Root exudates drive interspecific facilitation by enhancing nodulation and N₂ fixation

Significance

Plant diversity often leads to an increase in ecosystem productivity, but the underpinning mechanisms remain poorly understood. We found that faba bean/maize intercropping enhances productivity, nodulation, and N₂ fixation of faba bean through interspecific root interactions. We provide a mechanism explaining how maize promotes N₂ fixation of faba bean, where root exudates from maize increase root hair deformation and nodulation in faba bean, double exudation of flavonoids (signaling compounds for rhizobia), and up-regulate the expression of a chalcone–flavanone isomerase gene involved in flavonoid synthesis, and genes mediating nodulation and auxin responses. Our results provide a mechanism for facilitative root–root interactions explaining how species diversity may enhance ecosystem productivity with important implications for developing sustainable agriculture.

Abstract

Plant diversity in experimental systems often enhances ecosystem productivity, but the mechanisms causing this overyielding are only partly understood. Intercropping faba beans (Vicia faba L.) and maize (Zea mays L.) result in overyielding and also, enhanced nodulation by faba beans. By using permeable and impermeable root barriers in a 2-y field experiment, we show that root–root interactions between faba bean and maize significantly increase both nodulation and symbiotic N₂ fixation in intercropped faba bean. Furthermore, root exudates from maize promote faba bean nodulation, whereas root exudates from wheat and barley do not. Thus, a decline of soil nitrate concentrations caused by intercropped cereals is not the sole mechanism for maize promoting faba bean nodulation. Intercropped maize also caused a twofold increase in exudation of flavonoids (signaling compounds for rhizobia) in the systems. Roots of faba bean treated with maize root exudates exhibited an immediate 11-fold increase in the expression of chalcone–flavanone isomerase (involved in flavonoid synthesis) gene together with a significantly increased expression of genes mediating nodulation and auxin response. After 35 d, faba beans treated with maize root exudate continued to show up-regulation of key nodulation genes, such as early nodulin 93 (ENOD93), and promoted nitrogen fixation. Our results reveal a mechanism for how intercropped maize promotes nitrogen fixation of faba bean, where maize root exudates promote flavonoid synthesis in faba bean, increase nodulation, and stimulate nitrogen fixation after enhanced gene expression. These results indicate facilitative root–root interactions and provide a mechanism for a positive relationship between species diversity and ecosystem productivity.

Many ecosystems, including grasslands (1, 2), forests (3), phytoplankton communities (4), and cropping systems (5), show a positive relationship between plant diversity and ecosystem productivity. Several mechanisms have been proposed to explain this relationship. A “sampling effect” occurs, because more diverse mixtures have a higher probability of containing a species with higher productivity (6). Complementarity effects occur when species vary in resource use and niche differentiation in space or time, leading to greater resource utilization (6–9). Facilitation occurs when one species increases the growth of other species through a wide range of processes (10). Facilitative effects may be direct (e.g., by shade or protection from harsh conditions) or indirect (e.g., when one species reduces attack by pathogens or herbivores on other species) (11–13).

Legume/cereal intercropping systems have been widely studied in the context of diversity and ecosystem function and commonly overyield, because dinitrogen (N₂) fixation by legumes increases ecosystem nitrogen (N) supply (7, 8), an example of facilitation. This facilitation is important to agriculture on a large scale, because application of chemical N fertilizer decreases biological N₂ fixation by legumes. Intercropping with maize increases N₂ fixation by faba bean, even with high input of N fertilizer (7), but such a stimulatory effect on faba bean has not been observed in all legume/cereal intercropping systems (7), suggesting species-specific facilitative relationships (14). Plants communicate by an “underground highway”—root exudates, which inhibit or facilitate their neighbors (15)—and part of this communication is through flavonoids, which are key signals in nodulation of legumes. How exactly maize promotes faba bean nodulation and the potential role of flavonoids remain unclear. Here, we explore cross-talk between maize and faba bean through rhizosphere processes and physiological and molecular mechanisms underpinning this communication.

