Azospirillum brasilense promotes increases in growth and nitrogen use efficiency of maize genotypes

The development of cultivars with an improved nitrogen use efficiency (NUE) together with the application of plant growth-promoting bacteria is considered one of the main strategies for reduction of fertilizers use. In this sense, this study: i) evaluated the effect of Azospirillum brasilense on the initial development of maize genotypes; ii) investigated the influence of A. brasilense inoculation on NUE under nitrogen deficit; and iii) sought for more NUE genotypes with higher responsiveness to A. brasilense inoculation. Twenty-seven maize genotypes were evaluated in three independent experiments. The first evaluated the initial development of maize genotypes with and without A. brasilense (strain Ab-V5) inoculation of seeds on germination paper in a growth chamber. The second and third experiments were carried out in a greenhouse using Leonard pots and pots with substrate, respectively, and the genotypes were evaluated at high nitrogen, low nitrogen and low nitrogen plus A. brasilense Ab-V5 inoculation. The inoculation of seeds with A. brasilense Ab-V5 intensified plant growth, improved biochemical traits and raised NUE under nitrogen deficit. The inoculation of seeds with A. brasilense can be considered an economically viable and environmentally sustainable strategy for maize cultivation.


Introduction
The world yield and productivity of maize (Zea mays L.) doubled in the last three decades, resulting in an output of 1,034.8 million tons of grain in 2017/2018 [1]. This significant yield increase is attributed mainly to chemical fertilizers, breeding and crop management [2]. However, the dependence of modern agriculture on chemical fertilizers is alarming, since the indiscriminate use of these inputs has been causing serious environmental problems, e.g., water eutrophication, soil acidification and air pollution [3][4][5]. different maize commercial hybrids. The bacterial strain A. brasilense Ab-V5 was used in the experiments. This strain is derived from a selection program that evaluated N 2 -fixing capacity in vitro and under field conditions in Paraná State, Brazil, being highly efficiency in promoting growth of maize in several trials, mainly due to capacity of producing phytohormones, increasing root growth and nutrients uptake [29]. The A. brasilense Ab-V5 is registered for commercial use in Brazil by the Ministry of Agriculture, Livestock and Food Supply (MAPA), and is part of the "Collection of Diazotrophic Bacteria and Plant Growth Promoters" of Embrapa Soybean, Londrina, Paraná, Brazil.

Inoculant preparation
The inoculant was prepared from a pre-inoculum in DYGS liquid medium [30] and incubated on a rotary shaker (180 rpm) at 28±2˚C for 24 h. The pre-inoculum was multiplied in Erlenmeyer flasks with 250 mL of Form 15 culture medium [31] and incubated in an orbital shaker (180 rpm) at 28±2˚C for 24 h. After the growth period, the bacterial population density was diluted to a concentration standard of 1 × 10 8 mL -1 cells.

