Evaluation of the Interactions between Water Extractable Soil Organic Matter and Metal Cations (Cu(II), Eu(III)) Using Excitation-Emission Matrix Combined with Parallel Factor Analysis

The objectives of this study were to evaluate the binding behavior of Cu(II) and Eu(III) with water extractable organic matter (WEOM) in soil, and assess the competitive effect of the cations. Excitation-emission matrix (EEM) fluorescence spectrometry was used in combination with parallel factor analysis (PARAFAC) to obtain four WEOM components: fulvic-like, humic-like, microbial degraded humic-like, and protein-like substances. Fluorescence titration experiments were performed to obtain the binding parameters of PARAFAC-derived components with Cu(II) and Eu(III). The conditional complexation stability constants (logKM) of Cu(II) with the four components ranged from 5.49 to 5.94, and the Eu(III) logKM values were between 5.26 to 5.81. The component-specific binding parameters obtained from competitive binding experiments revealed that Cu(II) and Eu(III) competed for the same binding sites on the WEOM components. These results would help understand the molecular binding mechanisms of Cu(II) and Eu(III) with WEOM in soil environment.


Introduction
Dissolved organic matter (DOM) is ubiquitous in soil system. Although being small in weight percentages, DOM is the most labile and reactive fraction of the multicomponent soil organic matter pool. As DOM plays a significant role in the biogeochemical cycling of trace metals, remarkable efforts have been made to classify the processes involved in trace metal interacting with DOM, through both experimental and modeling approaches [1][2][3]. Nevertheless, the processes are not fully understood. For example, determination of metal binding constants with DOM is still hampered, due to the intrinsic complexity of DOM, the lack of stoichiometric information, and analytical limitations [4]. Trace metals complexation with DOM remains poorly defined at the molecular scale under relevant environmental conditions, such as low concentrations of metals relative to DOM [5].
Excitation-emission matrix combined with parallel factor analysis (EEM-PARAFAC) is a selective, high sensitive and non-destructive technique. The EEM fluorescence spectra can be seen as an overall "fingerprint" of DOM bulk samples that include the key components [6]. However, the overlapping of the fluorescence spectra between complex DOM components may lead to confusion in subsequent data analysis. PARAFAC analysis, a multiway data analysis method, can model the complex EEM landscape and extract chemically meaningful spectral and concentration components [7,8]. Moreover, this combined technique is applicable to low DOM (less than 10 mg/L) and metal ion concentrations of environmental relevance. The EEM-PARAFAC derived soil DOM components mainly included humic-and fulvic-, and protein-like substances [9][10][11]. The humic and fulvic components, which carry a large number of functional groups, can play a dominant role in interacting with metals in terrestrial environments [12]. EEM-PARAFAC has been applied to study the interaction between metals such as Al(III) and Fe(III) with soil DOM components [10,13].
Dissolved organic matter in soil samples, is often extracted with water or dilute aqueous salt solutions in laboratory experiments [14,15]. In this paper, we will emphasize the interaction of metal cations (Cu(II) and Eu(III)) with water extractable organic matter (WEOM) from soil. Cu is a plant nutrient at low concentrations but harmful at high concentrations [16]. Increased application of wastewater, sludge and manure helped Cu accumulation in agricultural soils worldwide [17][18][19]. Eu(III) is often used as an analogue for trivalent lanthanium, a member of the rare earth elements (REEs). REEs may reach soil system through mining production [20], application of REEs-enriched fertilizers [21] and animal manure containing the lanthanides [22]. As an emerging pollutant, the accumulation of REEs in soil, their bioaccumulation in crops, and entry to the food chain are of growing ecological and human health concerns in China [23][24][25].
In order to gain a better understanding of the behavior of the REEs and heavy metals in the soil environment, detailed investigation is warranted to identify the mechanisms of REEs interacting with soil WEOM, with the existence of competing heavy metal cations. Therefore, the primary objectives of the present study were: (1) To identify the main binding components in soil WEOM and evaluate the binding constants of each component for Cu(II) and Eu(III) using EEM combined with PARAFAC analysis; and (2) To explore the binding sites and the competing effect of Cu(II) and Eu(III) with soil WEOM.

