Combinative effects of Azospirillum brasilense inoculation and chemical priming on germination behavior and seedling growth in aged grass seeds

Germination of seeds during the transportation or after prolonged storage naturally and inevitably decreases because of ageing, but germination potential can be partially restored with seed priming treatments. A novel attempt was made to investigate the effects of combined treatments and to optimize the conditions for naturally aged seeds of tall fescue (Festuca arundinacea Schreb.), orchardgrass (Dactylis glomerata L.) and Russian wild rye (Psathyrostachys juncea (Fisch.) Nevski) using an orthogonal activity level experimental design [factor A: Azospirillum brasilense concentration, factor B: three seed priming treatments (H2O, MgSO4 and H2O2) and factor C: different priming times]. Multivariate regression model analysis was applied to determine the interactive effects of pairwise factors (A and C) and to optimize experimental conditions. The results showed that the mixed treatments positively affected seed germination and seedling growth. The three seed priming treatments were the dominant factors for germination promotion, whereas the bacterial concentration had the largest effect on seedling growth and the activity of superoxide dismutase (SOD), especially root elongation. The malondialdehyde content was reduced, the activities of SOD, peroxidase and catalase were triggered, and ascorbate peroxidase activity was also affected by the co-treatment. The combined results of all determined attributes showed that A. brasilense bio-priming with H2O2 priming constituted the optimal combination. The optimal bacterial concentration of A. brasilense and the time of seed priming were 52.3 × 106 colony forming units (CFU) mL-1 and 17.0 h, respectively.


Introduction
The successful completion of seed germination and the establishment of a normal seedling determine the survival and propagation of plant species [1]. However, seeds inevitably and irreversibly deteriorate and age during storage. This deterioration and aging lead to substantial PLOS

Seed treatments and germination tests
Seeds were surface-sterilized with 2% NaOCl solution for 5 min followed by 70% ethanol for 10 min, after which the seeds were rinsed five times with sterile distilled water. According to the orthogonal design [L 9 (3 4 )], each group of seeds that were coated with gauze were primed at ambient temperature. Inoculation was performed by soaking the sterilized seeds in prepared bacterial suspensions or only sterile phosphate-buffered saline for control seeds, for 2 h [18]. After hydropriming, osmopriming, or redox priming followed by A. brasilense inoculation, the primed seeds were air-dried to reduce their moisture content. Three replicates were used for each treatment and 50 healthy seeds from each replication were evenly spaced on two layers of moistened Whatman No. 1 filter paper in each 150-mm Petri dish. Equal volumes of distilled water were added daily. Petri dishes were placed in a climate incubator (RQH-250, Jing Hong Laboratory Instrument Co., Ltd, Shanghai, China) that maintained an alternating diurnal regimen of 16 h of light at 25 ± 0.5˚C and 8 h of dark at 20 ± 0.5˚C at 85% relative humidity for 28 days. Seed germination was recorded daily, and the seed was considered germinated when its radicle extended at least 0.5 mm. The shoot and root lengths of 10 randomly selected seedlings from each replication were measured in accordance with the rules recommended by the International Seed Testing Association (ISTA) at 28 th day after sowing (DAS). When the total number of seedlings was fewer than 10, all seedlings were measured. To more precisely reflect the vigor of seeds, the germination index (Gi) was estimated as follows: where Gt is the is the number of germinated seeds within a day, Dt is the corresponding number of germination days and t is the number of total germination period (28 days). The seed vigour index (SVi) was determined using the following formula: where R is length of root.

