Fractionation of metal complexes with dissolved organic matter in a rhizosphere soil solution of a humus-rich Andosol using size exclusion chromatography with inductively coupled plasma-mass spectrometry

Abstract

To better understand the behavior of metals in soil–plant systems, their physicochemical forms in rhizosphere soil should be elucidated. The dissolved organic matter (DOM) in a soil solution influences the mobility and bioavailability of metals in soil. The present study examined the effects of plant growth on DOM–metal complexes in a rhizosphere soil solution using size exclusion chromatography combined with an inductively coupled plasma-mass spectrometer system (SEC–ICP-MS) and an ultrafiltration technique. Humus-rich volcanic ash soil from the surface of an agricultural field was used in a pot cultivation experiment. Brassica rapa nothovar. was cultivated in a pot in which rhizosphere soil (R) was separated by a nylon net screen from non-rhizosphere soil (NR). Soil solutions were collected using a high-speed centrifugation method 3 weeks after sowing and analyzed using SEC–ICP-MS. Some peaks of DOM with a high molecular size were detected in the ultraviolet-absorbing chromatograph (280 nm) of the soil solution samples. Their concentrations were much higher in the R solution than in the NR solution. Metals including Al, Fe, Cu, Zn, Pb, Y, La and U were detected at the ultraviolet peak positions of the DOM. The ultrafiltration experiment showed that the size distributions of the organic materials to which the metals were combined differed between the R and NR soil solutions. These results suggest that plant growth enhanced the dissolution of metals adsorbed with organic matter from the solid phase in rhizosphere soil.

Key words:

• dissolved organic matter
• high-performance liquid chromatography
• inductively coupled plasma-mass spectrometry
• speciation
• trace metals

INTRODUCTION

Dissolved organic matter (DOM) is an important constituent of soil solution and plays an important role in many chemical and biological processes in soil. In particular, DOM regulates the bioavailability, toxicity and mobility of metals in soil solution because of its ability to complex with metals. To better understand the behavior of metals in soil–plant systems, the fate and role of DOM–metal complexes in soil needs to be elucidated. The rhizosphere is an extremely important microenvironment for the transport of metals from the soil to plants, and soil characteristics in the rhizosphere differ greatly from bulk soil because of plant–soil–microbial interactions (Hinsinger 2001; Lombi et al. 2001). A number of studies have shown that changes to the dissolved organic carbon (DOC) concentration are closely related to the bioavailability or speciation of metals in rhizosphere soil (Legrand et al. 2005; Tao et al. 2005). However, there is little information about DOM–metal complexes in the rhizosphere and their behavior and status needs to be elucidated.

Size exclusion chromatography (SEC) has been widely used to estimate the molecular size of humic substances (Nobili and Chen 1999; Piccolo et al. 2001). In recent metal speciation studies, inductively coupled plasma-mass spectrometry (ICP-MS) has been used in direct combination with SEC (Piatina and Hering 2000), particularly for the determination of DOM–metal complexes in natural waters (Haraguchi et al. 1995, 1998; Itoh et al. 2000; Vogl and Heumann 1997; Wu et al. 2004). The SEC–ICP-MS system has also been used for analyses of various materials, for example, for water-soluble boron compounds in radish roots (Matsunaga et al. 1996), for iodine associated with humic and fulvic acid in soil (Yamada et al. 2002), for trace elements in citrate extracts from soil (Kimura 2004), in soil water (Casartelli and Miekeley 2003) and in extracts from solid waste (Sadi et al. 2002), and for uranium (U) in a water extract from sediment (Jackson et al. 2005).

In a previous report, we showed that concentrations of DOC and U in a soil solution in the rhizosphere of Brassica rapa nothovar. increased with time during cultivation, implying that plant growth enhances U solubility in relation to DOM (Takeda et al. 2008). In the present study, DOM–metal complexes in soil solution under plant growth conditions were observed using a SEC–ICP-MS system and an ultrafiltration technique to elucidate the influence of plant root activity on metal speciation in a rhizosphere soil solution.