Fluorescence Properties of Water Extractable Organic Matter (WEOM)
An example is shown in Figure 1 the fluorescence EEM spectrum of a diluted WEOM sample without metal titration. Visual inspection of the WEOM spectrum suggests the presence of three fluorophores at Excitation/Emission (Ex/Em) wavelength pairs of 230/440, 330/400, and 350/450 nm. The 230/440 and 350/450 nm peaks, presented in the ultraviolet range and in the visible range, were designated as A and C, respectively, and were attributed to humic-like substances [26]. The 330/400 nm peak presented in the ultraviolet range was designated as a microbial degraded humic (M) peak [9].

Parallel Factor Analysis (PARAFAC) Derived Components
Four fluorescent components were identified by PARAFAC analysis based on the split-half validation procedure ( Figure 2). The largely overlapping excitation and emission loadings of the four components, modeled with the halves of the dataset and on the whole dataset are also shown in Figure 2. All fluorescent components had single emission maximum and single or multiple excitation maxima. Component 1 (Ex/Em = 250/460) was classified as fulvic-like substance, and component 2 (Ex/Em = (235, 330)/440) was designated as humic-like substances. Component 3 (Ex/Em = (240, 300)/400) was ascribed as microbial degraded humic-like substances. Component 4 (Ex/Em = (<235, 275)/340) was protein-like, containing mainly tryptophan-like substances. In the case of the high fluorescence intensity of the fulvic and humic peaks masking the protein peak presented in the bulk sample spectra, the PARAFAC can decompose the complex mixture effectively. These components had been previously identified in the terrestrial environment [27,28].
The fluorescence intensity per unit TOC [QSU/(mg/L)] of components 1-4 were 0.74, 1.39, 0.56 and 0.27, and their relative abundance to total fluorescence were 25.1%, 46.8%, 19.0% and 9.1%, respectively. Fulvic-like and humic-like components are most commonly abundant in terrestrial DOM, such as DOM from soil extractions, soil solution, wetlands, and forested streams. Components derived from PARAFAC in previous studies that fall within the microbial degraded humic-like component were variable in occurrence and abundance. However, it was found that the identity of this component increased following biodegradation of plant biomass and manure [28][29][30].

