Label-Free Quantitative Proteomics Reveal the Involvement of PRT6 in Arabidopsis thaliana Seed Responsiveness to Ethylene

In Arabidopsis thaliana, the breaking of seed dormancy in wild type (Col-0) by ethylene at 100 μL L−1 required at least 30 h application. A mutant of the proteolytic N-degron pathway, lacking the E3 ligase PROTEOLYSIS 6 (PRT6), was investigated for its role in ethylene-triggered changes in proteomes during seed germination. Label-free quantitative proteomics was carried out on dormant wild type Col-0 and prt6 seeds treated with (+) or without (−) ethylene. After 16 h, 1737 proteins were identified, but none was significantly different in protein levels in response to ethylene. After longer ethylene treatment (30 h), 2552 proteins were identified, and 619 Differentially Expressed Proteins (DEPs) had significant differences in protein abundances between ethylene treatments and genotypes. In Col, 587 DEPs were enriched for those involved in signal perception and transduction, reserve mobilization and new material generation, which potentially contributed to seed germination. DEPs up-regulated by ethylene in Col included S-adenosylmethionine synthase 1, methionine adenosyltransferase 3 and ACC oxidase involved in ethylene synthesis and of Pyrabactin Resistance1 acting as an ABA receptor, while DEPs down-regulated by ethylene in Col included aldehyde oxidase 4 involved in ABA synthesis. In contrast, in prt6 seeds, ethylene did not result in strong proteomic changes with only 30 DEPs. Taken together, the present work demonstrates that the proteolytic N-degron pathway is essential for ethylene-mediated reprogramming of seed proteomes during germination.


Introduction
At harvest, seeds of numerous species are considered dormant when they cannot germinate even under apparently favorable germination conditions [1][2][3][4]. In Arabidopsis thaliana, dormant Col-0 seeds have marked difficulty germinating in darkness at temperatures higher than 15 • C [5,6]; however, they are able to germinate at 25 • C under continuous light [7,8]. This dormancy is progressively broken during dry storage and the germination of dormant seeds is promoted after the application of dormancy-breaking agents, such as chilling, gibberellins (GAs), jasmonates, brassinosteroids, reactive oxygen species (ROS) and reactive nitrogen species (RNS) [5,[9][10][11]. The hormonal balance between abscisic acid (ABA) and GAs plays an essential role in the regulation of seed dormancy [11][12][13][14][15]. In addition, it has been demonstrated that the plant hormone, ethylene, could also release seed dormancy of numerous species through crosstalk with ABA and GAs and the ABA/GAs balance [16][17][18]. Interestingly, the actions of these dormancy agents are associated with ethylene production [19], as well as by the activation of their responsive

Effects of Ethylene Treatment on the Germination of Dormant Seeds
At harvest, seeds from both genotypes (Col and prt6) were dormant and did not germinate at 25 • C and in darkness [6]. Seed incubation in the presence of ethylene at 100 µL L −1 promoted the germination of Col seeds, while the prt6 seeds were insensitive to ethylene (Figure 1). Continuously in the presence of this hormone, dormant Col seeds started to germinate after 30 h, and about 55% and 96% of the seed population harvested in 2016 became able to germinate after 48 and 72 h, respectively. Seeds harvested in 2016 germinated better at 25 • C without ethylene (11.9%) than those harvested in 2018 (0%), they could be considered less dormant (Table 1). In addition, Table 1 shows that breaking of dormancy required an application of ethylene for at least 24-30 h. When the batch was deeply dormant like those harvested in 2018, only 7.7% of the seed population can germinate after 16 h of incubation in the presence of ethylene, while a pretreatment for 30 h resulted in 54.6% germination. Except for 48 h-application of ethylene, seeds harvested in 2016 which were less dormant than those harvested in 2018, were more sensitive to ethylene; half of the population was able to germinate in air at 25 • C after a 24 h or 30 h pretreatment with ethylene, respectively, for the batch harvested in 2016 and 2018 (Table 1).
proteomic changes in both genotypes incubated at 25 °C in air and in the presence of ethylene.

Effects of Ethylene Treatment on the Germination of Dormant Seeds
At harvest, seeds from both genotypes (Col and prt6) were dormant and did not germinate at 25 °C and in darkness 6. Seed incubation in the presence of ethylene at 100 μL L −1 promoted the germination of Col seeds, while the prt6 seeds were insensitive to ethylene (Figure 1). Continuously in the presence of this hormone, dormant Col seeds started to germinate after 30 h, and about 55% and 96% of the seed population harvested in 2016 became able to germinate after 48 and 72 h, respectively. Seeds harvested in 2016 germinated better at 25 °C without ethylene (11.9%) than those harvested in 2018 (0%), they could be considered less dormant (Table 1). In addition, Table 1 shows that breaking of dormancy required an application of ethylene for at least 24-30 h. When the batch was deeply dormant like those harvested in 2018, only 7.7% of the seed population can germinate after 16 h of incubation in the presence of ethylene, while a pretreatment for 30 h resulted in 54.6% germination. Except for 48 h-application of ethylene, seeds harvested in 2016 which were less dormant than those harvested in 2018, were more sensitive to ethylene; half of the population was able to germinate in air at 25 °C after a 24 h or 30 h pretreatment with ethylene, respectively, for the batch harvested in 2016 and 2018 (Table 1).