MATERIALS AND METHODS

Soil and soil solution

The soil used in the present study was collected from the plow layer of an Andosol in Aomori Prefecture, Japan (40°51′N, 141°06′E). The collected soil was dried at 50°C and passed through a 2-mm sieve. The particle size distribution of the soil samples was determined using sieving and sedimentation methods. The water-holding capacity was determined using a Hilgard soil cup (Daiki Rika Kogyo, Kounosu, Japan). The pH (H₂O) and pH (KCl) values were measured at a soil : water ratio of 1:2.5. The cation exchange capacity (CEC) of the soil samples was determined using the semi-micro Schollenberger method. Exchangeable cations (Ca, Mg, K and Na) were extracted by 1 mol L⁻¹ ammonium acetate (pH 7) and determined using ICP atomic emission spectrometry (ICP-AES). Acid-oxalate-extractable Si (Si_o), Al (Al_o) and Fe (Fe_o) were determined using the methods described by Blakemore et al. (1981). The soil samples were finely ground with an agate mortar and pestle. The total carbon and nitrogen contents in the soil were measured using a dry combustion method. The concentrations of metals in the soils were determined by ICP-AES or ICP-MS after decomposition with a mixed acid solution (HNO₃, HF and HClO₄) using a microwave digestion technique.

The physical and chemical properties are shown in Table 1. The soil was rich in poorly crystalline clay minerals and organic matter. An X-ray diffraction analysis showed the occurrence of 2:1–2:1:1 intergrade mineral, illite and kaolinite in the soil clay fraction. The soil had normal levels of metal concentrations without evident pollution in comparison with the average concentrations in Japanese soils (Takeda et al. 2004).

Table 1 Soil properties and concentrations of the elements in the soil

Particle size distribution (wt %)			Texture	Water-holding capacity	Soil pH		Oxalate extractable (g kg^−1)			Allophane^†	Ferrihydrite^†	CEC	Exchangeable cations				Total C	Total N	Concentration of elements in the soil
													(cmolc kg^−1)						Al	Fe	Mn	Cu	Zn	Y	Cd	La	Pb	U
Sand	Silt	Clay		(g g^−1)	H2O	KCl	Sio	Alo	Feo	(g kg^−1)		(cmolc kg^−1)	Ca	Mg	K	Na	(g kg^−1)		(g kg^−1)			(mg kg^−1)
48	34	18	Clay loam	1.0	5.5	4.2	10	29	14	71	24	32	9.7	1.8	0.2	0.2	85	6.3	90	35	1.6	20	146	34	0.48	31	21	2.4
^†Allophane and ferrihydrite contents were estimated by Sio × 7.14 and Feo × 1.7, respectively. CEC, cation exchange capacity.

Soil (400 g) was placed into a plastic pot (11-cm diameter and 7-cm high) after mixing with NH₄NO₃ and KH₂PO₄ (60 mg N pot⁻¹; 17 mg P pot⁻¹). The rhizosphere (R) soil in the pot was separated from the non-rhizosphere (NR) soil by a nylon mesh screen (with a mesh size of 20 µm) through which water, but not plant roots, could pass. The R compartment (upper part of the pot) and the NR compartment (bottom part of the pot) contained 80 g and 320 g of soil, respectively. The soil water content was adjusted to 70% of the water-holding capacity by adding deionized water 2 days before sowing and everyday after sowing. Brassica rapa nothovar. (komatsuna, Japanese mustard spinach) was used for the experiment. After five germinated seeds were sown into each pot, the soil surface in each pot was covered with polypropylene balls (5-mm diameter) to reduce water evaporation. The plants were cultivated in a controlled environmental growth chamber (20°C, 70% relative humidity, 12 h day length and 20,000 lx light intensity). Two seedlings from each pot were culled 6 days after sowing, leaving three plants in each pot. The pot experiments were conducted in duplicate.

At 0 days and 21 days after sowing, the R and NR soils were collected from two pots 4 h after adjusting the water content in the soil. Most plant roots were removed from the R soil. The soil solution was collected immediately after soil sampling using a high-speed centrifuge (H-1400pF; Kokusan Corporation, Tokyo, Japan) up to –1.5 M Pa (10,000 g, 90 min, 20°C) and filtered through a membrane filter (pore size 0.45 µm). The R and NR soil solutions obtained from two soils were mixed to obtain composite samples.

An ultrafiltration technique was used to acquire a rough size distribution of elements in the soil solution. The R and NR soils were collected from three different pots 26 days after sowing, and the soil solutions were collected from the soils immediately by centrifugation (as outlined above). The soil solutions were filtered through a 0.45-µm pore size membrane filter and a portion of the solution was filtered using a centrifugal filter device (Vivaspin 20; Sartorius AG, Goettingen, Germany) with a cutoff molecular weight of 10,000 Da. The concentrations of metals in the soil solution with and without the ultrafiltration process were determined.

