•The direct analysis of phytosiderophores (PSs) and their metal complexes in plants is critical to understanding the biological functions of different PSs. Here we report on a rapid and highly sensitive liquid chromatography-electrospray ionization-quadrupole-time of flight-mass spectrometry (LC-ESI-Q-TOF-MS) method for the direct and simultaneous determination of free PSs and their ferric complexes in plants.
•In addition to previously reported PSs – deoxymugineic acid (DMA), mugineic acid (MA) and epihydroxymugineic acid (epi-HMA) – two more PSs, avenic acid (AVA) and hydroxyavenic acid (HAVA), were identified by this method in roots of Hordeum vulgare cv Himalaya and in root exudates under iron (Fe) deficiency.
•The two identified PSs could be responsible for Fe acquisition under Fe deficiency because of their relative abundance and ability to form ferric complexes in secreted root exudates.
•This LC-ESI-Q-TOF-MS method greatly facilitates the identification of free PSs and PS–Fe complexes in one plant sample.
The release of certain metabolites from roots is one of the major responses of plants to a range of biotic and abiotic stresses (Gobran et al., 2000). Under nutrient deficiency, plant roots secrete specific compounds such as organic acids (Jones et al., 1996) and phytosiderophores (PSs) (Takagi, 1976; Takemoto et al., 1978) into the rhizosphere to increase metal bioavailability by changing the soil pH or by binding to minerals (Badri & Vivanco, 2009). PSs are nonproteinogenic amino acids that belong to the mugineic acid (MA) family. The biosynthesis and secretion of PSs are specific to graminaceous plants (Ma & Nomoto, 1996; Kobayashi et al., 2010). The acquisition of iron (Fe) from soil by grasses via PSs and a chelation-based mechanism is called strategy II, whereas nongrass plants use a reduction-based mechanism termed strategy I (Marschner et al., 1986; Kim & Guerinot, 2007).
To chelate Fe (III), grasses produce and secrete PSs into the rhizosphere through the TOM1 transporter (Nozoye et al., 2011), and the resulting PS–Fe (III) complexes can be transported into plants through Yellow-Stripe 1 (YS1) and YS1-like family transporters (Curie et al., 2001, 2009).
Phytosiderophores are known to mobilize and chelate a wide range of metals in addition to Fe, including zinc (von Wiren et al., 1996; Suzuki et al., 2006), cadmium (Shenker et al., 2001), copper, nickel, and manganese (Gobran et al., 2000; Meda et al., 2007). Therefore, highly sensitive and simple analytical methods for simultaneous identification of different PSs and their metal complexes are critical for elucidating the detailed functions of PSs in plant nutrition and toxicology.
Since the first identification of PSs, several techniques have been used for the study of PSs. High-performance liquid chromatography (HPLC) has been mainly used with fluorescence detection after post-column o-phthaldialdehyde derivatization (Kawai et al., 1987), with UV detection after pre-column phenylisothiocyanate derivatization (Howe et al., 1999), and with pulsed amperometric and atomic absorption detection (Weber et al., 2001, 2002). In recent years, electrospray ionization mass spectrometry (ESI-MS)-coupled approaches have been widely used to detect free PSs and PS–metal complexes. Direct separation and identification of PSs and their metal complexes involve zwitterionic hydrophilic interaction chromatography (ZIC-HILIC) coupled with ESI-MS (Xuan et al., 2006). With this method, PSs and their complexes are separated on a ZIC-HILIC column. In addition, capillary electrophoresis-MS (CE-MS) with UV or conductivity detection has revealed the importance of electrophoretic analysis in the study of PS–metal complexes without complex stability problems (Xuan et al., 2007). PSs and their different metal complexes were identified efficiently by CE-ESI-MS and CE-inductively coupled plasma (ICP)-MS (Dell’mour et al., 2010). Although these electrophoretic methods have good sensitivity and separate metal species well, traditional LC-based approaches are still useful for studies of PSs, nicotianamine (NA, a precursor of PSs) and their metal complexes.