Ethylene-Induced Proteome Changes Observed at 30 h but Not 16 h after Treatments
To identify the global proteomic changes in prt6 and Col seeds in response to ethylene, proteome data were collected from dry seeds and imbibed seeds for 16 h and 30 h in the presence or absence of ethylene. A total of 1737 proteins were detected across samples from dry and 16 h-imbibed seeds and 1002 of them were selected for further quantitative analysis by integration of the extracted ion current (XIC) (Supplementary dataset S1, sheets 1 and 2). Two-way ANOVA analysis showed that no differentially expressed proteins (DEPs) were present in these proteins (adjusted p value < 0.05) (Figure 2A). In 30 h-imbibed seeds, 2552 proteins were identified and 2202 proteins were selected for further analysis (Supplementary dataset S1, sheets 3 and 4). Except for 38 out of 1002 quantified proteins that were only present in the 16 h-proteome, all the remaining identified proteins were also present in the 30 h-proteome (Supplementary dataset S1, sheet 5). After 30 h, 619 proteins showed a significant interaction between the genotype and the ethylene treatment, indicating that the two genotypes responded differently to ethylene (adjusted p value < 0.05) ( Figure 2B) (Supplementary dataset S1, sheets 6, 7 and 8). Scatter plots ( Figure 2C,D) illustrated that ethylene did not significantly affect the majority of proteins of both genotypes in 16 h-imbibed seeds ( Figure 2C). However, in the 30 h-imbibed seeds, it is obvious that ethylene affected the abundance of a large number of proteins in Col seeds, while ethylene had only a limited effect on the seed proteome of the prt6 mutant ( Figure 2D). Protein expression levels for each replicate were assessed using PCA analysis ( Figure 2E,F). In the "dry" and "16 h" proteomes, we Scatter plots ( Figure 2C,D) illustrated that ethylene did not significantly affect the majority of proteins of both genotypes in 16 h-imbibed seeds ( Figure 2C). However, in the 30 h-imbibed seeds, it is obvious that ethylene affected the abundance of a large number of proteins in Col seeds, while ethylene had only a limited effect on the seed proteome of the prt6 mutant ( Figure 2D). Protein expression levels for each replicate were assessed using PCA analysis ( Figure 2E,F). In the "dry" and "16 h" proteomes, we noticed that prt6 was more predominantly located at the positive side of PC1 (13% of the variation) while Col was at the negative side of PC1. Furthermore, PC2 was able to separate dry seeds from 16 h-imbibed seeds, but the four conditions of seed imbibition could not be separated as shown by the clustered ellipses ( Figure 2E). However, after 30 h of imbibition, each condition was clearly separated in the plot ( Figure 2F). It was also shown that the PC1 from 30 h-imbibed seed proteome was 46.8% that was higher than 13.3% from "dry" and "16 h" proteomes.

