Identification of metal species by ESI-MS/MS through release of free metals from the corresponding metal-ligand complexes

Electrospray ionization-mass spectrometry (ESI-MS) is used to analyze metal species in a variety of samples. Here, we describe an application for identifying metal species by tandem mass spectrometry (ESI-MS/MS) with the release of free metals from the corresponding metal–ligand complexes. The MS/MS data were used to elucidate the possible fragmentation pathways of different metal–deoxymugineic acid (–DMA) and metal–nicotianamine (–NA) complexes and select the product ions with highest abundance that may be useful for quantitative multiple reaction monitoring. This method can be used for identifying different metal–ligand complexes, especially for metal species whose mass spectra peaks are clustered close together. Different metal–DMA/NA complexes were simultaneously identified under different physiological pH conditions with this method. We further demonstrated the application of the technique for different plant samples and with different MS instruments.

ions in different MS instruments such as matrix-assisted laser desorption ionization (MALDI) imaging showed the potential use of this approach in metal speciation studies 18 .
Analysis of natural metal chelators such as the phytosiderophores (PSs) and their precursor nicotianamine (NA) and their metal complexes in plant samples is of particular interest because of their critical involvement in the acquisition and translocation of different nutrients in crops 6,19,20 . Metal-PSs and metal-NA complexes have been identified in several plant species by ESI-time-of-flight (TOF)-MS 21 and LC-ESI-TOF-MS 22,23 based on the isotopic signatures of the complexed metals.
In this study, we present tandem mass spectrometry (ESI-MS/MS) as an accurate and sensitive analytical procedure for identifying metal species. The method allows for identifying metal species by the release of free metals from the corresponding metal complexes. We illustrate the analytical procedures by analysis of different metal-deoxymugineic acid (DMA)/NA complexes as examples. The procedure may be applied in studies of metal species in a wide range of biological samples.

ESI-MS/MS identification of metal species by release of free metals from metal-ligand complexes.
To date, the identification of metal species with ESI-MS has relied directly on metal-specific isotopic signatures or isotopic patterns of the metal complex spectra 21,22 . Here, we report a method for identifying metal-ligand complexes by using ESI-MS/MS. Henceforth, "MS1" and "MS2" will be used to denote the spectra of primary ESI-MS and secondary ESI-MS/MS, respectively. In metal complex studies, we used five transition metals, Fe, Cu, nickel (Ni), Zn and cobalt (Co). The exact atomic masses and abundance of their natural isotopes (available at www.ciaww.org) are shown in Table 1. Analysis of these metal complexes in high-resolution ion-trap MS involved prepared standard metal-DMA and metal-NA complexes in positive ESI mode.
In MS1, metal-DMA/NA complexes can be identified with singly charged metal species in a metal-ligand stoichiometry of 1:1 21,22 . Formulations of the complexes were obtained as [M-H + Me(II)] + and [M-2H + Me(III)] + for di-valence and tri-valence metal-ligand complexes, respectively, as shown in the elemental composition in Table 2. Metal complexes were primarily identified in MS1 with high accuracy (<5 ppm), calculated from the comparisons of the calculated and observed m/z of metal-DMA/NA complexes (Table 2 and Supplementary  Table S1). In MS1, the complex isotopic signatures also showed high accuracy, with <10% errors in relative isotopic abundance (RIA) 24 (Table 2 and Supplementary Table S2). With this method, in MS2, during the fragmentations of isotopic spectra of metal complexes, we obtained product ion spectra that corresponded to the released free metals from the corresponding metal-DMA/NA complexes (Table 2 and Fig. 1). The identification of all the released free metals was confirmed with an exact atomic mass measurement of free metals and their isotopic signatures with high accuracy except for the low-abundant 57 Fe isotope (Table 2 and Supplementary Table S3). Although the mass accuracy of free metal isotopes showed relatively low values (<10 and <14 ppm for major and minor isotopes, respectively) compared to those obtained in MS1, the values could still provide an accurate mass measurement 25 in MS2. The measured mass accuracy increased under high fragmentation energies.
