID DIOXYGENASEs (NCEDs) (13, 14). A disruption in ABA metabolism can lead to a loss of dormancy and premature germination, underscoring the need for precise regulation of ABA levels during this crucial developmental phase. Seed germination involves highly dynamic epigenetic and transcriptional changes (15, 16). Much focus has been directed toward transcriptional and posttranscriptional regulation in seed germination, the role of translational regulation has also attracted increasing attention because of its direct function in controlling protein production.

Ribosome biogenesis and translation represent essential mechanisms for regulating gene expression in response to developmental cues and environmental stresses (17–20). Previous studies have revealed that early germination is more dependent on de novo protein synthesis instead of newly synthesized mRNA, evidenced by treatment with transcriptional or translational inhibitors (21–23). During seed development, a large amount of mRNA accumulates in seeds, offering an available pool ready for protein synthesis required for germination (24–26). Translation is kept in an off state in dry seeds, but after seed imbibition, translation is initiated and the stored mRNAs are selectively translated (24, 25). This dynamic process involves the coordination of multiple molecular pathways, including ribosome activity and the activation and repression of specific mRNAs (27–31). Specific sequence features and structural elements within the mRNA molecules strongly influence their translational dynamics, such as the internal ribosome entry site, upstream open reading frames (uORFs), and the Kozak sequence (18, 32), but their specific roles during germination remain less reported.

In this study, we investigated the critical role of translational regulation during seed germination. Through transcriptome deep sequencing (RNA-seq) and polysome profiling analyses, we identified dynamic changes in transcript and translation levels that highlight the importance of specific protein synthesis during early germination. We used some ribosome-defective mutants as example to find that their delayed-germination phenotype is in part regulated by ABA2, which contains a uORF that represses ABA2 translation. Additionally, we explored the role of the orthologous uORF in rice; disrupting this uORF alleviates PHS. We also identified natural sequence variations in the OsABA2 uORF, the observed variation in PHS linked to the haplotype at OsABA2 uORF provides a potential strategy for crop improvement. Our findings underscore the significance of translational regulation and genetic diversity in seed germination and crop adaptation.

Results

Ribosome Is Essential during Seed Germination.

Seed germination represents a critical phase during seed-to-seedling transition, which necessitates the synthesis of specific proteins that are vital for establishing metabolic activities and cellular functions. The timing and efficiency of germination are optimized with multiple regulatory pathways. In this study, we investigated the regulation of seed germination at translational level. Specifically, we treated seeds from the Arabidopsis accession Col-0 with the translation inhibitor cycloheximide (CHX) at the time of seed imbibition. We observed a significant inhibition of germination when seeds were treated with CHX (SI Appendix, Fig. S1). The severity of the germination delay was positively associated with the CHX concentration: Higher concentrations resulted in a more pronounced inhibition of seed germination (SI Appendix, Fig. S1). These observations underscore the essential role of translational regulation in early germination, which is widely existed in different species (21, 23, 33).

To explore the importance of translation in germination, we collected different genes essential for ribosome biogenesis and translation and compared the germination phenotypes of their mutants. For example, AtPRMT3 is an Arabidopsis protein arginine methyl transferase, RPS2A2B is a member of Ribosomal protein S2, both are involved in ribosome biogenesis by regulating preribosome assembly, and thereby affecting ribosome translation (34–36); APUM23 is a pumilio protein involved in preribosomal RNA processing (37); RPS6A and RPS6B are members of Ribosomal protein S6 essential for protein translation (38). Our examination of these mutants revealed that they all exhibit delayed germination compared to wild-type seeds (Fig. 1 A and B). This delayed-germination phenotype reinforces the crucial role of translational regulation in seed germination, highlighting the necessity for proper ribosomal function and protein synthesis during this critical developmental phase.

Fig. 1.