Determination of physiological parameters in aged grass seeds
Lipid peroxidation was assessed in terms of the malondialdehyde (MDA) content in fresh seedlings according to the thiobarbituric acid protocol [25]. Each group of sample containing 0.5 g crumbled seedlings was mixed with 5 mL trichloroacetic acid (TCA; 0.5%) and was centrifuged at 10,000 × g for 25 min. For every 1 mL of the aliquot of the supernatant, 4 mL of 20% TCA containing 0.5% thiobarbituric acid was added. The mixture was heated at 95˚C for 30 min and then the tubes were cooled immediately in an ice-bath. After the tube was centrifuged at 10 000 × g for 10 min, the absorbance of supernatant was measured at 450, 532, and 600 nm. The MDA content was expressed as μmol�g −1 FW. The activities of the antioxidant enzymes SOD, POD, CAT and ascorbate peroxidase (APX) in fresh seedlings were determined. The extraction of antioxidant enzymes was carried out on seedling samples in accordance with the procedures described by Zhang and Nan, with minor modifications [26]. Each group of seedlings (0.5 g) was homogenized in 5 mL of 50 mM phosphate buffer (pH 7.8 for SOD and POD containing 1% (w/v) polyvinylpyrrolidone, pH 7 for CAT and APX containing 0.1 mM Na 2 EDTA). The homogenate was centrifuged at 10000 × g for 15 min at 4˚C, after which the supernatant was collected for the determination of enzyme activities. The activities of SOD, POD, CAT and APX were analyzed as reported by Zhang and Nan [26].

Statistical analysis
The data were statistically analyzed by means of variance analysis (ANOVA) using the software SPSS version 20.0 and were compared using the Duncan test (P < 0.05).
The below-mentioned analyses and graphical procedures were performed using SAS version 8.2. The germination percentage (GP) was simulated using the following logistic model [27,28]: where a, b, and c are constants and X represents the germination days. The models were significant at P < 0.05.
For the generic results, the bacterial concentration of A. brasilense and seed priming time (factors A and C) were denoted by X 1 and X 3 . The dependent variables were denoted by Y i , which included GP, Gi, SVi, root length, shoot length, seedling length, the shoot/root ratio (the index of seed vigor during aging), MDA content, and the activity of SOD, POD, CAT and APX. These dependent variables were approached and analyzed via two-variable (X 1 and X 3 ) quadratic regression models described as follows: where β is a constant. The critical values of X 1 and X 3 were obtained from the quadratic models that were significant at P < 0.05.

The effect of co-treatment on the vigor of aged seeds
None of the seeds of tall fescue 2006 germinated in the whole nested experiment, only 2-4% of seeds germinated in treatments 1-4 (T1-T4) and T7 for orchardgrass 2006 (data not shown). The models regarding germination dynamics (Fig 1) showed that the germination of T1-T5, and T8 in tall fescue 2015 occurred earlier than it did for the control (CK), and the greatest germination percentage (28 DAS) occurred in T1 (98.7%). The germination from all treatments occurred earlier than it did from the controls for both orchardgrass 2013 and Russian wildrye 2009. The other 8 treatments increased germination of orchardgrass 2013 except for T3, the germination percentages from T2 and T6 (92.0%) were the highest. The germination percentages of all treatments were higher than that of the CK, the highest germination percentage was in T1 (62.7%) among them in Russian wildrye 2009. Whereas, the co-treatments had a negative effect on germination of tall fescue 2014 except for T7 (72.0%) compared with the CK. But the germination started earlier in T2 and T7. The lowest germination percentages of tall fescue 2015, tall fescue 2014 and orchardgrass 2013 occurred in T9, T3 and T3, respectively. While the lowest germination percentage of Russian wildrye 2009 occurred in T9, which increased by 2.7 times compared with that of the CK. The range analyses revealed that factor C (seed priming time) exhibited the largest effect on germination percentage among three factors, with the exception that factor B (seed priming approaches) had the highest range value (R value; 4.0) for tall fescue 2015 (Table 2). And the most effective factors on the Gi and SVi of tall fescue 2015, orchardgrass 2013 and Russian wildrye 2009 were factors B, A (the bacterial concentration) and C, individually, which were the same for every species. The most important factors on the Gi and SVi of tall fescue 2014 were factors C and B, which were different. The analysis of variance indicated that the germination percentage was significantly (P < 0.05) affected by both grass species and factor C, and their interactive effects were also significant (P < 0.05) both in pairwise factors and among three factors, with the exceptions of the grass species × factor A and grass species × factor B interaction (Table 3). Both factors B and C significantly (P < 0.01) affected the Gi and SVi in different-aged seeds, also their interactive effect was significant (P<0.05).

