Anion exchange liquid chromatography–inductively coupled plasma-mass spectrometry detection of the Co2+, Cu2+, Fe3+ and Ni2+ complexes of mugineic and deoxymugineic acid

Journal of Chromatography A, 1129 (2006) 208–215

Anion exchange liquid chromatography–inductively coupled plasma-mass spectrometry detection of the Co2+, Cu2+, Fe3+ and Ni2+ complexes of mugineic and deoxymugineic acid Estelle Bakkaus a , Richard N. Collins a,∗ , Jean-Louis Morel b , Barbara Gouget a a

b

Laboratoire Pierre S¨ue, CEA-CNRS UMR 9956, 91191 Gif sur Yvette, France Laboratoire Sols et Environnement – ENSAIA-INRA-INPL, 54505 Vandoeuvre-les-Nancy, France Received 12 April 2006; received in revised form 28 June 2006; accepted 4 July 2006 Available online 31 July 2006

Abstract Phytosiderophores, such as mugineic and deoxymugineic acid, are produced by graminaceous plant species in response to Fe deficiency conditions normally experienced in calcareous and alkaline non-calcareous soils. As these phytosiderophores have the ability to form thermodynamically stable complexes with other metal cations present in the growing medium, they have also been implicated in the transport and bioavailability of these metals in the environment. However, routine analytical methodology to detect the various metal complexes formed by these phytosiderophores is lacking. Therefore, as these complexes are negatively charged over a wide range of pH values, anion exchange liquid chromatography (AE LC) coupled to inductively coupled plasma-mass spectrometry (ICP-MS) was investigated as a means to separate and quantify these complexes. The metal–phytosiderophore complexes were prepared at pH 7 and separated by NaOH or NH4 NO3 gradient elution on a Dionex AS11 anion exchange column. Of the metals tested only the Co2+ and Ni2+ complexes of mugineic and deoxymugineic acid were detected when using a 0–20 mM NaOH gradient elution profile. However, the phytosiderophore complexes of Cu2+ and Fe3+ were also detected when using NH4 NO3 as the mobile phase at pH 7. Base-assisted hydrolysis of the latter two complexes is proposed to explain their apparent ‘instability’ in the high pH NaOH mobile phase. The absolute detection limits of the developed methodologies for these metal complexes ranged from 0.1 to 2.8 pmol. As phytosiderophore complexes with Cd2+ and Zn2+ were not detected, it was concluded that the dissociation kinetics of these metal–phytosiderophore complexes were too rapid for these complexes to be observed in the present chromatographic conditions. © 2006 Elsevier B.V. All rights reserved. Keywords: Phytosiderophores; Metal-coordinated complexes; Dissociation kinetics; HPLC–ICP-MS

1. Introduction Mugineic and deoxymugineic acid (Fig. 1) are two phytosiderophores released by graminaceous plant roots in response to Fe deficiency. Phytosiderophores increase the phytoavailability of Fe through the (1) chelation-induced dissolution, or desorption, of insoluble forms of Fe and (2) subsequent uptake of the intact Fe3+ –phytosiderophore complex via specific membrane proteins. Several genes involved in the biosynthesis of phytosiderophores have been isolated and some rice (Oryza ∗

Corresponding author. Present address: Centre for Water and Waste Technology, School of Civil and Environmental Engineering, the University of New South Wales, Sydney, NSW 2052, Australia. Tel.: +61 2 9385 5045; fax: +61 2 9313 8624. E-mail address: [email protected] (R.N. Collins). 0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2006.07.004

Sativa L.) cultivars have already been genetically engineered to stimulate deoxymugineic acid production during conditions of Fe deficiency [1]. Similarly, the gene encoding the Fe3+ deoxymugineic acid transporter (yellow stripe 1) in maize (Zea mays L.) has also been identified [2]. More recent work suggests that this transporter can also facilitate plant uptake of the deoxymugineic acid complexes of Cd2+ , Cu2+ , Ni2+ and Zn2+ [3] as well as the Cu2+ - and Co2+ -mugineic acid complexes [4]. Despite the tremendous advances in isolating the genetic and physiological controls of phytosiderophore production and metal–phytosiderophore uptake by graminaceous plants, comparatively little has been reported on routine analytical methodology suitable to analyse metal–phytosiderophore complexes. In fact, there exists only one HPLC study which was restricted to the analysis of Fe3+ –phytosiderophore complexes [5]. As phytosiderophores also form complexes with a number of other

E. Bakkaus et al. / J. Chromatogr. A 1129 (2006) 208–215

209

[16]. The dissociation kinetics of metal–phytosiderophore complexes have not been experimentally determined, however, they appear to be sufficiently slow for the Fe3+ -, Zn2+ -mugineic acid and Fe3+ -deoxymugineic acid complexes to be detected with high-voltage paper electrophoresis [10]. In addition, successful anion exchange liquid chromatographic separation of the Fe3+ complexes of mugineic and deoxymugineic acid has been reported [5]. Therefore, after an initial consideration of published thermodynamic and theoretically predicted kinetic data, the aim of the present experimental work was to assess the use of AE LC, coupled to ICP-MS, to separate and quantify the stable metal complexes of mugineic acid (MA) and deoxymugineic acid (DMA). 2. Methods and materials 2.1. Collection, preparation and quantification of phytosiderophores