Effects of Ethylene on the Expression of DEPs
As shown in Figure 2B, 619 proteins showed genotype/ethylene interaction. Quantification of these DEPs on the basis of log XIC values in the four genotype/treatment combinations (Col+, Col−, prt6+, prt6−) was used to present expression patterns by a hierarchical clustering heatmap ( Figure 3A). It was revealed that Col+ clustered separately from others. The sample clustering fits nicely with the observation that Col seeds, but not others, could germinate in the presence of ethylene. On the other hand, the color of each column indicated that ethylene had a more pronounced effect in Col than in prt6. Tukey analysis (p value < 0.05) and fold change filtering (fold change >1.3 or <0.7) were then employed to select DEPs that showed a significant effect of ethylene in Col and in prt6 (Supplementary dataset S2). In Col+/Col−, 311 down-regulated and 276 up-regulated DEPs were identified, while we identified only 15 down-regulated and 15 up-regulated DEPs in prt6+/prt6− ( Figure 3B,D). In addition, 23 down-regulated and 77 up-regulated DEPs show genotype-dependent differences in protein abundances in the absence of ethylene ( Figure 3C) (Table S1).
A significant number of DEPs from the comparison groups, Col+/Col− and prt6+/prt6−, were found in the SeedNet, a transcriptional topological model [45] ( Figure 4A,B). It was revealed that 234 out of 311 down-regulated DEPs and 234 out of 276 up-regulated DEPs in Col+/ Col− were included in the SeedNet model (Supplementary dataset S2, sheet 1). As shown in Figure 4A, down-regulated DEPs were predominantly located within region 1 where genes were associated with dormancy. Conversely, upregulated DEPs were primarily presented within region 3 where genes were associated with germination. These results indicated that the regulation of the protein abundances by ethylene in wild-type (Col) seeds were consistent with the general transcriptional regulation related to seed germination. In parallel, 10 out of 15 down-regulated DEPs and 13 out of 15 up-regulated DEPs in prt6+/prt6− were also found in the SeedNet ( Figure 4B) (Supplementary dataset S2, sheet 2). As shown in Figure 4B, although the numbers were small, up-regulated DEPs were located within regions 2 and 3, while down-regulated DEPs were in regions 1 and 3. The Venn diagram showed 29 DEPs belonging to both groups Col+/Col− and prt6+/prt6−, which only PHYTOCHROME A (AT1G09570) decreased in prt6 after ethylene treatment ( Figure 4C). The relative abundance of the 29 common DEPs were used to present expression patterns by a hierarchical clustering heatmap ( Figure 4D). Four clusters were divided according to the expression profiles. Cluster I illustrated the reduction in protein expression in Col and prt6 in the presence of ethylene. In contrast, Cluster IV showed that ethylene enhanced the protein expressions in both genotypes, even the reduction and induction were much more pronounced in Col than that in prt6. Cluster II included two proteins SAD6 (stearoyl-acyl carrier protein 9-desaturase, AT1G43800) and an adenine nucleotide alpha hydrolases-like superfamily protein (AT3G03270), both of which showed more protein in prt6− than in Col− in the absence of ethylene and both were up-regulated by ethylene. Cluster III presented the opposite changes in which ethylene increases the expressions of the five proteins LDAP2 (lipid droplet associated protein 2, AT2G47780), EXP2 (expansin 2, AT5G05290), GLL22 (GDSL lipase-like protein 22, AT1G54000), JLL30 (jacalin-related lectin, AT3G16420) and PYK10 (beta-glucosidase, AT3G09260) in Col, while the five proteins were down-regulated in prt6 with ethylene treatment. We noticed that the five proteins were also increased in prt6 in the absence of ethylene in comparison with Col.
Quantification of these DEPs on the basis of log XIC values in the four genotype/treatment combinations (Col+, Col−, prt6+, prt6−) was used to present expression patterns by a hierarchical clustering heatmap ( Figure 3A). It was revealed that Col+ clustered separately from others. The sample clustering fits nicely with the observation that Col seeds, but not others, could germinate in the presence of ethylene. On the other hand, the color of each column indicated that ethylene had a more pronounced effect in Col than in prt6. Tukey analysis (p value < 0.05) and fold change filtering (fold change >1.3 or <0.7) were then employed to select DEPs that showed a significant effect of ethylene in Col and in prt6 (Supplementary dataset S2). In Col+/Col−, 311 down-regulated and 276 up-regulated DEPs were identified, while we identified only 15 down-regulated and 15 up-regulated DEPs in prt6+/prt6− ( Figure 3B,D). In addition, 23 down-regulated and 77 up-regulated DEPs show genotype-dependent differences in protein abundances in the absence of ethylene ( Figure 3C) (Table S1). . X-axis and Y-axis represents the log 2 (fold ratio) and −log 10 (p), respectively. Red dots indicate that the proteins have significant differences (p < 0.05, and fold change < 0.7 or fold change > 1.3), but the threshold . X-axis and Y-axis represents the log 2 (fold ratio) and −log 10 (p), respectively. Red dots indicate that the proteins have significant differences (p < 0.05, and fold change < 0.7 or fold change > 1.3), but the threshold of fold change was shown as |log2 (fold ratio)| > 0.4 in the plot), while the black dots correspond to proteins with no significant differences. The horizontal dashed line represents p = 0.05, namely, −log 10 (p) = 1.3. The two vertical dashed lines correspond to log 2 (fold change) = 0.4. Seeds harvested in 2016.
GO enrichment analysis was performed to interpret the functional modules of DEPs. Down-and up-regulated DEPs in Col+/ Col− group were analyzed separately. As shown in Figure 5 [46,47], the network of biological process enrichment analysis indicated that down-regulated DEPs were enriched in the seed development and dormancy, secondary metabolic process, glutathione metabolism, various responsive processes, such as the response to ABA, lipid, water, light, heat and H 2 O 2 . While up-regulated DEPs were enriched in the amino acids biosynthetic process, gene expression and protein metabolism, RNA modification, lipid catabolic process and also some responsive processes, such as response to ABA, alcohol and H2O2. To sum up, the signal perception and transduction from down and up-regulated proteins, the energy mobilization from up-regulated proteins, new material generation from up-regulated proteins and some other assisting processes work together to contribute to seed germination of Col seeds. GO enrichment analysis was performed to interpret the functional modules of DEPs. Down-and up-regulated DEPs in Col+/ Col− group were analyzed separately. As shown in Figure 5 [46,47], the network of biological process enrichment analysis indicated that down-regulated DEPs were enriched in the seed development and dormancy, secondary metabolic process, glutathione metabolism, various responsive processes, such as the response to ABA, lipid, water, light, heat and H2O2. While up-regulated DEPs were enriched in the amino acids biosynthetic process, gene expression and protein metabolism, RNA modification, lipid catabolic process and also some responsive processes, such as response to ABA, alcohol and H2O2. To sum up, the signal perception and transduction from down and up-regulated proteins, the energy mobilization from up-regulated proteins, new material generation from up-regul proteins and some other assisting processes work together to contribute to s germination of Col seeds.