A quick and sensitive method designed to quantify NA and deoxymugineic acid (DMA) involves LC-ESI-time-of-flight (TOF)-MS with 9-flourenyl-methoxycarbonyl chloride derivatization (Wada et al., 2007; Kakei et al., 2009; Schmidt et al., 2011). In addition, NA, DMA and their Fe (II and III) complexes in a complex mixture were separated by nano-ESI-Fourier transform ion cyclotron resonance (FTICR)-MS (Weber et al., 2006). The technique was also used to study different metal complexations with NA in different pHs, such as typical xylem and phloem pH (Rellan-Alvarez et al., 2008).
Liquid chromatography-ESI-Q-TOF-MS is mainly used to target certain PSs such as DMA and NA in plant samples, but here we used it for a simple and sensitive identification of different PSs in their ferric complexes. We developed a rapid and highly sensitive LC-ESI-Q-TOF-MS method for use with barley to simultaneously analyze free PSs and their ferric complexes.
Materials and Methods
Nicotianamine (95%) and DMA (95%) were from Toronto Research Chemicals (North York, Canada). Acetonitrile (analytical grade, Fluka), nicotine and leucine enkephalin (LC-grade, Sigma) were from Sigma. Fe (III) chloride (99.2%, J.T. Baker), Fe (II) chloride (99.2%, J.T. Baker), ammonium bicarbonate (98%, J.T. Baker), formic acid (analytical grade, J.T. Baker) and methanol (LC-grade, J.T. Baker) were from the Mallinckrodt Baker Institute (Phillipsburg, NJ, USA). Water was purified through a Milli-Q ultrapure water purification system.
Plant materials and growth conditions
Barley seeds (Hordeum vulgare L. cv Himalaya) (Lin & Ho, 1986) were germinated for 5 d at room temperature (25°C) on a plastic net support floating on distilled water. The germinated seedlings were further cultured hydroponically in 400 ml plastic containers (four seedlings per container) containing modified half-strength Hoagland nutrient solution with the following composition: 3 mM KNO3, 2 mM Ca(NO3)2, 0.5 mM NH4H2PO4, 1 mM MgSO4, 46.2 μM H3BO3, 9.1 mM MnCl2, 0.8 μM ZnSO4, 0.3 μM CuSO4, 0.08 μM Na2MoO4 and 5 μM NaFeEDTA. To induce Fe deficiency, NaFeEDTA was omitted, and 5 μM ferrozine was added to the nutrient solution. The pH of the nutrient solution was adjusted to 5.7 using KOH, which was renewed every 3 d under controlled growth conditions with 100 μmol m−2 s−1 light intensity, 16 : 8 h light : dark cycles at 22°C.
Extraction of PSs and collection of root exudates
For the extraction of PSs from root tissue, c. 50 mg of freeze-dried roots, dried in vacuum freeze-dryer (FD24-20PL, KingMech, New Taipei, Taiwan; at −60 to −80°C for 8–10 h), was extracted with 1 ml of 80% methanol for 30 min under sonication at room temperature (25°C). Root extracts were microfiltered (0.45 μm Millipore, Millex-HA) and used directly for LC-ESI-Q-TOF-MS analysis.
For collecting root exudates, 10-d-old plants were transferred to nutrient solution with or without Fe. After 5 d of Fe-deficiency treatment, root exudates were collected. Before collection of root exudates, the plants were removed from the nutrient solution 2 h after the onset of the light period and roots were washed a few times with distilled water. Plant root systems were then submerged in 400 ml distilled deionized water for 4 h for the exudate collection. After collection, plants were placed back into the new culture solution. The root exudates were collected every 2 d on five occasions. Care was taken not to damage the roots during exudate collection to avoid the artificial secretion of phytocompounds from damaged roots. The sterility of the nutrient solutions and root exudates during the experiments was examined by growing the plants on medium containing the collected nutrient solution or root exudates (von Wiren et al., 1994). The collected clean root exudates were combined and filtered immediately through filter paper (Whatman), concentrated to 10 ml in a vacuum evaporator at 35°C and kept at –20°C until the analyses. After the addition of the internal standard (nicotine at a final concentration of 100 ng ml−1) into the concentrated exudates, root exudates were microfiltered (0.45 μm Millipore, Millex-HA) and subjected to LC-ESI-Q-TOF-MS.