Active ribosomal function is required for seed germination. (A and B) Representative photographs of seed germination (A) and seed germination rates (B) of Col-0, atprmt3-2, rps2a2b-1, apum23, rps6a, and rps6b mutant seeds. Germination rates were recorded every 12 h after release from stratification until complete germination. The photographs were taken at 36 h after release from stratification. (Scale bar, 1 mm.) (C) Scatterplot showing the correlation between changes in transcript levels and translation levels. Log₂-FC of FPKM from RS data (x-axis) and polysome profiling (PS; y-axis) are shown. Genes with FPKM > 0 were calculated. Pearson’s correlation coefficients (r) and two-tailed P-values are indicated. (D) Heatmaps showing genes with differentially expressed TE for imbibed Col-0 seeds collected at 24 h and 48 h compared with 0 h. The representative enriched GO terms are shown for the indicated groups.

Dynamic Gene Regulatory Programs during Seed Germination.

To gain comprehensive insights into mRNA translational regulation, we conducted both RNA-seq (RS) and polysome profiling (polysome-seq [PS] hereafter) to evaluate the dynamic changes in transcript and translation levels during different seed germination stages (SI Appendix, Fig. S2). To this end, we collected seeds of Col-0 imbibed in water for 0 h, 24 h, or 48 h for RS and PS analysis. An assessment of independent replicates of the RS and PS libraries demonstrated a high degree of reproducibility, indicating the reliability of our experimental approach (SI Appendix, Fig. S3).

We calculated the fold-change (FC) for both transcript (RS) and translation (PS) datasets at the 24 h and 48 h time points relative to the 0 h reference point (Dataset S1). Initially, we focused on the genes with upregulated transcript levels during seed germination at 24 h and 48 h compared to 0 h. Gene Ontology (GO) term enrichment analysis indicated that these upregulated genes are primarily associated with various cellular metabolic processes, ribosome biogenesis, and photosynthesis, suggesting heightened metabolic activation throughout germination (SI Appendix, Fig. S4 A and B). By contrast, the downregulated genes were mainly linked to nucleic acid metabolic process and respond to chemicals, these genes are involved in regulation of transcription, and response to abscisic acid, etc.; for example, ABA is a negative regulator during seed germination, there is decreased ABA demand during seed germination (SI Appendix, Fig. S4 A and B). Similarly, we analyzed the genes showing a differential translational output level, as determined by their binding to polysome translation. The genes in this analysis that experienced greater translation was predominantly involved in a range of cellular metabolic and biosynthetic processes, whereas those genes being less translated were primarily associated with stress–response pathways, highlighting the adaptive mechanisms plants employ to respond to internal and external signals to satisfy germination process (SI Appendix, Fig. S5 A and B). These patterns enable seed to manage the significant physiological, morphological, and chemical changes that occur during germination.

To investigate the relationship between regulation of the transcript and translation levels, we examined the correlation between RS and PS data, observing an increased correlation as germination progressed (Fig. 1C). This trend suggested that transcript levels and translation outcomes tend to be gradually synchronized during early seed germination, highlighting the dynamic interplay between these two regulatory axes. Additionally, we calculated translational efficiency (TE), defined as TE = PS_FPKM/RS_FPKM, by using the Fragments Per Kilobase of exon model per Million mapped fragments (FPKM) values for each gene from the RS and PS data. We compared TE of 24 h and 48 h to 0 h in Col-0, GO term analysis revealed that as germination progressed, genes with decreasing TE are mainly associated with ribosome biogenesis, indicating that cells fine-tune ribosome amount to keep them at appropriate level for translation function, whereas those with increasing TE are mainly linked to various growth and developmental processes, reflecting the increased biological demands during seed germination (Fig. 1D and Dataset S2). The comparison of differential TE between 48 h and 24 h reveal that the down-regulated genes are mainly related to some metabolic process, such as nucleobase, nucleoside, nucleotide and nucleic acid metabolic process, RNA metabolic process, RNA processing, etc. The up-regulated genes are still mainly associated with some developmental process (SI Appendix, Fig. S6). The pattern indicates the dynamic gene expression during germination progress.

We conclude that the dynamic regulation of transcript levels and translation ensures the timely production of the appropriate proteins, enabling seeds to meet the demands of seed-to-seedling transition.