Seedling growth
Compared with the CK, T7 resulted in the biggest increase in root length, and the shoot length increased in T5 and T6 in tall fescue 2015 group (Fig 2). The root length in T8 was the longest significantly, and the shoot lengths in both T8 and T9 were markedly improved in tall fescue 2014. In group of orchardgrass 2013, the root and shoot lengths were increased by T3 and T6, respectively. While in Russian wildrye 2009, the changes of root elongation was not significant. The root length was the largest in T8, the shoot length in T6 was significantly higher than that of the CK.
Factor A markedly affected the root length, and the effect of factor A was the largest for tall fescue 2014 and orchardgrass 2013 (Tables 3 and 4). The interactive effects were also  Table 2 Table 3. Variance analyses for the model, the effects of each experimental factors and their interaction effects on germination percentage, gemination index, seed vigor index, root, and shoot length. Columns GS, X 1 ,X 2 and X 3 represent grass species, bacteria concentration, seed priming and priming duration, respectively.  Table 3).

Characterization of physiological responses
The MDA content of CK in Russian wildrye 2009 was significantly (P < 0.05) higher than those of CKs in tall fescue 2015, tall fescue 2014 and orchardgrass 2013. And the SOD activity of CK in Russian wildrye 2009 was noticeably lower than those in the three other controls. These results indicated that the seed of Russian wildrye 2009 was the most severely damaged by aging (Fig 3A; Table A in S1 File). In all treatments, the MDA content sharply decreased relative to that of the CKs in Russian wildrye 2009 and tall fescue 2015. The lowest MDA content in tall fescue 2014 and orchardgrass 2013 occurred in T6 and T7 (Fig 3A). The changing trends regarding the MDA content and SOD activity were opposite in Russian wildrye 2009 and tall fescue 2014 (T1, T4, T5 and CK). The highest activities of SOD and CAT were observed in T6 for tall fescue 2015 and in T7 for tall fescue 2014. While the highest SOD and CAT activities were in T3 and T2 for orchardgrass 2013, and were in T8 and T9 for Russian wildrye 2009, respectively (Fig 3). The highest POD values for tall fescue 2015 and Russian wildrye 2009 were found in T3 and T2, separately ( Fig 3C).
Factor A and the interactive effects significantly (P < 0.05) influenced SOD activity, and factor A was the most important factor for those of tall fescue 2015, orchardgrass 2013 and Russian wildrye 2009 (Tables 4 and 5). The experimental factors did not significantly affect the MDA content, except seed species (Table 5). The experimental factors markedly affected the POD and CAT activities with the exceptions of factor A, A × B, B × C and grass species ×A × B on the POD activity as well as factors A and B on the CAT activity.

Model analysis for interactive effects and optimized conditions
Given the interactive effects of factors A and C and exploring optimized conditions, regression models were constructed. As shown in Fig 4, the root and seedling length in the four groups of aged seeds showed via three-dimensional parabolic trends that there were maximum values

MDA content (A), activity of SOD (B), POD (C) and CAT (D) of tall fescue (2015 and 2014), orchardgrass (2013) and Russian wildrye (2009).
Data are presented as mean ± SD. The different lowercase letters indicate significant differences at P < 0.05 within the years among the treatments.
https://doi.org/10.1371/journal.pone.0210453.g003 Azospirillum brasilense inoculation and aged grass seeds less than 24.0 h, although the SVi of orchardgrass 2013 exhibited a completely opposite trend (Fig 4E-4H). The Gi of Tf 2015 continuously increased with the increase in factor A under short priming time (less than 8.0 h), and the Gi of three other seed groups showed similar trends as those of the SVi of orhardgrass2013 (Fig 4A-4D). The MDA content in Russian wildrye 2009 and the APX activity in tall fescue 2015 had minimum values at 52.8 × 10 6 CFU mL -1 , 19.2 h and 52.1 × 10 6 CFU mL -1 , 15.6 h, respectively (Figs 5B and 6F and Table B in S1 File). The MDA content in tall fescue 2015 decreased with the increase in factor A from 60.0 × 10 6 CFU mL -1 and showed opposite changing pattern as that of SOD activity. When factor A was less than 60.0 × 10 6 CFU mL -1 , the MDA content in tall fescue 2014 was sharply reduced with the increase in factor A (Figs 5A, 5C and 6D). The POD activity in Russian wildrye 2009 exhibited a parabolic trend that peaked at 61.1 × 10 6 CFU mL -1 at 3.9 h; the CAT activity in Russian wildrye 2009 was positively correlated with bacterial concentration when factor A was larger than 30.0 × 10 6 CFU mL -1 and when the priming time was longer than 16.0 h (Figs 5E and 6B and S2 Table in S1 File). The CAT activity in tall fescue 2015 was the highest within the factor A range of 45.0-90.0 × 10 6 CFU mL -1 and when the priming time was less than 4.0 h (Fig 5D). The APX and CAT activities showed analogous trends, and the APX and SOD activities were nearly complementary in orchardgrass 2013 (Figs 5F, 6C and 6E).
A total of 48 pairs of critical values composed of factors A and C were obtained from the model analysis, as shown in Fig 7. The mode of critical values was considered the largest value and was used as the final optimal value [4,25]. The final optimal conditions for factors A and C were the bacterial concentration of A. brasilense of 52.3 × 10 6 CFU mL -1 and the seed priming time of 17.0 h.

