Soybean Metabolites Regulated in Root Hairs in Response to the Symbiotic Bacterium Bradyrhizobium japonicum^,,

Abstract

Nodulation of soybean (Glycine max) root hairs by the nitrogen-fixing symbiotic bacterium Bradyrhizobium japonicum is a complex process coordinated by the mutual exchange of diffusible signal molecules. A metabolomic study was performed to identify small molecules produced in roots and root hairs during the rhizobial infection process. Metabolites extracted from roots and root hairs mock inoculated or inoculated with B. japonicum were analyzed by gas chromatography-mass spectrometry and ultraperformance liquid chromatography-quadrupole time of flight-mass spectrometry. These combined approaches identified 2,610 metabolites in root hairs. Of these, 166 were significantly regulated in response to B. japonicum inoculation, including various (iso)flavonoids, amino acids, fatty acids, carboxylic acids, and various carbohydrates. Trehalose was among the most strongly induced metabolites produced following inoculation. Subsequent metabolomic analyses of root hairs inoculated with a B. japonicum mutant defective in the trehalose synthase, trehalose 6-phosphate synthase, and maltooligosyltrehalose synthase genes showed that the trehalose detected in the inoculated root hairs was primarily of bacterial origin. Since trehalose is generally considered an osmoprotectant, these data suggest that B. japonicum likely experiences osmotic stress during the infection process, either on the root hair surface or within the infection thread.

The interaction between legumes (Leguminosae) and bacteria in the family Rhizobiaceae leads to the development of a nitrogen-fixing symbiosis. The development of this symbiotic process is complex and is coordinately regulated by the mutual exchange of diffusible signal molecules. Flavonoids, either in root exudates or released from the seed coat during germination, act as specific inducers of the nodulation genes in compatible rhizobia (Subramanian et al., 2007). These bacterial nodulation genes are involved in the production of lipochitooligosaccharide Nodulation (Nod) factors that are essential for inducing root hair deformation and subsequent nodule formation (Oldroyd and Downie, 2008). In addition to these key regulatory molecules, a variety of other compounds, including exopolysaccharides, plant hormones, and vitamins, have been implicated as regulators of the nodulation process (Matamoros et al., 2006; Gibson et al., 2008; Ding and Oldroyd, 2009).

A significant change in the metabolism of both the symbiont and host clearly occurs during the infection process and subsequent nodule development. During the earliest stages of the infection process, the invading rhizobia trigger root hair deformation, depolarization of the cell membrane, cytoskeleton reorientation, calcium spiking, cortical cell division, and the expression of a variety of plant genes (Stacey et al., 2006). Formation of the infection thread, by which the rhizobia gain entry into the plant cell, also clearly requires a drastic change in plant cell metabolism to support the formation of additional cell wall and membrane components. Roth and Stacey (1989) previously estimated that membrane biosynthesis must be elevated 35-fold to support the formation of new membranes during the infection process and nodule development. Plant metabolism must also change to provide carbon sources to support ATP synthesis and to respond to the provision of fixed nitrogen by the symbiont.

Previous studies utilized transcriptomic and proteomic approaches to examine the global changes occurring during legume nodulation. Collectively, these studies identified numerous genes and proteins involved in carbon and nitrogen metabolism, plant defense responses, nutrient exchange, and signal transduction that are significantly regulated in Medicago truncatula, Lotus japonicus, soybean (Glycine max), and Phaseolus vulgaris colonized by their respective rhizobial symbionts (Lohar et al., 2006; Kuster et al., 2007; Benedito et al., 2008; Brechenmacher et al., 2008; Meschini et al., 2008; for review, see Stacey et al., 2006).

Metabolomic approaches have been successfully used to study a variety of plant-pathogen interactions. For example, published studies profiled the metabolomic response of plants infected by pathogenic fungi (Magnaporthe grisea, Sclerotinia sclerotiorum, Penicillium strains, Mycosphaerella graminicola), oomycetes (Pythium sylvaticum), bacteria (Candidatus liberibacter, Pseudomonas syringae, phytoplasma), viruses (Tobacco mosaic virus), and insects (Frankliniella occidentalis; Schmelz et al., 2004; Overy et al., 2006; Jobic et al., 2007; Cevallos-Cevallos et al., 2009; Leiss et al., 2009; Parker et al., 2009; for review, see Allwood et al., 2008). However, few studies have examined the plant metabolome in response to rhizobial infection. Various amino acids (Asn, Gln, Pro), organic acids (threonic acid), sugars (Rib, maltose), and polyols (mannitol) were shown to be more abundant in L. japonicus and M. sativa nodules than in the corresponding roots (Colebatch et al., 2004; Desbrosses et al., 2005; Barsch et al., 2006). However, to our knowledge, no previous studies have sought to characterize metabolites responding during the earliest stages of the rhizobial infection process (e.g. within the first 48 h), when key signaling events are occurring and both host and symbiont are transitioning into the symbiotic state.

In this study, metabolites extracted from soybean roots and root hairs either mock inoculated or inoculated with Bradyrhizobiun japonicum were analyzed by gas chromatography-mass spectrometry (GC-MS) and ultraperformance liquid chromatography-quadrupole time of flight-mass spectrometry (UPLC-QTOF-MS). This analysis resulted in the identification of 166 small molecules whose accumulation was significantly affected by B. japonicum inoculation. Among those compounds showing the most significant response to inoculation was trehalose, which was produced by the invading rhizobium, likely as a response to conditions of osmotic stress.

RESULTS AND DISCUSSION

The aim of this research was to identify metabolites that were significantly regulated in root hairs in response to B. japonicum inoculation. The metabolomic profiles of root hairs and stripped roots, roots from which root hairs were removed, were compared from 0 to 48 h after inoculation with B. japonicum. Mock-inoculated root hairs and stripped root samples served as comparative controls.

Light microscopy was used to estimate the purity and quality of the root hair preparations, as described by Wan et al. (2005) and Brechenmacher et al. (2009). Little or no contamination by epidermal cells was observed. Four independent biological replicates were produced for each time point, and extracted metabolites were analyzed using GC-MS and UPLC-QTOF-MS.

Metabolites Detected in Root Hairs and Stripped Roots

Polar and nonpolar extracts of root hairs and stripped roots were analyzed by GC-MS. A total of 422 and 459 compounds were detected in stripped roots and root hairs from the soluble and lipophilic fractions, respectively (Table I; Supplemental Table S1). However, only 199 of these compounds could be tentatively identified based on available standards and literature data (Supplemental Table S1). Polar metabolites included nucleotides, amino acids, alcohols, organic acids, and sugars, while nonpolar metabolites included fatty acids, sterols, and long-chain alcohols.

Table I. Total number of known and unknown metabolites identified in root hairs and stripped roots by GC-MS (polar and nonpolar) and by UPLC-QTOF-MS (LC-MS).

Metabolite	GC-MS Polar	GC-MS Polar	GC-MS Nonpolar	GC-MS Nonpolar	LC-MS	LC-MS	Total
	No.	%	No.	%	No.	%	No.
Known	123	29.1	76	16.6	435	25.1	634
Unknown	299	70.9	383	83.4	1,294	74.9	1,976
Total	422	100	459	100	1,729	100	2,610

Due to their relatively low abundance and labile nature, secondary metabolites were examined by UPLC-QTOF-MS, since these compounds are not easily identified by GC-MS. This approach identified 1,729 small molecules in root hairs and stripped roots, of which about 25% could be identified using analytical standards (Table I; Supplemental Table S1). The secondary metabolites identified included flavonoids, isoflavonoids, and triterpenoids.

