Mutation in the LONGIFOLIA1 Gene Resulted in Suppressed Insensitivity of Arabidopsis thaliana proteolysis6 Mutant to Ethylene During Seed Germination

Abstract

Seed dormancy and germination is regulated by internal hormones and exogenous environment cues. Ethylene is one of the hormones that break seed dormancy and induce seed germination. Our previous study showed that N-degron pathway gene, proteolysis6 (PRT6) was involved in dormancy release by ethylene, the defection of which exhibiting ethylene-insensitivity in Arabidopsis thaliana. In the present study, through screening an ethyl methyl sulfonate-mutagenized (EMS) population of prt6−1, we isolated a recessive mutant that acted as a suppressor of prt6 that rescued its insensitivity to ethylene as well as a phenotype of shorter silique length. Further bulk segregant analysis on F2 population identified a premature termination located in the third exon of LONGIFOLIA1 (LNG1), previously reported in the regulation of longitudinal cell elongation. Mutation of LNG1 in prt6−1 background by CRISPR-Cas9 confirmed that LNG1 was epistatic to PRT6 in seed responsiveness to ethylene. Our finding proposed the pleiotropic effect of LNG1 in seed dormancy breakage by ethylene via PRT6, providing novel functional component at the downstream of the coordinated PRT6 and ethylene signaling pathway.

1. Introduction

Primary seed dormancy refers to the inability of seeds to germinate in apparently favorable conditions at harvest, while secondary dormancy is considered as a quiescent state of the seed induced by unfavorable conditions [1,2]. The transition from dormancy to germination is critical for the generative reproduction in the plant life cycle. The induction and breakage of seed dormancy depends on genetic and environmental factors [1,2]. Ethylene, a gaseous hormone, has been reported to release seed dormancy of many species via fine-tuning of the dominant hormone balance of abscisic acid (ABA)/gibberellins (GAs) and reactive oxygen species (ROS) [3,4,5,6]. Moreover, ethylene promotes the endosperm cap weakening and possibly enhances water transport after seed imbibition [7,8]. In several species, germination completion is concomitant with the endogenous ethylene production [6,9,10].

Classical ethylene signaling is identified from the “triple response”. It starts from ethylene perception and binding by receptors, which inactivates and inhibits the function of CTR1 (CONSTITUTIVE TRIPLE RESPONSE 1), thus allowing EIN2 (ETHYLENE INSENSITIVE 2) to act as a positive regulator of ethylene and facilitating the accumulation of EIN3. EIN3 as the central transcription factor binds to the promoters of some ERF (ETHYLENE RESPONSIVE FACTORS) to regulate transcription of downstream genes [6,11,12]. The effect of ethylene in seed dormancy control was proven by the significant dormancy level alterations in ethylene receptor mutants [13,14,15,16]. It was proposed that increased seed dormancy was present in the gain-of-function mutant, like ethylene response1−2 (etr1−2), while the reduced seed dormancy in loss-of-function mutant, such as etr1−6, etr1−7 or reduced dormancy3 (rdo3) [16]. It was recently elucidated that ERF12 functioned downstream of the ethylene receptor ETR1 and formed a repressive complex with TOPLESS (TPL) that inhibited the expression of DELAY OF GERMINATION1 (DOG1), a conserved central regulator of seed dormancy [16]. Dormancy release by ethylene required the production of reactive oxygen species (ROS), a prerequisite for radicle elongation during the germination process, in which mitochondria was suggested to be the major intracellular source [17].