Preparation of NA/DMA–Fe complexes
The standard Fe complexes were obtained by mixing 50 μM NA or DMA with 1 mM Fe (III) chloride or Fe (II) chloride in 20 mM aqueous ammonium bicarbonate solution, pH 7.5, as described (Weber et al., 2006; Rellan-Alvarez et al., 2008). Mixtures were incubated in the dark at room temperature (25°C) for 12 h. The samples were subjected to LC-ESI-Q-TOF-MS analysis after microfiltration.
Samples of root exudates and root extracts (8–10 μl) were injected into an LC system (ACQUITY Ultra-performance LC (UPLC), Waters, Millford, MA, USA) coupled with a hybrid Q-TOF-MS (Synapt HDMS, Waters, Manchester, UK). The samples were further separated on a UPLC column (T3 column: 1.7 μm, 2.1 mm × 150 mm) by gradient elution at a column temperature of 40°C. In UPLC-MS analysis, acetonitrile (ACN) and formic acid (FA) were used in a mobile phase. The mobile phases for positive electrospray ionization (ESI+) consisted of 0.1% FA in 2% ACN (buffer A) and 0.1% FA in 100% ACN (buffer B). The mobile phases for the negative electrospray ionization (ESI–) consisted of 2% ACN (buffer A) and pure 100% ACN (buffer B). The gradient conditions on the T3 column were as follows: buffer A, 99.5–5.0% in 0–4.0 min, 5.0–5.0% in 4.0–4.5 min, 5.0–99.5% in 4.5–4.8 min, and 99.5–99.5% in 4.8–6.3 min. The mobile-phase flow rate was 0.3 ml min−1.
The MS and MS-MS were equipped with a locked ESI probe and operated in both the ESI+ and ESI– modes. The electrospray capillary voltage was 3 kV in the ESI+ mode and 2.5 kV in the ESI– mode, with a cone voltage of 40 V. The cone and desolvation gas flows were 50 and 600 l h−1, respectively. The desolvation temperature was 250°C. The acquisition method was set to one full MS scan (50–990 m/z) with 0.2 s scan time in both centroid and continuous data modes. The collision energy of the MS-MS analysis was 10–30 eV.
MS data processing
The LC-ESI-Q-TOF-MS data were processed using a MarkerLynx XS 4.1 SCN639 within MassLynx 4.1 (Waters). In the LC-MS, data were generated by intensities of the peaks, with retention time (RT) and m/z data pairs used to identify each peak. The extracted peak data were further processed to remove the peak-to-peak noise, deisotope and to filter MS peaks with intensity < 30 counts. The processed data for each sample were combined and aligned for each of the RT–m/z pairs to generate the final data table. The ion intensities for each peak were then normalized within each sample. The three-dimensional data, peak identifier (RT–m/z pair), sample name and ion intensity were analyzed by principal component analysis.
Limit of detection and calibration
Limit of detection (LOD) refers to the amount of analytes that give an signal-to-noise (S/N) ratio of 3 (Rellan-Alvarez et al., 2011). We calculated the limits by using NA and DMA standard solutions prepared in water and extraction buffer. The calibration curves of the standards were generated using peak areas of 10 different concentrations (from 0.25 to 250 pmol) of standards solutions. All contents were calculated by the standard linear equation, where y is the peak area and x is the concentration of the analyzed standard materials.
Recovery assays of NA and DMA standards and their Fe complexes, NA–Fe (II) and DMA–Fe (III), during sample preparation such as freeze-drying and sample extraction involved standard solutions prepared in water and extraction buffer. Recoveries of standard analytes and standard-Fe complexes were analyzed with nicotine and leucine enkephalin as internal standards (ISs) in ESI+ and ESI– modes, respectively. The recovery values were calculated by dividing the normalized peak areas of the standard analytes after procedure by the normalized peak of the standard analytes before the procedure.