Collectively, the GC-MS and UPLC-QTOF-MS analyses detected a total of 2,610 small molecules in root hairs and stripped roots, of which 634 could be identified as known compounds. Each compound was characterized by its distinct retention time and mass-to-charge ratio (m/z). No metabolites that were specifically expressed only in root hairs or stripped roots were identified using these analyses.

Metabolites Significantly Regulated in Root Hairs in Response to B. japonicum Inoculation

Statistical analysis was performed to identify metabolites significantly regulated in root hairs colonized by B. japonicum. Bacterial inoculation significantly modified the accumulation of 166 metabolites in root hairs (114 were more abundant after inoculation, 50 were less abundant, and two were more or less abundant at different time points of the time course; Table II; Supplemental Table S2). Out of these 166 small molecules, 48 could be identified using standards (Table III). The analytical methods used could not easily distinguish those metabolites produced by the bacterium from those produced by the plant. However, a bacterial or plant origin was hypothesized for the identified metabolites based on their expression levels, the B. japonicum and soybean genome annotation, and literature mining (Supplemental Table S2). A metabolomic analysis of a bacterial liquid culture cannot identify the origin of the small molecules from this study, since B. japonicum metabolism is greatly affected during the development of the symbiosis. All 48 of the metabolites responding significantly to bacterial inoculation were detected in mock-inoculated root hairs and, therefore, are likely of plant origin. However, the B. japonicum genome encodes for enzymes capable of catalyzing the synthesis of 11 metabolites found significantly regulated in B. japonicum-colonized root hairs (Supplemental Table S2). Except for the trehalose that was analyzed in more detail (see below), none of the bacterial genes possibly involved in the synthesis of the 10 compounds was found regulated in B. japonicum treated with genistein, an isoflavone able to induce nod gene expression, and during nodule development (Pessi et al., 2007; Lang et al., 2008), indicating that these 10 compounds are most likely of plant origin. The fold change for each metabolite after inoculation, the P value, and the se for each regulated metabolite at each time point are indicated in Supplemental Table S2. Principal component analysis clearly segregated metabolites that were significantly regulated in root hairs and stripped roots, providing further evidence for the purity of the root hair preparations analyzed (Supplemental Fig. S1). These data also argue for a distinct set of metabolites that are regulated in root hairs in response to inoculation.

Table II. Distribution of metabolites significantly regulated in roots hairs in response to B. japonicum inoculation identified by GC-MS in polar (p) and nonpolar (np) extracts and by UPLC-QTOF-MS (LC-MS).

The number of known and unknown compounds is indicated. The activation (up), the repression (down), or dual regulation (activation at one time point and repression at another time point; both) are mentioned.

Method	All Compounds	Total Known	Total Unknown	Up Known	Down Known	Both Known	Up Unknown	Down Unknown
LC-MS	93	31	62	22	7	2	48	14
GC-MS np	33	3	30	3	0	0	27	3
GC-MS p	40	14	26	8	6	0	6	20
Total	166	48	118	33	13	2	81	37

Table III. Metabolites significantly regulated in root hairs in response to B. japonicum.

The predicted compound origin, the P value (pVal; comparison of inoculated versus uninoculated condition), and the fold change ratio (FC; inoculated/uninoculated) for each time point (t0, t12, t18, t24, t36, and t48 hours after inoculation with the symbiotic bacterium) are indicated. Metabolite abundance was also statistically analyzed by combining all the time points and comparing the metabolite abundance in all inoculated conditions versus all uninoculated conditions (pVal Comb). Significant changes are indicated in boldface.