Analytical method

The soil solution samples collected from the R and NR soil 21 days after sowing were analyzed using SEC–ICP-MS. The SEC–ICP-MS analysis used in the present study was based on a method developed for lake water (Takaku et al. 2003). The LC/ICP-MS system consisted of a high-performance liquid chromatograph system (LC1100 LC system; Agilent Technologies, Santa Clara, CA, USA) combined with an ICP-MS instrument (VG-PQ-Excel; Thermo Elemental, Winsford, UK). The organic materials were measured at an absorbance of 280 nm ultraviolet (UV) using an UV/Vis absorption detector. The UV absorbance at 280 nm is commonly used to estimate the apparent molecular size and aromaticity of DOM (Chin et al. 1994; Kalbits et al. 2003). A gel permeation column measuring 10.7 mm internal diameter and 300 mm length (GL-W540; Hitachi Chemical, Tokyo, Japan) was used as an SEC column. A buffer solution of 0.01 mol L⁻¹ Tris–HNO₃ (pH 7.3) was selected as the mobile phase. The sample volume injected into the column was 100 µL, and the flow rate was 1.0 mL min⁻¹. The molecular weights were calibrated using a Gel Filtration Standard (Bio-Rad Laboratories, Hercules, CA, USA), which was a lyophilized mixture of molecular weight markers, including thyroglobulin (670,000 Da), γ-globulin (158,000 Da), ovalbumin (44,000 Da), myoglobin (17,000 Da) and vitamin B-12 (1,350 Da). Metals that might be environmental pollutants (Cu, Zn, Cd, Pb and U) and metals with a high ability to complex with organic ligands (Y and La) were selected for determination, along with metals that were abundant in the soil (Al, Fe and Mn). Signals of ²⁷Al, ⁵⁵Mn, ⁶⁵Cu, ⁶⁶Zn, ⁸⁹Y, ¹¹⁴Cd, ¹³⁹La, ²⁰⁸Pb and ²³⁸U in the eluate from the UV detector were monitored using ICP-MS. Collision cell technology with helium gas was used to reduce the spectral interference resulting from ⁴⁰Ar¹⁶O for ⁵⁶Fe measurements in the ICP-MS. The time lag between the UV and ICP-MS detection was corrected for by using the signal of ⁵⁹Co in the vitamin B-12.

The concentrations of the elements in the soil solution before and after the ultrafiltration treatment were measured using ICP-MS and ICP-AES. The DOC in the soil solution was measured using a total organic carbon analyzer (TOC-5000; Shimadzu, Kyoto, Japan).

RESULTS AND DISCUSSION

The soil solution pH at the beginning of the experiment was 5.8, resembling that in the NR soil solution after 21 days. In contrast, the soil solution pH in the R soil was significantly higher than that in the NR soil (Table 2). The pH change in the rhizosphere compensates for the imbalance of anions and cations taken up by the plant roots (Hinsinger et al. 2003; Nye 1981). The electrical conductivity in the soil solution collected from the R soil was much lower than that in the NR soil, probably because of nutrient uptake (Table 2). The DOC concentration in the soil solution at the beginning of the experiment was high, and decreased with time, particularly in the NR soil solution because the DOC was readily available to microorganisms (Cook and Allin 1992; Marschner and Bredow 2002). The concentration of DOC in the R soil solution was approximately threefold higher than that in the NR soil solution, indicating that root activity elevates the DOC in the rhizosphere. Plant roots release various organic substances, including both high and low molecular weight compounds, which can increase the DOC in the rhizosphere soil solution. In the experimental pot, the root compartment is mixed with the soil. Therefore, both root exudate and soil compounds might have contributed to the increase in DOC (Wenzel et al. 2001). In addition, the root hairs cannot be completely removed from the R soil before the collection of the soil solution, and this can cause disruption to the roots during the centrifugation process and root sap might have contaminated the solution samples (Lorenz et al. 1994). In this experiment, the R soil was placed in the upper part of the pot. Therefore, differences in the results between the R and the NR soils might be caused by a position effect as well as a plant effect. However, the water content in the soils was less than 70% of the maximum water-holding capacity during the experimental period. The plastic balls on the soil surface prevented the influence of water evaporation and exposure to light. Therefore, the results obtained in the experiment strongly reflect a plant effect rather than a position effect.