For Fe-DMA/NA complexes, we generated the product ions by fragmenting three isotopic Fe(III)-DMA complex spectra, m/z 356.051, 358.046 and 359.046, and three Fe(II)-NA complex spectra, m/z 356.074, 358.070 and 359.072, with different collision energies. We then acquired the product ion spectra for the release of free Fe isotopes, m/z 53.939, 55.934 and 56.942 for 54 Fe, 56 Fe and 57 Fe, respectively, from their isotopic precursor spectra (Fig. 1a,e). The signal intensity of Fe spectra increased with increasing fragmentation energy applied to the precursors (HCD: 70-90%). Notably, the identification of the Fe(III)-NA complex in close isotopic spectra of the Fe(II)-NA complex was clearly separate from the release of 56 Fe with m/z 55.934 from its precursor Fe(III)-NA, m/z 357.062 (Fig. 1e).
For Cu-DMA/NA complexes, we acquired the MS2 spectra from both isotopic precursors m/z 366.048 and 368.047 for the Cu(II)-DMA complex and m/z 365.065 and 367.063 for the Cu(II)-NA complex. The released free Cu isotopes, m/z 62.929 and 64.927 corresponding to 63 Cu and 65 Cu, from the corresponding isotopic parent spectra increased in abundance under different collision energies (Fig. 1b,f). For Ni-DMA/NA complexes, we generated the product ions from two isotopic precursors, m/z 361.055 and 363.051 for the Ni(II)-DMA complex and m/z 360.071 and 362.070 for the Ni(II)-NA complex. The product ion spectra for the released free Ni isotopes, m/z 57.935 and 59.930 for 58 Ni and 60 Ni, respectively, corresponded to the precursor spectra (Fig. 1c,g). For Co-DMA/NA complexes, we fragmented the parent spectra of m/z 362.053 and 361.068 for the Co(II)-DMA and Co(II)-NA complexes, respectively. The product ion spectrum identified as the released free 59 Co single isotope was detected with m/z 58.933 in MS2 (Fig. 1d, (Fig. 2a,b). The detection sensitivities were lower for the free Zn isotopes than for other metals. We obtained Zn isotopic spectra by applying very high-energy collisional dissociation (HCD: 120-150%) with high concentrations of Zn-DMA/NA prepared solutions; other metal isotopes were generally detected at HCD: 50-90% with low concentrations of prepared solutions.
Of note, the released free metals from the metal complexes were detected only in positive but not negative ESI mode. Free metals were not released in MS2 spectra from free DMA and NA samples (Supplementary Figures  S1 and S2).
Together, our data demonstrate a highly accurate method involving ESI-MS/MS for identifying metal species by the release of free metals from the metal-ligand complexes.   (MRM) 26 analysis and/or for analyzing the capacity of certain fragments that are essential for metal-binding and complex stability.
In addition, we tested the applied fragmentation energies that are required for releasing the free metals from the corresponding metal complex ions by using six different dilutions of Cu(II)-NA and Ni(II)-NA complex standard solutions. Results in MS2 spectra further showed the requirement for increased fragmentation energy for the released free Cu and Ni isotopes with decreased abundance of the Cu(II)-NA and Ni(II)-NA complexes in the series of diluted samples. In the same sample, the intensity of free Cu/Ni isotopes was increased under high collision energies, HCD: 80-90%, as compared with lower collision energies, HCD: 40-50%. The fragmentation energies also differed for different metal complexes because the Cu(II)-NA complex required lower percentages of collision dissociation energies than the Ni(II)-NA complexes (Table 3). Analysis of metal-DMA and metal-NA complexes in physiological pH conditions. We next analyzed different metal-DMA/NA complexes simultaneously in one sample. The identification of metal complexes based on the metal-specific isotopic signatures is complicated in the presence of other metal complexes with peaks positioned close together, especially with low-abundance metal complexes in biological samples. In this analysis, we prepared a mixture of five different metal-DMA and metal-NA complexes. Compared to the isotopic spectra identified in Fig. 1 for separate metal complexes, with mixtures of different metal-DMA/NA complexes (Fig. 4, upper panels), the spectra of metal isotopic patterns were diminished by the other closely detected complex spectra. However, the use of MS2 with release of free metals from the metal complexes allowed for clear identification of the corresponding metal complex spectra. The abundance of the released free metals was also comparable to that of their precursor metal complex spectra (Fig. 4, lower panels).