Results and Discussion
Direct LC-ESI-Q-TOF-MS analysis of PSs: identification of avenic acid (AVA) and hydroxyavenic acid (HAVA) in roots of H. vulgare under Fe deficiency
Under Fe deficiency, the biosynthesis of PSs increases markedly in barley roots (Kawai et al., 1988). Therefore, we chose H. vulgare cv Himalaya roots with Fe deficiency for PS analysis. Previously, LC-ESI-TOF-MS was used to identify PSs in plant samples with 9-fluorenyl-methoxycarbonyl chloride derivatization (Kakei et al., 2009). In our study, we used LC-ESI-Q-TOF-MS in both ESI+ and ESI– ionization modes without any chemical derivatization to directly analyze small amounts (8–10 μl) of crude extracts extracted from barley roots with 80% methanol. PSs were further identified by tandem mass spectrometry (MS-MS). The sensitivity of PS detection was higher in the ESI+ than in the ESI– mode, probably because the low pH of the LC buffer resulted in low dissociation of the carboxyl group. In the ESI+ mode, DMA, MA and epi-HMA were easily detected, with major peaks at 305.1355, 321.1301 and 337.1252 m/z, respectively, as expected from previous reports (Howe et al., 1999; Suzuki et al., 2006). Interestingly, we also detected two minor peaks at 323.1463 and 339.1407 m/z, close to the peaks for MA and epi-HMA, respectively, and in the range of molecular weights of known PSs (Fig. 1). Although these two peaks look like the third peaks of MA and epi-HMA, the accurate exact mass measurement and detection of these peaks in secreted root exudates (to be discussed later) showed them to be independent of MA and epi-HMA peaks. The exact mass measurement indicated that 323.1463 and 339.1407 peaks might have the elemental composition of C12H23N2O8 and C12H23N2O9, respectively.
We later identified the peaks as AVA and HAVA from their mass fragmentation patterns (Fig. 2a,b). In addition to the PSs, NA (m/z, 304.1523) and 3’’-oxo acid (m/z, 303.1198, Fig. 2c), an intermediate in the conversion from NA to DMA, were identified in barley roots by their MS-MS spectra. Detection of 3’’-oxo acid is thought to be difficult because it can be converted to DMA immediately (Kanazawa et al., 1994; Kakei et al., 2009). However, we confirmed the observation in five independent biological repeats. The calculated and observed m/z values of the identified compounds and their fragmentation patterns are given in Table 1. Proposed mass fragmentations of NA, DMA, MA and epi-HMA are shown in Supporting Information, Fig. S1.
Table 1. Calculated and observed m/z values of phytosiderophores (PSs) in barley (Hordeum vulgare cv Himalaya) roots
Estimation of m/z values and mass fragmentations were processed in positive electrospray ionization (ESI+) mode. iFIT, isotope fit value in MassLynx (Waters); NA, nicotianamine; DMA, deoxymugineic acid; MA, mugineic acid; Epi-HMA, 3-epihydroxymugineic acid; AVA, avenic acid; HAVA, hydroxyavenic acid.
304, 286, 268, 240, 185, 141, 114
303, 197, 186, 169, 84
305, 287, 241, 197, 186, 169, 142, 114
321, 303, 186, 178, 130, 88
337, 319, 220, 202, 178, 158, 130, 112
323, 305, 269, 211, 204, 186, 158, 142
339, 321, 204, 178, 158, 130, 114, 100
Although several different barley species have been reported to synthesize the only three types of PSs – DMA, MA and epi-HMA – in response to Fe deficiency (Howe et al., 1999; Suzuki et al., 2006), we identified AVA and HAVA for the first time in this barley species using this direct LC-ESI-Q-TOF-MS analysis without chemical derivatization. Previously, AVA and HAVA were also observed in oat (Fushiya et al., 1980) and Poa pratensis cv Baron under Fe deficiency (Ueno et al., 2007). Alternatively, the explanation for AVA and HAVA not previously identified in barley samples is the lower sensitivity of the analytical methods used as compared with our method. Our data suggest that we have developed a simple, direct and highly sensitive analytical method for identifying PSs in plant samples. Using LC-ESI-Q-TOF-MS without chemical derivatization, we demonstrated that H. vulgare cv Himalaya synthesizes two PSs, AVA and HAVA, in addition to DMA, MA and epi-HMA, under Fe deficiency.