Fig. 1. Chemical structures of mugineic and deoxymugineic acid. The molecules are shown in the fully protonated state and the values in parentheses represent the pKa values for these functional groups. Values are derived from von Wiren et al. [10].

metal cations, phytosiderophore exudation into the rhizosphere does not necessarily ensure that Fe3+ will be complexed. Indeed, there have been numerous reports where phytosiderophores have been implicated in the enhanced plant uptake of other metals, such as Cd, Cu, Ni and Zn, from soils [6–8]. As such, the development of routine analytical methodology to quantify the other metal–phytosiderophore complexes would effectively complement future experiments conducted in situ in soils as well as support further similar physiological studies to those cited above. It has been reported that a number of transition metal cations form negatively charged (−1) complexes with mugineic and deoxymugineic acid at pH values >4 [9]. More recently it has been established that a neutral complex consisting of Fe3+ and deoxymugineic acid predominates over the negatively charged complex at pH values <6 [10]. It is, however, agreed that mugineic and deoxymugineic acid only form stoichiometric (1:1) hexadendate complexes with all these metal cations [9–15]. In addition, phytosiderophores are released in response to Fe deficiency, which is primarily observed in calcareous or alkaline non-calcareous soils at pH values ≥7. As such, it is reasonable to presume that metal–phytosiderophore complexes will be negatively charged in the soil environment around the root zone. Given this shared characteristic it is, therefore, logical that anion exchange liquid chromatography (AE LC) would be suitable to separate metal–phytosiderophore complexes before quantification with a sensitive detection system, such as inductively coupled plasma-mass spectrometry (ICP-MS). Nevertheless, chromatographic techniques are generally limited to thermodynamically stable metal-coordinated complexes, which do not rapidly dissociate during the separation process

The mugineic acid used in these experiments was obtained in 1999 as a lyophilised powder from Dr. Satoshi Mori (The University of Tokyo, Japan). Sterile stock solutions of MA were obtained by simply dissolving the powder in a small amount of high purity water (>18 M
cm−1 resistivity) and filtering at 0.22 ␮m. These solutions were stored at approximately 4–6 ◦ C between experiments. Deoxymugineic acid was collected as root exudates from wheat (Triticum aestivum L.) plants experiencing Fe deficiency in nutrient solutions that lacked a source of Fe. The plants were initially germinated using vermiculite and sand moistened with high purity water. Approximately 30 seedlings were then transferred to 2.5 L of aerated nutrient solution having the following composition: 2 mM Ca(NO3 )2 , 0.7 mM K2 SO4 , 0.5 mM MgSO4 , 0.1 mM KH2 PO4 , 0.1 mM KCl, 10 ␮M H3 BO3 , 0.5 ␮M MnSO4 , 0.5 ␮M ZnSO4 , 0.2 ␮M CuSO4 , 0.07 ␮M Na2 MoO4 . The solutions were changed every 3–4 days and the pH was maintained at 6.5 with the use of 2 mM 2-(N-morpholino)ethanesulfonic acid (MES). The wheat plants were grown in this solution until the onset of Fe deficiency (necrosis, yellowing, etc.), approximately 21–28 days, with the following controlled climatic conditions: day/night photoperiod, 16/8 h; light intensity, 310 ␮mol m−2 s−1 ; temperature (day/night) 24/20 ◦ C; relative humidity, 70/75%. Root exudates were then collected daily during a 2-week period. This was achieved by removing the seedlings from the nutrient solution, 2 h after the commencement of the photoperiod, washing the roots with high purity water and then immersing the roots in an exudate collection solution consisting of 100 mL of high purity water. Root exudates were collected during a 3–4 h period at which time the plants were returned to the nutrient solution until the following day. The solutions containing the exudates were immediately filtered (0.22 ␮m) into sterilised bottles and also stored at 4–6 ◦ C. No further pre-treatment of the root exudates was conducted as electrospray QqTOF MS/MS analyses identified the unique phytosiderophore released by this plant as DMA (cf. Acknowledgements). Furthermore, anion exchange liquid chromatography, effectuated with NaOH gradient elution on an AS11 Dionex analytical column, followed