The N-degron pathway corresponds to the ubiquitin-proteasome system (UPS) degradation by targeting specific N-terminal residue of a protein [18]. In Arabidopsis, two N-recognins, namely, PROTEOLYSIS 1 (PRT1) and PRT6 acted as E3 ligase identifying specific N-degrons, the aromatic residues (Phe, Trp, Tyr) and the basic residues (Arg, His, Lys), respectively. These N-degrons could be generated by proteolytic cleavage such as N-terminal Met excision by methionine aminopeptidase (MAP) or endoproteolytic cleavage by endopeptidase (EP). In addition, N-degrons is also able to be produced through successive process, firstly, post-translational modification of tertiary residues (Asn, Gln, Cys) to secondary residues (Asp, Glu, Cys-sulfinic acid) via deamidation or oxidation; finally, a peptide initiating with Arg via Arginyl-tRNA transferase (ATE) is subjected to identification by PRT6 [18,19]. The N-degron pathway has been reported to regulate many aspects of plant development and adaptation to the environment. In recent years, particular attention has been paid to the involvement of PRT6 in oxygen sensing with the identification of the first substrate, Ethylene Response Factor transcription factors VII (ERFVIIs), including RELATED TO APETALA (RAP) 2.12, RAP2.2, and RAP2.3, HYPOXIA-RESPONSIVE ERF (HRE) 1 and HRE2 [20,21]. The second and third substrates of PRT6, LITTLE ZIPPER 2 (ZPR2) and VERNALIZATION 2 (VRN2) were proposed in the regulation of plant development, also in an oxygen dependent manner [22,23,24]. All the substrates mentioned above begin with Met-Cys and functioned as N-degrons in their pre-pro-protein forms, a feature that has facilitated their discovery despite the diversity and complexity of neo-N-termini. The effect of PRT6 in the regulation of seed germination was first reported in regard to the involvement of ABA signaling with the enhanced sensitivity of prt6 to ABA, compromised sensitivity of prt6aba insensitive3 (abi3) or prt6abi5 to ABA in seed germination [25]. Further studies showed that PRT6 promoted seed germination by destabilizing ERFVIIs through nitric oxide, a key molecule in dormancy release. ERFVIIs, in turn, directly bind to the ABI5 promoter and enhance its expression, thereby mediating the crosstalk between nitric-oxide and ABA signaling during seed germination [26]. Our previous study showed that prt6 mutant were ethylene-insensitive during seed germination. And their canonical substrates, the ERFVIIs transcription factors RAP2.2, RAP2.3, and RAP2.12, functioned downstream of PRT6 in the ethylene response with the enhanced sensitivity to ethylene in prt6rap2.2rap2.3rap2.12 [27,28].

In order to discover a novel component of PRT6−involved pathway in the control of seed responsiveness to ethylene other than the well-known substrates, we launched out a genetic screening for the prt6−1 suppressors from an ethyl methane sulfonate (EMS) mutant library. A recessive mutant, prt6−r, was subsequently isolated that exhibited increased seed sensitivity to ethylene and produced shorter siliques. Our study not only described the identification of the suppressor but also provided insights for future molecular-mechanism investigations.

2. Materials and Methods

2.1. Plant Materials and Growth Condition

T-DNA insertion mutant prt6−1 (SAIL−1278−H11) in Arabidopsis thaliana Columbia background was obtained from the Nottingham Arabidopsis Stock Center (NASC). For seed multiplication, prt6−r (isolated from the EMS mutant library of prt6−1), the F1 and F2 hybrid populations of prt6−1 and prt6−r seeds were firstly cold stratified at 4 °C for 3 days to break dormancy then transferred into the potting mixture (2:1 mixture of Pindstrup substrate and vermiculite) placed in a growth chamber under a 16 h light/8 h dark cycle (23 °C/20 °C). Siliques were harvested at maturity and seeds were stored at −20 °C before experiment.

2.2. EMS Mutagenesis

Arabidopsis prt6−1 seeds were imbibed in 0.1% KCl (MedChemExpress, Shanghai, China) at 4 °C for 16–20 h, rinsed with tap water, and treated with 0.1 mM EMS (Biorigin, Beijing, China) at room temperature for 6–8 h. After discarding the EMS solution, seeds were neutralized in 0.1 M sodium thiosulfate (Macklin, Shanghai, China) for 30 min, rinsed in running tap water for 12 h, suspended in 0.3% agar (Forscience, Beijing, China), and immediately sown onto the potting mixture.