Results and Discussion

In 2 y of field experiments (field study 1), plastic partitions were used to separate the roots of intercropped faba bean and maize (Fig. S1). We found that, when root interactions were possible (no root partition), faba bean biomass and grain overyielded by 35% and 61%, respectively, with no N fertilization (P < 0.0001) and when fertilized with 150 kg ha⁻¹ N, overyielding percentages were 28% and 35%, respectively (P < 0.0001) (Table S1). These results are similar to those reported previously (7). Also, intercropped faba bean in the absence of root partition produced 32% more dry weight of nodules (F = 11.6; P = 0.004) and a 36% increase in N derived from air (Ndfa; F = 20.1; P < 0.0001). N fertilization decreased the dry weight of nodules per plant by 27% (F = 26.1; P = 0.004) (Fig. 1A), and Ndfa decreased by 22% (F = 23.6; P < 0.0001) (Fig. 1B). However, root interactions between faba bean and maize enhanced N₂ fixation, even under high N fertilization.

Fig. S1.

Diagram showing (A) root partition, (B) no root partition between faba bean and maize roots, and (C) crop row arrangements in field study 1. The partition was 2-m long and 1-m deep with 0.12-mm-thick plastic. One intercropping combination included a 0.4-m faba bean strip (two rows of faba bean with a 0.2-m interrow distance) and a 0.8-m maize strip (two rows of maize with a 0.4-m interrow distance). Interplant distance within the same row was 0.2 m for faba bean and 0.3 m for maize.

Table S1.

Biomass and grain yield of intercropped maize and faba bean with and without root barriers under field conditions (field study 1)

N application (kg ha−1)	Faba bean				Maize
	Biomass (kg ha−1)		Grain yield (kg ha−1)		Biomass (kg ha−1)		Grain yield (kg ha−1)
	Barrier	No barrier	Barrier	No barrier	Barrier	No barrier	Barrier	No barrier
0	5.77a	7.81b	3.96a	6.39b	6.70a	6.37a	6.47a	6.98a
150	5.77a	7.40b	4.22a	5.69b	10.54a	9.61a	11.99a	10.77a

Values for grain yield are averages of 2 y (2006 and 2007). Values in the same row followed by different letters are significantly different (P < 0.05).

Fig. 1.

Dry weight of (A) nodules and (B) biological N fixation of faba bean as affected by root partition and N fertilization (field study 1). Values are means (n = 3) ± SE. White bars indicate root partition, and black bars indicate no root partition. *Significant differences between solid root partition and no root partition treatments.

Most studies of leguminous monoculture systems show that N application inhibits nodulation and N₂ fixation (16). When intercropped with cereals, rhizosphere N availability decreases (17), and this decrease presumably drives greater nodulation and N₂ fixation of legumes (7, 8). However, the effect of cereals on nodulation is not consistent. Nodulation and N₂ fixation are not enhanced when faba bean is intercropped with wheat (7), which is considered to be more strongly competitive for N than maize (18). This inconsistency suggests that a low soil N availability is not the only driver of increased nodulation by legumes in mixtures.

Wheat and barley, both of which are stronger competitors for N than maize (18), cause a greater decrease in rhizosphere N availability than maize. They are, therefore, expected to increase nodulation by faba bean if a low soil N availability is the primary mechanism driving increased nodulation rather than show fewer effects than shown by intercropping with maize (19). Therefore, we conducted a greenhouse study in which faba bean was intercropped with three cereals (maize, wheat, or barley), which differ in their ability to compete for N and reduce its availability in soil (greenhouse study 1). When grown with maize, faba bean nodulation increased by 69% for nodule number per pot (F = 10.5; P < 0.05) and 63% for nodule dry mass per pot (F = 6.91; P < 0.05) compared with faba bean grown with wheat and by 58% for nodule number per pot (F = 7.53; P < 0.05) and 46% for dry nodule mass per pot (F = 4.06; P = 0.059) compared with faba bean grown with barley (Fig. S2). However, soil nitrate levels in the rhizosphere of faba bean grown with maize were similar to those when grown with barley or wheat because of the level of N application in the experiment (Fig. S2). Thus, the facilitative effect of maize roots on faba bean productivity and nodulation by intercropping is specific to maize and not solely related to decreasing rhizosphere N availability. Therefore, we tested if root exudates from maize are an essential factor in this facilitative effect.