Experiment on germination paper
For the experiment on germination paper (E1), a completely randomized design with four replications was used, evaluating 27 maize genotypes with (+Azo) and without (−Azo) inoculation with A. brasilense Ab-V5. The seeds were initially disinfected by immersion in 95% (v/v) ethanol solution for 30 sec, followed by soaking in 5% (v/v) H 2 O 2 solution for 10 min, and then washed six times with sterile deionized water [32]. Thereafter, maize seeds from the +Azo treatments were inoculated by briefly soaking the seeds on inoculant solution to a final concentration of 3.3 × 10 −6 cells of A. brasilense per seed. After inoculation, 30 seeds per treatment were placed on germination paper moistened with sterilized distilled water and incubated in a growth chamber at 25±2˚C and 70% relative humidity.
Ten days after sowing (growth stage V1), the roots of five seedlings were scanned at 300 dpi and the images treated and analyzed with software GiA Roots [33]. The total root surface area (RSA, in cm 3 ) and total root length (RL, in cm) were evaluated. The shoot part and root system of the seedlings were oven-dried separately under forced ventilation at 60˚C for 72 h to determine shoot dry mass (SDM, in g) and root dry mass (RDM, in g).
The traits IAA and PRO were evaluated by methodologies described by Bautista and Gallardo [35] and Bradford [36], respectively. For IAA, a 600 μL aliquot of the supernatant was mixed with 200 μL sodium borate buffer solution (50 mM) and 1.2 mL Salkowski solution and maintained for 30 min in the dark. To determine PRO, an aliquot of supernatant (50 μL) was added with 950 μL sodium borate buffer solution (50 mM) and 1 mL Coomassie Brilliant Blue G-250 reagent, after gentle shaking and a rest period of 5 min. Readings on an Agilient 8453 spectrophotometer (Agilient Technologies, USA) were performed at wavelengths of 540 and 595 nm, respectively, for IAA and PRO.
The activities of the enzymes PAL and PPO were assessed by the methodologies described by Kamdee et al. [37] and Sommano [38], respectively. To determine PAL, an aliquot of the supernatant (150 μL) was mixed in 3 mL sodium borate buffer solution (50 mM) and 350 μL L-phenylalanine (100 mM). Subsequently, the test tubes were vortexed and incubated at 40˚C for 1 h. The PPO activity was determined by adding 100 μl supernatant to 250 μl 4-methylcatechol (10 mM) and 650 μl potassium phosphate buffer (0.1 M). Thereafter, the test tubes were vortexed and incubated at 30˚C for 30 min. Spectrophotometric readings were performed at wavelengths of 290 and 410 nm, respectively, for PAL and PPO.

Experiment in Leonard pots
The experiment was arranged in a completely randomized design with four replications and the 27 maize genotypes were evaluated in three conditions: cultivation at high nitrogen (HN), low nitrogen (LN) and LN plus A. brasilense Ab-V5 inoculation (LN+Azo). After disinfestation, the seeds on moist germination paper were incubated in a growth chamber at 25±2˚C and 70% relative humidity. After five days, the seedlings were selected for uniformity of length, and one seedling per pot was transplanted into independent Leonard pots [39]. In condition LN+Azo, A. brasilense Ab-V5 inoculation with 1 mL inoculant per pot containing 1 × 10 8 mL -1 cells was performed immediately after transplanting.
The upper part of the Leonard pots was filled with 450 cm 3 perlite as inert substrate and the lower part (saucer) with 100 mL nutrient solution. The pots were arranged on tables under greenhouse conditions and the nutrient solution was replaced every five days. After 28 days (growth stage V4), the total root volume (RV, in cm 3 ) was determined as the difference between the water volume within a graduated cylinder before and after insertion of the fresh roots. Afterwards, the shoot part and root system of the plants were stored separately in paper bags and dried in a forced ventilation oven at 60 o C for 72 h for subsequent determination of the shoot dry mass (SDM, in g) and root dry mass (RDM, in g). The SDM samples were ground and used to determine total shoot nitrogen by Kjeldahl digestion method [42] using a Tecnal TE-0371 digester. The nitrogen use efficiency (NUE, in mg mg -1 ) was determined as described by Moll et al. [43] by the following formula: where: NUE ijk is the nitrogen use efficiency of genotype i in replication j under condition k; TSN ijk is the total nitrogen contained in the shoot of genotype i in replication j under condition k; TAN ijk is the total amount of nitrogen available for genotype i in replication j under condition k; and SDM ijk represents the shoot dry mass of genotype i in replication j under condition k.

Experiment in pots with substrate
In the experiment in pots with substrate (E3) we used the same treatments and experimental design as in experiment E2. However, after selecting healthy seedlings grown on germination paper, a 3:1 (v/v) mixture of sand and soil (Eutrophic Red Latosol) was filled in 1 L plastic containers, and one seedling per pot was planted. The pots were placed on tables in a greenhouse and fertigation was applied every five days consisting of 100 mL per pot of the nutrient solution of Hoagland and Arnon [40], modified by Chun [41]. After 28 days (growth stage V6), the traits RV (in cm 3 ), SDM (in g), RDM (in g) and NUE (in mg mg -1 ) were evaluated.