Discussion
Seed aging has become an intractable problem in agricultural science, which negatively affects the external morphology and internal physiology of seeds and seedlings [11]. Seed pre-treating with PGPR is increasingly favored by scholars because of its eco-friendly and high efficiency characteristic. The present study showed that the combined treatments of A. brasilense inoculation and the three seed priming treatments counteracted the adverse effects from natural aging on four groups of seeds. The results revealed that the three seed priming treatments played the dominant role in improving the seed germination and vigor. The pairwise interactions among three factors were significant, especially between bacterial concentration of A. brasilense and three seed priming treatments for the germination percentage and germination index. However, the sole function of bacterial concentration was not significant. In this regard, Carrozzi et al. noted that hydration treatment might affect only the early stages of imbibitions, while the function of A. brasilense is additive to the hydration process, which could result in repair processes for the promotion of germination in aged seed [10]. It was also found that the germination of Russian wildrye 2009 during the early phases in all treatments lagged far behind that of the other seeds. This phenomenon indicates that the germination percentage might have been intimately linked to the reparation of seed deterioration during the lag phase of germination [10]. Accordingly, for the older seed, a longer lag phase was required to reverse deterioration by seed priming. Previous studies have reported that hydropriming, osmopriming and redox priming can invigorate aged seed associated with the initiation of repair-and germination-related processes, such as the activation and resynthesis of some ROS scavengers, increased α-amylase activity, crosstalk with phytohormones, and the repair of DNA damage [6,29,30]. On the other hand, A. brasilense may contribute to phytohormone communication between PGPR and aged seeds, mainly GA and IAA produced by PGPR. IAA might assist GA biosynthesis which could activate some genes responsible for α-amylase mRNA transcription in aged seeds during germination [31,32]. Therefore, it is speculated here that A. brasilense could synergize the reinvigorating action of three seed priming treatments for seed germination and vigor modulation in aged seeds; the latter treatments would play the main role in the early stages.
The combined treatments were more effective for boosting the growth of relatively aged seed (Russian wildrye 2009) compared with that of new seed. This result could confirm the repair effect of the co-treatments on aged seeds. The repair effect of co-treatments on the antioxidant enzyme activities, mitochondria, or associated proteins might be less/negative in new seeds with respect to old seeds, which resulted from different physiological states such as seed moisture levels, water transport and redox status [6]. And seeds are expected to exhibit differences in experimental factors, which are manifested by the different genes in the different aged seeds. For tall fescue 2014, the co-treatment had a small positive impact, even negative effect on germination. But it is worth noting that the germination in T2 and T7 for tall fescue 2014 occurred earlier, and T1, T2, T5, T6 and T8 significantly increased the root length, which means the co-treatment had potentially positive effect on seeds of tall fescue 2014. Therefore, the co-treatments have to be considered carefully in this species and to require further study.
The co-treatments expressed remarkable root elongation-promoting activity, particularly under the bacterial concentration of 10 7 CFU mL -1 and H 2 O 2 -priming for 18.0 h (T8) for all three species. The study revealed that the effect of bacterial concentration of A. brasilense was the greatest among the factors on root length. The root elongation response might be mainly due to the positive feedback loop between root and bacterial IAA. As reported previously, hormones produced by A. brasilense could affect root cell wall dynamics [33]; A. brasilense inoculation can stimulate root carboxylate exudation acting as precursors of phytohormones, which indirectly and positively affected maize root length and area [34]. In addition, H 2 O 2 -priming has been reported to promote the adventitious root development under drought stress. H 2 O 2 is necessary for the formation and growth