Validation of the LC-ESI-Q-TOF-MS method
To validate our developed method, we determined the LODs, and a linear range of the method using NA and DMA standards because of commercial availability of only these two standard analytes. Standard solutions were prepared in water and extraction buffer. Results of validation assays of standards dissolved in water and extraction buffer slightly differed. However, LODs for NA and DMA standards were the same as 0.25 pmol in water and 0.5 pmol in extraction buffer with similar calibration curves (Table S1). We also obtained the same linear ranges for these two standards as 0.25–250 pmol in water and 0.5–250 pmol in extraction buffer (Table S1). The detected amounts of these two standards in tested biological samples are in the ranges of the calibrations. Although we do not know the detection limits and saturation points of the other identified PSs, we expect that the linear ranges of these PSs might be similar to ones of NA and DMA, as we obtained the similar calibrations for these two standards. Because the PSs have similar structure and chemical properties, their detection efficiencies, including LC separations and electrospray ionizations in LC-ESI-Q-TOF-MS, could be similar in tested samples. Therefore, we tried to compare the relative abundance of the identified PSs using peak areas to assess the relative quantification in further studies.
To further evaluate the sample preparation procedures such as freeze-drying and sample extraction process on the stabilities of the PSs and PS–Fe complexes, we performed recovery assays using NA and DMA standards and their prepared standard-Fe complexes, NA–Fe (II) and DMA–Fe (III), in water and sample buffer. To control the changes in ionization efficiency during day-to-day operation, we used nicotine and leucine enkephalin as ISs in the ESI+ and ESI– modes, respectively. We obtained good recoveries, from 84 to 94%, for NA and DMA standards and their standard-Fe complexes during freeze-drying and sample preparation (Table S2).
The standard-Fe complexes were stable during ESI as we did not detect the dissociated free standards from the analysis of standard-Fe complexes. Furthermore, the standard-Fe complexes were also stable during LC separation, as validated by the complex stability after keeping the standard complexes on the chromatographic column for several minutes by delaying the LC buffer gradient. The standard-Fe complex peaks showed almost no stability problem after staying on the column, as complex peaks showed no peak tailing and broadening with the same peak areas and intensities as detected on the normal method gradient (data not shown). The recovery and the complex stability examinations suggest that PSs and PS–Fe complexes are stable during our sample preparation and analysis procedure.
To validate the matrix signal suppression, we analyzed the linearity of the detections in the series of diluted samples for five PSs together with NA and 3’’-oxo acid. The dilution of the samples before instrumental analysis is one of the strategies to decrease the sample matrix effect from co-eluting compounds (Villagrasa et al., 2007; Stahnke et al., 2012). In this analysis, we observed good linearity of detections in all PSs, NA and 3’’-oxo acid (Table S3), suggesting that there were no considerable matrix effects in the sample analysis.
Secretions of PSs increase under Fe deficiency
Some grasses, such as barley, wheat and rye, secrete high amounts of PSs and are more tolerant to low-Fe conditions than are grasses secreting smaller amounts of PSs, such as rice, maize and sorghum (Kawai et al., 1988). To determine the secreted PSs from the barley roots in response to Fe deficiency by LC-ESI-Q-TOF-MS, we collected the root exudates from Fe-sufficient and Fe-deficient roots. The secreted PSs in root exudates were further analyzed by LC-ESI-Q-TOF-MS in the ESI+ mode and identified by MS-MS. Under Fe-sufficient conditions, epi-HMA and MA were secreted at low concentrations, as evidenced by small peak areas and intensities, whereas the other three PSs were not detected in root exudates, even though they were identified at high concentrations in roots (Fig. 3a). However, all five PSs – DMA, MA, epi-HMA, AVA and HAVA – were identified in root exudates collected from Fe-deficient roots (Fig. 3b). In contrast to PSs, the precursors NA and 3’’-oxo acid were not detected in root exudates under both Fe-sufficient and Fe-deficient conditions, which suggests that they do not function in Fe acquisition from the soil in barley.