Compound	Metabolite Origin	pVal t0	FC t0	pVal t12	FC t12	pVal t18	FC t18	pVal t24	FC t24	pVal t36	FC t36	pVal t48	FC t48	pVal Comb
(Iso)flavonoid
Amino-flavonoid	Plant	5.2E-02	2.5	2.4E-02	4.0	4.4E-04	4.8	5.2E-05	3.6	1.5E-03	2.9	1.6E-04	3.7	2.0E-13
Aureusidin	Plant	9.3E-01	1.2	3.5E-01	2.8	7.1E-01	1.3	7.7E-01	1.5	6.4E-01	1.3	2.8E-03	0.5	9.1E-01
Isoliquiritigenin	Plant	8.9E-01	1.0	2.8E-01	1.1	3.2E-03	1.7	3.3E-03	1.6	1.4E-03	1.4	1.1E-02	1.4	1.2E-08
Apigenin 7-glucoronide	Plant	3.8E-01	2.6	9.1E-01	1.4	1.3E-02	1.4	5.3E-01	1.0	7.4E-01	1.2	1.9E-02	0.7	3.8E-01
Dihydrokaempferol b	Plant	2.2E-01	0.8	1.7E-01	2.3	3.1E-01	1.4	7.8E-01	1.1	3.3E-01	1.7	1.3E-01	1.9	7.9E-03
Afrormosin	Plant	3.5E-01	0.9	3.0E-01	1.3	1.2E-02	1.5	5.5E-01	1.3	7.7E-02	1.4	1.4E-01	1.4	5.5E-03
Genkwanin a	Plant	9.0E-01	1.2	7.8E-02	1.5	1.4E-03	2.9	5.9E-01	1.6	1.7E-01	1.4	1.5E-01	1.3	5.2E-04
Prunetin a	Plant	3.0E-01	1.9	8.4E-02	1.5	5.6E-04	2.0	5.0E-01	1.6	1.8E-01	1.3	1.7E-01	1.3	9.2E-04
Liquiritigenin	Plant	8.2E-01	1.0	1.6E-01	1.1	8.7E-03	1.3	2.9E-01	1.2	4.0E-01	1.1	1.7E-01	1.2	2.7E-03
Liquiritin	Plant	8.1E-01	1.0	1.4E-01	1.1	7.5E-03	1.3	2.8E-01	1.2	3.9E-01	1.1	1.8E-01	1.2	2.4E-03
Sissotrin	Plant	1.1E-01	0.9	4.8E-01	1.0	1.4E-01	1.2	1.5E-01	1.3	2.3E-01	1.2	1.9E-01	1.5	4.1E-03
Isorhamnetin 7-α-d-glucosamine	Plant	6.7E-01	1.3	5.5E-01	1.0	2.5E-01	1.6	7.2E-03	0.5	5.5E-01	1.6	2.3E-01	2.1	7.8E-01
Apigenin	Plant	7.7E-01	1.0	2.0E-07	5.0	4.1E-02	4.4	8.8E-01	1.1	9.2E-01	1.0	3.1E-01	0.9	1.8E-01
Prunetin b	Plant	4.4E-02	0.6	8.4E-01	1.1	5.9E-05	4.7	5.4E-01	1.5	5.4E-01	0.8	4.0E-01	1.9	3.2E-02
Delphinidin 3-rhamnoside-5-glucoside	Plant	5.6E-01	0.9	2.0E-01	0.7	2.0E-01	0.5	1.7E-02	0.3	2.6E-01	0.7	6.0E-01	1.2	6.5E-03
Naringenin 4′-O-glucoside	Plant	2.0E-01	9.6	1.5E-02	1.4	3.4E-02	1.6	4.1E-01	1.1	4.0E-02	1.4	6.9E-01	1.1	6.1E-02
Fatty acid
Lignoceric acid	Plant	2.8E-02	0.8	3.2E-01	1.1	8.7E-02	1.4	3.2E-01	1.1	3.8E-01	1.1	3.7E-03	1.3	1.1E-02
α-Eleosteric acid	Plant	3.7E-01	1.2	6.1E-01	1.0	5.6E-01	1.1	6.7E-01	1.1	8.3E-01	0.9	6.0E-03	1.5	4.5E-01
α-Linolenic acid	Plant	3.5E-01	1.1	8.0E-01	1.0	3.7E-01	1.2	4.0E-01	0.9	6.6E-01	0.9	8.4E-03	1.5	5.7E-01
Octadecenoic fatty acid	Plant	6.2E-02	15.1	3.2E-02	2.6	5.2E-04	4.3	1.1E-02	2.6	3.5E-03	3.2	8.5E-03	3.3	1.0E-10
Lauric acid	Plant/bacteria	2.5E-01	0.8	5.9E-02	0.8	7.2E-01	1.0	2.8E-01	0.9	1.8E-01	0.9	3.3E-01	0.9	6.7E-03
Myristic acid	Plant/bacteria	1.8E-01	0.9	1.8E-02	0.9	3.5E-01	0.9	4.2E-01	0.9	2.2E-01	0.9	4.0E-01	0.9	6.2E-03
Amino acid
l-Aspartic acid, N, O, O-TMS	Plant/bacteria	7.1E-02	15.5	2.9E-05	5.0	6.4E-02	15.4	1.6E-03	7.4	3.8E-03	7.0	2.3E-03	6.4	1.0E-07
l-Aspartic acid, O, O-TMS	Plant/bacteria	7.2E-02	3.6	6.8E-05	5.4	5.9E-02	7.3	1.3E-02	9.7	1.7E-05	6.3	2.2E-02	7.9	5.7E-09
l-Aspartic acid, 4TMS	Plant/bacteria	4.3E-01	1.3	1.7E-01	1.6	7.8E-02	2.8	2.0E-02	3.5	9.9E-04	2.3	6.9E-02	3.6	8.2E-05
4-Aminobutyric acid, N, O, O-TMS	Plant	1.3E-01	1.6	2.9E-01	0.8	7.6E-01	1.0	2.6E-03	0.7	2.9E-01	0.9	1.2E-01	0.8	2.1E-02
l-Glutamic acid, O, O-TMS	Plant/bacteria	1.3E-01	1.6	9.7E-01	1.2	6.8E-01	1.4	2.6E-03	0.7	5.5E-01	1.0	1.2E-01	0.8	1.2E-01
Proline, N, O-TMS	Plant/bacteria	7.8E-01	1.2	7.0E-01	1.0	1.5E-01	0.8	4.7E-01	1.1	4.8E-03	0.7	1.4E-01	0.8	1.0E-01
Carboxylic acid
Lactic acid, O, O-TMS	Plant/bacteria	6.6E-01	5.0	6.3E-02	70.1	5.0E-03	54.9	1.2E-02	19.0	1.2E-01	10.8	7.6E-03	64.9	3.2E-08
Benzoic acid, O-TMS	Plant	9.8E-01	1.1	2.4E-01	1.5	4.9E-01	1.2	7.7E-03	1.4	7.3E-01	1.0	2.4E-01	0.9	2.2E-01
Sydnic acid	Plant	1.4E-01	0.8	5.6E-01	1.3	4.1E-02	0.6	7.6E-01	2.0	3.4E-02	2.4	5.5E-01	1.3	6.6E-01
Threonic acid, O, O, O, O-TMS	Plant	2.4E-01	1.7	2.3E-02	1.6	2.5E-02	1.6	6.0E-01	1.1	7.7E-01	1.2	9.4E-01	1.0	1.6E-01
Disaccharide
Trehalose, 8TMS	Plant/bacteria	1.2E-01	1.4	5.2E-05	15.6	4.9E-05	17.6	3.8E-03	7.8	3.0E-03	32.7	9.9E-04	27.4	6.1E-11
Maltose monohydrate MEOX1, TMS	Plant	5.7E-03	1.8	2.1E-01	2.9	1.1E-01	7.0	4.6E-01	1.7	2.8E-03	32.6	4.9E-02	19.3	1.9E-04
Maltose methoxyamine	Plant	2.1E-03	1.6	3.6E-03	1.6	6.2E-03	1.8	6.0E-01	1.0	9.4E-03	1.8	2.3E-01	1.3	4.2E-03
Glucosinolate
4-Hydroyxindol-3-yl-methyl-glucosinolate	Plant	3.2E-01	1.8	2.3E-02	0.6	7.9E-01	1.2	2.2E-01	2.5	6.0E-01	1.1	3.0E-02	2.1	2.7E-01
7-Methylsulfinyl-n-heptyl-glucosinolate	Plant	7.7E-01	1.6	7.7E-01	1.3	9.1E-03	5.5	2.0E-02	0.5	8.4E-01	1.2	9.1E-02	0.6	4.4E-01
Other
Beiwutine	Plant	6.1E-01	0.9	1.0E-01	0.6	6.1E-01	1.4	1.9E-01	0.7	1.8E-01	0.8	2.1E-04	2.7	2.3E-01
O-Methyl inositol, 5TMS	Plant	6.0E-02	0.6	9.2E-01	4.8	5.8E-03	0.2	8.9E-01	6.0	2.6E-01	0.9	2.0E-01	0.4	3.0E-02
Ascochalasin	Plant	8.4E-01	1.0	3.8E-01	0.7	9.1E-01	1.2	3.6E-02	1.6	6.0E-01	1.1	2.8E-02	0.7	7.5E-01
3,3′,4,4′5-Pentahydroxystilbene	Plant	7.5E-01	1.4	8.8E-01	1.1	3.9E-04	2.4	4.8E-01	1.4	4.2E-01	1.3	2.9E-01	1.6	9.7E-03
4’,5′-Trihydroxy-3′,6′′,6′′-trimetylpyranol (2′′,3′′:6′′,5′′)dihydrochalcone	Plant	1.2E-01	0.8	1.7E-01	3.0	6.2E-03	2.3	2.7E-01	0.8	3.5E-01	0.8	7.8E-01	0.9	8.9E-01
Neoxanthin	Plant	4.7E-01	1.3	5.2E-01	2.3	3.8E-01	0.8	1.7E-01	0.6	3.2E-03	0.6	1.1E-01	0.6	1.5E-02
Medicagenic acid 3-O-triglucoside	Plant	4.4E-01	2.6	1.2E-01	7.2	4.1E-01	1.8	3.6E-01	0.8	8.1E-03	2.3	4.9E-01	1.5	3.6E-02
α-Glycerophosphate ester, TMS	Plant	8.1E-01	1.2	2.4E-02	1.7	3.0E-01	1.6	5.4E-01	1.7	3.7E-02	3.1	5.9E-01	1.2	1.9E-02
Cytosine, N, N-TMS	Plant/bacteria	2.2E-01	0.6	2.2E-01	0.8	1.6E-02	0.5	3.3E-01	1.0	9.0E-02	0.7	5.8E-01	0.9	2.5E-03
Xylose methoxyamine	Plant/bacteria	4.5E-02	2.0	4.4E-01	0.9	3.4E-01	1.2	5.4E-01	1.0	6.7E-01	1.0	8.6E-03	0.8	3.3E-01
Rha-Hex-Hex-Hex-gypsogenic acid	Plant	2.5E-01	1.3	2.7E-01	1.1	7.0E-03	1.3	9.4E-01	1.0	1.6E-02	1.3	2.1E-01	1.1	5.9E-02

Metabolites were classified into various functional categories. Many compounds, including various flavonoids and isoflavonoids, amino acids, carboxylic acids, fatty acids, and carbohydrates, were found to be significantly regulated in root hairs colonized by B. japonicum (Fig. 1). Their possible functions during the nodulation process are discussed below. The fold change ratios and statistical values of all these compounds are summarized in Table III.