We further analyzed the complex formations of different metals with DMA and NA under two physiological pH conditions, pH 5.5 and 7.5, as near-typical plant xylem and phloem pH values, respectively. In this in vitro experiment, different pH conditions were applied in background buffer to estimate the complexation nature of different metal-ligand complexes. The binding abilities of DMA and NA to different metals such as Fe, Zn, Cu, Ni, Co and manganese (Mn) were previously reported 16,21 and the stabilities of the metal-NA complexes were also calculated 27 . However, the simultaneous identification of different metal-DMA and metal-NA complexes under physiological pH conditions to directly detect their dominant complex forms has not been described.  The direct infusion MS analysis we describe shows that among the identified metal-DMA complexes, DMA formed the most abundant complex with Cu(II) at both pH 5.5 and 7.5 (Fig. 4a,b). The relative abundances of the Cu(II)-DMA precursor complex and its released free Cu spectra were both higher than that of other metal-DMA complexes including Fe(III)-DMA. As noted previously, because of the poor detection sensitivity of the released free Zn spectra from Zn-DMA/NA complexes, we did not include it among the released free metals. The abundance of metal-DMA complexes did not change apparently under the two different pH conditions. Similar to results obtained from the metal-DMA complexes, among the analyzed metal-NA complexes, the highest abundance was for Cu(II)-NA and its released free Cu spectra under the two pH conditions (Fig. 4c,d). The relative abundance of metal-NA complexes such as Fe(III)-NA, Cu(II)-NA and Ni(II)-NA varied slightly under the two different pH conditions. Cu(II)-DMA/NA complexes were also the most abundant forms under limited DMA and NA conditions (see Methods). Previously, the possible metal-exchange properties of NA were suggested during long-distance metal-NA transport in plants 6 . Thus, MS2 analysis of different metal-NA (and metal-DMA) complexes can be used to identify the different formed complexes in plant samples. Together, these data demonstrate the potential use of this method for the accurate and simultaneous identification of different metal complexes in a single sample.
Application of the methodology in biological samples. We used the method for analysis of a real sample of plant shoot extracts. We used a hydrophilic-interaction HILIC column before ESI-MS. With HILIC column separation in an Ultra Performance Liquid Chromatography (UPLC)-ESI-Q-TOF-MS instrument, we identified three metal-DMA complexes, Fe(III)-DMA, Cu(II)-DMA and Zn(II)-DMA, and three metal-NA complexes, Fe(III)-NA, Fe(II)-NA and Zn(II)-NA, in shoot extracts of japonica rice (Fig. 5a,b). Metal-DMA/ NA complexes were first identified with the exact mass measurement on MS1 analysis (Table 1) and further confirmed in MS2 with the release of free metals from the corresponding metal complexes at fragmentation energy 50 eV. As described in Table 3, the relative abundance of the free metals released from the identified metal-DMA complexes appeared to differ at the same fragmentation energy. As well, under the conditions applied in MS2, the major isotopes were identified only from the free metals.