The Delayed Germination of Some Ribosome-Defective Mutants Is Partially Dependent on ABA2.

Then, we used atprmt3-2 mutant with defective ribosome biogenesis and translation, as an example to explore the underlying translational mechanism during seed germination in more detail (34–36). Similarly, we conducted RS and PS with Col-0 and the atprmt3-2 mutant at 48 h into germination, when Col-0 seeds have all germinated but only ~30% of atprmt3-2 mutant seeds have germinated. We then compared the genes with different TE values between these two genotypes, followed by GO enrichment and Kyoto Encyclopedia of Genes and Genomes pathway analysis to identify genes involved in this AtPRMT3-dependent regulatory step. The genes with lower TE in the atprmt3-2 mutant than in Col-0 were predominantly associated with multiple developmental process (SI Appendix, Fig. S7 A and B and Dataset S3). The genes with higher TE in the atprmt3-2 mutant were primarily linked to ribosome biogenesis and translation, consistent with the known role of AtPRMT3 in the ribosome assembly (SI Appendix, Fig. S7 A and B and Dataset S3).

We observed that genes associated with carotenoid biosynthesis pathway exhibit higher TE in atprmt3-2 mutants (SI Appendix, Fig. S8A). Given the role of ABA in inhibiting seed germination, we focused on the carotenoid biosynthesis and ABA biosynthesis pathways. We hypothesized that the elevated TE of genes related to ABA biosynthesis might enhance ABA production, then we quantified the content of ABA, finding that it is up-regulated in atprmt3-2 mutants (Fig. 2A). This led us to propose that the delayed seed germination phenotype in atprmt3-2 could be due to higher ABA levels. To test this hypothesis, we treated Col-0 and atprmt3-2 seeds with ABA inhibitor fluoridone: Indeed, the fluoridone treatment partially rescued the delayed-germination phenotype of the atprmt3-2 mutant (SI Appendix, Fig. S9).

Fig. 2.

The delayed germination of these ribosome-defective mutants partially depends on ABA level. (A) Accumulation of ABA level in atprmt3-2 mutants. P-values were calculated by a two-tailed Student’s t test. (B) Accumulation of ABA2 protein in atprmt3-2 mutants. Seeds of Col-0 and atprmt3-2 mutants germinated for 0 h, 24 h, 48 h, 60 h, and 72 h were collected and subjected for immunoblotting with anit-ABA2 antibodies. HSC70 served as a loading control. (C and D) The loss of ABA2 function partially rescues the delayed germination in the selected ribosome-defective mutants. The photographs were taken at 36 h and 48 h after release from stratification. Germination rates of Col-0, atprmt3-2, rps2a2b-1, rps6a, and rps6b mutant seeds and their double mutants with aba2 were recorded every 12 h following release from stratification until full germination was achieved. (Scale bar, 1 mm.)

Then, we focused on ABA2, a key gene encoding an enzyme responsible for the conversion of xanthoxin to abscisic aldehyde during ABA biosynthesis (39, 40), which showed higher TE in the atprmt3-2 mutant than in Col-0 (SI Appendix, Fig. S8B). We detected that the protein level of ABA2 is gradually decreased in Col-0 as germination progress, however, ABA2 keeps at a constant higher level in atprmt3-2 mutants (Fig. 2B). To further explore the role of ABA2, we crossed atprmt3-2 to aba2 to obtain the atprmt3 aba2 double mutant, whose seeds also germinated earlier than those of atprmt3-2, indicating that loss of ABA2 partially rescues the delayed seed germination of atprmt3-2 (Fig. 2 C and D).

We wondered whether this pharmacological rescue by fluoridone and genetic rescue by aba2 were consistent across other ribosome-defective mutants. Accordingly, we investigated the germination rate of seeds from the mutants rps2a2b-1, rps6a, and rps6b upon imbibition in the presence of fluoridone: As with the atprmt3-2 mutant, fluoridone treatment partially rescued the delayed germination of these mutants (SI Appendix, Fig. S9). Moreover, we generated the rps2a2b aba2, rps6a aba2, and rps6b aba2 double mutants, which all exhibited earlier seed germination than their respective single mutants (Fig. 2 C and D). The above results indicate that the delayed germination seen in these ribosome-defective mutants is partially dependent on higher ABA2 expression.