To further assess the functions of PSs in Fe acquisition under Fe deficiency, we examined the relative abundances of secreted free PSs normalized to an internal standard (nicotine, 100 ng ml−1) using peak areas. We found greater abundance of AVA and HAVA in Fe-deficient root exudates. Therefore, we further examined the regulation of biosynthesis and secretion of AVA, HAVA and other PSs in response to Fe deficiency. We compared the relative abundances of each PS in barley roots and root exudates (Fig. 3). The abundance of all free PSs was higher in roots under Fe-deficient than under Fe-sufficient control conditions (Table 2). Although little MA and epi-HMA were detected in exudates under Fe-sufficient control conditions, they showed significantly increased abundance under Fe deficiency. Notably, DMA, AVA and HAVA were only secreted under Fe deficiency (Table 2). The data imply that the regulation of PSs secretion in addition to biosynthesis plays a role in the control of Fe acquisition under deficiency.
Table 2. Normalized peak areas of phytosiderophores (PSs) in barley (Hordeum vulgare cv Himalaya) roots and root exudates in response to iron (Fe) deficiency
Increase ratio −Fe/+Fe
Data are mean ± SD from three replicates with three technical repeats. Fifteen-day-old plants were treated under conditions of Fe deficiency for 2 wk. The root exudates were collected and combined on days 5, 7, 9, 12, and 14 after Fe-deficiency treatment. Root samples were harvested after 2 wk of Fe deficiency. The normalized peak areas of different PSs in positive electrospray ionization (ESI+) mode were obtained by normalizing the peak areas of free PSs to peak areas of internal standard (nicotine, 100 ng ml−1). Amounts of NA and DMA shown in parentheses are given in μmol g−1 DW and nmol per plant in roots and exudates, respectively.
The highly increased abundance of DMA and MA in roots and the low abundance in secreted root exudates may indicate that these two PSs are primarily involved in the translocation of Fe to other parts of plants rather than in Fe uptake from soil under Fe deficiency. Epi-HMA was the most abundant PS in barley roots and in root exudates under Fe-deficient conditions. Compared with the abundances of epi-HMA, DMA and MA, those of AVA and HAVA were low in both Fe-sufficient and Fe-deficient roots, even though their relative abundance increased 1.3–3.3 fold under Fe deficiency. However, their abundance was greatly increased in root exudates under Fe deficiency relative to other secreted PSs such as DMA (Table 2). This finding suggests the possible roles of these two PSs in Fe uptake through secretions to the rhizosphere.
Hydroxyl groups on the PS skeleton were previously found to increase Fe (III) chelate stability in acidic environments such as the root apoplasm and rhizosphere because of decreased acid-induced protonation of the PS–Fe complex (von Wiren et al., 2000). In general, PSs differ from each other in their hydroxylation patterns. Epi-HMA contains two substituent hydroxyl groups and binds Fe (III) at least 15 times more tightly than does DMA, which contains no hydroxyl group, at pH 6.0. Furthermore, the epi-HMA–Fe (III) complex is highly stable over the pH range 5.0–8.0 (von Wiren et al., 2000). Similar to MA, AVA contains one substituent hydroxyl group at the C-4 position, and HAVA contains two substituent hydroxyl groups at C-2’ and C-4, like epi-HMA. The difference between AVA and MA is an alkyl side chain on the terminal amino-carboxylate region, similar to the difference between HAVA and epi-HMA (Fig. S2). The importance of this alkyl side chain on the terminus of PSs for Fe (III)-binding affinity was suggested previously (von Wiren et al., 2000). When the azetidine ring on the amino-carboxylate terminus was replaced with an alkyl side chain, the affinity of PSs to Fe (III) was increased (von Wiren et al., 2000). The Fe-chelating capacities of AVA and HAVA were found to be similar to those of DMA and 3-hydroxy-2’-deoxymugineic acid by ion-exchange and gel-filtration chromatography (Kawai et al., 1994; Ueno et al., 2007). In addition, a stereospecific Fe uptake mediated by PSs in grasses was suggested in a study involving enantiomers of AVA and deoxy-distichonic acid (Oida et al., 1989). These data imply that AVA and HAVA contribute to Fe acquisition under Fe deficiency on the root surface by making a stable Fe (III) complex in the rhizosphere through their high degree of hydroxylation or specific chemical structures. Therefore, we examined the formation of PS–Fe complexes.