210

E. Bakkaus et al. / J. Chromatogr. A 1129 (2006) 208–215

by eluent-suppression conductivity detection (Product Manual, Ionpac AS11 analytical column, Dionex, Sunnyvale, CA, USA), revealed the absence of high concentrations of other polycarboxylic acids that may also potentially form stable metal complexes (e.g.<17 ␮M citric acid). Preliminary experiments indicated that phytosiderophores cannot be detected with eluent-suppression conductivity detection and, in addition, have extremely low molar absorptivities in the UV–vis light spectrum. Therefore, the concentration of DMA in the root exudate solutions was assessed indirectly by measuring the aqueous concentration of Fe that the solution could maintain at pH 7—a slightly modified Fe-solubilizing assay to that published by Takagi et al. [17]. Due to the unavailability of pulsed amperometric [5] or fluorimetric [18] detection systems, these more accurate analytical methodologies were not employed for these analyses. 2.2. Preparation of metal–phytosiderophore complexes Stock solutions of the metal–MA complexes were prepared by mixing equimolar concentrations of the metal cations (as the nitrate or chloride salt) and MA, adjusting the solution pH to 7 and finally diluting the solutions to a metal–MA complex concentration of 1 ␮M (10 ␮M for the Fe3+ –MA complex). The metal–DMA complexes were prepared in a similar fashion with the exception that the phytosiderophore was possibly added in slight molar excess over the metal cation concentrations as its concentration in the collection solutions may have been slightly overestimated by the Fe-solubilizing assay. However, this was not considered problematic as DMA (and MA for that matter) forms stoichiometric 1:1 metal complexes, and, therefore, only quantification of the metal concentration is necessary to calculate the concentration of the metal–phytosiderophore complex after chromatographic separation, if, of course, the retention time of the complex is different to that of the metal cation. The stock solutions were allowed to equilibrate overnight, to ensure complete metal complexation, and were only further diluted with high purity water before anion exchange liquid chromatography–inductively coupled plasma-mass spectrometry (AE LC–ICP-MS) experiments. New stock solutions were prepared on a weekly basis. 2.3. Anion exchange liquid chromatography–inductively coupled plasma-mass spectrometry Chromatographic separation of the metal–phytosiderophore complexes was examined with a Dionex 4-mm AS11 anion exchange column. The active anion exchange functional group of this column is the alkanol quaternary ammonium ion. Anion exchange liquid chromatography was examined with NaOH or NH4 NO3 (adjusted to pH 7) gradient elution. The mobile phases were not degassed, however, use of the same mobile phase during a 5-day period did not influence the retention time of the metal complexes indicating that further CO2(g) solubilization was insignificant (notably for the NaOH mobile phase). The mobile phases were delivered to the column via a Gilson (Villiers Le Bel, France) HPLC system. Regardless of the mobile

phase being examined, a flow rate of 2 mL min−1 and a sample injection volume of 25 ␮L were used in all experiments. Due to the potential of metal contamination in these types of analyses, the column was cleaned daily, and sometimes twice daily, to remove any precipitated metals. The column cleaning procedure for the AS11 column consisted of consecutively passing solutions of 100 mM NaOH, 100 mM HCl and 100 mM oxalic acid and finally 100 mM NaOH (to remove oxalic acid from the column). The column was then re-equilibrated with the mobile phase used for the AE LC experiments. The outlet of the anion exchange column was coupled directly to the nebuliser of the ICP-MS (Series X7, Thermo Electron Corporation, Cergy-Pontoise, France), via PEEK tubing, and chromatograms of counts per second (cps), of the metal isotope ions, against time were recorded. Collision cell technology (CCT) was used to suppress the isobaric interferences of 40 Ar16 O+ with the 56 Fe+ isotope ion—the most abundant isotope of Fe (94%). Detection limits for a number of other metal isotope ions were also greatly improved with the use of CCT and was, therefore, used for all experiments. The flow rates of the collision cell gases into the hexapole collision cell were generally 1.7 mL min−1 and 6.5 mL min−1 , respectively, for 8% H2 in He and He. The principal differences between the operating parameters of the Series X7 ICP-MS when using CCT are the potentials of the pole bias (reduced to a negative value (e.g. −5 V)) and the focus (reduced from 13 V to approximately 8 V). With the exception of the monoisotopic element Co, numerous isotopes of an element were recorded to observe any possible mobile phase-, Ar plasma-, or analyte-induced isobaric interferences. The isotope ions were monitored using a dwell time of 10 ms and transient peaks were integrated with a 1.4 s integration time (0.7 s on each side of the peak maxima). All peaks were background corrected using an average integrated-cps calculated on a 5 s section of the chromatogram where neither transient peaks nor gradient elution-enhanced background counts were observed. The response of the most abundant metal isotope ion was used to relate the metal concentration in the transient peaks to the metal–phytosiderophore complex concentration in solution. In addition, the EDTA complexes of the metal cations were also used as internal standards, due to their well-studied separation behaviour in similar anion exchange applications [19,20] and were prepared in an identical fashion to the metal–MA complexes. The chromatographic behaviour of these complexes not only aided the interpretation of the chromatograms but also assisted in the quantification of the metal–phytosiderophore complexes as metal cation complexation by EDTA is complete at equimolar metal cation and EDTA concentrations [19]. Quantification of the metal concentrations in the transient metal ion peaks, produced by all the metal–ligand complexes, was also performed off-line to verify the degree of metal complexation in equimolar 1 ␮M metal–ligand stock solutions. These metal recovery experiments were accomplished by comparing the total metal concentration in 600 ␮L fractions collected at the outlet of the anion exchange column at the retention time of the metal–ligand complexes with the metal concentrations of the injected stock solutions. Solutions that contained only one of the ligands were used in order to avoid any interference with

E. Bakkaus et al. / J. Chromatogr. A 1129 (2006) 208–215

the other metal–ligand complexes. The collected fractions were acidified with 10 ␮L of 65% HNO3 and diluted to 2.5 mL with high purity water. A 25 ␮L sample of the stock solutions was treated in an identical fashion. The metal concentrations in these solutions were determined directly by ICP-MS. On-line detection limits for the metal complexes were calculated from linear calibration curves, developed from the response of 4–5 standards varying in metal–phytosiderophore complex concentration (0.008–10 ␮M, depending on the metal being examined), using unweighted least-squares regression analysis (α = 0.05, Student’s t distribution on n − 2 degrees of freedom). 3. Results and discussion