2.3. Screening of Suppressors of prt6−1 Mutant

EMS-mutagenized prt6−1 seeds (M0) were grown and the seeds were harvested altogether as M1 without evaluating any parameters. M2 seeds were harvested from the mother plant of each M1 individual. Newly harvested M2 seeds were used to perform a seed germination test with/without the treatment of ethylene (See Section 2.7). The M2 seeds with enhanced sensitivity (>80% germination percentage) to ethylene in germination were kept for further propagation. Around 30 individuals for each family of M2 population were grown and each individual was harvested as M3. A seed germination test with/without the treatment of ethylene (See Section 2.7) was also carried out on the M3 population. The stable mutant individuals were confirmed by the stability of the phenotype of significant sensitivity to ethylene in germination (>80% germination percentage) carried out continuously in M4 and M5.

2.4. Population Construction and Bulk DNA Preparation

Seeds of prt6−1 and the prt6−r mutant were planted and hybridized to generate F1, one single of F1 were grown to develop a F2 population. A total of 523 F2 individuals and the two parental lines were planted. Siliques from each individual were harvested at maturity and seeds were stored at −20 °C to keep dormancy for ethylene treatment. Genomic DNA was isolated from fresh young leaves of each F2 plant using the procedure of the CTAB method [29]. The grinded leaves were added with CTAB buffer (Macklin, Shanghai, China) and incubated at 65 °C to break down the cells. The supernatant was collected and mixed with chloroform–isoamyl alcohol to separate DNA. The upper aqueous phase was added with isopropanol to precipitate DNA and the final pellets were washed by ethanol. Two genomic DNA bulks were established for the F2 population with extremely contrasting differences in the phenotypes of seed germination and silique length. Each bulk consisted of around 20 individuals with equal DNA concentration, which were evaluated by Qubit 4 TM Fluorometer (Thermo Fisher Scientific, Wilmington, NC, USA) with dsDNA HS Assay Kit (Vazyme, EQ121, Nanjing, China) as the manufacturer’s user guide. The DNA bulks were then used for whole-genome sequencing.

2.5. BSA-Seq Analysis

The DNA bulks and DNA samples from parents were sent to the MGI DNBSEQ-T10 platform for library construction and sequencing. Raw reads were firstly cleaned by fastp [30]. The subsequent cleaned reads were mapped to the Tair10 genome using bwa [31]. The mapping files were processed into BAM files using the SAMtools software version 1.6 [32]. Meanwhile, the duplicates were removed by picard version 2.18.29 (https://broadinstitute.github.io/picard/, accessed on 1 August 2025). The software gatk version 4.2.3 was used for variant calling and filtering across all samples simultaneously using standard hard filtering parameters [33]. The generated VCF files were then used to calculate the SNP index via the sliding window method. The delta (SNP index) was obtained by subtracting SNP-index (1) from that of SNP-index (2), which was integrated into the QTLseq pipline [34]. G prime statistic algorithm method analysis was carried out with the utilization of QTLseqr [35]. SNP annotation was performed based on the Tair10 genome using the annovar software version 2020.06.08 [36].

2.6. CRISPR-Cas9 Editing

LNG1 genome-edited lines were generated using the egg cell specific promoter derived CRISPR/Cas9 vector, pHK2-AtCas9-U6 (BioRun, Wuhan, China), which contained a ccdB cassette flanked by two BsaI sites. The forward and reversed oligos (Sangon, Guangzhou, China) were synthesized with BsaI overhangs and 20 bp spacer (Primer sequences shown in Supplementary Data). The single-stranded primers were firstly annealed, and the products were digested and ligated together with pHK2-AtCas9-U6 vector by BsaI enzyme (NEB) and T4 DNA ligase (NEB). The positive construction was selected by transformation into E.coli DH5α and confirmed by colony sequencing. For plant genetic transformation, the fused vector was transformed into Arabidopsis thaliana prt6−1 with Agrobacterium tumefasciens strain GV3101 by floral dip method [37]. The T0 generations were sprayed with glufosinate ammonium (Yeason, Shanghai, China) in early seedling stages to select positive transformation. In the T1 and T2 generations, the homozygous lines were identified by genome PCR and sequencing near the spacer site.