of lateral and adventitious root and it might act as downstream signal molecules in the auxin signaling cascade during rooting [35]. The combinative effects of Azospirillum brasilense inoculation and H 2 O 2 priming on rooting might involve in IAA production. It is noteworthy that under the interactive effects of factors A and C, the root and seedling lengths first increased but then decreased with the increase in factor A. There were somehow critical thresholds between the factors for the co-treatments. A moderate bacterial concentration (50.0-65.0 × 10 6 CFU mL -1 ) and priming duration (5.8-26.3 h) could effectively facilitate the root system development in aged seeds. Similarly, A. brasilense Sp 7-S and Sp 245 at 10 7 CFU mL -1 stimulated root and seedling growth in cucumber and lettuce [36].
To further elucidate the underlying physiological mechanisms of the combined treatments, both antioxidant enzymatic activities and lipid peroxidation were investigated, which are generally considered as the primary indicators to evaluate viability and vigor of aged seeds [6]. Seed aging inevitably generates the overproduction of ROS, which accelerate the disintegration of membranes [2]. In the present study, the corresponding co-treatments effectively increased the activities of SOD, POD and CAT and lowered the MDA contents in four groups of aged seeds, especially, in severely aged seed of Russian wildrye 2009. Its maximum POD activity occurred at 61.1 × 10 6 CFU mL -1 and 3.9 h, and the minimum value of MDA content was at 52.8 × 10 6 CFU mL -1 and 19.2 h, under the strong interactive effect. The result also suggested that the germination percentage and root length of aged seed were negatively correlated with MDA content, and were positively correlated with SOD activity (Table C in S1 File), which is in line with the finding of Pereira et al. [37]. Wojtyla et al. pointed out that in aged seeds, the deleterious effects derived from excessive ROS can hinder cell wall loosening to allow cell elongation in root apical meristems [38]. Coincidently, the increased activities of antioxidant enzymes confer the ability of plants to scavenge excessive ROS [39]. SOD can directly modulate the amount of ROS [6]. It was clarified here that A. brasilense inoculation was the major modulator to improve the SOD activity in aged seeds. And this mode of action is consistent with the promoting effect on root length mentioned above. Recent studies have shown that PGPR regulate the expression of the antioxidant-responsive genes MnSOD and APX1 [15]. Wang et al. also indicated that a consortium of PGPR that induce drought tolerance involves SOD and cAPX, not POD and CAT, in cucumber [40]. In this regard, the present results indicated that the three seed priming approaches were the most effective factors for enhancing POD and CAT activities. Among a number of protective enzymes, SOD contributes in catalyzing O 2 •− into H 2 O 2 , which in turn can be neutralized by CAT [41]. A. brasilense and three seed priming approaches took co-allocation action to improve antioxidant system for repairing aged seed. It was observed that the activities of APX and SOD were complementary in orchardgrass 2013 and that the APX activity in tall fescue 2015 had a minimum value under the interactive effects of factors A and C, which contradicts the results of some studies. However, this finding suggested that an operation mode of co-allocation might exist among antioxidant enzymes during the repair process in aged seeds. APX might complement the activity of SOD when the amount of ROS over-accumulated in aged seeds and that it might not play a relevant role in antioxidant enzyme system in relatively new seeds. This assumption is supported by the study from Bosco de Oliveira et al. [42]. Additionally, the reduction in MDA content after the combined treatment suggested the capability for attenuating the extent of membrane lipid peroxidation under aging stress. Although the experimental factors did not significantly affect the MDA content, the changing trends were complementary between the MDA content and SOD activity in Russian wildrye and tall fescue. Any sort of treatments reduced the MDA levels in the seedlings, which could be explained by the enhancement of antioxidant enzyme activities [37]. Although minor inter-and intra-species differences existed for the effect of factor B, all the greatest average root length for tall fescue 2015, the highest average GP and the activity of SOD for orchardgrass 2013, those