Figure 1.

Distribution of root hair metabolites regulated in response to B. japonicum and identified by GC-MS and UPLC-QTOF-MS into eight functional categories.

(Iso)flavonoids

Sixteen metabolites belonging to the (iso)flavonoid family were clearly responsive to inoculation by B. japonicum. Bacterial inoculation triggered the induction in root hairs of afrormosin (18 h), amino-flavonoid (12, 18, 24, 36, 48 h), apigenin (12, 18 h), dihydrokaempferol b, genkwanin a (18 h), naringenin 4′-O-glucoside b (12, 18, 36 h), liquiritin (18 h), liquiritigenin (18 h), isoliquiritigenin (18, 24, 36, 48 h), sissotrin, and prunetins (18 h; Fig. 2; Table III; Supplemental Fig. S2; Supplemental Table S2). In contrast, the accumulation of apigenin 7-glucoronide (48 h), aureusidin (48 h), isorhamnetin 7-α-d-glucosamine (24 h), and delphinidin 3-rhamnoside-5-glucoside (24 h) was reduced in root hairs upon B. japonicum inoculation. Moreover, none of these compounds was significantly regulated in stripped roots in response to bacterial inoculation.

Figure 2.

Relative abundance of lactic acid, amino-flavonoid, trehalose, Asp, octadecenoic acid, and an unknown compound in root hairs (RH) and stripped roots (STR) 0, 12, 18, 24, 36, and 48 h after inoculation with the symbiotic bacterium B. japonicum (I) or mock inoculated (NI). Asterisks indicate significantly regulated compounds colonized by B. japonicum. se values are also indicated.

The (iso)flavonoid class of compounds is of particular interest since they are essential for the infection process and are known to induce the nodulation genes in Bradyrhizobium (Subramanian et al., 2007). Daidzein and genistein are soybean isoflavones required for the establishment of the nitrogen-fixing symbiosis (Subramanian et al., 2006). Although these two isoflavones were detected in root hairs and stripped roots, their levels were not significantly elevated in response to B. japonicum inoculation. In contrast, several other (iso)flavonoids (listed above) were significantly increased in root hairs inoculated by B. japonicum (Fig. 2; Table III; Supplemental Fig. S2; Supplemental Table S2). Liquiritigenin, isoliquiritigenin, apigenin, and prunetin were reported to induce nod gene expression in Sinorhizobium meliloti, Sinorhizobium medicae, Rhizobium leguminosarum, or B. japonicum (Kape et al., 1992; Zuanazzi et al., 1998; Broughton et al., 2000; Zhang and Cheng, 2006; Yokoyama, 2008; Watkin et al., 2009). The abundance of liquiritigenin and isoliquiritigenin was increased in P. vulgaris root exudates when common bean was inoculated with R. leguminosarum in comparison with the mock-inoculated control (Bolanos-Vasquez and Werner, 1997). The two latter flavonoids also possess antifungal and antibacterial activities (Machado et al., 2005; Kusuma and Tachibana, 2007; Cui et al., 2008) and might be compounds involved in controlling B. japonicum development. Apigenin levels were also elevated in alfalfa (Medicago sativa) and soybean grown under conditions of elevated atmospheric CO₂ (Agrell et al., 2004; Kretzschmar et al., 2009) and increased in response to herbivore damage in alfalfa (Agrell et al., 2003). Conjugation of apigenin with glycosides inactivates the molecule (Scervino et al., 2006). This may explain the observation that apigenin 7-glucuronide was down-regulated in B. japonicum-inoculated root hairs 48 h after inoculation (Table III; Supplemental Fig. S2; Supplemental Table S2). Conjugates of delphinidin and isorhamnetin were also repressed in root hairs 24 h after bacterial inoculation (Table III; Supplemental Fig. S2; Supplemental Table S2).

The naringenin 4′-O-glucoside conjugate accumulated to higher levels in root hairs 12, 18, and 36 h after inoculation with B. japonicum compared with the mock-inoculated control (Table III; Supplemental Fig. S2; Supplemental Table S2). Naringenin is known to be an inducer of nod genes in Rhizobium, to stimulate nodulation, and to directly interact with NodD, the key regulator of nod gene expression (Begum et al., 2001; Novak et al., 2002; Li et al., 2008). However, further study is required to determine the role of naringenin 4′-O-glucoside in the early events of soybean nodulation.

Afrormosin levels were also up-regulated in root hairs inoculated with B. japonicum (Table III; Supplemental Fig. S2; Supplemental Table S2), and this compound was also found in M. truncatula cell cultures treated with a yeast elicitor and methyl jasmonate (Farag et al., 2008). Afrormosin was also implicated in soybean insect resistance (Caballero et al., 1986).

Taken together, the metabolomic data indicated that only a subset of (iso)flavonoid compounds were induced in root hairs in response to B. japonicum inoculation. While these compounds play a key role in the symbiotic process by inducing the bacterial nodulation genes, they likely have a variety of other biological functions. For example, isoflavonoids are also strong inhibitors of polar auxin transport (Subramanian et al., 2007). However, this function does not appear to be essential for soybean nodulation (Zhang et al., 2009).

Flavonoid compounds are precursors of dihydrochalcones (Nazareno et al., 2000; Kamara et al., 2005). The 2′,4′,5′-trihydroxy-3′,6′′,6′′-trimetylpyranol(2′′,3′′:6′′,5′′) dihydrochalcone was up-regulated in root hairs 18 h after inoculation with B. japonicum (Table III; Supplemental Fig. S2; Supplemental Table S2). Dihydrochalcone derivatives are known to possess antifungal and antibacterial properties (Awouafack et al., 2008; Pontais et al., 2008; Poumale et al., 2008). Therefore, an interesting question is whether the production of this compound adversely affects B. japonicum growth during the infection process.