To analyze the existence of different metal-DMA and metal-NA complexes in rice plants in response to Fe deficiency, we collected shoot samples from two rice varieties, indica and japonica, treated with Fe deficiency for 7 days. In the shoot of indica rice, we identified the same metal-DMA complexes as in japonica rice, and two metal-NA complexes, Fe(III)-NA and Fe(II)-NA (Supplementary Figure S6). In both rice varieties, Fe(III)-DMA and Fe(II)-NA complexes showed the highest abundance for DMA and NA chelators, respectively, with similar relative abundance (Fig. 6a,b). After their identification, the concentrations of free DMA and NA in the analyzed shoot samples were quantified by using authentic standards (Supplementary Figure S7). Both rice shoots showed lower concentrations of NA than DMA (Fig. 6c,d), which suggests the high stability of the Fe(II)-NA complex in these plants. We found a slight decrease in abundance of Fe(III)-DMA and Fe(III)-NA complexes under Fe deficiency, similar to the concentration of Fe in shoots (Fig. 6a,b,e). Compared with Fe complexes, the Cu(II)-DMA complex showed relatively low abundance (8 to 12 times lower). However, the concentration of Cu in shoots was significantly lower than that of Fe and Zn, especially 16 to 37 times lower than Fe content (Fig. 6e,f,g). Therefore, the data still suggest the possible high stability of Cu(II)-DMA complex in rice shoots.
Fe(III)-DMA, Cu(II)-DMA and Zn(II)-NA complexes were previously identified in tissues of rice 28 , barley 22 and wheat 16 , with a role for DMA and NA suggested in the translocation of these nutrients in these grasses. In agreement with this suggestion, we identified the Cu(II)-DMA complex in addition to Fe(III)-and Zn(II)-DMA complexes in roots of wheat and barley plants with our method. Of note, in plant sample analysis, we used a different MS instrument (UPLC-ESI-Q-TOF-MS) rather than Orbitrap-MS to confirm the applicability of the method with different MS instruments. The data indicate the effective application of our method for biological samples and with different MS instruments.

Discussion
In certain environments, the availability and reactivity of metal ions can be controlled by metal-binding ligands. Therefore, precise analysis of metal speciation is critical for studies related to metal biofunctions 6,29 . To date, analyses of metal species in a range of samples have generally involved ESI-MS 11 . However, the accurate identification of metal complex spectra to reveal metal ions was missing with this method. Here, we used MS/MS to detect product ion spectra that correspond to free metal release from five different precursor metal complexes formed by two natural organic ligands, DMA and NA.
We here present a multi-elemental approach compared with previously described MS/MS methods for the identification of NA, DMA and their metal complexes those mainly restrict to certain metals such as Fe 22 , Fe and Zn 28 , Ni 30,31 and Cu 32 in biological samples. Our method may offer two advantages over the traditional means of identifying metal species in MS1. In MS2, the relative abundance of free metal spectra increased with increasing fragmentation energies applied to the precursor spectra. This finding further indicates the potential use of the method to identify metal species in minute biological samples by applying high dissociation energies for fragmentation. Depending on the type of mass spectrometer used, fragmentation energies can be adjusted for target precursor ions 33,34 . The second advantage is the simultaneous identification of metal species in one sample. Because Figure 6. Concentrations of DMA and NA and nutrients in rice shoots. The relative abundance of the metal-DMA (a) and metal-NA (b) complexes in shoots of rice, indica (Ind) and japonica (Jp), was obtained by normalizing the peak area of the complexes to peak area of internal standard (leucine enkephalin, 2 ppm) in positive ESI mode by UPLC-ESI-Q-TOF-MS with a mean ± SD of three replicates. Shoot samples were collected from rice plants treated with control (C) and Fe deficiency (-Fe) for 7 days. The concentrations of DMA (c) and NA (d) in shoots samples were calculated in positive ESI mode by UPLC-ESI-Q-TOF-MS with a mean ± SD of three replicates. The concentrations of iron (e), zinc (f) and copper (g) in shoots of rice varieties were determined by ICP-OES. Data are the mean ± SD of three replicates with pooled 8 plants. Asterisks denote significant differences compared with control. * P < 0.05; ** P < 0.01 by Student's t test. FW, fresh biomass; DW, dry biomass. the method does not require a metal isotopic pattern in MS1, the identification of different metal complexes can be confirmed by the release of free metals from the precursor spectra. Simultaneous analysis of metal species in biological samples is particularly important to study the effect of one metal and/or one ligand on homeostasis of other metals 2,6 . However, an appropriate separation of different metal species with suitable chromatographic conditions in biological samples is still critical to avoid the mass interferences from isobaric co-eluted molecules.