Interaction of Fluorescent Components with Cu(II) and Eu(III)
The fluorescence quenching curves of PARAFAC-derived components with Cu(II) and Eu(III) are shown in Figure 3. As initial levels of fluorescence intensity were different among four components, the fluorescence quenching curves were shown as percent changes from initial levels (F/F0 × 100, where F and F0 are the fluorescence intensity with and without titrated metals, respectively).
Fluorescence quenching occurred for all the PARAFAC derived components investigated in this study, indicating the occurrence of electronic structural changes in the WEOM components by forming complexes with Cu(II) and Eu(III). Along with the addition of metal cations, the fluorescence identities of four components firstly decreased sharply in a low range of metal concentration (0-10 μmol/L), and then decrease slowly in the high range of metal concentration (20-80 μmol/L) (Figure 3a,c). However, the quenching effects of specific components with Cu(II) and Eu(III) were different. Table 1 summarizes the conditional binding parameters of PARAFAC derived components with both cations.
The conditional complexation stability constants (logKM) of Cu(II) for WEOM components are close to those reported by other investigators using fluorescence quenching titration (FQT). For instances, the Cu(II) complexation stability constants were reported ranging from logKM of 4.65 to 5.55 for the soil-and compost-borne humic acid [31,32]. The Cu(II) logKM for the fulvic-like component and microbial humic-like component, which derived from surface water-borne DOM, were reported ranging from 4.26 to 4.69, and 4.20 to 6.76, respectively [33]. Previous studies mostly focused on the binding capacity of Eu(III) with specifically extracted fulvic acid and humic acid. The Eu(III) logKM were reported ranging between 4.90 to 5.03 for soil fulvic acid, 5.98 to 6.09 for soil humic acid, and 5.61 to 5.71 for Aldrich humic acid [34]. This study is a step forward in that Eu(III) stability constants with four components of soil WEOM were derived simultaneously (Table 1).  In the case of complexation with the fulvic-, humic-like, and microbial degraded humic-like components, the logKM values for Eu(III) were larger than those for Cu(II). On the contrary, the logKM values for Eu(III) with protein were smaller than that for Cu(II). These results indicate that, within the entire range of metal concentrations in this study, Eu(III) had a stronger binding capability than Cu(II) with the fulvic-, humic-like, and microbial degraded humic-like components, but not with protein-like component. The logKM values for Cu(II) with four components were in the order: protein-like > fulvic-like > microbial degraded humic-like > humic-like component, and for Eu(III) the order was fulvic-like > microbial degraded humic-like > humic-like component > protein-like. Besides, the fraction of the initial fluorescence that corresponds to the binding fluorophores (f) and the stoichiometric concentration of the ligand (CL) values varied between four components interacting with Cu(II) and Eu(III). These results suggested that the binding processes varied with cations and WEOM components.
For Cu(II) and Eu(III), binding occurs mostly with the carboxylic and phenolic groups of the natural organic matter [35]. Luster et al. [36] have investigated the binding of Cu(II) to leaf litter extracted DOM using fluorescence spectrometry and electron spin resonance spectroscopy. The results showed that stable inner-sphere complexes of Cu(II) and the DOM could be assigned mainly to binding sites formed by carboxylic and/or phenolic O ligand atoms, and the weak Cu(II) binding may be caused by out-sphere complexes involving ketonic, phenolic and carboxylic groups. Fourier transform infrared spectra responses on Cu(II) with DOM revealed that carboxylic groups were most likely the dominant binding site involved in complexation [37,38]. The important roles played by carboxylic and phenolic groups in forming Eu(III) complexes of humic and fulvic aicd were confirmed using various spectroscopic method [39,40]. Thus, Cu(II) and Eu(III) may bind with WEOM components through its carboxylic and phenolic groups. Compared with humic acid, fulvic acid contains a larger number of carboxylic sites and relatively few phenolic sites [41]. The various types and amounts of functional groups in fulvic acid and humic acid substances might cause the differences in cation binding behavior between the two components. Extended study is needed to evidence the above hypothesis. It should also be noted that the nonfluorescent substances (i.e., polysaccharides, lipids and lignin) in WEOM can interact with cations. Further study is needed to look into the binding strength and sequencing of functional groups with cations, such as using two-dimensional FTIR correlation spectroscopy [10].
It is well-known that fluorescence of protein-like substances is quenched or enhanced due to interaction with metal cations. However, there were few studies using the FQT to characterize the heavy metal binding capacity of soil protein-like substances. When charactering the interaction between surface water-borne DOM with heavy metals, it was found that the utility of FQT combining synchronous and EEM spectra might be limited, because the fluorescent peaks of protein-like substances in bulk DOM EEM were often overlapped with those of fulvic-and humic-like peaks [33,42,43]. In our study, the FQT combining EEM PARAFAC allowed evaluating the binding potential of Cu(II) and Eu(III) with specific soil protein-like components. The degree of quenching protein-like component decreased rapidly at low metal concentration (0-10 μmol/L), while reached steady soon compared with those of the other three components. It was reported that the fluorescence intensity of surface water-borne protein decreased at the earlier stage of Cu(II) addition, but sharply increased with further increasing of Cu(II) [33]. The exact reason for these findings is not yet clear, but it may be related to the different sensitivity of protein-like substances to different metals and changes in molecular environment of proteins.

Competitive Reactions between Cu(II) and Eu(III) with WEOM
Data from competition experiments between metal cations with WEOM may help understand the behaviors of metal cations in soil systems. At the beginning of the competitive experiment the WEOM was loaded with Eu(III). After stepwise addition of the Cu(II), the quenching effects of four components with Cu(II) (Figure 3b) were all less pronounced than the controls (Figure 3a). Furthermore, the logKM, f, and CL values for Cu(II) with four components all decreased (Table 1). Similarly, the quenching effects and corresponding binding parameters of fulvic-, humic-and microbial degraded humic-like components with Eu(III) reduced after loaded with Cu(II). The quenching curve of protein-like component in Figure 3d was not modeled because of the poor fitness with testing data. Our results suggest that the binding sites within each WEOM component for both cations are probably the same. This is in agreement with Marang et al. [44] who evaluated the Eu(III) binding behavior with humic acid in the presence of competitive Cu(II). Their results demonstrated that the strong Cu(II) competitive binding led to increased concentration of free Eu(III) cation in solution, and the competitive binding sites were mainly carboxylic and phenolic groups. Similarly, Konstantinou et al. [35] showed that after stepwise addition of the Eu(III), Cu(II) adsorbed on the olive cake-derived DOM were partly replaced. The conditional complexation stability constants of the Cu(II)-DOM and Eu(III)-DOM complexes they evaluated were 5.3 and 6.3 at pH 6, respectively.
The relatively strong interaction of Cu(II) and Eu(III) cations to the soil WEOM could have significant implications on the bioavailability and mobility of these and homologue trace metal cations in natural environment, especially when the application of sludge and manure increased amount of soil WEOM. The trace metals pre-bound to soil WEOM could be replaced by competing metals, and then released into soil solution as free metal cations, becoming bioavailable. Additionally, the mobility of these trace metals can be enhanced due to their strong binding to the WEOM. Further experiments and model calculations are still needed to better understand the competition of Cu(II) and Eu(III) cations with functional group-sites of WEOM, and their influences on Cu(II) and Eu(III) behaviors in the soil system.