Table 2 pH, electrical conductivity and concentrations of the elements in the soil solution (< 0.45 µm)

		I		R		NR		R/NR
pH		5.8 ab	(0.2)	6.2 b	(0.3)	5.6 a	(0.2)
EC	(mS cm^−1)	0.90 b	(0.1)	0.09 a	(0.05)	1.0 b	(0.3)
DOC	(mg L^−1)	104 c	(0.4)	67 b	(4.5)	26 a	(3.1)	2.6
Al	(µg L^−1)	166 a	(4.7)	766 b	(131)	173 a	(36)	4.4
Fe	(µg L^−1)	12 a	(0.6)	131 b	(18)	5.2 a	(0.1)	25
Mn	(µg L^−1)	247 b	(47)	27 a	(20)	71 a	(56)	0.4
Cu	(µg L^−1)	49 a	(3.3)	36 a	(18)	28 a	(12)	1.3
Zn	(µg L^−1)	14 a	(0.7)	37 a	(31)	41 a	(21)	0.9
Y	(µg L^−1)	0.69 ab	(0.03)	1.2 b	(0.3)	0.44 a	(0.01)	2.8
Cd	(µg L^−1)	0.094 ab	(0.002)	0.033 a	(0.002)	0.17 b	(0.04)	0.2
La	(µg L^−1)	0.51 a	(0.01)	0.63 a	(0.20)	0.29 a	(0.04)	2.2
Pb	(µg L^−1)	0.15 a	(0.01)	0.42 a	(0.12)	0.47 a	(0.15)	0.9
U	(µg L^−1-)	0.10 ab	(0.004)	0.16 b	(0.06)	0.019 a	(0.004)	8.6
I, initial soil solution obtained at the beginning of the experiment; R, soil solution obtained from the rhizosphere on day 21 after sowing; NR, soil solution obtained from the non-rhizosphere on day 21 after sowing. Data are the means and standard deviations (in parentheses) (I, n = 2; R and NR, n = 3). Different letters within a column indicate significant differences using Scheffe's multiple comparison tests (P = 0.05).

The concentrations of the metals in the soil solution are presented in Table 2. The concentrations in the R soil solution were higher than those in the NR soil solution for some metals, such as Fe, U, Al, Y and La. In particular, the Fe concentration in the R soil solution was more than 20-fold higher than that in the NR soil solution.

The results of the UV absorbance in the soil solutions are shown in Fig. 1. In both the R and the NR soil solutions, three broad UV-absorption peaks were observed over a retention time of 16–20 min, which corresponded to molecular weights of approximately 5–30 kDa. The peaks for the R soil solution were higher than those for the NR soil solution. A small peak was observed only in the R soil solution at a retention time of approximately 11 min, which would be larger than the exclusion limit of the column. The results suggest that plant roots enhance the dissolution of high molecular weight organic substances into the rhizosphere soil solution. Plant roots release low molecular weight organic acids into the rhizosphere soil. Some Brassica plants exude considerable amounts of organic acids, such as citric acid (Chiang et al. 2005; Gerke and Meyer 1995; Hoffland et al. 1989; Jones 1998; Zhang et al. 1997). The UV-absorbing organic substances in the soil solutions are considered to be humic substances (Aoyama 1996; Berdén and Berggren 1990), mainly fulvic acid (Aoyama 2002). Humic substances are modified by the organic acids of the root exudates (Albuzio and Ferrari 1989). Therefore, the low molecular weight organic acids released from the root might solubilize the humic substances in the R soil.

The signal intensities detected by ICP-MS are shown in Fig. 1. In both the R and NR soil solutions, Al and Fe were detected at 16–20 min, which corresponds to UV-absorbing organic substances. The peaks of Al and Fe in the R soil solution were much higher than those in the NR soil solution. These results suggest that Al and Fe were dissolved from the soil solid phase as forms of DOM complexes. Humic substances are known to be solubilized by citrate, and humic–Al and humic–Fe complexes are important species in the presence of humic substances and citrate (Gerke 1993, 1997).

Figure 1  Size exclusion chromatography–inductively coupled plasma-mass spectrometry (SEC–ICP-MS) chromatograms of the soil solution samples. (A) Ultraviolet absorbance and (B) ICP-MS detection of the metals. The arrows indicate the retention time for the molecular weight markers. R, rhizosphere soil solution; NR, non-rhizosphere soil solution.

Figure 2  Percentages of metal concentrations in the low molecular weight fraction (< 10 kDa) compared with the soil solution (< 0.45 µm). Lowercase letters indicate significant differences between R and NR in each metal using a two-sample t-test (P < 0.05). Error bars indicate standard deviation (n = 3). R, rhizosphere soil solution; NR, non-rhizosphere soil solution; n.d, not detected.

The exclusion limit for the SEC column was 700 kDa, and molecules or colloids larger than this size should elute in the void volume at 11 min. For both Al and Fe, a large peak was found only in the R soil solution at 11 min, and this might correspond to metals mainly in an inorganic colloidal form. The lower ionic strength in the R soil solution might enhance the dispersion of colloidal substances. Free metal cations or complexes with low molecular weight organic acids less than the effective sizable range of the SEC column (approximately 1 kDa) did not elute and were presumably bound to anionic surface groups of the gel phase (Jackson et al. 2005).