Amino Acids

The amino acids Pro, Glu, Asp, and 4-aminobutyric acid (GABA) were significantly regulated in root hairs colonized by B. japonicum (Fig. 2; Table III; Supplemental Fig. S2; Supplemental Table S2). Asp was significantly more abundant in root hairs 12, 24, 36, and 48 h after inoculation, as well as in stripped roots at similar time points. In contrast, Pro, Glu, and GABA were less abundant in root hairs 24 and 36 h after inoculation. Pro is known to have an osmoprotective function during the plant response to drought, cold, salt, and heavy metal treatment (Cook et al., 2004; Singh et al., 2004; Gong et al., 2005; Charlton et al., 2008; Shulaev et al., 2008; Urano et al., 2009). Furthermore, Pro was also shown to accumulate in nodules of Cicer arietinum and Vicia faba during salt stress (Trinchant et al., 1998; Nandwal et al., 2007). Nitrogen fixation was less sensitive to osmotic stress in M. truncatula plants overaccumulating Pro (Verdoy et al., 2006). As discussed below, the production of trehalose by B. japonicum during infection of root hairs suggests that the bacteria may be responding to osmotic stress. In contrast, the reduction of Pro in root hairs after inoculation suggests that the plant cell may not experience osmotic stress or that B. japonicum colonization triggers a different regulation of compounds involved in osmotic stress tolerance.

Glu is a precursor of GABA. Accumulation of the latter compound was reduced in root hairs 24 h after inoculation (Table III; Supplemental Fig. S2; Supplemental Table S2). Interestingly, a Glu dehydrogenase gene involved in Glu synthesis was also found down-regulated in root hairs inoculated by B. japonicum (Libault et al., 2010). GABA accumulated very quickly in rice (Oryza sativa) during the germination process and was suggested to function as a signaling molecule in plants, as it does in animal systems (Bouche and Fromm, 2004; Howell et al., 2009). GABA was also implicated in regulating pollen tube development (Yang, 2003; Yu and Sun, 2007). GABA is known to accumulate in response to abiotic stresses, including drought, cold, heat, salt, ozone exposure, mechanical damage, and anoxia (Kinnersley and Turano, 2000; Cho et al., 2008). GABA is also involved in the plant defense reaction against insects (Shelp et al., 1999; Bouche and Fromm, 2004) and was induced in rice cells in response to fungal elicitor (Takahashi et al., 2008). The reduction of Glu and GABA in response to B. japonicum inoculation could be interpreted as the result of an active suppression of a plant innate immunity response. This conclusion is supported by our recent transcriptomic results showing that transcripts known to be associated with the plant innate immunity system were initially elevated but then repressed within 48 h after B. japonicum inoculation (Libault et al., 2010).

In contrast to Glu, GABA, and Pro, which were repressed in root hairs colonized by B. japonicum, Asp levels were significantly elevated in root hairs and stripped roots after inoculation (Fig. 2; Table III; Supplemental Fig. S2; Supplemental Table S2). Prell and Poole (2006) suggested that Asp is produced by bacteroids in nodules and plays a critical part in the nutrient exchange process between symbiont and host. The current data, however, do not allow firm conclusions as to whether the Asp seen in root hairs is of bacterial or plant origin.

Fatty Acids

Six fatty acids were significantly regulated only in root hairs in response to bacterial inoculation. Octadecenoic acid was more abundant in root hairs from 12 to 48 h after inoculation with B. japonicum (Fig. 2; Table III; Supplemental Fig. S2; Supplemental Table S2). α-Eleosteric, α-linolenic, and lignoceric acids accumulated in root hairs 48 h after B. japonicum inoculation (Table III; Supplemental Fig. S2; Supplemental Table S2). The release of fatty acids from the membrane affects its fluidity and is involved in the plant tolerance to biotic and abiotic stresses (Upchurch, 2008). Linolenic acid is the most abundant fatty acid in plant membranes, and its accumulation in B. japonicum-colonized root hairs may be involved in modifying membrane fluidity necessary for the bacteria to colonize the cell. Another possibility is that elevation of this fatty acid, as well as the others, may simply be a reflection of an up-regulation of metabolic pathways necessary for the synthesis of new membranes required during the infection process.

However, fatty acids also have other biological activities. For example, free α-linolenic acid possesses antifungal activity and was found to accumulate in potato (Solanum tuberosum) leaves infected with Pseudomonas syringae, in parsley (Petroselinum crispum) cells treated with a fungal elicitor, and in Silybum marianum cell cultures exposed to yeast extract (Kirsch et al., 1997; Gobel et al., 2002; Walters et al., 2004; Sanchez-Sampedro et al., 2007). These data suggest that the accumulation of free α-linolenic acid may be an indicator of an active plant defense response. Trienoic fatty acids also play a role in the plant response to abiotic stresses, including cold and heat, as well as wounding (Conconi et al., 1996; Matsuda and Iba, 2005). Linolenic acid is a precursor of jasmonic acid, which plays a critical role in plant responses to biotic and abiotic stresses (Wasternack, 2007). Incubation of B. japonicum with either jasmonate or methyl jasmonate was reported to induce expression of the nod genes and enhance nodulation and nitrogen fixation (Mabood and Smith, 2005).

Root hairs colonized by B. japonicum accumulated more octadecenoic acid (Fig. 2; Table III; Supplemental Fig. S2; Supplemental Table S2). Accumulation of oleic acid, which is an octadecenoic acid, was detected in mycorrhizal M. truncatula roots (Stumpe et al., 2005; Schliemann et al., 2008). Mycorrhizal and nitrogen-fixing symbioses require the synthesis of plant membranes surrounding the fungi and bacteria, respectively. Given the projected need for a 35-fold increase in membrane biogenesis for infection thread formation and subsequent symbiosome membrane production in nodules (Roth and Stacey, 1989), it is more likely that the noted increases in the presence of free fatty acids may reflect a cell in the process of up-regulating membrane biosynthetic pathways.

In contrast to those free fatty acids that accumulated to higher levels after inoculation, lauric and myristic acids were repressed in root hairs 12 h after inoculation with B. japonicum (Table III; Supplemental Fig. S2; Supplemental Table S2). Lauric acid exhibits antifungal properties, and its repression in root hairs colonized by B. japonicum may reflect a down-regulation of the soybean defense reaction (Walters et al., 2003). To our knowledge, changes in myristic acid levels in response to plant-microbe interaction have not been previously reported. N-Myristoylation regulates weak interactions of proteins with the cell membrane as well as protein-protein interactions. Many proteins involved in signal transduction are N-myristoylated (Farazi et al., 2001), which suggests that the repression of myristic acid, leading to a presumed decrease of protein N-myristoylation, could have an effect on the cell signaling processes at work during nodulation.

Carboxylic Acids

Lactic acid accumulated specifically in root hairs 18, 24, and 48 h after B. japonicum inoculation, but not in stripped roots (Fig. 2; Table III; Supplemental Fig. S2; Supplemental Table S2). Lactic acid was previously reported to be 4.8 times more abundant in alfalfa nodules than in roots and accumulated in nodules during salt stress (Swaraj and Bishnoi, 1999; Barsch et al., 2006). Polymers of lactic acid were shown to stimulate plant growth in corn (Zea mays) and soybean (Kinnersley et al., 1990; Chang et al., 1996). However, further studies are required to determine a possible role for lactic acid in root hairs colonized by B. japonicum.