211

can be estimated if the thermodynamic stability constant of the complex is known. For example, at thermodynamic equilibrium the dissociation rate constant (kd ) of a metal–ligand complex (ML) is related to the formation rate constant (kf ) and the thermodynamic stability constant of the complex (KML ): KML =

kf kd

(1)

As observed rates of metal–ligand complex formation are generally consistent with the rate limiting step being the initial exchange of one functional group of the ligand with an innersphere water molecule coordinated to the metal cation [21], kf can be estimated using the following equation: kf = kM-H2 O × KOS

3.1. Thermodynamic and theoretical kinetic properties of metal–phytosiderophore complexes As stated earlier, chromatographic techniques are generally limited to the analysis of thermodynamically stable metalcoordinated complexes, which do not rapidly dissociate during the separation process. Although labile metal–ligand complexes can be detected in certain conditions using AE LC, quantification of these complexes is complicated by competition between the metal cation and the positively-charged functional groups of the anion exchange column for complexation with the negativelycharged ligand (i.e. as labile metal–ligand complexes have rapid dissociation kinetics). Given that the apparent concentration of a labile metal–ligand complex in a sample will, therefore, depend upon the interactions with the functional groups of the anion exchange resin, accurate quantification of the complex will only arise if this interaction is quantified or, alternatively (if standards are used to quantify the complexes), if the metal and ligand concentrations in the sample and standard are identical. Thus, it should be clarified that this experimental work is focused solely on determining stable metal–phytosiderophore complexes that do not dissociate during the chromatographic separation process and, therefore, do not elute at, or near, the retention time of the uncomplexed ligand. The dissociation kinetics of metal–phytosiderophore complexes have not been experimentally determined, however, they

(2)

where kM-H2 O is the characteristic rate constant of water exchange for a metal cation and KOS is the thermodynamic stability constant of the outer-sphere metal–ligand complex. In the present context, some experimentally determined values of KML and kM−H2 O are available [9,22]. Furthermore, values of KOS can be generated based solely on electrostatic considerations [23] and these values have been conveniently tabulated by Morel and Hering [24] for a large range of metals and ligands varying in charge. As both mugineic and deoxymugineic acid are zwitterions, their overall molecular charge varies from +2 to −3 depending on the pH of the solution (Fig. 1). A charge of −1 has been adopted for these calculations as this is the overall molecular charge of mugineic and deoxymugineic acid at the pH of the stock solutions. As such, the KOS values for the metal–phytosiderophore complexes are 100.74 and 101.36 M for, respectively, the divalent and trivalent metal cations [24]. Calculations of the kd for the metal–phytosiderophore complexes listed in Table 1 (complexes of Cu2+ , Fe3+ , Ni2+ and Zn2+ ) indicate that the dissociation kinetics of these metal–phytosiderophore complexes should be slow enough so that the complexes remain intact during a normal AE LC separation time period (e.g. 5–15 min). This is assuming, of course, that the dissociation rates will not be dissimilar when the thermodynamic equilibrium of the sample solution is disturbed upon injection into the HPLC system—which is valid, regardless of

Table 1 Estimated dissociation rate constants (kd ) for (1:1) metal complexes with mugineic and deoxymugineic acid, based on the kM−H2 O and KOS data tabulated by Morel and Hering [24] and the KML data of Murakami et al. [9] kM−H2 O (s−1 )

Fe3+ Ni2+ Cu2+ Zn2+

2 × 102 3 × 104 1 × 109 7 × 107

Mugineic acid

Deoxymugineic acid (s−1 )a

KML (M)

kd

1015.4

2 × 10−12

1014.9 1018.1 1012.7

2 × 10−10 3 × 10−9 8 × 10−5

Complexation (%)

KML (M)

kd (s−1 )a

Complexation (%)

100 100 100 97

1016.3

2 × 10−13

97b 99 100 93

1014.8 1018.7 1012.8

3 × 10−10 7 × 10−10 6 × 10−5

The percent metal complexation, at thermodynamic equilibrium, for solutions (I = 0.01 M, pH 7) containing equimolar concentrations (1 ␮M) of the metals and phytosiderophores were calculated using GEOCHEM-PC, version 2 [25] with a database that included the pKa values of the phytosiderophores [10] and the KML values of the metal–phytosiderophore complexes. Both of the Fe3+ -deoxymugineic acid complexes, (Fe3+ L)− and (Fe3+ L(H−1 ))− , were considered in the calculations. a The dissociation kinetics of the Fe3+ –phytosiderophore complexes should be considered as indicative only as the ligand exchange reactions involving this cation are principally associative [22,30]. Nevertheless, considering the extremely slow predicted dissociation kinetics, when it is assumed that the reactions are dissociative, large variations in this rate constant would still suggest that this complex would be amenable to chromatographic separation techniques. b The Fe3+ -deoxymugineic acid complex of the type (Fe3+ L(H ))− accounted for > 99.99% of the complexed Fe3+ . −1