Fig. S2.

Relationship between (A) nodule number and (B) nodule weight of faba bean and NO₃⁻ concentration in soil (greenhouse study 1): ● and — refer to faba bean intercropping with maize, * and ▬▬▬ refer to faba bean intercropping with barley, and ▲ and ─ ─ refer to faba bean intercropped with wheat.

Nodule-forming rhizobia initiate the deformation of root hairs followed by penetration of rhizobia and formation of nodules by cortical cells (20). Such responses to inoculation occur within 2–6 min of contact (21) and reach the most intensive root–hair deformation after 12 h (22). We examined the early response of faba bean root hairs to rhizobia with a root–hair deformation assay (greenhouse study 2). Maize root exudates significantly stimulated the deformation 12 h after inoculation with rhizobia, with 33% deformed root hairs of faba bean when treated with exudates vs. 13% deformed root hairs in the control (Fig. 2). Exudates from barley or wheat roots had no effects on root hair deformation (Fig. 2).

Fig. 2.

Faba bean root hair deformation 12 h after inoculation with rhizobia (R. leguminosarum bv. viciae; greenhouse study 2). Values with different letters are significantly different (P ≤ 0.05). Error bar is an SD (n = 4). CK−, control (sterilized water) without inoculation; CK+, control (sterilized water) with inoculation; CK+B, sterilized barley root exudates with inoculation; CK+M, sterilized maize root exudates with inoculation; CK+W, sterilized wheat root exudates with inoculation.

We also conducted a two-layer pot experiment (greenhouse study 3) to determine whether the effect of root exudates on initial nodulation responses was manifest in mature nodule size and N₂ fixation rates. When 1 mM nitrate was supplied to the quartzite root medium, maize root exudates significantly increased the number of medium-sized (1–3 mm in diameter) nodules on faba bean roots (Fig. 3A) as well as their fresh weight (Fig. 3B) compared with the control and wheat or barley root exudates. Also, the rate of N₂ fixation by faba bean was enhanced by maize root exudates (Fig. 3 C and D). Neither barley nor wheat root exudates affected faba bean nodulation. When N application was increased from 0 to 1 mM, faba bean roots treated with maize root exudates retained the same high numbers of medium-sized nodules, but in contrast, medium-sized nodule numbers in the control were reduced by more than 50% (Fig. 3A).

Fig. 3.

Effects of root exudates and N supplies on (A) medium-sized nodule number and (B) fresh weight, (C) acetylene reduction activity (ARA) of root nodules, and (D) Ndfa of faba bean (greenhouse study 3). Bar groups with different capital letters indicate a significant difference (P < 0.05) between N supply levels. Bars with different lowercase letters indicate a significant difference (P < 0.05) among four root exudate addition treatments within the same N supply level. Error bar is an SD (n = 4). B+, sterilized barley root exudates with inoculation; CK, control (sterilized water) with inoculation; M+, sterilized maize root exudates with inoculation; N, 2 mM nitrate; N0, no nitrate application; N1/2, 1 mM nitrate; W+, sterilized wheat root exudates with inoculation.

Plants may release large amounts of root exudates (23), which may provide carbon for bacteria or act as signal molecules with the potential to play a role in root–root interactions among plant species (24). For instance, both extracts and exudates of maize roots induce lipochitooligosaccharide production in Bradyrhizobium japonicum, which initiates nodulation on legume roots (24). Our results (greenhouse study 4) indicate that root exudates from maize did not increase rhizobial growth (Fig. S3). In contrast, root exudates from wheat did enhance rhizobial growth in pure cultures (Fig. S3). Of the three species, only maize enhanced nodulation through root exudates. Therefore, signal molecules in the root exudates of maize likely play an important role in activating rhizobia.

Fig. S3.

Rhizobia growth curve (OD at 600 nm; greenhouse study 4). Results are the mean value of six replicates. *OD with wheat root exudates was different significantly (P < 0.05) from those treated with water, barley, or maize root exudates at a specific time point.