Results for the other trace metals are presented in Fig. 1. The chromatograms observed for Cu, Zn, Y, La, Pb and U resemble those for Al and Fe. These metals would also be mobilized as DOM complexes and/or colloidal forms in the rhizosphere soil. Using SEC–ICP-MS, similar results have been found for metals associated with DOM in natural water samples (Casartelli and Miekeley 2003; Haraguchi et al. 1995, 1998; Itoh et al. 2000; Jackson et al. 2005; Takaku et al. 2003; Wu et al. 2004). In contrast, Mn was only found in the fraction at the exclusion limit in the R soil solution, indicating that Mn did not exist as a DOM-binding form in the soil solution, but was present as a colloidal form in the rhizosphere solution. No obvious peaks were observed for Cd in either the R or NR soil solution.

The concentrations of metals in the filtrate of the ultrafiltration with a cutoff molecular weight of 10 kDa were compared to those in the original soil solution (< 0.45 µm). The metal concentrations in the low molecular weight fraction (< 10 kDa) are shown in Fig. 2 as percentages of those in the whole soil solution (< 0.45 µm). In the NR soil solution, the metal concentrations before and after ultrafiltration did not differ greatly. Therefore, the low molecular weight fraction occupied more than 90%, except for U (87%) and Cu (76%). These results indicated that the metals in the soil solution were mainly in smaller labile forms, including free ions or organic and inorganic complexes. In contrast, the percentages of metals in the low molecular weight fraction in the R soil solution were significantly lower than those in the NR soil solution for Al, Mn, Zn, Pb, Y, La and U. In other words, the percentages of the larger size fraction (0.45 µm–10 kDa) would be higher in the R solution than in the NR solution. The percentages of the low molecular weight fraction of metals in the R solution were in the following order: Cd, Zn, Cu and Y (> 80%); Al, La and U (80–60%); Mn (49%); Fe and Pb (27%). The results obtained in the filtration experiments were consistent with the observations obtained using SEC–ICP-MS (Fig. 1), and suggested that the high molecular weight DOM complexes or colloidal forms of the metals would be more abundant in the rhizosphere soil solution.

For Cd, free ion and labile complexes smaller than 10 kDa would be dominant in both the R and NR soil solution, according to the SEC–ICP-MS and ultrafiltration analysis (Figs 1, 2). Although a higher concentration of a DOM–Cu complex in the R solution was found in the SEC–ICP-MS analysis (Fig. 1), significant differences were not found for the concentration and size distribution of the DOM–Cu complex between the R and the NR solutions (Table 2; Fig. 2). It is important to note that the SEC and ultrafiltration techniques sometimes yield different results for molecular weight estimations, possibly because of filtration effect artifacts during the ultrafiltration process (Zsolnay 2003), non-size exclusion effects (Perminova 1999) and dissociation of metal complexes during chromatographic separation (Piatina and Hering 2000).

The results obtained in the present study suggest that root activities can influence not only the concentrations of metals, but also the size distribution of their complexes with DOM in the rhizosphere soil solution. Many studies have examined the relationship between root exudation and metal mobilization in the rhizosphere, particularly in relation to the chelation of metals by organic acids (Lombi et al. 2001). In contrast, low molecular weight organic acids, such as citrate, can break bridges between Fe or Al and humic molecules by complexing the metals (Gerke 1992). Therefore, Al, Fe and other trace metals that are strongly associated with humic substances might be released to the soil solution as forms of DOM–metal complexes in the rhizosphere. The partitioning of metals between humic substances and organic acids should depend on the concentrations and species of the organic acids. Kimura (2004) reported that most transitional metals in 0.01 mol L⁻¹ citric acid extracts formed complexes with citric acid, and it is implied that these metals were desorbed from humic complexes by citric acid.

Although the separation of the R soil was operationally defined, the results show that the R and NR soils clearly differ with respect to the concentration and speciation of metals in the soil solution, and our results suggest that rhizosphere effects can modify the molecular size distribution of DOM–metal complexes in soil solution.

ACKNOWLEDGMENTS

The authors are grateful to Ms J. Kogawa and K. Taneichi (ZAX) for their assistance with our work. We also thank the analytical chemistry staff at the Environmental Research Center for their assistance with the elemental analysis and JGC Plantech Aomori for their assistance with plant cultivation in the environmentally controlled chamber. This work was carried out under contract with the Aomori Prefectural Government, Japan.