Seed Germination of Different Rice Hybrid Combinations Following Aging Treatment Differs Significantly
Using 16 hybrid rice combinations as experimental materials, the harvested seeds were subjected to natural aging and artificial aging treatments, and the seed germination before and after aging was calculated (Fig. 1). There was no significant difference in the seed germination of the different combination before the treatment. However, after aging, the germination percentage of all combinations, regardless of the aging treatment, demonstrated a significant downward trend. There was also a significant difference in the degree of the decline in seed germination among the different combinations, indicating that the seed storage tolerance or seed vigor among the tested 16 hybrid rice combinations differed (Fig. 1).
After two years of natural aging, the seed germination of BIIIY-122, TY-122, IIY-122, QY-122, and TY-3618 was greater than 80%. Therefore, these five combinations were thus regarded as being resistant to storage under natural conditions, and those seeds were considered as high-vigor. By contrast, the germination percentages of IIY-998, IIY-368, QY-368, BIIIY-3618, IIY-3618 and QY-3618 were 45%, 53%, 58%, 52%, 40% and 50% respectively (Fig. 1). These six combinations were thus regarded as less resistant to natural aging, and were considered as low-vigor seeds in this study. The other combinations whose germination percentages between 60 and 80% were considered to be moderately vigorous.
The germination dynamics of the artificially aged seeds were similar to the naturally aged seeds. The germination percentage of the five combinations, BIIIY-122, TY-122, IIY-122, QY-122 and TY-3618, that were resistant to natural storage, was higher than 80% after artificial aging. For IIY-998, IIY-368, BIIIY-3618, IIY-3618 and QY-3618, which were not resistant to natural storage, the seed germination under artificial aging was also lower than 80% (Fig. 1).
The Maintenance of Vigor During Seed Aging Varies According to the Parental Sterile Line and Restorer Line
Before storage or aging, no difference in the average germination percentage of the seeds among the sterile lines was detected. Conversely, the germination percentage differed significantly following natural aging or artificial aging, indicating that the sterile line affected the seed vigor retention ability or life span (Fig. 2A). The hybrid rice combinations of the II-32A sterile line had poor storage resistance except IIY-122, being lower than the average value of the other sterile lines after natural aging (Figs. 1, 2).
The overall trend of the restorer lines was similar to that of the sterile lines. There was no difference in the average germination percentage of the seeds of the four restorer lines before aging, but there was a significant difference after aging (Fig. 2B). The restorer line Guanghui 122 showed the highest germination percentage under both artificial and natural aging treatments, being significantly higher than that of the other three restorer lines, thus demonstrating good vigor maintenance capacity. Its germination after natural aging and artificial aging was 91% and 88%, respectively (Fig. 2B). The germination percentage of the hybrid combinations between Guanghui 122 and sterile lines BIIIY-122, TY-122, IIY-122 and QY-122 were 87%, 87%, 88%, and 87%, respectively (Fig. 1). However, the seeds of Guanghui 998 and Guanghui 368 showed only moderate storability, and their derived hybrid combinations presented weak storability (Figs. 2B, 1).
Comparative Analysis of the Metabolomes of Hybrid Rice Seeds Before and After Natural Aging
To assess if the chemical composition of the hybrid rice seed was associated with its seed vigor or storability, we firstly compared the metabolome of 32 seed samples (16 before storage and 16 after 24-month storage) using a GC–MS-based metabolomics approach. A total of 89 metabolite peaks were detected, and 56 metabolites were identified based on our in-house database (Yan et al. 2018). Among them, 24 were identified as sugar-related compounds, 20 were amino acid-related compounds, 2 were free fatty acids, 6 were tricarboxylic acid cycle-related intermediates, and 4 were other compounds. Most of them were primary metabolites, which may reflect the physiological states of the seeds.
To generate an overview of the metabolic difference between all the samples, we performed multivariate statistical analyses of metabolomic data, including principle component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA). First, the score plot derived from the PCA model indicated that the non-aged and aged seed samples were distributed in separate groups for all hybrid rice combinations based on the first two principal components, and there was no outlier sample (Additional file 1: Fig. S1), thus indicating distinct metabolic differences occurred during 24-month storage. Further, OPLS-DA model was also established to assess the metabolic patterns of these samples accurately, which can filter out irrelevant orthogonal signals. It is shown that the overall trends were similar to those observed in the PCA model, but more obvious in the score plot (Fig. 3A–D).
Finally, pairwise comparisons of the seed metabolome using orthogonal OPLS-DA were further performed to identify potential metabolite biomarkers that differed significantly in each comparison group. As shown in Fig. 3, 12 metabolites were significantly different between BIII A-0M and BIII A-24M (Fig. 3E; Additional file 4: Table S3); 13 were significantly different between II-32A-0M and II-32A-24M (Fig. 3F; Additional file 4: Table S4); 11 were significantly different between Qiu A-0M and Qiu A-24M (Fig. 3G; Additional file 4: Table S5); and 10 were significantly different between Tianfeng A-0M and Tianfeng A-24M (Fig. 3H; Additional file 4: Table S6). Among these metabolites, most of them were sugar metabolism-related metabolites, such as galactose, fructose, gluconic acid, glycerol (Additional file 4: Table S3–S6).
Changes in Metabolites Related to Seed Aging During Storage
The differential metabolites were analyzed using univariate Student’s t-tests. Although we did not find any differently abundant amino acids and lipids related to seed vigor or seed aging, we did detect differential soluble sugars and their derivatives.