Data analysis
The data were analyzed based on restricted maximum likelihood (REML) and best linear unbiased prediction (BLUP) with software Selegen-REML/BLUP [44]. The predicted genotypic means were calculated after testing for data normality and homogeneity by the tests of Shapiro and Wilk [45] and Hartley [46], respectively. Deviance analyses (ANADEV) were performed based on the following statistical model: where y is the data vector; u the scale for the general mean (fixed effect); g the vector of the genotypic effects (assumed as random); e the vector of errors or residues (random); and X and Z represent the incidence matrices for u and g, respectively.
The predicted genotypic means were used in Pearson's correlation coefficient, principal component analysis (PCA) and a heatmap based on standardized data. For the heatmap analysis, Ward's clustering method [47] based on the Euclidean distance was used. The inoculation efficiency index (IEI, in %) was calculated by the following formula: where: IEI i is the inoculation efficiency index of genotype i; GMLN i is the predicted genotype mean of genotype i in the low nitrogen (LN) condition; and GMI i is the predicted genotype mean of genotype i under LN plus inoculation with A. brasilense Ab-V5 (LN+Azo). For the statistical analyses, software R (http://www.r-project.org) was used with the packages FactoMi-neR [48], heatmaply [49] and ggplot2 [50].
The formation of three large groups was detected by heatmap analysis (Fig 1A). Principal component analysis (PCA) explained 82.5% of the total variation by the first two components, and the resulting groups coincided with those of the heatmap (Fig 1B). Group I (blue) comprised most of the +Azo treatments, aside from the genotypes L22, L23, L24 and 2B587PW in condition −Azo. Thirteen inbred lines were clustered in group II (green), eight of which in condition +Azo and five in −Azo. On the other hand, group III (pink) consisted of 18 inbred lines in condition −Azo. In general, the mean genotype values of group I were highest for SDM, RDM, RSA and RL, and those of group II for IAA, PAL and PPO. On the other hand, the means of group III were the lowest for all evaluated traits.

Experiment in Leonard pots
The predicted genotypic values under HN, LN and LN+Azo, as well as their respective IEI, are shown in Table 3. The highest general means were observed under HN for all evaluated traits Table 1

Genotype
IAA (mg g -1 ) PRO (mg g -1 ) PPO (μmol min -1 mg -1 ) PAL (μmol min -1 mg -1 ) The heatmap was used to distinguish the lines in six groups (Fig 2). The genotypes under LN were distributed in the groups I (purple), II (dark blue), III (light blue) and IV (green), while genotypes under HN were allocated to groups V (yellow) and VI (pink). The inbred lines clustered in groups II and III had the highest mean values for NUE, especially those allocated in group II, in which the means were also high for the traits RDM and RV. Groups V and VI had the lowest NUE means; however, the inbred lines in group VI had high means for the other evaluated traits.

Experiment in pots with substrate
The predicted genotypic values under HN, LN and LN+Azo, as well as their respective IEI, are shown in Table 4. The overall means were highest in the condition HN for the traits SDM and RDM, while in LN+Azo, the overall means were highest for RV and NUE. In relation to the IEI, positive general means were observed for all evaluated traits, ranging from 12.05 (NUE) to 26.03% (RV). In general, most of the genotypes had positive IEI values for the traits SDM, RDM, RV and NUE, mainly inbred lines L1, L6, L7, L8, L13 and L24. The heatmap showed the formation of six groups (Fig 3). The genotypes under HN were all allocated in groups I (purple) and II (dark blue). Group III (light blue) was formed by the genotypes in condition LN+Azo, except for the genotypes L1, L3 and 2B587PW at LN. With the exception of inbred line L2, group IV (green) consisted only of lines under LN, whereas the groups V (yellow) and VI (pink) were formed by inbred lines in the conditions LN and LN+Azo. In general, the genotypes under HN (groups I and II) had a lower NUE and higher SDM. Group III was characterized by the highest means for RDM, RV and NUE, while in group IV, the mean values for SDM, RDM and RV were the lowest. Group IV can be characterized by high means for NUE, and group VI by median values for all evaluated traits.