212

E. Bakkaus et al. / J. Chromatogr. A 1129 (2006) 208–215

the composition and pH of the mobile phase, if dissociation is predominantly disjunctive [24]. However, as adjunctive dissociation mechanisms may possibly be important during AE LC, in particular the unknown effect of outer-sphere complexation with the anion exchange functional groups, these calculations only serve as an indication of the slowest possible dissociation rates of the metal–phytosiderophore complexes during the chromatographic separation process. Due to the similarity of the coordination chemical and kinetic properties of Co2+ with Ni2+ [22] and the environmental toxicological significance of Cd2+ , these two metal cations were also included in the experiments, despite the lack of thermodynamic stability data on their phytosiderophore complexes. 3.2. Separation of Co2+ - and Ni2+ –phytosiderophore complexes by NaOH gradient elution A NaOH mobile phase was initially examined to separate the metal–phytosiderophore complexes as it has reportedly been successful for the AE LC separation of Fe3+ –phytosiderophore complexes [5]. In the present experiments, however, transient metal ion peaks were only observed for solutions containing Co2+ or Ni2+ and the phytosiderophores when using NaOH gradient elution to a final concentration of 200 mM. No metal ion peaks were observed for the solutions containing the phytosiderophores and Cd2+ , Cu2+ , Fe3+ or Zn2+ at equimolar 1 ␮M concentrations, despite off-line ICP-MS analyses indicating that these metal cations were actually present in the solutions. It should be noted that the very low detection limits of these metal ions by ICP-MS (0.1–0.005 ␮M) should facilitate detection of any of these metal–phytosiderophore complexes even if metal complexation is as low as 0.5–10%. These results, therefore, indicate that the dissociation kinetics of these phytosiderophore complexes are too rapid for the complexes to be detected and that the dissociated metal cations rapidly precipitate (or their negatively charged hydroxyl complexes are strongly retained in the column) in these chromatographic conditions. However, considering that no transient peaks for the metal cations were produced during gradient elution to 200 mM NaOH, precipitation seems the more likely mechanism. When analysed separately, the stock solutions containing either Co2+ or Ni2+ and one of the ligands (DMA, MA or EDTA) resulted in the production of only one transient peak. In contrast, however, transient peaks of the metal ions (59 Co+ , 58 Ni+ , 60 Ni+ ) were not observed when solutions containing only the metal cation or only the ligand were analysed. The precipitation of the Co2+ and Ni2+ cations would be expected with these chromatographic conditions. Example chromatograms, of the 59 Co+ and 58 Ni+ isotope ions, produced by solutions containing only the metal cation or the metal cation and the three ligands (MA, DMA and EDTA) are shown in Fig. 2 when using the following optimised gradient elution conditions: 0 mM NaOH (0–0.5 min), 0–20 mM NaOH (0.5–3.5 min), and re-equilibration at 0 mM NaOH (3.5–6.5 min). The phytosiderophore complexes of Co2+ and Ni2+ were well separated using this gradient elution and eluted before their respective EDTA complexes.

Fig. 2. AE LC–ICP-MS chromatograms, using NaOH gradient elution, of (a) the 58 Ni+ isotope ion showing the transient peaks for Ni2+ –DMA, Ni2+ –MA and Ni2+ –EDTA (in order of elution) as well as (b) the response of a solution containing Ni2+ only and (c) the 59 Co+ isotope ion showing the transient peaks for Co2+ –DMA, Co2+ –MA and Co2+ –EDTA (in order of elution) as well as (d) the response of a solution containing Co2+ only. The concentration of each complex in the solutions was 0.33 ␮M. The small earlier eluting peak in chromatogram (c) was attributed to Co3+ –EDTA. The y-axis represents relative counts per second (cps) for the respective metal isotope ions and the x-axis represents time (min) after the injection of the sample.

Anion separation with the Dionex AS11 anion exchange column used in these experiments is primarily via anion exchange mechanisms [19]. Therefore, as the metal–EDTA complexes have a −2 charge, this result supports the assignment of the earlier eluting transient metal ion peaks to the metal–phytosiderophore complexes, which have a −1 charge. In addition, at the pH of the NaOH mobile phase, the uncomplexed phytosiderophores have a −3 charge, which also suggests that these ligands will elute after the metal–EDTA complexes. Furthermore, Weber et al. [26] have reported that DMA and MA elute, respectively, at approximately 7.2 min and 8.2 min when using isocratic 12.5 mM NaOH elution from 0–8 min followed by gradient elution to 25 mM NaOH during 8–18 min (when using the identical analytical column). In the present experiments this methodology (including the use of an AG11 guard column and a mobile phase flow rate of 1 mL min−1 ) resulted in the Co2+ and Ni2+ complexes of the phytosiderophores to elute between 1.6 and 1.7 min, thus, experimentally verifying that the metal–phytosiderophore complexes elute before the uncomplexed ligands. As 20 mM NaOH was used as the final concentration in the gradient elution profile used in the present experiments, the uncomplexed phytosiderophores will also be eluted during each chromatographic run. Therefore, it was concluded that the transient metal ion peaks of Co and Ni were, in fact, due to the presence of stable metal–phytosiderophore complexes in the sample solutions. The results for the Ni2+ –phytosiderophore complexes are in agreement with the reported thermodynamic and theoretically calculated kinetic data. Given the similarity between the coordination chemical and kinetic properties of Co2+ and