2.7. Germination Assays with Ethylene Treatment

Germination assays were carried out in a 2.6 L tightly closed container injected with or without gaseous ethylene (5%) (HDX, Shenzhen, China). The seed responsiveness to ethylene test was conducted with 100 µL/L concentration. For the germination test, seeds were placed on three layers of filter paper moistened with deionized water at 25 °C. Seeds were considered to have germinated as soon as the radicle protruded through the seed coat. Germination percentages were calculated after 7 days. Results corresponded to the mean of the germination percentages obtained with at least three biological replicates of 50–100 seeds from independent mother plants.

2.8. Silique Length Measurement

In order to count the average silique length of each individual from F2 population, we adopt a more convenient statistical method. After the silique maturation, the first six developed siliques from the main inflorescence axis were taken and the average silique length was then calculated.

3. Results

3.1. Phenotypic Characterization of prt6−r and Bulk Construction

The prt6−r mutant was identified from EMS-based mutagenesis in the genetic background of T-DNA mutant prt6−1, which was reported the insensitivity to ethylene in seed germination from our previous research [27]. The mutant prt6−r not only repressed the insensitivity to ethylene, but also exhibited shorter siliques (Figure 1A,B).

For rapid identification of the gene controlling the two phenotypes, especially seed sensitivity to ethylene, we crossed the prt6−r mutant to original line, prt6−1 to obtain F1 progeny. F1 plants were self-pollinated to generate F2 progeny for bulk construction. Silique lengths from each individual of the F2 population were firstly measured as described in Section 2 The silique length distribution in a total of 243 accessions exhibited a positively skewed distribution pattern (Figure 1D). Subsequently, freshly harvested seeds from 243 individuals were firstly subjected to the test of the dormant level, that is, germination assay at 25 °C in the dark without ethylene. Only non-germinated accessions were kept for further ethylene treatment. This procedure excluded 6 lines with relatively lower dormancy. The germination percentage distribution in the presence of ethylene in a total of 237 accessions was shown a negatively skewed distribution (Figure 1C). Both of the distribution patterns indicated that the germination and silique length of prt6−r might be a recessive trait. Based on the phenotype of the progenies, the first 15% and the last 15% in the population of “germination percentage” and “silique length” were taken into consideration for bulk construction. Two bulks were developed for sequencing, bulk1 habor 33 progenies with germination percentage with ethylene < 20% and average silique length > 10 mm, and bulk2 habor 17 progenies with germination percentage with ethylene > 80% with ethylene and their silique length were less than 7.5mm. Therefore, the preliminary dormancy test without ethylene treatment and the recruitment of the evaluation on two quantitative traits improved the accuracy of BSA (Bulked-segregant analysis)-seq.

3.2. Quality Control of Whole-Genome Sequencing Data

Genomic DNA from bulk1, bulk2, and the two parents were subjected to whole-genome sequencing using MGI-DNBSEQ-T10 platform (BGI, Shenzhen, China). The generated raw data was firstly removed from the adaptor sequence. Meanwhile, low-quality reads that had a mass value of Q ≤ 5 and accounted for more than 50% of the entire read were also removed. Thus, quality control of the sequencing data was conducted to retain only the qualified data before performing any further analyses. Following the trimming and filtering, clean data were obtained with an average Q30 of 88.50% (

Table 1. Quality of the sequencing data and mapping data for the bulked samples and parents.

	Clean Base/G	Q30/%	GC Content/%	Mapped Reads	Mapped Rate/%	Coverage/%	Average Depth
bulk1	23.84	87.87	37	154,147,804	99.16	99.69	137
bulk2	19.75	86.45	36	125,074,857	99.42	99.68	113
prt6	24.41	87.23	36	151,180,273	99.75	99.68	152
prt6−r	30.71	92.46	41	124,972,640	63.44	99.69	126

), signifying that the bases were sequenced precisely and hence retained a high quality that allows for the identification of candidate genomic variations between the two bulks. The sequence data were compared with the reference genome of Tair10 to determine the genome coverage and mapping rate. The bulk1 covered a mapping rate of 99.69% for the whole genome with 99.16% mapping rate, while the bulk2 covered 99.68% for the whole genome with 99.42% mapping rate. These values indicated high-quality mapping results and the better homogeneity of the sequencing data with the reference sequence (Table 1).