Correlation between experiments
By means of a correlation analysis between the experiments (Fig 4), a positive and significant correlation was observed between experiments E1 × E2 for trait RDM (r = 0.49 � ). Between the experiments E1 × E3, positive and significant correlations were found for RDM (r = 0.63 �� ), as well as for E2 × E3 for SDM (r = 0.62 �� ) and RDM (r = 0.57 �� ).

Discussion
The results of this study indicated that maize inoculation with A. brasilense Ab-V5 improved plant growth and biochemical traits and increased NUE under N limiting conditions. Metabolic changes in maize plants in response to A. brasilense inoculation were described previously, e.g., an improved root architecture [51], increase in plant biomass [52] and N assimilation [53], as well as mitigation of abiotic stresses [54][55][56]. In this way, the results show Table 3 the powerful effect of A. brasilense inoculation on maize, mainly under limiting nutritional conditions, and also reinforce the importance of the plant microbiota as an extension of the maize genome to beat developmental restrictions under limiting-growth conditions [57].

HN LN+Azo LN IEI (%) 1/ HN LN+Azo LN IEI (%) HN LN+Azo LN IEI (%) HN LN+Azo
In most maize genotypes inoculated with A. brasilense Ab-V5, the IAA concentration increased, possibly favoring plant growth and development. This beneficial effect can be related to the observed increases in the plant biomass and modifications on the root architecture in experiment E1. The initial effect of Azospirillum inoculation on the promotion of seedling growth can be mimicked a phytohormone treatment [58,59]. However, modifications in the plant development pattern during an extensive growth period require the uninterrupted entry of exogenous phytohormones, which occurs when Azospirillum colonizes the plants Influence of Azospirillum on the nitrogen use efficiency for maize crop [21]. Although the IAA biosynthesis by Azospirillum is influenced by endogenous and exogenous factors, it is produced during all phases of bacterial development, which is a highly relevant characteristic for plant growth promotion, since benefits can already be observed in the first days or months after inoculation [60]. According to Bashan and de-Bashan [23], phytostimulation of Azospirillum by means of IAA biosynthesis is extremely important in the early growth stages (germination and initial seedling growth) and is considered complementary to other mechanisms at more advanced plant growth stages.
Increases in the traits related to plant growth were also observed in the experiments E1 and E2, reinforcing the role of A. brasilense in promoting structural changes that are essential for plant growth and development. Changes in the root system of Azospirillum-inoculated plants have already been observed, such as root elongation [53,61], development of lateral and adventitious roots [62,63] and root hair development [64,65]. These modifications were associated to increases in plant biomass and nutrient uptake, increasing the tolerance to limiting Table 4 Influence of Azospirillum on the nitrogen use efficiency for maize crop nutritional conditions [24,29]. The structural morphological changes in inoculated plants are partly caused in response to phytohormone production and release by Azospirillum [15]. The IAA, for example, is related to the division, extension and differentiation of plant cells and tissues and closely linked to the differentiation of the vascular system of plants [66,67]. In a study on the effect of A. brasilense Ab-V5 on maize, Calzavara et al. [52] observed a higher number of elements of the metaxylem of inoculated plants than of the control plants. This resulted in a thicker vascular cylinder of the plants, which is favorable for water and nutrient transport, resulting in higher root and shoot biomass production. Although the efficiency of PGPB inoculation may vary according to the plant genotype, bacterial strain and environmental conditions [68], the influence of N fertilization Influence of Azospirillum on the nitrogen use efficiency for maize crop management on inoculation efficiency has been considered more relevant [69]. According to Rozier et al. [70], in a study on the effect of the different levels of N fertilization associated with A. lipoferum inoculation, N fertilization induced higher maize grain yields. However, no influence of A. lipoferum inoculation on this increment was detected, suggesting that these technologies are not additive. In the same context, a meta-analysis of the effect of Azospirillum spp. on maize yield showed a mean increase of 651.58 kg ha -1 in inoculated over uninoculated treatments [71]. However, the same study observed a strong influence of the levels of N topdressing (absence, low, moderate and high) on inoculation efficiency, since the positive effects of inoculation were only significant in the absence of N topdressing, which confirms the theory of non-additivity of the two technologies. Thus, the use of Azospirillum as biofertilizer can be considered a promising technology, in particular under N stress [31,72,73].