Identification of PS–Fe (III) complexes
Simultaneous identification of PSs and PS–Fe complexes in plant samples with LC-ESI-Q-TOF-MS is a prerequisite to studying the function of different PSs under Fe deficiency. To check the applicability of our LC-ESI-Q-TOF-MS method for studying the PS–Fe complex, we used the same Fe-deficient samples with which we identified the free PSs, and we analyzed the PS–Fe (III) complexes in both ionization modes. We calculated theoretical mass and isotope distribution patterns for the Fe complexes using fluorine chemistry online software (University of Manchester). Fe complexes can be easily identified by MS because of the isotopic patterns of Fe. The relative abundance of naturally occurring stable Fe isotopes is approximately 54Fe (5.8%), 56Fe (91.7%), 57Fe (2.2%), and 58Fe (0.3%). We detected PS–Fe (III) complexes with high abundance in roots in the ESI– mode but found no complexes in the ESI+ mode, which may be because the PS–Fe (III) complexes are negatively charged at a metal : ligand ratio of 1 : 1.
Fe-deficient roots showed singly charged DMA–Fe (III) and epi-HMA–Fe (III) complexes at 356.031 and 388.027 m/z, respectively, and the MA–Fe (III) complex was detected at 372.027 m/z (Fig. 4a). The observed and calculated m/z values of PS–Fe (III) complexes are given in Table 3. We did not detect AVA–Fe (III) or HAVA–Fe (III) complexes in roots, perhaps because of their relatively low abundance in roots as compared with the other three PSs, DMA, MA and epi-HMA. Furthermore, we did not detect the 3”--oxo–Fe (III) complex in roots but detected free 3”-oxo acid with high abundance, which suggests that it functions differently from Fe chelation as a biosynthetic intermediate of PSs.
Table 3. Calculated and observed m/z values of phytosiderophore-iron (PS-Fe (III)) complexes in barley (Hordeum vulgare cv Himalaya)
Compounds for complexes
Calculated m/z (M – 4H + 56Fe(III))
m/z values are from liquid chromatography-electrospray ionization-quadrupole-time of flight-mass spectrometry (LC-ESI-Q-TOF-MS) in negative electrospray ionization (ESI–) mode.
For NA, we examined the existence of both ferric and ferrous complexes because the NA–Fe (II) complex is more stable than the NA–Fe (III) complex. Because the difference in monoisotopic mass between NA and DMA is 0.9841 amu, distinguishing their Fe complexes that eluted closely on UPLC is difficult, especially the NA–Fe (II) and DMA–Fe (III) complexes. Therefore, we used NA and DMA standards for more accurate Fe-complex determination by incubating them with Fe (II) and Fe (III) stock solutions as previously described (Weber et al., 2006). In contrast to the detection of PS–Fe (III) complexes, no NA–Fe (III) complex was detected, but the NA–Fe (II) complex was detected at 356.055 m/z (data not shown).
The AVA–Fe (III) and HAVA–Fe (III) complexes were detected at 374.047 and 390.026 m/z, respectively, in Fe-deficient root exudates (Fig. 4b). These data show that the high affinity of secreted AVA and HAVA to Fe (III) enables the compounds to form ferric complexes with residual Fe (III) in the Fe-deficient treatment. UPLC detection of the PS–Fe (III) complexes in root and root exudates is shown in Fig. S3. During our experiments, there were no considerable microbial contaminations in the nutrient solution and collected root exudates as it was validated (Fig. S4).
Together, our data support the good applicability of the use of direct LC-ESI-Q-TOF-MS for determining free PSs and their Fe complexes in plant samples. The workflow for the method is shown in Fig. S5.
In conclusion, we report on a simple and sensitive LC-ESI-Q-TOF-MS method for the direct and simultaneous determination of PSs and their Fe complexes in plant samples. Using this method, we identified five different PSs and their biosynthetic precursors and intermediates, together with their Fe complexes, in H. vulgare cv Himalaya. This analytical method could contribute greatly to determining the detailed roles of PSs in different aspects of plant life such as nutrition and toxicology.
This work was supported by grants from DPIAB and Academia Sinica. We thank Dr Tuan-Hua David Ho for seeds of Hordeum vulgare cv Himalaya. The mass spectrometry analysis was supported by the Metabolomics Core Facility, Scientific Instrument Center at Academia Sinica.