3.3. Genomic Variation Distribution and Annotation

The software gatk was used for SNPs and insertion/deletion (InDel) calling for the two pools and parents following the alignment of the reads to Tair10 genome. The further annovar algorithm was used to perform annotation and classification of all the variations that were identified. The variation in SNP occurs when a single nucleotide of A, T, C, or G differs between samples or individuals, while InDels are the insertions and deletions of small DNA fragments. In total, 21,185 variants were detected with 15,745 (74.32%) SNPs and 5440 (25.68%) InDels, respectively (Supplementary Dataset S1) (Figure 2). The distribution of the variants tended to be located at the downstream, upstream, and intergenic regions, whereas the variants located at the exon or splicing site only accounted for 4.16%, among which a total of 2125 variants resulted in missense or silent mutation (Supplementary Dataset S1). In addition, the identified base pair substitution patterns indicated that the most common nucleotide change was C:G > T:A, which was consistent with the previous report of the preference of EMS-induced mutation [38].

3.4. BSAseq Based on G’ Value Method and Delta SNP Method

The G’ value is a modified statistic obtained after smoothing the G statistic which was proposed in the BSA studies [39]. We mapped putative genomic regions with a q-value at the false discovery rate of 0.05 for the approach of G’ value. The distribution of the G’ value across five chromosomes revealed that four genomic regions, 17.3–18.2 Mb on chromosome 1, 14.0–14.6 Mb on chromosome 3, 4.8–5.6 Mb, and 12.4–13.2 Mb on chromosome 5, were highly linked to seed sensitivity to ethylene and silique length (Figure 3) (Supplementary Dataset S2). In addition, among them, one peak on chromosome 5, spanning from 4.8 to 5.6 Mb, exhibited the highest mean G′ value, indicating that it was major QTL with the application of the G’ statistic (Figure 3). Based on the approach of sliding window of the whole genome, we calculated SNP index, followed by plotting the △SNP index for all five chromosomes according to the QTLseq pipeline [34] (Figure 4). The SNP index represents frequencies of parental alleles in the population of bulked individuals. In this case, the Tair10 genome was used as the reference, where an SNP-index of 1 indicates that reads in the population are derived from the EMS mutant allele prt6−r genome, whereas SNP-index = 0 indicates the reads are derived only from original prt6 genome. It was shown that the SNP index of bulk2 from one region of chromosome 5 was close to 1, while the SNP index of bulk1 from this region was much lower than the remaining regions, which resulted in a significant candidate region at 4.1–7.8 Mb on chromosome 5 with the △SNP index above 99% confidence interval (Figure 4, Supplementary Figure S1). The finding here coincided with one of the regions using the G’ statistic method (Figure 3 and Figure 4). Likewise, a third method, Mutmap was also recruited, which used the sequence of the original cultivar to polarize the site frequencies of neighboring markers with an unexpected site frequency. It confirmed again that the region (4.3–5.5 Mb) on chromosome 5 harbored the gene responsible for the mutant phenotype (Supplementary Figure S2).

3.5. Annotation of Candidate Variants and Candidate Gene Confirmation

Subsequent variants annotation at 4.1–7.8 Mb on the chromosome 3 were classified according to gene distance and location such as downstream, upstream, exonic, intronic, and intergenic, and the variants located in the exonic region were further annotated into stopgain, splicing stie, nonsynonymous, synonymous and so on. After filtering, a total of 12 variants in this region had resulted in higher effect by impact, such as non-synonymous or stopgain (

Table 2. Variant annotation in the candidate region of chromosome 5 with exonic variant function.