E. Bakkaus et al. / J. Chromatogr. A 1129 (2006) 208–215

Ni2+ it is not surprising to also obtain successful chromatographic separation of the Co2+ –phytosiderophore complexes. Although the calculated values listed in Table 1 may be considered as a first estimate, the extremely slow dissociation kinetics predicted for the Cu2+ - and Fe3+ –phytosiderophore complexes would suggest that these complexes should not dissociate during the chromatographic separation process—even if errors spanning 5–6 orders of magnitude are considered in the calculations (in contrast to the Zn2+ –phytosiderophore complexes). As such, there must be another mechanism(s) occurring, which significantly increases the dissociation kinetics of the Cu2+ - and Fe3+ –phytosiderophore complexes in these chromatographic conditions. We propose base-assisted hydrolysis of the secondary amine group of the phytosiderophore to explain the apparent faster dissociation kinetics of the Cu2+ and Fe3+ complexes as these types of conjugate base reactions can increase dissociation kinetics by up to eight orders of magnitude [27,28]. 3.3. Separation of Cu2+ - and Fe3+ –phytosiderophore complexes by NH4 NO3 gradient elution Upon obtaining the results with the NaOH mobile phase, it was considered that further experiments were warranted at circumneutral pH values to eliminate the interferences of high OH− concentrations on the thermodynamic stability and dissociation kinetics of the metal–phytosiderophore complexes. As such, AE LC–ICP-MS experiments were conducted using NH4 NO3 as the mobile phase at pH 7. Similar to the results obtained with NaOH gradient elution, transient peaks of the Co and Ni metal isotope ions could be attributed to the Co2+ - and Ni2+ –phytosiderophore complexes (data not shown). However, the limits of detection were slightly higher and it was, therefore, concluded that NaOH gradient elution is more suited to the AE LC–ICP-MS detection and quantification of these metal complexes. Transient metal ion peaks of Cu (63 Cu+ , 65 Cu+ ) and Fe 54 ( Fe+ , 56 Fe+ ) were also produced when solutions containing Cu2+ or Fe3+ and the phytosiderophores were analysed by gradient elution with NH4 NO3 as the mobile phase. As observed for Co2+ and Ni2+ with the NaOH mobile phase, no transient metal ion peaks were detected for a solution that contained only 1 ␮M Cu2+ , indicating that uncomplexed Cu2+ is deposited within the analytical column. The examination of a solution containing only Fe3+ was not attempted as this metal cation is highly insoluble, e.g. −log(Ksp ) > 37.3 [29]. The gradient elution profile that provided the optimum separation conditions for the Cu2+ –phytosiderophore complexes, as well as elution of the Cu2+ –EDTA complex, consisted of 0–20 mM NH4 NO3 (0–3 min), 20–50 mM NH4 NO3 (3–5 min), and re-equilibration at 0 mM NH4 NO3 (5–8 min). Example chromatograms, of the 65 Cu+ isotope ion, produced by solutions containing only the metal cation or the metal cation and the three ligands (MA, DMA and EDTA) are shown in Fig. 3. Although the transient peaks of the Cu2+ –phytosiderophore complexes could not be baseline separated, neither the use of other mobile phases containing different counter anions (e.g.

213

Fig. 3. AE LC–ICP-MS chromatograms, using NH4 NO3 gradient elution (pH 7), of the 63 Cu+ isotope ion showing (a) the transient peaks for Cu2+ –DMA, Cu2+ –MA and Cu2+ –EDTA (in order of elution) as well as (b) the response of a solution containing only Cu2+ . The concentration of each complex in the solution was 0.16 ␮M. The y-axis represents relative counts per second (cps) for the 63 Cu+ isotope ion and the x-axis represents time (min) after the injection of the sample.

gradient elution 0–50 mM NaNO3 , K2 HPO4 or Na2 B4 O7 ) nor the variation of mobile phase pH from 6.8 to 7.5 further improved the separation of the two peaks. Nevertheless, this was eventually not considered to be problematic for quantifying the Cu2+ –phytosiderophore complexes as the two peaks could still be integrated using a 1.4 s integration time when using NH4 NO3 gradient elution. Regardless of the mobile phase used, no metal ion peaks were detected for the stock solutions containing Cd2+ or Zn2+ and the phytosiderophores. These observations, therefore, confirm that the dissociation kinetics of these phytosiderophore complexes are too rapid, regardless of the composition and pH of the mobile phase, to be detected using the current AE LC separation system. The gradient elution profile used to separate the Cu2+ –phytosiderophore complexes was not suitable for the Fe3+ –phytosiderophore complexes. The peaks were extremely broad, although distinguishable, eluting over a 30 s period. The peak shapes were neither improved by altering the mobile phase, as described for the Cu2+ –phytosiderophore complexes, nor by adding an organic solvent (e.g. methanol and acetonitrile up to 5% (v/v)). In fact, peak shape could only be improved by increasing the ionic strength of the initial mobile phase (i.e. the concentration of the mobile phase at the time of sample injection). As this was a common observation regardless of the mobile phase tested, it was tentatively concluded that the transient peaks were produced by the Fe3+ –phytosiderophore complexes and not by complexation of dissociated Fe3+ by components of the mobile phases (e.g. NH3(aq) , PO4 2− , etc.). The resultant AE LC conditions used to produce the 56 Fe+ ion chromatograms depicted in Fig. 4 consisted of 5–50 mM NH4 NO3 (0–3 min) gradient elution, and re-equilibration at 5 mM NH4 NO3 (3–6 min). As was the case for the Cu2+ –phytosiderophore complexes, the Fe3+ –phytosiderophore complexes were not baseline separated but could still be integrated using a 1.4 s integration time. Further experimental support for the assignment of the transient peaks to stable Cu2+ - and Fe3+ –phytosiderophore complexes was not as simple as for the Co2+ - and Ni2+ –phytosiderophore complexes as DMA, MA and their Cu2+ and Fe3+ complexes all have a molecular charge of −1 at pH