of root length, shoot length and CAT activity for Russian wildrye 2009, as well as those of GP, SVi, root length, shoot length, the activity of SOD and CAT for tall fescue 2014 occurred in the H 2 O 2 -priming. Consistently, Liu et al. reported that H 2 O 2 priming could improve the oxygen bioavailability, imbibition rate and metabolism of aged seeds [43]. H 2 O 2 was originally recognized mainly as a reactive oxygen species that plays deleterious role in cell organization and viability. But pretreatment with it can protect plants from oxidative damage caused by different abotic stresses, such as heat, salinity, drought, waterlogging and low temperature, which is related to higher activities of antioxidative enzymes [35,44,45]. The effects of H 2 O 2 were also correspondingly manifested by T8 (H 2 O 2 -priming) stimulating root elongation for all three species mentioned above. H 2 O 2 of an appropriate level could function as a common stress and developmental signaling molecule that regulates a common set of transcription factors related to the antioxidant defense system and can eventually restore redox homeostasis and oxidative membrane damage [46]. However, the germination percentage was the lowest in T9 (H 2 O 2 priming) among nine treatments. This result implied that the level of ROS in T9 in Russian wildrye 2009 that already have an abundance of ROSwas not within the optimum range of concentrations, but the range was still within the "oxidative window" [47]. It is worth noting that the germination percentage of Russian wildrye 2009 in T8 was relatively higher than the remaining six treatments. And the seeds of orchardgrass 2006 germinated in T7. Thus, the relatively lower H 2 O 2 concentration or priming duration shorter than 24h may be required for severely aged seeds. Taking into account the overall analysis of results of all experimental indicators, A. brasilense bio-priming combined with H 2 O 2 -priming constituted the optimal combination for repairing aged seeds. Similarly, H 2 O 2 priming has improved the germination and seedling growth of pea, rice, dried Quercus fabri (Fagaceae) seeds as well as aged maize, squash, tomato and switchgrass seeds [48][49][50][51]; A. brasilense inoculation with H 2 O 2 -priming also enhanced the grain yield of wheat under dryland conditions [22]. Except that H 2 O 2 -priming enhanced the promoting effect of Azospirillum on the activities of antioxidant enzymes, Jafariyan and Zarea reported that exogenously supplied H 2 O 2 could induce decrease in ABA content which is increased in aged seed during germination [22]. While Lukasz et al. suggested that H 2 O 2 could activate GA signaling and/or biosynthesis to alleviate seed dormancy of barley, which could also reinvigorate aged seeds [30,52]. These possible modulations of phytohormone balance by H 2 O 2 pretreatment might synergistically act with A. brasilense bio-priming during germination in aged seeds. The mechanism of interplays between A. brasilense and H 2 O 2 priming during repairing aged seeds requires more precise experiment.

Conclusion
Azospirillum brasilense inoculation and the three seed priming treatments repaired aged grass seeds by the specific operation mode of co-allocation during the different developmental stages. The three seed priming methods were the most effective factors for stimulating germination, whereas the bacterial concentration had the largest effect on root elongation and SOD activity. The combined treatments that elicited aging-related repair reduced the MDA content and improved antioxidant enzyme activities. Overall, A. brasilense bio-priming with H 2 O 2priming was the optimal combination for reinvigorating aged seeds. The final optimal conditions consisted of the bacterial concentration of A. brasilense of 52.3 × 10 6 CFU mL -1 and a seed priming time of 17.0 h, given the interactive effects. However, the interactional mechanisms of combined treatments are just beginning to be unraveled. Whether the co-treatments affect phytohormone balance, aging-related proteins or genes as well as antioxidant enzyme genes still requires further exploration.
Supporting information S1 File. Table A Table B. Canonical response surface analysis for germination percentage, germination index, seed vigor index, root, shoot and seedling length, ratio of S/R, MDA content, the activity of SOD, POD, CAT and APX of the seedlings based on coded data. Table C. Pearson correlation coefficients for activity of SOD, APX, POD, CAT, and MDA content. (PDF)