Effect of prt6 on the Expression of DEPs in the Absence of Ethylene after 30 h of Imbibiti
As shown in Table S1, 100 DEPs were selected in the group prt6−/Col−. This gr was selected in order to determine the impact of the mutation of prt6 in seed dorma and proteome, independently of the regulation by ethylene. Table 2 shows, except fo proteins which were miscellaneous, classified DEPs from this group. Three prot encoded by "core hypoxia" genes characterized by transcriptionally up-regulatio hypoxia conditions [44], increased in prt6 mutant in the absence of ethylene for 3 (prt6−). Among them, the protein level of HB1 (hemoglobin 1, AT2G16060) and SA (stearoyl-acyl-carrier-protein desaturase family, AT1G43800) were around 20 ti higher in prt6− than in Col−. In addition, we noticed that several proteins involve nucleobase synthesis, transcription, translation, protein modification and transport w moderately increased in prt6−, most of which showed an increase of less than two tim Three seed storage proteins LEAs were found with less abundance in prt6− compare Col−. Other proteins involved in ABA synthesis, such as AO4 (aldehyde oxidas AT1G04580) was in reduced abundance in prt6, when proteins involved in methio metabolism, such as MTO3 (S-adenosylmethionine synthetase, AT3G17390) and M (methionine adenosyltransferase 3, AT2G36880) had higher abundance in prt6.

Effect of prt6 on the Expression of DEPs in the Absence of Ethylene after 30 h of Imbibition
As shown in Table S1, 100 DEPs were selected in the group prt6−/Col−. This group was selected in order to determine the impact of the mutation of prt6 in seed dormancy and proteome, independently of the regulation by ethylene. Table 2 shows, except for 19 proteins which were miscellaneous, classified DEPs from this group. Three proteins encoded by "core hypoxia" genes characterized by transcriptionally up-regulation in hypoxia conditions [44], increased in prt6 mutant in the absence of ethylene for 30 h (prt6−). Among them, the protein level of HB1 (hemoglobin 1, AT2G16060) and SAD6 (stearoyl-acylcarrier-protein desaturase family, AT1G43800) were around 20 times higher in prt6− than in Col−. In addition, we noticed that several proteins involved in nucleobase synthesis, transcription, translation, protein modification and transport were moderately increased in prt6−, most of which showed an increase of less than two times. Three seed storage proteins LEAs were found with less abundance in prt6− compared to Col−. Other proteins involved in ABA synthesis, such as AO4 (aldehyde oxidase 4, AT1G04580) was in reduced abundance in prt6, when proteins involved in methionine metabolism, such as MTO3 (Sadenosylmethionine synthetase, AT3G17390) and MAT3 (methionine adenosyltransferase 3, AT2G36880) had higher abundance in prt6.

Functional Class Scoring Analysis without Filtering DEPs in 30 h-Proteome
Because of the low variance in the comparison group, especially in prt6+/prt6−, fewer DEPs were achieved after cut-off selection, so over-representation analysis failed to give us an informative result. GSEA (gene set enrichment analysis) [48] and Pathifier [49] were then used to overcome this limitation. The subtle but coordinated changes in the group prt6+/prt6− were captured by GSEA analysis (Figure 6). Most of the up-regulated BP (Biological Process) terms were enriched in protein synthesis, transport, localization and modification. The most significantly enriched BP term represented by the highest count was "peptide biosynthetic process". Some BP terms related to tissue development or morphogenesis (class 2, Figure 6) were improved in the presence of ethylene, but at the same time, BP terms involved in development growth (class 5, Figure 6) were inhibited in the presence of ethylene. These results suggested that BP terms in class 5 might partially lead to the dormancy of prt6 when treated with ethylene. In parallel, GSEA analysis was also carried out in Col+/Col− group to compare the different effects of ethylene between prt6 and Col ( Figure S1A). It seems that lipid transport and localization, and response to hydrogen peroxide were down-regulated in both groups. However, some BP terms, such as fatty acid oxidation, lipid biosynthetic process, mitochondrial transport, hormone biosynthetic process coenzyme biosynthetic process, were specific in the Col+/Col−, which contributed to the diversity and complexity of the affected BP categories. It was shown that even without ethylene treatment, the genotype effect had revealed some distinct BP terms ( Figure S1B). We can also notice that the term, lipid localization, was down-regulated in both Col+/Col− and prt6−/Col− groups ( Figure S1). also notice that the term, lipid localization, was down-regulated in both Col+/C prt6−/Col− groups ( Figure S1).
As shown in Figures 6 and S1A, the enrichment of lipid localization genotypes could not distinguish which one (Col or prt6) was more affected by e Therefore, Pathifier analysis was chosen to give us a quantified deregulation le specific biological process pathway. At the same time, some categories that are k be associated with seed germination or some related to the function of PRT6 w included, such as hormone regulation, protein metabolism (folding, local molecule transport, development growth and response to hypoxia were sel perform Pathifier analysis with Col− acting as a reference. Figure 6. A dot plot of enriched BP terms in prt6+/prt6− determined from GSEA results. T the dot represents count of the core proteins in each set, the color represents p va GeneRatio represents the number of core genes in gene set/the total number of genes in set. The enriched BP terms were classified into several groups depending on functional co and similarity, that is, class 1, response to stimuli; class 2, plant morphogenesis; class Figure 6. A dot plot of enriched BP terms in prt6+/prt6− determined from GSEA results. The size of the dot represents count of the core proteins in each set, the color represents p value, and GeneRatio represents the number of core genes in gene set/the total number of genes in this gene set. The enriched BP terms were classified into several groups depending on functional connection and similarity, that is, class 1, response to stimuli; class 2, plant morphogenesis; class 3, energy metabolism; class 4, protein synthesis, localization and modification; class 5, development growth; class 6, response to stimuli; class 7, anion and Golgi transport; class 8, lipid transport and localization. As shown in Figure 6 and Figure S1A, the enrichment of lipid localization in both genotypes could not distinguish which one (Col or prt6) was more affected by ethylene. Therefore, Pathifier analysis was chosen to give us a quantified deregulation level of a specific biological process pathway. At the same time, some categories that are known to be associated with seed germination or some related to the function of PRT6 were also included, such as hormone regulation, protein metabolism (folding, localization), molecule transport, development growth and response to hypoxia were selected to perform Pathifier analysis with Col− acting as a reference.
As shown in Figure 7A, Pathifier successfully discriminated between dormant and non-dormant seeds. Among the four samples, only Col+ possessed germination potential, and PDS values of each BP category were particularly higher than that in the other three samples. Secondly, some biological processes were deregulated between Col− and prt6−, especially the BP term, response to hypoxia, which was much more affected than others. In addition, with the treatment of ethylene, response to H 2 O 2 , response to stress were also deregulated in prt6 to some degree, but the inductive effect could not be compared with the changes observed in Col in the presence of ethylene ( Figure 7B).