Likewise, benzoic acid was significantly up-regulated only in B. japonicum-infected root hairs 24 h after inoculation (Table III; Supplemental Fig. S2; Supplemental Table S2). Benzoic acid itself exhibits antifungal activity, and pretreatment of plants with this compound increases plant resistance to fungi (Khan et al., 2001; Williams et al., 2003; Shabana et al., 2008). Similarly, benzoic acid levels increase in response to both bacterial and viral infection (Shapiro and Gutsche, 2003; Matsuura et al., 2009). The level of benzoic acid was also shown to increase in plants treated with manganese and cadmium and was correlated with improved plant resistance to cold, drought, and heat (Senaratna et al., 2003; Pal et al., 2005; Fuhrs et al., 2009). Benzoic acid was shown to be metabolized to salicylic acid in tobacco (Nicotiana tabacum) during ozone stress (Ogawa et al., 2005). However, salicylic acid abundance was not modified in root hairs after B. japonicum inoculation (Supplemental Table S1).

Threonic acid is another carboxylic acid that was significantly up-regulated in root hairs 12 and 18 h after inoculation with B. japonicum (Table III; Supplemental Fig. S2; Supplemental Table S2). Threonic acid is enriched in L. japonicus nodules in comparison with other plant organs (Desbrosses et al., 2005) and arises from ascorbic acid, which is a strong antioxidant (Loewus, 1999). Increased activities of antioxidant enzymes as well as accumulation of hydrogen peroxide occur during the early colonization of M. sativa by S. meliloti (Bueno et al., 2001). Hydrogen peroxide was also shown to be critical for the development of the infection thread within the root hair (Jamet et al., 2007). The levels of reactive oxygen species increase very rapidly in root hairs treated with Nod factor (Cardenas et al., 2008). Altogether, these data suggest that a tight regulation of the oxidative burst is required for soybean root hair colonization by the symbiotic bacteria. The accumulation in B. japonicum-colonized root hairs of threonic acid, which is degradation product of ascorbic acid, might be part of this regulation.

Carbohydrates

The pentose Xyl was repressed in root hairs in response to B. japonicum 48 h after inoculation (Table III; Supplemental Fig. S2; Supplemental Table S2). Xyl can be used as a carbon source by B. japonicum and other rhizobia under both normal and salt stress conditions (Wagner et al., 1995; Ghosh et al., 2005; El Idrissi and Abdelmoumen, 2008). The level of Xyl was also shown to decrease in wheat (Triticum aestivum) roots treated with calcium (Hossain et al., 2006). Interestingly, Xyl exudation was reduced from wheat roots infected by Helminthosporium sativum, the agent responsible for root-rot disease (Jalali and Suryanarayana, 1971). Further experiments are required to determine whether the repression of Xyl in B. japonicum-colonized root hairs is due to its use as a carbon source by the symbiotic bacterium, to a plant stress response, or to an elevation in intracellular calcium levels, which is known to occur in root hairs during the infection process (Talukdar et al., 2009).

Trehalose: A Disaccharide Required for Root Hair Colonization

The inoculation of root hairs by B. japonicum triggered the accumulation of the disaccharides trehalose (12, 18, 24, 36, 48 h) and maltose (36, 48 h; Table III; Fig. 2; Supplemental Fig. S2; Supplemental Table S2). Trehalose is a nonreducing disaccharide of Glc whose metabolic pathway is implicated in regulating embryo, leaf, and flower development. In many cases, an accumulation of trehalose is correlated with a stress response, especially in response to osmotic stress (Iordachescu and Imai, 2008).

During the early events of the nitrogen-fixing symbiosis, trehalose can be synthesized by both plant and bacterial cells. In order to determine if the trehalose accumulation seen in root hairs was of plant or bacterial origin, root hairs were inoculated with a B. japonicum mutant strain defective in trehalose synthase (treS), trehalose 6-phosphate synthase (otsA), and maltooligosyltrehalose synthase (treY) activities. This mutant produced little or no trehalose in culture (Sugawara et al., 2010). Root hairs were collected 48 h after inoculation, since this coincided with the highest accumulation of trehalose when plants were inoculated with wild-type B. japonicum (Fig. 2). The amount of trehalose in root hairs inoculated with the wild type was 16 times greater than those inoculated with the triple mutant, clearly indicating that the accumulation of trehalose in root hairs after inoculation is most likely due to bacterial synthesis (Fig. 3). However, we cannot rule out some plant synthesis of trehalose, since a higher accumulation of this compound was observed in root hairs inoculated with the B. japonicum triple mutant than in mock-inoculated root hairs (Fig. 3). Furthermore, a soybean trehalose phosphatase gene was found up-regulated in root hairs colonized by B. japonicum (Libault et al., 2010), suggesting that a small amount of the accumulated trehalose in root hairs might be of plant origin. It is also possible that the bacterial triple mutant synthesizes a limited amount of trehalose.

Figure 3.

Average trehalose abundance (bars) determined by GC-MS in root hairs (RH) and stripped roots (STR) inoculated with B. japonicum wild type (WT) or the triple mutant for trehalose synthesis (TM) and in the uninoculated control (UN). The abundance of trehalose for each of the four replicates and each condition is also included (diamonds, squares, triangles, crosses). Asterisks indicate significant differences in trehalose accumulation. se values are also indicated.

Trehalose is known to accumulate and/or confer resistance to plants in response to drought, cold, and salt stress (Garcia et al., 1997; Garg et al., 2002; Pramanik and Imai, 2005; Iordachescu and Imai, 2008) and is induced in Arabidopsis (Arabidopsis thaliana) and cucumber (Cucumis sativus) infected by fungal pathogens (Brodmann et al., 2002; Abood and Losel, 2003). Trehalose also accumulated in ectomycorrhized and endomycorrhized roots, particularly during drought stress (Schellenbaum et al., 1999; Pfeffer et al., 2004; Lopez et al., 2007). Genetic engineering of tobacco, potato, and tomato (Solanum lycopersicum) transformed with trehalose biosynthetic genes conferred higher resistance to drought, low and high temperature, salt, and oxidative stresses (Yeo et al., 2000; Almeida et al., 2005; Cortina and Culianez-Macia, 2005). Trehalose was detected in nodules of M. truncatula, P. vulgaris, and soybean, where it accumulated in response to water stress (Muller et al., 1996; Farias-Rodriguez et al., 1998; Streeter and Gomez, 2006; Lopez et al., 2008; Suarez et al., 2008). This disaccharide is also produced by bacteroids in soybean nodules (Muller et al., 1996). Trehalose synthesis by soybean during the early events of the nitrogen-fixing symbiosis might be part of a stress response triggered by bacterial colonization or infection, perhaps indicating conditions that confer osmotic stress. Unfortunately, the methods used cannot distinguish between trehalose produced by B. japonicum cells attached to the surface of the root hair and those cells contained within the infection threads.

It was previously described that three different pathways, showing different regulation in in vitro culture and in nodules, exist in B. japonicum for the synthesis of trehalose and involved three genes: treS (encoding trehalose synthase), otsA (encoding trehalose 6-phosphate synthase), and treY (encoding maltooligosyltrehalose synthase; Streeter and Gomez, 2006). However, it was recently suggested that treS is involved in the catabolism of trehalose to maltose (Sugawara et al., 2010). The expression levels of treS, otsA, and treY were quantified by real-time PCR in sprayed root hairs 12 h after inoculation with B. japonicum and in liquid culture. All three genes were strongly expressed in B. japonicum sprayed root hairs. The elevated expression of these genes is also consistent with the conclusion that trehalose accumulation in sprayed root hairs is primarily of bacterial origin. Little or no expression of treS, otsA, and treY was found in B. japonicum liquid cultures (Fig. 4).