Several flavonoids, including naringenin, hespertin, genistein, and 7,4′-dihydroxyflavone, induce Nod genes in Rhizobium leguminosarum (25, 26). Therefore, we investigated the effects of root exudates in faba bean/maize and faba bean/wheat intercropping systems and their corresponding monoculture systems on these flavonoids (greenhouse study 5). We found that the concentration of genistein in the solution grown with maize root exudates alone was similar to that of faba bean exudates alone. In contrast, the genistein concentration in root exudates collected from a mixture of maize and faba bean was 73% greater than that of the mean of the two crops grown alone (Fig. S4). Thus, the enhancement of nodulation was likely caused by the enrichment with genistein in root exudates when faba bean was intercropped with maize.

Fig. S4.

Concentration of genistein in agar medium-grown maize alone (M), maize/faba bean mixture (M/F), faba bean alone (F), wheat/faba bean mixture (W/F), and wheat alone (W; greenhouse study 5). The slash between different crops indicates the mixture of the crops; otherwise, they are grown alone. Bars with different superscript letters are significantly different (P < 0.05). Error bar is an SD (n = 3).

To unravel the underlying molecular mechanism of the stimulatory effects of maize root exudates on faba bean nodulation, we analyzed expression patterns of a series of key genes regulating flavonoid biosynthesis, nodulation, and N₂ fixation in faba bean roots (greenhouse study 6). Chalcone–flavanone isomerase (CFI) catalyzes flavanone synthesis, resulting in flavonoid accumulation after a wide array of external stimuli (27, 28). We found that expression of the CFI gene showed an 11-fold increase in faba bean roots after the addition of root exudates from maize compared with that without addition of root exudates (Fig. 4A), indicating a major up-regulation of flavonoid synthesis in faba bean roots 12 h after being treated with maize root exudates. Such a rapid response supports a critical signaling role of flavonoids in root–root communications between intercropped faba bean and maize. In legumes, early nodulin-like (ENODL) genes are important for cell differentiation and cell wall reorganization during nodulation (29). The expression of the gene [nodulin-like 4 (NODL4)] encoding a nodulin-like protein was also enhanced 1.8-fold at this early stage (Fig. 4B), indicating that the nodulation program was triggered as early as 12 h after root–root interrecognition by root exudates; 24 h after treatment, an auxin-responsive gene (GH3.1) and ENODL2 displayed 1.5-fold increases (Fig. 4 C and D), implying nodulation initiation under the regulation of auxin signaling at this stage. Consistent with our field and greenhouse studies, a key nodulation gene, early nodulin 93 (ENOD93), showed a 3.6-fold increase in its expression level (Fig. 4E) together with a 4.5-fold increase in expression of another nodulin-like gene (NODL4) (Fig. 4F), clearly indicating that nodulation was strongly stimulated in faba bean roots when treated with maize root exudates compared with that in the control. Importantly, we detected a 2.4-fold increase in the expression of a putative N₂ fixation gene (FixI) (Fig. 4G), providing strong evidence for physiologically enhanced N₂ fixation in treated faba bean roots at the molecular level.

Fig. 4.

Relative gene expression levels of (A) a CFI, (B) a nodulin-like gene (NODL4), (C) an auxin-responsive GH3 family gene (GH3.1), (D) ENODL2, (E) ENOD93, (F) a nodulin-like gene (NODL4), (G) N fixation gene FixI of faba bean treated with maize root exudates (12 h for A and B, 24 h for C and D, and 35 d for E–G; greenhouse study 6). White bars represented gene expression levels in the faba bean root treated with water (FC), and black bars denoted gene expression levels in the root treated with maize root exudates (FM). Error bar is an SE (n = 4). *Significant difference between two treatments (P < 0.05; t test).

Our results provide another example of beneficial effects of species diversity in agricultural systems and underpin the fundamental signaling role of flavonoids in early interspecies communication in an intercropping system (Fig. 5). Maize root exudates stimulate flavonoid biosynthesis partially through up-regulating CFI gene expression and trigger nodulation processes very rapidly after exposure; nodulation continues under the regulation of auxin signaling and the nodulin-like protein (Fig. 5). At nodule maturity, maize root exudates continue to up-regulate key nodulation genes, such as ENOD93, and promote N₂ fixation (Fig. 5). Collectively, our results provide a previously unidentified mechanism for how maize promotes N₂ fixation in faba bean in a two-crop intercropping system: maize root exudates contain significant flavonoids and promote flavonoid synthesis in faba bean, thus triggering nodulation with the involvement of hormone signaling and stimulating N₂ fixation through enhanced activities of proteins involved in N₂ fixation at the gene expression and physiological levels (Figs. 2, 3, 4, and 5 and Fig. S4).