localization.
As shown in Figure 7A, Pathifier successfully discriminated between dormant and non-dormant seeds. Among the four samples, only Col+ possessed germination potential, and PDS values of each BP category were particularly higher than that in the other three samples. Secondly, some biological processes were deregulated between Col− and prt6−, especially the BP term, response to hypoxia, which was much more affected than others. In addition, with the treatment of ethylene, response to H2O2, response to stress were also deregulated in prt6 to some degree, but the inductive effect could not be compared with the changes observed in Col in the presence of ethylene ( Figure 7B).

Effect of prt6 on the Fate of Storage Proteins during Germination
Abundant proteins in seeds such as seed storage proteins (SSPs), late embryogenesis abundant proteins (LEAs) and oleosins were selected to compare their changes in abundance as related to ethylene treatment or the genotype. In 30 h-proteome, 47 abundant proteins were identified while in the dry and 16 h-proteome, the quantity was only 36, all of which were present after 30 h-proteome (Supplementary dataset S3). The common 11 SSPs in dry seeds, 16 h-and 30 h-imbibed seeds, and their abundance ratios (prt6−/Col−) ( Table 3) showed that cruciferin A (AT5G44120.3) and 2S albumins (AT4G30880) were most significantly up-regulated in the prt6 in dry seeds when compared with Col dry seeds, suggesting that PRT6 was involved in the stabilization of some SSPs during seed development. After 16 h of imbibition, levels of all SSPs were slightly higher in prt6, but after 30 h, most were not altered in their protein abundance, indicating that during seed imbibition, the N-end rule affected the mobilization of SSPs.  In the heatmap, blue to red represents the PDS score from low to high.

Effect of prt6 on the Fate of Storage Proteins during Germination
Abundant proteins in seeds such as seed storage proteins (SSPs), late embryogenesis abundant proteins (LEAs) and oleosins were selected to compare their changes in abundance as related to ethylene treatment or the genotype. In 30 h-proteome, 47 abundant proteins were identified while in the dry and 16 h-proteome, the quantity was only 36, all of which were present after 30 h-proteome (Supplementary dataset S3). The common 11 SSPs in dry seeds, 16 h-and 30 h-imbibed seeds, and their abundance ratios (prt6−/Col−) ( Table 3) showed that cruciferin A (AT5G44120.3) and 2S albumins (AT4G30880) were most significantly up-regulated in the prt6 in dry seeds when compared with Col dry seeds, suggesting that PRT6 was involved in the stabilization of some SSPs during seed development. After 16 h of imbibition, levels of all SSPs were slightly higher in prt6, but after 30 h, most were not altered in their protein abundance, indicating that during seed imbibition, the N-end rule affected the mobilization of SSPs. In 30 h-proteome, the effect of ethylene was also compared between wild type and prt6 mutant. It was shown that overall ratios of Col+/Col− were much lower than that of prt6+/prt6− (Supplementary dataset S3, sheet 2), indicating that the prt6 mutation probably inhibited ethylene-mediated degradation of SSPs, LEA and oleosins.