Figure 4.

B. japonicum treS, otsA, and treY gene expression levels determined by real-time PCR in root hairs (RH) 12 h after inoculation and in liquid culture (LC). se values are also indicated.

Maltose levels were elevated in root hairs in response to B. japonicum inoculation. The enzyme trehalose synthase (treS) was suggested to convert trehalose to maltose (Sugawara et al., 2010). Therefore, it is likely that the accumulation of maltose seen in the inoculated root hairs reflects the catabolism of trehalose.

Trehalose accumulation in B. japonicum was shown to be crucial for its survival during desiccation (Streeter, 2003). Free-living rhizobia were shown to synthesize trehalose in response to phenanthrene, hyperosmolarity, high temperature, salt stress, very low oxygen levels, and desiccation (Hoelzle and Streeter, 1990; Ghittoni and Bueno, 1996; Dardanelli et al., 1997, 2000; Cytryn et al., 2007; Essendoubi et al., 2007; McIntyre et al., 2007; Keum et al., 2008). Furthermore, the resistance of Rhizobium etli to salt stress was improved in a strain overexpressing a trehalose 6-phosphate synthase gene and decreased in a strain mutated in the same gene (Suarez et al., 2008). Trehalose accumulation in rhizobia appears to be a general bacterial response to stress, suggesting that trehalose induction in B. japonicum-sprayed root hairs during the early events of the nitrogen-fixing symbiosis is due to a perceived stress during the infection process. Consistent with this hypothesis, nodule formation by the B. japonicum ΔtreS, ΔotsA, ΔtreY mutant is significantly reduced in comparison with the wild type (Sugawara et al., 2010). These findings are also consistent with the report by Ampomah et al. (2008), who demonstrated that S. meliloti strains mutated in genes involved in trehalose catabolism (presumably allowing trehalose accumulation) were more competitive than the wild type, particularly at the stage of infection thread development.

Other Compounds

O-Methyl inositol, a polyol, was significantly less abundant in root hairs 18 h after inoculation with B. japonicum (Table III; Supplemental Fig. S2; Supplemental Table S2). Methyl inositol was implicated in drought or salt stress tolerance through the stabilization of membranes and proteins in soybean, alfalfa, and the ice plant (Mesembryanthemum crystallinum; Fougere et al., 1991; Muller et al., 1996; Guo and Oosterhuis, 1997; Streeter et al., 2001; Agarie et al., 2009). O-Methyl inositol is also secreted from soybean roots (Timotiwu and Sakurai, 2002) and exhibits antibacterial and antifungal activities (Tan et al., 1999; Agnese et al., 2001). Therefore, its repression in root hairs inoculated with B. japonicum may prevent the inhibition of bacterial growth.

Ascochalasin accumulated in root hairs 24 h after inoculation with B. japonicum, but levels decreased at 48 h (Table III; Supplemental Fig. S2; Supplemental Table S2). Ascochalasin is a member of the cytochalasin family, which includes molecules with antibiotic and antifungal properties (Bottalico et al., 1990), and therefore would also have the potential of affecting bacterial growth during the infection process.

The levels of 4-hydroxy-indole-3-yl-methyl-glucosinolate were reduced in root hairs 12 h after inoculation with B. japonicum but increased at 48 h (Table III; Supplemental Fig. S2; Supplemental Table S2). 7-Methylsulfinyl-n-heptyl-glucosinolate abundance increased in root hairs 18 h after B. japonicum inoculation (Table III; Supplemental Fig. S2; Supplemental Table S2). Glucosinolates are secondary metabolites that function in plant defense against insects and pathogens (Hopkins et al., 2009; van Dam et al., 2009). Indole glucosinolate was also suggested to be a precursor of auxin (Ludwig-Muller, 2009), a hormone implicated in nodule development (Mathesius, 2008).

Medicagenic acid 3-O-triglucoside levels also increased in root hairs 36 h after B. japonicum inoculation (Table III; Supplemental Fig. S2; Supplemental Table S2). Medicagenic acid and its glycoside derivatives exhibit antifungal and nemacidal activities (Zehavi and Polacheck, 1996; Argentieri et al., 2008). However, to our knowledge, they have never been tested for activity against B. japonicum.

CONCLUSION

Metabolomic analysis of soybean roots and root hairs following B. japonicum inoculation identified a variety of compounds, most of which were previously implicated in various plant-microbe interactions. The regulation of several metabolites (e.g. GABA, O-methyl inositol, Glu, lauric acid) suggests the B. japonicum may actively suppress plant defense pathways during infection. This conclusion is strongly supported by transcriptomic studies of root hair mRNA taken at similar time points and under identical treatment conditions (Libault et al., 2010). The regulation of other compounds, such as isoflavonoids and free fatty acids, may be due to their signaling role during the symbiosis or may reflect the need for increased cellular biosynthesis (e.g. membrane lipids) to create the structural components (e.g. infection thread) needed to support symbiotic development.

The methods used in this study did not allow a distinction between metabolites produced by the invading symbiont and those produced by the plant host. This is both a limitation and an advantage: the latter since it allows conclusions to be drawn about the interaction between host and symbiont. For example, the noted accumulation of trehalose in the inoculated root hairs likely reflects bacterial synthesis but strongly suggests that the symbiont is perceiving a stressful environment during the earliest stages of the infection process. Previous studies suggested that the response of bacteroids within the infected nodule cells may reflect a perception of osmotic stress or a related stress (Boscari et al., 2006). For example, a variety of protein chaperone proteins are induced in bacteroids (Djordjevic, 2004). However, our data suggest that the symbiont likely responds to the stress of a plant intracellular lifestyle very early during the infection process.

The soybean root hair is a single, differentiated cell type. The methods used in this work, as well as similar publications from our laboratory (Wan et al., 2005; Brechenmacher et al., 2009; Libault et al. 2010) seek to provide a thorough functional genomic description of this single-cell model system by using B. japonicum inoculation as a probe for physiological changes. It is hoped that the final integration of these data sets will provide a foundation for single-cell, systems biology studies of plant function.

MATERIALS AND METHODS

Bradyrhizobium japonicum Growth, Root Hair, and Stripped Root Isolation

Sterilization of the soybean seeds (Glycine max ‘Williams 82’) was performed by soaking seeds twice in 20% bleach for 10 min each. Seeds were then rinsed five times in sterile water, neutralized for 10 min in 0.1 n HCl, and rinsed five more times in sterile water. Sterilized seeds were germinated in a dark growth chamber (25°C, 80% humidity) for 3 d on nitrogen-free B&D agar medium (Broughton and Dilworth, 1971).

B. japonicum strain USDA110 was grown at 30°C for 3 d in HM medium (HEPES, 1.3 g L⁻¹; MES, 1.1 g L⁻¹; Na₂HPO₄, 0.125 g L⁻¹; Na₂SO₄, 0.25 g L⁻¹; NH₄Cl, 0.32 g L⁻¹; MgSO₄ 7H₂O, 0.18 g L⁻¹; FeCl₃, 0.004 g L⁻¹; CaCl₂ 2H₂O, 0.013 g L⁻¹; yeast extract, 0.25 g L⁻¹; d-Ara, 1 g L⁻¹; sodium gluconate, 1 g L⁻¹, pH 6.6) containing 20 μg mL⁻¹ chloramphenicol. Cells were centrifuged, washed twice with sterile water, diluted to an optical density at 600 nm of 0.8 in sterile water, and used to inoculate the 3-d-old seedlings using a perfume sprayer in order to consistently deliver a thorough inoculant dose. Control seedlings were mock inoculated with sterile water. The B. japonicum ΔotsAΔtreSΔtreY triple mutant defective in trehalose synthesis was grown as described above with the following additions: 100 μg mL⁻¹ streptomycin and 100 μg mL⁻¹ spectinomycin (Sugawara et al., 2010).