The Translation of ABA2 Is Regulated by a uORF in 5′ Untranslated Region (5′ UTR).

We noticed a predicted uORF in the 5′ UTR of the ABA2 transcript. uORFs are short protein-coding elements located in the 5′ UTR that repress translation of their downstream main ORFs (32). To evaluate the effect of the ABA2 uORF on ABA2 TE, we generated two dual-luciferase reporter constructs: one containing the intact AtABA2 uORF starting with an AUG start codon (uATG) and another with a mutated version in which the start codon was changed to AAA (uAAA) (SI Appendix, Fig. S10A). These two versions of the ABA2 uORF were cloned individually to upstream of the firefly luciferase (FLUC) gene, with the Renilla luciferase (RLUC) gene serving as an internal control. We transfected each reporter construct into Col-0 protoplasts and determined the relative mRNA abundance of FLUC/RLUC, as well as relative FLUC/RLUC activity. RT-qPCR analysis revealed that the ratio of FLUC to RLUC transcript levels was not significantly different between the uATG and uAAA constructs (SI Appendix, Fig. S10B). Importantly, the mutated uORF (uAAA) produced much higher relative FLUC/RLUC activity than did the intact uORF (uATG) (SI Appendix, Fig. S10C). The above results suggest that the uORF located upstream of the main AtABA2 ORF may repress downstream translation.

Next, we determined the translational status of ABA2 uORF in atprmt3-2 mutants. The dual-LUC reporter construct containing ABA2 uORF was introduced into both Col-0 and atprmt3-2 protoplasts (SI Appendix, Fig. S11A). The results revealed that the repressive function of the ABA2 uORF was attenuated in atprmt3-2 mutants (SI Appendix, Fig. S11 B and C). The above results indicate that ABA2 uORF tends to be less translated in atprmt3-2 mutants, which correlates with the observed increase in main ABA2 translation efficiency in atprmt3-2 mutants.

The Disruption of the OsABA2 uORF Prevents PHS.

ABA2 is a conserved protein widely found across plant species; in fact, examination of sequences in the uORFlight database revealed the presence of a uORF at the ABA2 locus in multiple plant species (Fig. 3A and Dataset S4) (41). We focused here on rice, a staple crop worldwide, to investigate the function of the uORF in OsABA2. To this end, we mutated its start codon from ATG to ATA (uATG to uATA) and evaluated their effects on relative FLUC transcript levels and relative FLUC activity in a dual-luciferase reporter assay in transfected rice protoplasts (Fig. 3B). We found that the mutated uATG increased the translation of relative FLUC/RLUC activity with no obvious effect on its transcription (Fig. 3 C and D), which is similar to the results in Arabidopsis. Above all, we revealed that there are uORFs that exist in 5′ UTR of the OsABA2 transcript, which functions in repressing the translation of its downstream ORF.

Fig. 3.

The Mutation of OsABA2 uORF alleviates PHS. (A) Left, phylogenetic tree illustrating the evolutionary relationships of ABA2 across different species. Right, gene models of ABA2 with its predicted uORFs. (B) Diagram of the reporters used for the dual-luciferase reporter assay. The 5′ UTR of the intact uORF from OsABA2 starting with uATG and its mutated version starting with uATA were individually cloned upstream of the FLUC reporter. (C and D) Effects of the uATG and uATA versions of the OsABA2 uORF on relative FLUC/ RLUC mRNA level (C) and relative FLUC/RLUC activity level (D), as measured by RT-qPCR and dual-luciferase assays, respectively. Data are means ± SD (n = 5). (E) Diagram illustrating the CRISPR/Cas9-mediated editing of the OsABA2 uORF. The mutated sequences are shown in red in osaba2-uorf (#6 and #9) mutants. (F) Relative OsABA2 expression levels as detected by RT-qPCR in wild type ZH11 and two osaba2-uorf mutants. Data are means ± SD (n = 4). (G) Immunoblot analysis of OsABA2 abundance. HSP82 was served as a loading control. (H and I) PHS phenotypes of wild-type ZH11 and osaba2-uorf mutants. The sprouting rate was calculated for preharvest panicles. Data are means ± SD (n = 4). In (C and D), P-values were calculated by a two-tailed Student’s t test. In (F and I), P-values were calculated by one-way ANOVA. ns indicates no significant difference.