Pos	Ref	Alt	Annotation	Annotation	SNP Index bulk1	SNP Index bulk2	deltaSNP	Short_Description
4,251,301	G	A	nonsynonymous	AT5G13280	0.33	0.88	−0.55	Aspartate kinase 1
4,469,182	G	A	Stopgain	AT5G13840	0.43	0.87	−0.43	FIZZY-related 3
4,922,519	G	A	nonsynonymous	AT5G15160	0.36	0.93	−0.57	BANQUO 2
5,070,109	G	A	Stopgain	AT5G15580	0.34	0.88	−0.53	Longifolia1 (LNG1)
5,361,835	G	A	nonsynonymous	AT5G16390	0.42	0.94	−0.52	Chloroplastic acetylcoenzyme A carboxylase 1
6,113,584	G	A	nonsynonymous	AT5G18440	0.36	0.91	−0.55	NULL
6,133,091	G	A	nonsynonymous	AT5G18480	0.41	0.89	−0.48	Plant glycogenin-like starch initiation protein 6
6,362,987	G	A	nonsynonymous	AT5G19040	0.28	0.97	−0.69	Isopentenyltransferase 5
6,467,895	G	A	nonsynonymous	AT5G19230	0.45	0.97	−0.52	Glycoprotein membrane precursor
6,854,241	G	A	nonsynonymous	AT5G20300	0.41	0.92	−0.51	Avirulence induced gene (AIG1) family protein
7,354,380	G	A	nonsynonymous	AT5G22190	0.35	0.94	−0.58	NULL
7,610,318	A	T	nonsynonymous	AT5G22794	0.99	0.99	0.00	NULL

). Based on the gene description of the 12 genes, only one gene AT1G15580 (LNG1), was speculated as a potential candidate considering the reported function in the regulation of cell expansion of several organs [40]. A point C to T mutation in LNG1 on the third exon led to a premature stop (Figure 5A). Moreover, we noticed that the SNP index from bulk2 on the candidate region was close to 1, indicating that most of the sequencing reads generated from the bulk2 harbored the variations; while the SNP index from bulk1 was not near 0, with an average value of 0.43, suggesting that the selection of the extreme individuals in this bulk was slightly biased. Even so, the case above was not that bizarre considering that it was often tricky to phenotype quantitative traits and the phenotypes we investigated in our study, seed dormancy and seed silique length, were prone to being affected by the environment.

In order to verify our speculation, we constructed the knock-out of LNG1 in prt6−1 background by CRISPR-cas9-mediated gene editing technology, generating two alleles with 1 bp insertion and 2 bp deletion in the 3rd exon of LNG1 (Figure 5D) (Supplementary Figures S3 and S4). Further germination assay and silique length measurement confirmed that both two mutants reversed the insensitivity of prt6 to ethylene and shortened the silique length as well (Figure 5B,C).

4. Discussion

Here, we identified a new allele, LNG1, that reversed the insensitivity of prt6−1 to ethylene with the forward genetic method, suggesting that LNG1 acted downstream of both ethylene signaling and the N-degron pathway to regulate seed dormancy. An EMS-mutagenized population was generated in the prt6−1 background, and a stable mutant displaying enhanced germination under ethylene treatment and shortened siliques was isolated. These dual phenotypes enabled quantitative trait analysis. An F2 population derived from this mutant was subjected to BSA-seq, and both the G’ statistic and ΔSNP-index approaches identified a significant interval on chromosome 5. Variant annotation within this interval revealed a C-to-T transition in LNG1 that introduces a premature stop codon. CRISPR/Cas9-mediated complementation of LNG1 fully restored ethylene sensitivity in prt6−1, confirming that the mutation in LNG1 is responsible for the observed phenotypes.