Soybean roots (about 1,500 per experiment) were collected 0, 12, 18, 24, 36, and 48 h after inoculation by cutting and allowing the roots to fall directly into liquid nitrogen. The roots were gently stirred for 20 min to break off root hairs from the roots. The liquid nitrogen slurry was filtered through a wire mesh to separate root hairs from stripped roots (roots with no root hairs). Root hairs and stripped roots were stored at −80°C until metabolite extraction. Four biological replicates were produced for each time point. This procedure was previously used to successfully isolate root hairs largely free of any other root cell contamination. (Wan et al., 2005; Brechenmacher et al., 2009).

GC-MS

GC-MS analyses of root metabolites were carried out as described previously (Broeckling et al., 2005). Root hairs and stripped roots were lyophilized until dry. The dried root tissues were homogenized using mortar and pestle, and 6 mg was transferred into a 4-mL glass vial. Metabolites were extracted by adding 1 mL of chloroform containing 10 μg mL⁻¹ docosanol (the standard for nonpolar metabolites), vortexed, and incubated for 45 min at 50°C. HPLC-grade water (1.5 mL) containing 25 μg mL⁻¹ ribitol (the standard for polar metabolites) was added and incubated a second time at 50°C for 45 min. The two phases were separated by centrifugation at 2,900g for 30 min at 4°C. One milliliter of each layer was collected into a vial and dried in a rotary evaporator for the polar layer or under nitrogen for the nonpolar layer. Methoxyamine hydrochloride (50 μL at 15 mg mL⁻¹) in pyridine was added to the polar metabolites. Compounds were sonicated until they were completely dissolved and incubated at 50°C for at least 1 h for methoximation. Silylation was performed by incubating the samples for 1 h at 50°C with 50 μL of N-methyl-N-trimethylsilyltrifluoroacetamide and 1% trimethylchlorosilane (Pierce Biotechnology). The sample was equilibrated to room temperature, transferred to a 200-μL glass insert, and analyzed using an Agilent 6890 GC apparatus coupled to a 5973 MSD scanning from m/z 50 to 650. One microliter of the sample solution was injected at a 15:1 split ratio, and the inlet and transfer line were held at 280°C. Separation was achieved on a 60-m DB-5MS column (J&W Scientific; 0.25 mm i.d., 0.25 μm film thickness) at a constant flow of 1.0 mL min⁻¹ with a temperature program of 80°C for 2 min, ramped at 5°C min⁻¹ to 315°C, and held for 12 min.

The nonpolar compounds were resuspended in 0.8 mL of chloroform and hydrolyzed for 4 h at 50°C by adding 0.5 mL of HCl (1.25 m) in methanol. HCl and solvent were removed by evaporation under nitrogen. Solubilization was achieved by sonication in 70 μL of pyridine and incubation at 50°C. Derivatization was performed, as described above, by using 30 μL of N-methyl-N-trimethylsilyltrifluoroacetamide and 1% trimethylchlorosilane. Metabolites were analyzed by GC-MS as described above for the polar compounds using a 1:1 split ratio.

UPLC-QTOF-MS

Root hairs and stripped roots were lyophilized until dry and ground to a fine powder. Ten milligrams of the powder was extracted for 2 h with 1 mL of 80% methanol containing 0.018 mg mL⁻¹ umbelliferone as an internal standard. Samples were centrifuged at 2,900g for 30 min at 4°C, and the supernatants were collected. Five microliters of the supernatant was injected into the Waters UPLC system coupled to a quadrupole-time of flight mass spectrometer (QTOF Premier). Separation of metabolites was achieved on a 2.1- × 150-mm i.d., 1.7-μm UPLC BEH C18 column (Waters Acquity) using the following gradient: mobile phase B (acetonitrile) increased from 5% to 70% over 30 min, then to 95% in 3 min, held at 95% for 3 min, and returned to 95% mobile phase A (0.1% acetic acid in water) for equilibration for 3 min. The flow rate of the mobile phases was 0.56 mL min⁻¹, and the column temperature and the autosampler temperature were maintained at 60°C and 4°C, respectively. Mass spectral data were acquired from m/z 50 to 2,000 in the negative ion electrospray mode, with the nebulization gas at 850 L h⁻¹ (350°C) and the cone gas at 50 L h⁻¹ (120°C). Raffinose (m/z 503.1612) was used as the reference compound in the independent lock-mass mode, with the lock mass scan (1 s) collected every 10 s for the accurate mass measurement. The concentration of raffinose was 50 fmol mL⁻¹, and the flow rate 0.2 mL h⁻¹.

Metabolomic Data Analysis

For the GC-MS data, metabolites were identified through spectral and retention time matching against a custom library compiled with authentic standards prepared in an identical manner and then against the National Institutes of Standards and Technology library for confirmation and/or identification of compounds not included in the custom database. For the UPLC-MS data, tentative identifications were made by matching the retention time and the accurate mass (m/z) of ions to those in a custom library compiled with authentic standards and/or plant metabolites previously reported in the literature. Relative metabolite abundances were calculated using custom MET-IDEA software to extract peak areas of individual ions characteristic of each component (Broeckling et al., 2006). Peak areas were normalized to the internal standard, and the normalized peak areas were used in statistical analyses. Principal component analysis was performed on normalized data sets using Pirouette (InfoMetrix). A t test was performed to identify metabolites significantly regulated in root hairs and stripped roots in response to B. japonicum inoculation. All metabolites having P < 0.01 for at least one time point or P < 0.05 for more than one time point were considered significantly regulated.

Gene Expression Quantification

Trizol reagent was used to extract RNA from soybean root hairs 12 h after inoculation with B. japonicum and from B. japonicum grown in liquid culture according to the manufacturer's protocol (Invitrogen). RNAs were subsequently treated with DNase (Ambion) to remove genomic DNA contamination according to the manufacturer's instructions. One microgram of total RNA was employed as template for cDNA synthesis as described by Brechenmacher et al. (2008), modified by the use of a mixture of random primers. Primers used to amplify treS (blr6767), otsA (bll0322), and treY (blr6771) genes are described by Cytryn et al. (2007). PCR conditions, control of the absence of genomic DNA, and the method to quantify the transcripts are identical to those in Brechenmacher et al. (2008) except for the use of a penicillin-binding protein (bll0910) to normalize the data (Chang et al., 2007).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. Principal component analysis of metabolites significantly regulated in root hairs and stripped roots.

• Supplemental Figure S2. Relative abundance of known compounds in root hairs and stripped roots after inoculation with the symbiotic bacterium B. japonicum or mock inoculated.

• Supplemental Table S1. Normalized data of metabolites detected in root hairs and stripped roots.

• Supplemental Table S2. Metabolites significantly regulated in root hairs in response to B. japonicum inoculation.