Fig. 5.

The root–root interactions driven by maize root exudates stimulate nodulation and N fixation of faba bean in the maize/faba bean intercropping system. Maize root exudates induce significant up-regulation of expression of CFI, NODL4, GH3.1, ENODL2, FixI, and ENOD93 in faba bean roots, which, at least partially, provides an explanation for deformation of root hairs, nodulation, and nodule maturation. Up-regulation of CFI expression enhances genistein synthesis and secretion. Genistein released from maize and faba bean roots promotes rhizobium enrichment and release of nod factors, which facilitate nodulation.

Flavonoids are ubiquitous and diverse secondary metabolites in plants. They play multiple roles in plants, including providing flower colors involved in attracting pollinators, protection against high levels of UV radiation, and defense against pathogens (30). Root exudates commonly contain flavonoids (31), which may be signals in the establishment of not only legume–rhizobium but also, plant–arbuscular mycorrhizal symbioses, and agents in plant defense and allelopathic interactions (32–35). Because the vast majority of terrestrial plants are arbuscular mycorrhizal (36), it is expected that roots of many plant species release flavonoids under at least some conditions. Therefore, the findings for Zea mays are likely more general in species mixtures, although there were no such effects of wheat and barley root exudates in this study.

These findings provide a perspective for ecosystem functioning as well as sustainable agriculture aimed at enhancing land utilization efficiency and reducing environmental pollution risk originating from inorganic fertilizer application in modern agriculture. Our results also enhance our understanding of the positive relationship between productivity and plant diversity in ecosystems. Plant communities comprise various functional groups, including grasses, legumes, and non–N₂-fixing forbs. N that is symbiotically fixed by legumes is often one of the important driving forces in positive relationships between productivity and biodiversity (1, 2). Previous studies have shown biologically fixed N enrichment in soil, the movements of N from legumes to nonlegumes, and stimulating effects of N competition leading to lowering of the soil N concentration, promoting nodulation and N₂ fixation in legumes. Our findings propose a previously unidentified mechanism underlying facilitation by an underground highway: chemical signals from root exudates.

Methods

Field Study 1 (Root Barrier Experiment).

A 2-y root partition field experiment was conducted in 2006 and 2007 at the Baiyun Experimental Site in Western Gansu Province, China. Soil characteristics have been reported previously (8). Our previous study showed that faba bean has relatively shallow roots and a limited lateral root distribution, and few roots spread under maize plants, whereas the maize roots spread under faba bean plants in an intercropping situation (37). Therefore, we used root partition, as control, to eliminate root interactions between faba bean and maize. Based on the root distribution of faba bean, we assumed that there was little influence of the partition on faba bean growth. We used a two-factor split plot design with three replications based on a faba bean (Vicia faba L. cv. Lincan No. 5) and maize (Z. mays L. cv. Shendan No. 16) intercropping system. The main plot had two N application rates, 0 and 150 kg N ha⁻¹, applied as urea, and the subplot was root-partitioned, including a plastic partition between faba bean and maize (Fig. S1A) (only aboveground interaction existed), and without root partition (Fig. S1B) (both above- and belowground interactions existed). The partition was 2-m long and 1-m deep with 0.12-mm-thick plastic. One intercropping combination included a 0.4-m faba bean strip (two rows of faba bean with a 0.2-m interrow distance) and a 0.8-m maize strip (two rows of maize with a 0.4-m interrow distance). Interplant distance within the same row was 0.2 m for faba bean and 0.3 m for maize (Fig. S1C). The individual plot area was 3.6 × 5.5 m, and there were three faba bean and maize strips on each intercropping plot. Faba bean was sown on March 18, 2006 and March 18, 2007 and harvested on August 1, 2006 and July 29, 2007, respectively. Maize was sown on April 19, 2006 and April 15, 2007 and harvested on October 4, 2007 and October 1, 2007, respectively. During the growing period, all plots were irrigated and weeded. All of plots received 75 kg P ha⁻¹ as triple superphosphate.