Hormone-Related Proteins in Seed
Proteins participating in hormone synthesis or hormone signaling were obtained from proteomic data. Four proteins involved in the ethylene synthesis were identified: MTO3 (S-Adenosylmethionine Synthase 2, AT3G17390), MAT3 (Methionine Adenosyltransferase 3, AT2G36880) and SAM (S-Adenosylmethionine Synthetase 1, AT1G02500) which catalyzed the formation of S-adenosylmethionine, and ACO (ACC oxidase 1, AT2G19590) which oxidized ACC in ethylene biosynthesis.
As shown in Figure 8, ethylene significantly resulted in an increase in protein level of MTO3 and MAT3 in Col. The abundances of ACO and SAM1 also increased to some degree in Col (although not significantly according to ANOVA analysis). However, ethylene did not markedly affect the four proteins in prt6. AO4 (Aldehyde Oxidase 4) catalyzed the final steps of carotenoid catabolism, while PYR (Pyrabactin Resistance) acted as one of the ABA receptors [50,51]. In Col, ethylene dramatically increased the abundance of PYR and decreased the abundance of AO4. The decrease in AO4 was consistent with the decrease in endogenous ABA in Col seeds when treated with ethylene [21]. On the other hand, the increased abundance of PYR was not expected, since it is involved in seed dormancy [52]. No significant difference in protein abundance was observed between prt6− and prt6+ for AO4 and a slight increase was observed for PYR.

Discussion
In the present work, a proteome analysis was performed to investigate the role of PRT6 in the ethylene response during seed germination. We demonstrated that only the application of ethylene longer than 30 h was able to break the dormancy of Arabidopsis