To explore the biological functions of the OsABA2 uORF in vivo, we employed base editor to mutate the uATG in uORF (Fig. 3E). We generated two lines with a disrupted uORF, with the ATG start codon replaced with ATA, OsABA2 transcript levels were similar among the wild-type Zhonghua 11 (ZH11) and the two edited lines (OsABA2-uorf #6 and #9) (Fig. 3F). Notably, OsABA2 protein levels in these mutants were elevated relative to that in the wild-type ZH11 (Fig. 3G). ABA is closely related to seed dormancy and germination, for example, PHS is a consequence of weak seed dormancy. We therefore investigated PHS in the wild type and in rice lines with edited OsABA2 uORF (#6 and #9): The two OsABA2-uorf lines showed delayed germination on freshly harvested panicles compared to the wild type (Fig. 3 H and I), suggesting the profound effect of OsABA2 uORF in regulating PHS.

Natural Variations at OsABA2 uORF Are Responsible for PHS.

Next, we evaluated the possible effects of variation in the OsABA2 uORF during rice domestication, haplotype analysis was performed using the genome sequences of more than 3,000 rice varieties available from the RiceVarMap (42). We identified six haplotypes, with Hap1 and Hap2 being the most prevalent. Notably, Hap1 was predominantly associated with japonica rice varieties, whereas Hap2 was primarily found in indica varieties (Fig. 4A). These two haplotypes differ at position −122 bp; Hap1 possesses the sequence GCA, whereas Hap2 has a 2-bp deletion and carries a G instead of GCA (Fig. 4A).

Fig. 4.

Natural variations in OsABA2 uORF are responsible for different PHS phenotypes. (A) Haplotype analysis of OsABA2 uORF. The polymorphic nucleotides, sequence variations, and positions are shown. Poi: Position, Hap: Haplotype. (B) Sequence alignment of OsABA2 uORF from the rice varieties ZH11 and 9311. (C) Diagram of the constructs used for the dual-luciferase reporter assay. The 5′ UTR of OsABA2 from ZH11 or 9311 was cloned to upstream of FLUC. (D and E) Effects of the OsABA2 uORF from ZH11 or 9311 on relative FLUC/RLUC mRNA levels (C) and relative FLUC/RLUC activity levels (D) as measured by RT-qPCR and dual-luciferase assays, respectively. Data are means ± SD (n = 5). (F) PHS phenotypes from representative Hap1 and Hap2 varieties. (G) The sprouting rate on preharvest panicles from Hap1 and Hap2 varieties. Data are means ± SD. (H) Diagram illustrating the prime editing of the OsABA2 uORF in ZH11. The mutated sequences are shown in red in ZH11-del mutants. (I) PHS phenotypes of wild-type ZH11 and ZH11-del mutants. The sprouting rate was counted for preharvest panicles. Data are means ± SD (n = 4). P-values were calculated by a two-tailed Student’s t test. ns indicates no significant difference.

Then, we chose two representative rice varieties: ZH11, from the japonica group, and 9311, classified as indica, to assess the biological implications of this sequence variation. The OsABA2 uORF^ZH11 possesses the GCA sequence (Hap1) and is 138 bp in length, whereas 9311 has the Hap2 sequence in its OsABA2 uORF⁹³¹¹, with a 2-bp deletion leading to a premature termination of the uORF, resulting in a much shorter uORF of 36 bp (Fig. 4B). We tested the effect of uORF^ZH11 and uORF⁹³¹¹ on transcript levels and translation by cloning them individually upstream of the FLUC reporter gene for a dual-luciferase reporter assay in rice protoplasts. Although the two uORF variants did not affect the ratio of FLUC to RLUC transcript levels, the shorter uORF⁹³¹¹ version yielded much higher relative FLUC/RLUC activity than that measured for uORF^ZH11 (Fig. 4 C–E), suggesting that the repressive function of uORF⁹³¹¹ is weaker than uORF^ZH11 on translation of the downstream ABA2.