Twenty faba bean shoots were sampled 21, 42, and 84 d after maize germination, then harvested, and separated into grain and vegetative parts. All samples were sundried and weighed. Grain and vegetative parts of the 20 plants were milled and mixed thoroughly, of which 20 g was used for N analysis and 2 g was further ground with a mortar and pestle for δ¹⁵N analysis after oven drying to constant weight. Twenty plants of maize were sampled and processed as described for faba bean above, but they were only used for dry weight analysis. N was analyzed with a micro-Kjeldahl procedure after digestion in a mixture of concentrated H₂SO₄ and H₂O₂. N derived from atmospheric N₂ fixation (percentage Ndfa) was measured by the natural abundance method (38, 39), and δ¹⁵N was determined using a mass spectrometer (DELTA^plus XP; Thermo Finnigan Electron Corporation).

Maize was selected as a reference crop. The δ¹⁵N value of soil (20-cm depth) in our experimental field was 7.56‰, which allowed us to use the natural abundance method in this study. The percentage Ndfa was calculated with the following equation (38, 39):

$$\%\text{Ndfa}=\frac{{\delta }^{15}{\text{N}}_{\text{maize}}-{\delta }^{15}{\text{N}}_{\text{fababean}}}{{\delta }^{15}{\text{N}}_{\text{maize}}-B}\times 100.$$

Whole-shoot δ¹⁵N values of faba bean obtained from the field experiment were calculated from the weighted values of the component parts. B is the δ¹⁵N value of faba bean grown within a nutrient solution free of N. After harvest and oven drying, the whole faba bean shoot was ground with a mortar and pestle for B value determination; for the faba bean cultivar used in this field experiment, B was −1.851.

Greenhouse Study 1 (Faba Bean Nodulation as Affected by Associated Barley, Maize, and Wheat).

PVC pots, 6.3 L in volume, were separated by a plastic sheet barrier or were not separated by a barrier, so that they provided two types of root interactions, including a plastic sheet barrier to eliminate root contact and solute movement and no barrier to permit root intermingling and exudate exchange between two species. The pots were filled with field soil, which contained 400 mg kg⁻¹ total N, 6 mg kg⁻¹ NH₄N, 27 mg kg⁻¹ NO₃N, 21 mg kg⁻¹ Olsen P, and 101 mg kg⁻¹ K, and pH was 8.3. There were two N application rates (50 and 150 mg kg^-1 soil) with four replicates for each treatment. Basal nutrients were mixed into the soil: P, 100 mg kg⁻¹; K, 126 mg kg⁻¹; Mg, 50 mg kg⁻¹; Fe, 5 mg kg⁻¹; Mn, 5 mg kg⁻¹; Cu, 5 mg kg⁻¹; Zn, 5 mg kg⁻¹; and Mo, 5 mg kg⁻¹ soil. Faba bean plants (V. faba L. cv. Lincan No. 5) were intercropped with maize (Z. mays L. cv. Zhengdan 958), wheat (Triticum aestivum L. cv. Yongliang No. 4), and barley (Hordeum vulgare L. cv. Ganpi No. 4). All faba bean plants were inoculated with rhizobium NM353 (R. leguminosarum bv. viciae was provided by Chen Wenxin, China Agricultural University, Beijing, China). After 43 d of growth, faba bean shoots and roots were harvested, and the biomass and N concentration of faba bean plants and nodulation were measured. Rhizosphere soil was collected for nutrient tests at harvest.

Greenhouse Study 2 (Root Hair Deformation Assay).

Root exudates of cereal plants were collected in advance. Plants of maize (Z. mays L. cv. Nongda No. 108), wheat (T. aestivum L. cv. Yongliang No. 4), and barley (H. vulgare L. cv. Ganpi No. 4) were cultured in sterilized quartzite from germination to the point when they had grown three leaves. Roots from 12 seedlings of each species were washed with sterilized water and transferred to a 50-mL container containing 12 mL Fåhraeus medium (40, 41) afterward for additional culture. After 2 h of culturing under light, plants were discarded, and the Fåhraeus medium was collected and divided into four parts as a root exudate treatment for each species. All procedures were carried out under sterilized conditions.