As shown in Fig. 4, the relative contents of galactose, fructose, gluconic acid, and glycerol in most of combinations increased significantly following 24 months of natural aging (Fig. 4A–D). Conversely, the levels of glucopyranoside and oxaloacetic acid in most of combinations decreased significantly after 24 months of natural aging in all samples (Fig. 4E, F).
Among these six sugar-related metabolites, both the levels of galactose and gluconic acid were significantly greater in the seeds after 24 months of natural storage, ranging between 4 and 50 times more for the former, and between 3 and 100 times more for the latter across the different combinations (Fig. 4C, D, Additional file 5: Table S7). Glycerol and fructose were two times higher in the seeds after 24 months of natural storage (Fig. 4A, B), while glucopyranoside and oxaloacetic acid decreased after the 24-month natural storage period (Fig. 4E, F, Additional file 5: Table S7). These results suggested that these six metabolites might constitute preliminary candidate markers for seed aging.
Among the other identified sugar-related metabolites, the levels of raffinose, galactinol, sucrose and myo-inositol remained relatively constant in all the 16 tested seeds during the 24-month natural storage period (Additional file 5: Table S7), which is similar to the findings in our previous study (Yan et al. 2018). Furthermore, we found that there was no significant correlation between the relative content of galactinol, sucrose, trehalose, and myo-inositol and seed germination (Additional file 6: Table S8). However, before storage, the correlation between raffinose content and seed germination under artificial aging (r = 0.5111) was significant (p < 0.05) (Additional file 6: Table S8). This confirmed our previous research that detected a positive correlation between raffinose content and seed vigor (Yan et al. 2018).
Galactose, Gluconic Acid, Fructose, and Glycerol are Potential Markers of Seed Vigor and Aging
To further verify the potential role of these metabolites as markers of seed vigor and aging, we selected two combinations, namely G8Y2156 and G8Y169, and subjected them to artificial aging treatment for 6 d and 15 d, respectively, to obtain seeds with different degrees of aging. The candidate metabolites of the above-mentioned six types of seed aging degree markers were determined in the seed embryo and endosperm (Fig. 5).
The results showed that as seed germination decreased with the increase in aging treatment duration (Fig. 5A, B), the content of fructose, glycerol, galactose, and gluconic acid in the endosperm also increased (Fig. 5C, E). The change trend of these substances in the endosperm was consistent with the above results in the seeds under natural aging (Fig. 4). However, the metabolite changes in the embryo were not completely consistent with the above trend, and only the fructose content increased (Fig. 5D, F). These results also suggested that the contents of these sugars should be determined only in the endosperm, which can provide a semi non-destructive means of testing, with potential broad application prospects. The other two candidates, oxaloacetic acid and glucopyranoside, exhibited little relationship with seed aging. Therefore, combining the results of the changes in the six metabolites in the rice seeds treated by natural aging and artificial aging, we suggest that galactose, fructose, gluconic acid, and glycerol may serve as candidate markers of seed vigor.
Regression Equations Based on Galactose and Gluconic Acid Content Predict the Seed Germination Percentage During Storage
We then used an absolute quantitative method to measure the galactose, gluconic acid, fructose, glycerol, oxaloacetic acid, glucopyranoside, and raffinose contents in the other rice combinations and restore lines (G8Y165, NXRZ, NYZ, G8B, HZ, RXZ, R534, M2YHZ, Y2Y1, JFY1002) stored at room temperature for different durations, comparing them against a galactinol control (a previous study showed that it is related to seed longevity or aging degree).The seed germination percentage of the 26 materials under natural storage for different durations ranged from 0 to 99% (Additional file 7: Table S9), indicating that the seeds differed in their aging degree and vigor. Among these metabolites, the correlation coefficient between galactose and gluconic acid and seed germination percentage was − 0.937 and − 0.935 respectively, indicating a highly significant negative correlation, and the correlation coefficient between the two metabolites was as high as 0.984 (Additional file 7: Table S9). In addition, glycerol contents was also negatively correlated with seed germination, with correlation coefficients between germination percentage and glycerol was − 0.723. Furthermore, the correlation coefficient between the sum of the absolute content of galactose, gluconic acid, and glycerol and seed germination reached − 0.940, indicating remarkably negative correlation (Additional file 7: Table S9).
The changes in oxaloacetic acid, glucopyranoside and fructose, were not significantly correlated with seed aging (Additional file 7: Table S9). Interestingly, although only the raffinose level was positively correlated with seed germination, this correlation coefficient was merely 0.230. Moreover, although the content of galactinol has been reported to be related to the storage capacity of seeds (de Souza Vidigal et al. 2016), it exhibited no significant correlation with the degree of natural aging in the rice seeds (Additional file 7: Table S9).
We then calculated the regression equation of galactose level, gluconic acid level and seed germination in order to predict the germination percentage of the seeds with various vigor during storage. The regression equation between the absolute content of galactose and the seed germination was: y = 89.174 − 0.2095x, and the determination coefficient (R2) was 0.8781 (Fig. 6A). The regression equation between the absolute content of gluconic acid and the seed germination was: y = 101.91 − 8.5837x, and the determination coefficient (R2) was 0.8748 (Fig. 6B). On this basis, we also determined a three-variable linear regression equation of seed germination (y) and galactose (× 1), gluconic acid level (× 2) and glycerol (× 3), which was y = 97.5831 − 0.1739 × 1 − 1.4513 × 2 − 2.2446 × 3, with a determination coefficient of 0.8966. The three-variable regression equation was considered to be more accurate than the single-variable equations.