To broaden our analysis, we included additional typical japonica and indica rice varieties to examine their PHS phenotypes. Varieties with Hap1 (GCA-type) of the OsABA2 uORF displayed a pronounced higher germination rate for grains on fresh panicles than did those with Hap2 (G-type) (Fig. 4 F and G and Dataset S5). To further test this observation, we employed prime editing to modify uORF^ZH11 from a Hap1-type to a Hap2-type by deleting the CA sequence. We successfully obtained the edited line, ZH11-del, with the desired 2-bp deletion in ZH11 background (Fig. 4H). A phenotypic analysis of this line revealed its lower sprouting rate compared to wild type ZH11 (Fig. 4I). This comprehensive analysis not only elucidates the functional consequences of uORF variation at the OsABA2 locus but also underscores the importance of genetic diversity in shaping the domestication and adaptive traits of rice varieties, providing more options for molecular design in breeding to prevent PHS.

Together, our results revealed a significant translational mechanism influencing seed dormancy and germination, particularly concerning ABA2, which contains a uORF in its 5′ UTR that represses translation of its downstream ORF. Importantly, our study in rice uncovered natural variation in the OsABA2 uORF between indica and japonica varieties and demonstrated that this variation contributes to differential OsABA2 translation, ultimately resulting in diverse PHS phenotypes (Fig. 5).

Fig. 5.

A proposed working model for ABA2 uORF in regulating seed-to-seedling transition. Model showing that the uORF in the 5′ UTR of ABA2 represses the translation of its downstream coding sequence. Natural variation in OsABA2 uORF confers different OsABA2 protein abundance, therefore exhibiting different PHS phenotypes.

Discussion

The transition from seed dormancy to germination is a pivotal phase in the plant life cycle, governed by intricate regulatory networks. In our study, we uncovered a dynamic translational regulation network during early seed germination, especially focusing on the role of uORFs at the ABA2 locus and how these elements are crucial for understanding genetic variation in seed dormancy and germination.

Ribosomes are integral to the translation machinery that produces proteins, which is essential for all living cells (43). Here, using translation inhibitors and ribosome-defective mutants, we obtained evidence that translation is prioritized during germination stages. In dry seeds, translation is largely inactive, and polysomes remain in an "off" state, however, as germination progresses, polysomes are gradually activated, making a shift toward active translation (27, 44). During early germination, stored mRNAs in seeds are primed for translation without the need for de novo transcription, suggesting that transcript and translation are not tightly coupled at the initial stage. Here, we revealed a progressive increase in the correlation between RS and PS data as germination advanced, highlighting dynamic transcript levels and translation regulation during these stages. As seeds germinate, they undergo various physiological and biochemical transformations, such as the mobilization of stored nutrients and activation of metabolic pathways (25). Our data indicate that during these early phases, the TE and transcription of genes related to ribosome biogenesis are dynamically changed, suggesting that the existing translation apparatus of dry seeds is sufficient for early seed germination, then ribosomes will be fine-tuned to satisfy further development. This reliance on the translation machinery loaded into seeds during their maturation allows the emerging seedling to conserve energy and prioritize the translation of mRNAs involved in metabolic activities, development, and stress responses, rather than producing more ribosomes. For example, the early translation of specific mRNAs enables a quick adaptation to environmental cues, such as water availability and changes in ambient temperature. Overall, effective and selective translation ensures that seeds undergo successful germination and seedling establishment (45).