Faba bean (V. faba L. cv. Lincan No. 5) seeds were surface sterilized with 15% (vol/vol) H₂O₂ for 15 min and 75% (vol/vol) alcohol for 5 min, then thoroughly washed with sterilized water, and germinated in a 50-mL sterilized centrifugal tube containing 20 mL 3% (wt/vol) agar medium with one seed per tube. Germinated seeds with 4- to 5-cm root length were transferred to a new tube containing 5 mL solid agar medium at the bottom and 12 mL liquid Fåhraeus medium containing two drops (0.1 mL) of suspension of rhizobium strain NM353 (∼10⁸ cells·L⁻¹) and 3 mL root exudates. The seedlings were grown at 22 °C with light for 12 h, after which the root hair deformation assay was conducted. Roots were microscopically examined, and the number of deformed root hairs in the susceptible zone was determined based on 60 root hairs in a microscope field. Four roots were used for each root exudation treatment.

Greenhouse Study 3 (Two-Layered Pot Experiment 1).

The experiment was a 3 (N application) × 4 (three root exudates and sterilized water) random factorial design with four replicates for each treatment. Faba bean (V. faba L. cv. Lincan No. 5) seeds were surface sterilized with 15% (vol/vol) H₂O₂ for 15 min and 75% (vol/vol) alcohol for 5 min, germinated, and then thoroughly washed with sterilized water. The seeds were inoculated with a suspension of rhizobium strain NM353 (∼10⁸ cells·L⁻¹) and sown in a sterilized quartzite bed. Plants were grown in a greenhouse under natural light conditions for 4 d; then, seedlings at the same growth stage were transplanted to a two-layered pot system (42), in which the upper pot was filled with 500 mL quartzite medium, and the lower pot was filled with 1 L nutrient solution (P, 250 μM; K, 1,850 μM; Fe, 100 μM; Mg, 100 μM; Mn, 1 μM; Zn, 1 μM; Cu, 0.1 μM; Mo, 0.005 μM). NO₃⁻ was applied in culture solution at levels of 0, 1, or 2 mM; 3 d later, 10 mL rhizobium suspension was added to each upper pot. Faba bean plants were cultured in this pot system for 28 d. Each day, 10 mL solution containing root exudates of maize, wheat, or barley was added separately to the upper pot as root exudate treatment, and deionized water was added as control. After harvesting, the biomass and N concentration of faba bean plants, nodulation, and acetylene reduction activity of root nodules were determined, and biological N₂ fixation was measured by the ¹⁵N natural abundance method (38).

Root exudates were collected as described above. Sterilized seeds of maize, wheat, and barley were sown in a sterilized quartzite bed; 5 d after sowing, 24 seedlings of each species were collected and transferred into 200 mL sterilized water for collecting root exudates.

Greenhouse Study 4 (Rhizobia Cultivation Experiment).

Ten-milliliter portions of yeast mannitol broth medium supplemented with 5 mL root exudates of maize, wheat, or barley were dispensed into 50-mL centrifuge tubes and then, inoculated with 25 μL suspension of rhizobium strain NM353. Control tubes were added with 5 mL sterilized saline solution and 25 μL suspension of rhizobium strain NM353. Six replicates were made for each treatment. Tubes were incubated at 28 °C in a rotary shaker (150 rpm) for 48 h. Growth curves were measured for rhizobium in each of the treatments.

Greenhouse Study 5 (Flavonoids from Root Exudates in Mixture and Monocropping).