214

E. Bakkaus et al. / J. Chromatogr. A 1129 (2006) 208–215

Fig. 4. AE LC–ICP-MS chromatogram, using NH4 NO3 gradient elution (pH 7), of the 56 Fe+ isotope ion showing the transient peaks for Fe3+ –DMA, Fe3+ –MA and Fe3+ –EDTA complexes (in order of elution). The concentration of each complex was 10 ␮M. The y-axis represents counts per second (cps) for the 56 Fe+ isotope ion and the x-axis represents time (min) after the injection of the sample.

values between 6.8 and 7.5 and, in addition, the retention times of the uncomplexed phytosiderophores have not been reported for these chromatographic conditions. Attempts to monitor 12 C+ to determine the retention time of the uncomplexed phytosiderophores were unsuccessful, as the Series X7 ICP-MS does not allow measurement of this ion. Nevertheless, the average recoveries of Cu and Fe in the transient peaks, as assessed offline on triplicate collected fractions, were >94% and >97% for, respectively, the Cu2+ - and Fe3+ –phytosiderophore complexes. As the sample solutions were prepared at equimolar metal and ligand concentrations (and as only 1:1 metal–phytosiderophore complexes are formed), this result suggests little, if any, competition from the functional groups of the analytical column with ligand complexation (which would not be observed with similar low concentrations of labile metal complexes, cf. Section 3.1). The transient metal ion peaks were, therefore, assigned to stable phytosiderophore complexes of Cu2+ and Fe3+ .

4. Conclusions

3.4. Quantification of the metal–phytosiderophore complexes It can be noted from the detection limits listed in Table 2 that the sensitivity of a metal’s complexes to ICP-MS detection was very similar. This can also be observed in the representative chromatograms shown in Figs. 2–4. This indicates that (1) neither metal cation complexation by the ligands nor the concentration of the mobile phase during gradient elution appreciably

Table 2 Limits of detection (LOD) for the various metal complexes in absolute mass (pmol) and, in parentheses, concentrations (nM)

Fe3+ Ni2+ Cu2+ Co2+

affected ICP-MS detection of the respective metal ions and (2) the degree of metal complexation by the phytosiderophores is comparable, if not identical, to EDTA. This is significant as this would suggest that metal–phytosiderophore complexes, which are not commercially available for preparing standards, may be quantified on-line in unknown samples by direct comparison with the response of metal–EDTA complexes which are commercially readily available. These conclusions were experimentally verified off-line by comparing metal concentrations in the collected fractions, at the retention time of the metal–ligand complexes, with those in the respective stock solutions. ICP-MS analyses of the collected fractions and stock solutions indicated that the average recoveries of the metals (Co, Cu, Fe and Ni) complexed to DMA, MA and EDTA in the transient peaks were 104% (n = 36, i.e. triplicate collected fractions for the 12 metal–ligand complexes examined). These data, therefore, confirm that all these metal–ligand complexes were stable during the chromatographic process and, as such, the detection limits of the metal complexes listed in Table 2 imply 100% metal cation complexation. Furthermore, the detection of the Cu2+ and Fe3+ complexes did not appreciably alter when the pH of the NH4 NO3 mobile phase was adjusted from 6.8 to 7.5, signifying that these complexes are also stable within this pH range. As a result, the detection limits for most of the metal complexes may be considered to be quite reasonable. Although the detection limits for the Fe3+ complexes were somewhat higher than the other metal–ligand complexes, due to the incomplete removal of the 56 ArO+ isobaric interferences by the CCT technology, these concentrations are still 10–100 times lower than those previously reported for these particular Fe3+ complexes [5,19].

DMA

MA

EDTA

2.8 (110) 0.13 (5) 0.35 (14) 0.15 (6)

2.3 (90) 0.1 (4) 0.13 (5) 0.1 (4)

2.0 (80) 0.13 (5) 0.25 (10) 0.1 (4)

The detection limits of the Co2+ (59 Co+ ) and Ni2+ (58 Ni+ ) complexes and the Cu2+ (63 Cu+ ) and Fe3+ (56 Fe+ ) complexes were determined, respectively, with the NaOH and NH4 NO3 gradient elution AE LC–ICP-MS conditions described in the text. The values were calculated from linear calibration curves using unweighted least-squares regression analysis (α = 0.05, Student’s t distribution on n − 2 degrees of freedom).