HN LN+Azo LN IEI (%) 1/ HN LN+Azo LN IEI (%) HN
Nitrogen limitation in maize can drastically reduce the photosynthetic activity of plants [74] and interfere with the transcription of genes related to the N and C metabolisms, causing a reduction in biomass production and, consequently, limiting grain yields [75]. In this sense, plants with a higher NUE can reduce the damages caused by N limitation, since they require a smaller amount of this nutrient for biomass and/or grain production [76]. In the experiments E2 and E3, a higher NUE of the inoculated genotypes could be observed in relation to the uninoculated genotypes, indicating that inoculation with A. brasilense Ab-V5 raised NUE under LN availability. In experiment E2, the genotypes with highest NUE were the same in the LN and LN+Azo conditions. However, this coincidence was not observed in experiment E3, indicating a differentiated NUE between genotypes under A. brasilense Ab-V5 inoculation.
Nitrogen use efficiency does not only depend on an efficient N uptake from the soil, but also on the internal transport, storage, recycling, remobilization and growth stage of the plants [77]. Several strategies have been used to improve NUE of plants [11,78]. However, since PGPB have the capacity to promote plant growth and nutrient uptake, they can be considered a promising solution to increase the efficiency of nutrient use, which is reinforced by the results obtained in this study. An increase in the efficiency of nutrient use by plants has been reported for several PGPB genera, since they are not only able to fix N 2 , but are also capable of solubilizing mineral and/or organic nutrients of the soil [14,79]. A meta-analysis addressing the benefits of PGPB in relation to NUE in several plant species identified a mean increment of 5.8±0.6 kg grain per kg fertilizer, reinforcing biofertilizers as a promising technology under limiting cultivation conditions [80].
In an evaluation of the response of greenhouse maize to A. brasilense inoculation in clayey and sandy soil, Ferreira et al.
[81] stated a positive response of maize to inoculation. However, these responses were dependent on the soil type and substrate, since increases in the evaluated traits were only observed in clayey soil. Similarly, Mehnaz et al. [24] observed differentiated responses among maize varieties inoculated with A. brasilense or A. lipoferum in pots with sand or soil, allowing the conclusion that, aside from the maize genotype and Azospirillum species, the type of substrate may also influence the effect of inoculation. In this study, although the experimental conditions of evaluation were contrasting, the observed results were similar under the three experimental conditions (E1, E2 and E3), mainly for RDM.
In general, the inbred lines L7 and L8 were the most responsive in relation to the efficiency of A. brasilense Ab-V5 inoculation, whereas line L16 was least responsive to inoculation. The identification of contrasting genotypes regarding inoculation response is fundamental in studies on the plant-Azospirillum interaction. In a population of 114 double haploid wheat (Triticum aestivum L.) lines, derived from the cross between two parents contrasting in terms of root adhesion of A. brasilense, De León et al. [82] identified six quantitative trait loci (QTL) responsible for 23.1% of the phenotypic variation of this trait. Among these, a QTL of greater effect was found to be responsible for 8.63% of this variation. The identification of genes/QTLs related to the plant-Azospirillum interaction may provide numerous molecular markers which, in the future, may be used in marker-assisted selection (MAS) for a successful plant-Azospirillum interaction, contributing to the breeding of plants associated with PGPB.