There were five treatments including maize/faba bean and wheat/faba bean mixtures and wheat, maize, and faba bean grown alone, and they were grown in containers with 0.3% algar and full nutrient solution for 14 d. Seeds for maize and faba bean were sterilized with 70% (vol/vol) alcohol for 10 min and rinsed with sterilized distilled water three times and then, sterilized with 15% (vol/vol) H₂O₂ for 20 min and rinsed with sterilized distilled water three times. Seeds of wheat were sterilized with 70% (vol/vol) alcohol for 2.5 min and rinsed with sterilized distilled water three times and then, sterilized with 2.5% (wt/vol) sodium hypochlorite solution for 15 min and rinsed with sterilized distilled water four times. Afterward, the sterilized seeds were soaked in sterilized distilled water for 24 h, and then, they were germinated in a growth chamber (22 °C for 16 h under light/18 °C for 8 h under dark cycles) for 72 h. The germinated seeds were transplanted into 1-L containers filled with 200 mL 0.3% agar culture medium containing full nutrients, in which the nutrient composition was the same as that in greenhouse study 3, with the exception of a single N level (1 mM nitrate). They included one faba bean and two maize plants, one faba bean and three wheat plants, two maize plants, three wheat plants, or one faba bean plant. After 14 d, all plants were removed from containers, and the agar solution was filtered through a 0.22-μm filter and frozen to dryness. The dry samples were dissolved in 30 mL methanol (chromatographically pure) and then, shaken in a rotary shaker (150 rpm) for 12 h. The solution was filtered through a 0.22-μm filter, and evaporated to dryness on a rotary evaporator (35 °C). The product was dissolved and made to 1.5 mL with methanol, and then, it was filtered through a 0.22-μm filter. The flavonoids (daidzein and genistein) in the filtered solution were examined by HPLC.

Greenhouse Study 6 (Nodulation-Related Gene Expression).

The device for culturing faba bean and the processes for collecting root exudates from maize are the same as those in the greenhouse study 3, with the exception of a single N supply level at half-strength N (1 mM nitrate). Faba bean plants were inoculated with rhizobium strain NM353.

Quantitative Real-Time PCR.

Total RNA was extracted from the faba bean roots with TRIzol Reagent (Invitrogen). Total RNA (5 µg) purified with the DNase I (TaKaRa Biomedicals) was used for first-strand cDNA synthesis following the M-MLV First-Strand Synthesis Method (Invitrogen). Primers used for gene amplification are listed in Table S2. Quantitative real-time PCR was carried out in an iQ5 Optical System (Bio-Rad) using the SYBR Premix Ex Taq Kit (TaKaRa Biomedicals). Briefly, the reaction included a preincubation at 95 °C for 2 min followed by 40 cycles of denaturation at 95 °C for 10 s, annealing at 60 °C for 30 s, and extension at 72 °C for 30 s followed by melting curve generation and a cooling step. The PCR products were loaded on 1.5% (wt/vol) agarose gels, separated, and photographed after staining with ethidium bromide. The relative expression level was calculated using the comparative C(t) (PCR cycle at which the fluorescent signal of the reporter dye crosses an arbitrarily placed threshold) method (43). Results were normalized according to the expression level of actin1.

Table S2.

Primer sequences used for quantitative real-time PCR (greenhouse study 6)

Primer name	Sequence (5′–3′)
CFI forward	TTCCATTTGGTGTTTGTCTGT
CFI reverse	CTGAACTCATTGAATCCCTTG
A nodulin-like protein forward	ACGCCTAGACCTCTTTGGAATG
A nodulin-like protein reverse	ATAACTCGGAAGCTGTTGGGAC
GH3.1 forward	TGCTTCCTCGGAATGCTACT
GH3.1 reverse	CGACTCGGTAACGGTTCAAT
ENODL2 forward	AACTCGAAGGCGCTATTGAA
ENODL2 reverse	CTCCATCACCGATGTCTCCT
ENOD93 forward	ATCCGAATCCACGATTGAAG
ENOD93 reverse	ACTATTGCCACTGCCATTCC
FixI forward	TCCTAGCTGCTTTTGCCAAT
FixI reverse	TACGGTGCTCAACTCCTTCC

Statistical Analyses.

Data in the figures were reported as means ± SD for eight replicate analyses, whereas all other data were reported as the mean only. A double-factor test was used for variance (ANOVA) analysis (SAS Institute Inc.). Significance level was set at P < 0.05. A Pearson correlation test was conducted to determine correlations among means (P < 0.05).