Discussion
In the present work, a proteome analysis was performed to investigate the role of PRT6 in the ethylene response during seed germination. We demonstrated that only the application of ethylene longer than 30 h was able to break the dormancy of Arabidopsis seeds placed at 25 • C (Table 1). Longer imbibition durations (30 h) also allowed us to observe the interaction effects between ethylene treatment and genotype (Figure 2). We aimed to find some genes that could rescue the insensitivity of prt6 seeds to ethylene, so the five genes in cluster III ( Figure 4D) were selected. However, germination tests showed that the five mutants germinated as well as Col in the presence of ethylene (data not shown), suggesting that one single gene was not sufficient to affect seed responsiveness to ethylene, or there might be due to functional redundancy.
The proteome data were analyzed using two classes of bioinformatics tools. Firstly, GO over-representation analysis was carried out using the defining DEPs. However, overrepresentation analysis could not be powerful when we identified only a small number of DEPs, like in our comparison group prt6+/prt6−. The second class of tools such as GSEA acted as a complementary method that could detect small but consistent changes in a set and then yielded the affected categories by using two enrichment analyses [53,54]. Furthermore, Pathifier belonging to the second class was also employed to improve the depth and breadth of our study.
Compared to leaf proteome or entire plant proteome, seed proteome is characteristic to over-represent the general categories such as SSPs, cell rescue, cell defense and protein fate [55]. For this reason, several selected GO terms such as response to stress, protein folding, protein transport and protein metabolism were included in the Pathifier analysis (Figure 7). At the same time, SSP analysis was launched by filtering all these proteins to compare their exact values (Table 3 and Supplementary dataset S3). One BP term, response to hypoxia, was also subjected to the Pathifier analysis, given that the effect of the N-end rule coordinated with oxygen sensing in the regulation of seedling survival or root growth under hypoxia [34,35,[56][57][58][59][60]. This would fit nicely with our result that the PDS score of response to hypoxia is significantly different between prt6− and Col−, meaning that many proteins involved in this process were affected in prt6 mutant seeds ( Figure 7B). Pathifier analysis evaluated the deregulation level of a specific pathway in the tested condition compared with the control, but it failed to indicate the direction of this change in comparison with control, that is, up-regulation or down-regulation. However, GSEA analysis was specialized to access the differential expression of gene sets for core genes in significant functional categories from a ranked list, which could meet the requirement to a certain degree ( Figure 6 and Figure S1). For example, the GSEA visualization plot of the BP category, lipid localization in the group prt6−/Col− showed that the leading edge gene set presented at the bottom of the ranked list. This suggested that this category was mostly suppressed in the prt6− mutant in the absence of ethylene ( Figure S2). As shown in Table  S2, all the proteins involved in lipid localization in 30 h-proteome were collected, which consisted of different classes of proteins, such as oleosins, lipid transfer proteins or some other binding proteins. It was shown that the reduced protein levels predominated through the list (Table S2). Considering that more than half of the proteins here were storage proteins or oleosins, it seemed logical that they are degraded and serve as energy support. In this case, many more carrier proteins or transfer proteins would be required to transport the "cargo", but in Table S2, there are few aiding proteins (AT3G05420, AT5G42890) with a ratio higher than 1. The predominance of the down-regulated proteins in this category might result from the lower abundance of transfer proteins. In the over-representation analysis of DEPs in Col+/Col−, we also found some BP terms concerning lipid metabolism, but which were more related to lipid catabolic process and fatty acid beta-oxidation ( Figure 4). It is also true for oleosins and oil body-associated proteins, most of which ratios in Col+/Col− were generally lower than that in prt6+/prt6− (Table S2 and Supplementary dataset S3, sheet 2). This indicated that in the presence of ethylene, the PRT6 mutation impaired the oil body metabolism, as the same phenotype was observed in the mutant seedlings [40,61].
SSPs accumulated during seed development could supply energy for the early stage of seed germination. SSPs are the most abundant proteins in the total seed proteome and the presence of extremely abundant SSPs decreases the sensitivity to detect other low abundant proteins in seeds. In our study, we demonstrated that the N-end rule pathway could regulate the accumulation or stabilization of SSPs during seed development or seed imbibition (Table 3). It is possible that the insensitivity of prt6 seeds to ethylene is related to the inhibited SSPs mobilization after 30 h (Supplementary dataset S3, sheet 2). It was reported that the abundance of SSPs was increased, especially cruciferins in prt6 etiolated seedlings (4 d under green light) [43]. Whereas, Choy et al. (2008) [32] reported that transcript abundance of genes encoding SSPs, oleosins, and LEA proteins, were much lower in the ged1 (an allele of prt6) seedlings (5 d in the dark followed by 2 d in the light). The opposite changes of SSPs at the transcript level to the protein level might be due to the different conditions of light in the two studies, as it was reported that the N-end rule coordinated with light perception to regulate seedling establishment after germination [62].
It was well documented that seed germination and dormancy were regulated via crosstalk of ABA, Gas and ethylene, whose hormone synthesis and signaling pathways could interact with each other to control seed dormancy [2,3,17,18,63]. The breaking of seed dormancy by exogenous ethylene was associated with activation of its own responsive genes as well as fine-tuning the hormonal balance between ABA and GA [6,16,18,20,21]. Here, the Pathifier analysis was carried out to calculate the deregulation score of seed sensitivity to hormones ( Figure 7A and Figure S3). The PDS scores of Col+ were significantly higher than that of Col− ( Figure S3). In contrast, with the treatment of ethylene, the PDS scores of response to the three hormones did not show a dramatic increase ( Figure S3). Particularly, almost no difference in PDS score of response to ethylene between prt6− and prt6+, indicating that the related proteins in this category were reluctant to alter in the presence of ethylene. Proteins related to ethylene synthesis such as MTO3, MAT3, ACO and SAM, were enhanced in Col in the presence of exogenous ethylene, which was in agreement with Liu et al. [64] who demonstrate that exogenous ethylene induces endogenous ethylene biosynthesis ( Figure 8). Furthermore, there is a negative interaction between ethylene and ABA in seed germination [17,18,21]. In Col, exogenous ethylene decreased ABA level [21] and the protein abundance of AO4, which was involved in ABA synthesis (Figure 8). In addition, the protein levels of PYR were stimulated by ethylene in Col, but considering that ethylene could suppress ABA signaling [65], we speculated that the higher abundance of PYR might be inactivated [66], or a feedback response due to the lower production of ABA in the presence of ethylene.
The N-end rule pathway target proteins to be degraded through proteasome at the N-terminal residues [22]. Considering the diversity of mechanisms of neo-N-termini generation, the targets of the N-end rule pathway were only achieved in the pre-pro-protein, while in plants, they are ERFVII proteins, all of which were starting with Met-Cys [34]. In our study, 100 proteins were significantly accumulated in prt6 mutant at 30 h ( Table 2, Supplementary dataset S3). All amino acid sequences of the 100 proteins were retrieved from the database to identify new putative substrates among Nt Met-Cys proteins. It showed that only one protein ASN3 (Asparagine Synthetase 3, AT5G10240) was a potential target, which was also suggested by Majovsky et al. (2014) [41] with the increased abundance in both ate1 ate2 seedlings and prt6 seedlings. To date, several transcriptomic or proteomic analyses of the effect of the N-end rule pathway by using the prt6 and ate1 ate2 mutants have been published [32,34,[41][42][43]. As shown in the Supplementary Dataset S4, there were few proteins overlapped, suggesting the difference in protein abundance between seedlings and seeds. Though Gibbs et al. (2011) [34] included imbibed prt6 seeds, the different imbibition duration and different statistical methods might also contribute to identifying fewer DEPs. It is worth noting that AT2G16060 (Hemoglobin 1) is represented across all the published data.
Here, the prt6 mutation was connected with ethylene signaling at the level of proteomes, which was a novel aspect further studying the effect of the N-end rule pathway in seed dormancy. In a future study, more attention could be paid to the transcriptomic changes of prt6 imbibed seeds and the involvement of ERFVIIs in the insensitivity of prt6 to ethylene.

Seed Materials
Arabidopsis thaliana seeds from Columbia-0 (Col-0) were used as the wild type of this study. Mutant prt6 (SAIL-1278-H11) was in the genetic background of Col-0 and obtained as a gift from Dr. M. J. Holdsworth (Nottingham University). For seed propagation, seeds were incubated in water at 4 • C for 3 days then transferred into the potting mixture (Jiffy ® substrates, Tref, Paris, France) and placed in a growth chamber at 21 • C under a photoperiod of 16 h light/8 h dark. Siliques were harvested at maturity in 2016 and 2018, and the seeds collected were stored at −20 • C until experiments.