In addition, mass fragmentation analysis such as MS2 and MS3 in MS/MS could be useful to study various metal-binding ligands including metabolites, proteins and enzymes to dissect their functional fragments and groups in the structures. Application of the method could also be extended in the screening of either targeted or untargeted metal complexes in a range of analytical samples and test platforms. Previously, the screening of isotopic pairs such as 54 Fe-56 Fe in LC-MS was used in the analysis of siderophores in culture extracts 35,36 . An open-source metal complex screening tool ChelomEx that identifies the isotope pattern-matched chromatographic peaks has recently been developed 37 . Our method could be useful to directly confirm the candidate species after such screening strategies for different metal complexes.
By using our method in in vitro experiments, we identified Cu(II)-DMA and Cu(II)-NA among other metal-DMA/NA complexes as major abundant complexes formed under two close physiological pH conditions. Analysis of metal complexes in shoots of two rice varieties further identified the Cu(II)-DMA complex in addition to Fe-and Zn-DMA/NA complexes in these plants. The relatively low content of Cu in the samples as compared with the Fe and Zn contents suggests the stable role of the Cu(II)-DMA complex that may be transported in vascular tissues. Previously, a study of the NA-free tomato chloronerva mutant reported the importance of NA for long-distance transport of Cu but not Fe 38 . In vitro analysis of Cu(II)-NA complex stability in a range of pH values has also suggested the role of NA in xylem transport of Cu(II) 21 . As well, Cu(II)-DMA was recently identified as the metal-DMA complex in rice xylem sap 32 .
In crops, NA and DMA are considered the crucial ligands for metal chelation 29,39 . Nearly half of the transgenic strategies used for biofortification in rice can manipulate the NA synthase gene (NAS) alone or with other genes 40 . Therefore, analyzing metal-DMA/NA complexes simultaneously in different tissues/parts of crops and transgenic plants would be of interest.
In conclusion, we have demonstrated an accurate method involving ESI-MS/MS for identifying metal species by the release of free metals from the corresponding metal-ligand complexes. This method could be useful with different MS instruments and different types of biological and environmental samples. MS2) acquired with high-energy collisional dissociation (HCD) were produced at different collision energies, 40-150%, and 2.0 ppm tolerance window. Data were analyzed by use of Thermo Scientific Xcalibur software (Thermo Fishers).

LC-ESI-Q-TOF-MS analysis. LC-ESI-Q-TOF-MS involved use of Ultra Performance Liquid
Chromatography (UPLC) with an electrospray ionization SYNAPT HDMS G1 hybrid quadrupole time-of-flight mass spectrometer (both from Waters, Millford, MA, USA). The SYNAPT HDMS G1 was operated in positive ionization mode with electrospray capillary voltage 3.0 kV. The cone and nitrogen desolvation gas flows were 50.0 and 500.0 L/h, respectively. Desolvation temperature was 250 °C. Mass data acquisitions were set at m/z 200-400 and 50-370 for MS1 and MS2 spectra analysis, respectively, in centroid data mode with scan time 0.2 s and tolerance widow 0.02 Dalton. The precursor isolation window was 3 Thomson and the collision energy of MS/MS analysis was set at 50.0 eV. A UPLC BEH HILIC column (1.7 μ m, 2.1 × 100 mm, Waters) was used at 25 °C with elution gradients of the mobile phase, which consisted of 0.1% FA in 2% ACN (buffer A) and 0.1% FA in 100% ACN (buffer B). The gradient conditions were buffer A, 5.0-5.0%, 0-1.