The initial assessment of thermodynamic data and theoretical kinetic calculations provides a valuable first estimation of the metal–ligand complexes that are amenable to chromatographic separation procedures. However, the results presented in this study also demonstrate that the apparent dissociation kinetics can be profoundly influenced by pH and, as such, various mobile phases at different pH values should be examined in the development of chromatographic methodology. The AE LC–ICP-MS techniques reported here were developed, and are only suitable, for the stable metal complexes of mugineic and deoxymugineic acid that do not dissociate during chromatographic separation. As such, in these experimental conditions, only the Co2+ , Cu2+ , Fe3+ and Ni2+ phytosiderophore complexes are considered to be stable. Acknowledgements The authors gratefully acknowledge Laurent Ouerdane and Dirk Schauml¨offel from the University of Pau/CNRS UMR 5034 for the electrospray-MS/MS analyses to identify DMA and to confirm the unique presence of Co2+ -, Cu2+ -, Fe3+ and Ni2+ –DMA complexes in the collected wheat root exudate solutions.

E. Bakkaus et al. / J. Chromatogr. A 1129 (2006) 208–215

The French national program ‘TOXNUC-E’ is also acknowledged for funding, in part, this work. References [1] M. Takahashi, H. Nakanishi, S. Kawasaki, N.K. Nishizawa, S. Mori, Nat. Biotechnol. 19 (2001) 466. [2] C. Curie, Z. Panaviene, C. Loulergue, S.L. Dellaporta, J.-F. Briat, E.L. Walker, Nature 409 (2001) 346. [3] G. Schaaf, U. Ludewig, B.E. Erenoglu, S. Mori, T. Kitahara, N. von Wiren, J. Biol. Chem. 279 (2004) 9091. [4] L.A. Roberts, A.J. Pierson, Z. Panaviene, E.L. Walker, Plant Physiol. 135 (2004) 112. [5] G. Weber, G. Neumann, V. Romheld, Anal. Bioanal. Chem. 373 (2002) 767. [6] M.J. Mench, S. Fargues, Plant Soil 165 (1994) 227. [7] V. Chaignon, D.D. Malta, P. Hinsinger, New Phytol. 154 (2002) 121. [8] M. Shenker, T.W.-M. Fan, D.E. Crowley, J. Environ. Qual. 30 (2001) 2091. [9] T. Murakami, K. Ise, M. Hayakawa, S. Kamei, S.-I. Takagi, Chem. Lett. (1989) 2137. [10] N. von Wiren, H. Khodr, R.C. Hider, Plant Physiol. 124 (2000) 1149. [11] Y. Sugiura, H. Tanaka, Y. Mino, T. Ishida, N. Ota, K. Nomoto, H. Yoshioka, Y. Takemoto, J. Am. Chem. Soc. 103 (1981) 6979. [12] T. Iwashita, Y. Mino, H. Naoki, Y. Sugiura, K. Nomoto, Biochemistry 22 (1983) 4842. [13] Y. Mino, T. Ishida, N. Ota, M. Inoue, K. Nomoto, H. Yoshioka, T. Takemoto, Y. Sugiura, H. Tanaka, Inorg. Chem. 20 (1981) 3440.

215

[14] Y. Mino, T. Ishida, N. Ota, M. Inoue, K. Nomoto, T. Takemoto, H. Tanaka, Y. Sugiura, J. Am. Chem. Soc. 105 (1983) 4671. [15] N. von Wiren, S. Klair, S. Bansal, J.-F. Briat, H. Khodr, T. Shioiri, R.A. Leigh, R.C. Hider, Plant Physiol. 119 (1999) 1107. [16] R.N. Collins, J. Chromatogr. A 1059 (2004) 1. [17] S.-I. Takagi, K. Nomoto, T. Takemoto, J. Plant Nutr. 7 (1984) 469. [18] G. Neumann, C. Haake, V. Romheld, J. Plant Nutr. 22 (1999) 1389. [19] A.A. Ammann, J. Chromatogr. A 947 (2002) 205. [20] R.N. Collins, B.C. Onisko, M.J. McLaughlin, G. Merrington, Environ. Sci. Technol. 35 (2001) 2589. [21] D.W. Margerum, G.R. Cayley, D.C. Weatherburn, G.K. Pagenkopf, in: A.E. Martell (Ed.), Coordination Chemistry, American Chemical Society, Washington, DC, 1978, p. 1. [22] L. Helm, A.E. Merbach, Coord. Chem. Rev. 187 (1999) 151. [23] R.M. Fuoss, J. Am. Chem. Soc. 80 (1958) 5059. [24] F.M.M. Morel, J.G. Hering, Principles and Applications of Aquatic Chemistry, John Wiley & Sons, New York, NY, 1993. [25] D.R. Parker, W.A. Norvell, R.L. Chaney, in: R.H. Loeppert, A.P. Schwab, S. Goldberg (Eds.), Chemical Equilibrium and Reaction Models, Soil Science Society of America, American Society of Agronomy, Madison, WI, 1995, p. 253. [26] G. Weber, G. Neumann, C. Haake, V. Romheld, J. Chromatogr. A 928 (2001) 171. [27] R.G. Wilkins, Kinetics and Mechanism of Reactions of Transition Metal Complexes, second thoroughly revised edition, VCH Publishers, New York, NY, 1991. [28] D.T. Richens, Chem. Rev. 105 (2005) 1961. [29] D. Langmuir, D.O. Whittemore, Adv. Chem. Ser. 106 (1971) 209. [30] J. Burgess, Ions in Solution: Basic Principles of Chemical Interactions, Ellis Horwood, Chichester, England, 1988.