Germination Assay
Germination assays were performed in darkness in 9-cm Petri dishes with 100 to 200 seeds per assay in 3 replicates. Seeds were placed on a filter paper laid on top of cotton wool moistened with deionized water at 25 • C. Germination assays in the presence of gaseous ethylene were carried out in tightly closed 360 mL containers, in which was injected gaseous ethylene (5%) (Air Liquide, Paris, France) to set the concentration of ethylene at 100 µL L −1 . A seed was considered to be germinated as soon as the radicle protruded through the seed coat. Germination counts were made every 24 h for 7 days, and the results presented are the means of the germination percentages obtained with 3 replicates ± SD.

Seed Treatment
Col and prt6 seeds were imbibed on a filter paper placed on the top of cotton wool moistened with deionized water in tightly closed 360 mL containers (50 mg per assay in 3 replicates) in darkness. Seeds were incubated with (+) or without ethylene (−) 100 µL L −1 for 16 h and 30 h. They were then frozen in liquid nitrogen and stored at −80 • C. Dry seeds from both genotypes were also analyzed.

Protein Extraction and LC-MS Analysis
The total protein was extracted by a modified phenol extraction method as described by Desjardin et al. (2012) [67]. The protein concentration was determined using 2D-QUANT kit (Amersham Biosciences, Little Chalfont, UK), then it was digested by trypsin (Promega, Madison, WI, United States). Desalting and concentration of the peptides were implemented before the operation of HPLC by NanoLC-Ultra (Eksigent, Dublin, OH, USA) coupled with an MS analysis by Qexactive PLUS (Thermo Fisher, Waltham, MA, USA). LC-MS analyses were performed on 3 independent batches: dry seeds, 16h-imbibed seeds and 30 h-imbibed seeds.

Protein Identification and Quantification
Arabidopsis FASTA format database was downloaded from TAIR (https://www. Arabidopsis.org/, accessed on 1 January 2019) and database searches were performed using the X!Tandem search engine [68] and protein inference and post-treatment filtering was performed by using X!TandemPipeline 02.28 [69]. Data filtering was achieved according to a peptide E-value < 0.01, protein log (E-value) < −4 and a minimum of two identified peptides per protein. Peptides were quantified based on eXtracted Ion Current (XIC) of their peptides using MassChroQ 2.2.12 Black Caïman [70]. Relative quantification of the proteins was calculated using R scripts by analyzing peptide intensities obtained from XIC. Firstly, reliable peptides were selected according to the criteria described in Belouah et al. (2018) [71], i.e., shared peptides, uncorrelated peptides, peptides showing retention time instability, and peptides showing more than 3 missing data out of 12 were filtered out. Only proteins represented by at least two reliable peptides were used for analysis. Protein relative abundances were computed as the sum of the peptide intensities. Protein abundance changes were detected separately for 16 h-and 30 h-imbibed seeds by a 2-way ANOVA (ethylene treatment and genotype) with interaction. The effect of dry seeds was estimated in comparison to 16 h-imbibed seeds in the absence of ethylene treatment.
Proteins showing a p value corrected for multiple comparisons < 0.05 were considered as showing a significant variation. Among them, the down and up-regulated proteins were then defined as those that significantly changed after Tukey test with p value < 0.05 and fold-change < 0.7 or > 1.3.

Bioinformatic Analysis
Principal component analysis was conducted by the R package factoextra [72]. Heatmap clustering was drawn by R package pheatmap [73] with values being centered in the row direction and clustering distance method being chosen as "Euclidean". Volcano plot was finished by EnhancedVolcano [74] with the cutoff of log 2 (fold change) at 0.4. Down-regulated and up-regulated proteins in the comparison group were input into the SeedNet [45], in the platform of Cytoscape to capture the overlapping proteins. Gene ontology (GO) analysis was launched by the R package clusterProfiler [75]. Biological Process (BP) terms of Arabidopsis were downloaded from the new release (TAIR Data 20180331) of GO annotation (https://www.Arabidopsis.org/index. Jsp, accessed on 1 January 2019) and automatically generated annotations, that is, Inferred from Electronic Annotation (IEA) were removed to improve the credibility of overrepresentation analysis. The redundant BP terms were discarded and by using the function "simplify" in R package clusterProfiler [75] to keep the term with a lower P value among all the similar terms. The connected BP terms were then input into the networkD3 [47] to present the functional module. Gene Set Enrichment Analysis (GSEA) of the comparison group was carried out by analyzing the rank-ordered protein lists based on abundance ratios Col+/Col−, prt6+/prt6− and prt6−/Col− on all quantified proteins, significant BP terms were kept after p value (< 0.05) selection and discarding the redundant BP terms [75]. Pathifier pipeline was applied to determine the extent to which individual BP terms are deregulated in every individual condition [49]. Before Pathifier analysis, the selected BP terms extracted from GO annotation data were mapped to ancestor BP terms by using the function "buildGOmap" to obtain all the genes included in a specific BP category [75]. Each BP category included a list of proteins whose expression data were available in all samples thus allowing the calculation of a deregulation score (PDS) measuring the deviation of a biological process from a reference condition (Col−).  Data Availability Statement: All data in this study can be found in public data bases and Supplementary Materials, as described in the Materials and Methods section.