Earlier studies have revealed the translation activation during seed germination in general and identified some sequence features correlating with translational shifts (27, 44). Here, we extend the study with ribosome defective mutants as examples to study seed germination and build up a link between translational regulation and ABA biosynthesis. Specifically, we demonstrate the specific role of uORF-mediated translational control in regulating ABA2 expression and seed dormancy. We find that the disruption of the OsABA2 uORF leads to higher translation of its downstream main ABA2, thereby influencing PHS in rice. Despite the clear importance of uORFs, natural variation in uORFs in plants remains less explored. To date, few examples of uORF variation in plants have been documented, with notable instances including PHOSPHATE1 (PHO1) in Arabidopsis (46), PHOSPHATE TRANSPORTER TRAFFIC FACILITATOR1 (GmPHF1) in soybean (Glycine max) (47), PETAL LOBE ANTHOCYANIN (PELAN) in monkeyflower (Mimulus) (48), etc. Notably, our data suggest that variation in the OsABA2 uORF between indica and japonica rice varieties contributes to their distinct PHS phenotypes. This result is an essential stride toward understanding how uORF diversity affects plant phenotypes. The underlying mechanism still needs to be explored, we propose that uORF function involves RNA secondary structure, which may substantially influence the regulation of translation by altering ribosome scanning patterns and overall translation efficiency. As RNA structures can evolve to play roles in transcription, processing, translation, and degradation (49). For instance, PHYTOCHROME-INTERACTING FACTOR 7 translation is regulated by a hairpin structure within the 5′ UTR under different temperatures, offering an example as how structural dynamics can modulate translation (50). Similarly, variation in uORFs may alter the RNA secondary structure, affecting ribosome scanning and overall translation efficiency. Integrating advances in gene-editing technology with the discovery and functional analysis of uORFs represents a promising approach to crop improvement (51–53). Targeted modulation of uORFs may enhance desirable traits such as crop yield and stress resilience, thereby contributing significantly to agricultural productivity and sustainability (54).

Materials and Methods

Plant Materials.

The Arabidopsis accession used in this study was Columbia 0 (Col-0). The mutants of atprmt3-2 (WISCDSLOX391A01), rps2a2b-1, and apum23 (SAIL_757_B08) were reported previously (34, 35). The mutants rps6a (SALK_048825) and rps6b (SALK_012147) were obtained from the Arabidopsis Biological Resource Center. The aba2-1 (CS156) seeds were a gift from Dapeng Zhang at Tsinghua University, Beijing, China.

Germination Assays.

For Arabidopsis, the seeds were harvested on the same day and dried at room temperature for 3 mo before conducting experiments. Surface-sterilized seeds were sown on Murashige and Skoog (MS) medium solidified with 1% (w/v) agar, stratified at 4 °C for 3 d in the dark, and then transferred to a growth chamber (BPC500-DH/C, JIUPO, China) with 16-h light/8-h dark photoperiod at 22 °C. The germination rate was determined at the designated time points. For rice, freshly harvested panicles were collected for PHS assays. The panicles were immersed in water and placed in a growth chamber (BPC500-DH/C, Jiupo, China) with 12-h light/12-h dark photoperiod at 35 °C. After 3 d of seed imbibition, the germination rates were scored from four biological plants.

Plasmid Construction.

For base editing, the A3A-PBE vector was used (55). For prime editing, the pH-ePPE-epegRNA vector was used (56). The corresponding 5′ UTR sequences from ABA2 of Arabidopsis or rice were synthesized by GenScript company (at Nanjing, China) and then cloned into a dual-luciferase assay reporter construct upstream of the FLUC reporter and downstream of the UBIQUITIN promoter. The specific primers used in this study are listed in SI Appendix, Table S1.

Other detailed methods including Protoplast transfection and dual-luciferase reporter assay, Polysome extraction and RNA isolation, Protein extraction and immunoblotting, Measurement of ABA, RT-qPCR, Data analysis, and Statistics and reproducibility are described in SI Appendix, Materials and Methods.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Dataset S05 (XLSX)

Data, Materials, and Software Availability

The sequencing data have been deposited in the Genome Sequence Archive at the National Genomics Data Center, Beijing Institute of Genomics, Chinese Academy of Sciences under the Bioproject number: PRJCA034204 (57). All other data are included in the manuscript and/or supporting information.