The conventional QTL mapping requires phenotyping and genotyping of individuals from a large population, which is labor-intensive and time-consuming. BSA, an elegant method that has been used by genotyping DNA pools with extreme phenotypes, which overcome the disadvantage of QTL mapping [39,41]. BSAseq reduced the cost, simplified the analytical process without compromising the statistical power. For the purpose of being more efficient and fruitful, comparatively high sequencing depth and coverage are required to distinguish significant SNP-trait associations [39]. In our study, the sequencing data of the bulks were well-qualified with more than 99% coverage and more than 100× depth, which were prerequisites for a productive BSA-seq experiment (Table 1). The calculation and comparison of allele frequencies in the pools allow us to identify genomic regions with frequencies differing significantly between bulks. The key to the BSA approach was that the alleles of a locus controlling the trait would be enriched in either bulk, which was determined by the well-constructed two extreme bulks [41]. In our case, one parent, prt6−1, exhibited insensitivity to ethylene during seed germination and displayed normal silique length, whereas the other parent, prt6−r, showed heightened sensitivity to ethylene and shortened silique length. Though the quantitative traits we investigated here were easily affected, the recruitment of the two traits and the preliminary exclusion of low dormant seeds that germinated at 25 °C without ethylene improved the accuracy of the bulk construction.

Some software packages, such as QTL-seq and QTLseqr have been developed to carry out BSA-seq in the population of F2 populations, recombinant inbred lines (RILs), and backcross populations, whereas the software Mutmap (version 2.3.9) is especially for the F2 progeny from a cross between the original cultivar and the mutant [34,35,42]. In order to localize genomic positions of Arabidopsis genes controlling seed traits including the responsiveness to ethylene in germination and seed silique length, several BSA-seq analysis methods were employed. The G-statistic value for each SNP was calculated through the G-test using both REF and ALT SNPs in each bulk [39]. The higher G-statistic value represented by SNP nearby was linked with the trait. Whatever the G-statistic or ∆SNP index, a sliding window algorithm is integrated for visualization. The calculated statistics for each window were compared to a threshold to determine if the window contains a statistically significant association. Windows with values exceeding the threshold are considered to harbor a trait-associated region [35]. All the BSAseq methods in our study repeatedly localize the significant QTL on the chromosome 5, which confirmed the accuracy of the analysis (Figure 3 and Figure 4, Supplementary Figure S2). In order not to miss any potential variants, the largest genomic region generated by QTLseq with the ∆SNP index was used for further variant annotation. The location of the final validated variant in the LNG1 was also included in the significant regions produced by G’ statistic method and the Mutmap method.

BSA-seq of the F2 generations of prt6 × prt6−r identified that LNG1 was epistatic to PRT6. Considering that LNG1 was not initiated with Met-Cys, we speculated that LNG1 was not able to be degraded by PRT6, at least in its form of pre-pro-protein, hinting that there might exist an unknown bridge between them, the mechanism of which awaited future exploration. It was previously reported that LNG1 was known to regulate cell size in the longitudinal direction, the over-expression of which elongated the size of leaves, petals, and siliques [40]. While in the present study, we proposed that LNG1 also participated in the regulation of seed germination under ethylene treatment. LNG1, as a multifunctional protein is implicated in both developmental patterning (cell elongation) and environmental adaptation (seed dormancy); its precise biochemical function remains a key open question. It is clearly an important integrator of hormonal signals, particularly ethylene, and proteolytic regulation via the N-degron pathway. Future research should focus on (1) identifying the molecular partners and substrates of the LNG1 protein; (2) determining how its activity or expression is regulated by the N-degron pathway and ethylene; (3) investigating whether its role in cell elongation is also connected to hormonal signaling networks.

5. Conclusions and Perspectives

Our finding provides a new response to ethylene and a component epistatic to PRT6 in seed germination control by forward genetic screening and reverse genetic verification. This result lays the foundation for future detailed molecular studies. In our study, we found that PRT6-R (LNG1) positively regulates both seed dormancy (at least in response to ethylene) and silique length in Arabidopsis thaliana. This raises the intriguing possibility that its homologues in other crops may exert similar pleiotropic effects. If this is confirmed, for example, in rapeseed, up-regulating the corresponding homolog via genome editing could simultaneously reduce pre-harvest sprouting and increase crop yield.