Soil Sampling and WEOM Extraction
Soils were collected from Nanjing, Jiangsu province, China (31°41ʹ N, 118°52ʹ E). The soil was classified as Ferri-Udic Argosols. Soil samples (0-20 cm) were collected using a 5 cm internal diameter auger from 12 random cores collected in three plots. The fresh soil was mixed thoroughly, air-dried, and sieved through 2 and 0.25 mm screens for further analysis [9]. The soil organic matter (SOM) content of the mixed sample was 32.36 ± 0.78 g/kg (mean ± standard deviation). The WEOM was extracted with deionized water (Milli-Q, Millipore, Billerica, MA, USA) (1:5 w/v soil-to-water ratio) by shaking air-dried soil samples (<2 mm sieved) for 24 h on a horizontal shaker at 25 °C. The WEOM solution was filtered using 0.45 μm polytetrafluoroethylene (PTFE) filters and stored at 4 °C for analysis.

Fluorescence EEM Determination and PARAFAC Modeling
The fluorescence EEMs were measured on a Hitachi F-7000 fluorescence spectrometer (Hitachi High Technologies, Tokyo, Japan) in scan mode. Scanning emission (Em) spectra from 250 to 600 nm were obtained in 1 nm increments by varying the excitation (Ex) wavelength from 200 to 450 nm in 5 nm increments. The spectra were recorded at a scan rate of 2400 nm/min, using excitation and emission slit band widths of 5 nm.
Water Raman scatter peaks were eliminated by subtracting a Milli-Q water blank from the EEMs. The Rayleigh scattering effects were removed [45]. Then, EEMs were calibrated and normalized in quinine sulfate units (QSU), where 1 QSU is the maximum fluorescence intensity of 0.01 mg/L of quinine (qs) in 1 N H2SO4 at the Ex/Em = 350/450.
The PARAFAC analysis was performed using the DOMFluor toolbox (www.models.life.ku.dk) for MATLAB R2011a [46]. We deleted excitation wavelengths from 200 to 235 nm, and emission wavelengths from 250 to 300 nm and from 550 to 600 nm. A total of 48 EEMs were applied for the PARAFAC analysis. The PARAFAC models with two to six components were computed for the EEMs. Determination of the correct number of components was primarily based on split half analysis, residual analysis, and visual inspection [46].

Complexation Modeling
For this study, the modified Ryan-Weber model [31,47] was applied to estimate the binding parameters between the PARAFAC derived components and metals (Cu(II) and Eu(III)). The major assumptions of this model are that 1:1 metal-ligand complexes are formed and that a linear relationship exists between the metal-bound ligand concentration and the quenched WEOM fluorescence intensity. Since PARAFAC can decompose the complex mixture of WEOM fluorophores into independent fluorescence components, the application of this model may provide more appropriate information than the fluorescence intensity from the peak maxima of EEM spectra.
The modified Ryan-Weber model equation is given by: where F, and F0 are the measured fluorescence intensities of a specific fluorophore at the metal concentration CM and at the beginning of the titration (in the absence of titrated metals), respectively. FML is the limiting fluorescence intensity below which the intensity does not change upon the metal addition (i.e., metal-saturated complex). KM and CL are the conditional stability constant and the stoichiometric concentration of the ligand, respectively. Using nonlinear regression analysis to estimate three parameters in Equation (1) could lead to unreasonably small values for CL [36]. The number of fitting parameters of Equation (1) was reduced using the modified Ryan and Weber model [36].
where FML/F0 − 1 and α are the fitting parameters. ORIGIN software (Version 8.0, OriginLab, Northampton, MA, USA) was used to solve for FML, KM, and CL. In addition, the fraction of the initial fluorescence that corresponds to a specific binding fluorophore (f) was determined using Equation (3) [33].

Conclusions
Application of EEM-PARAFAC divided soil WEOM into four components, namely fulvic-, humic-, microbial degraded humic-, and protein-like components. The main benefit of the EEM-PARAFAC to this study was to obtain quantitative information on the binding capacity of heterogeneous WEOM fluorescent components with Cu(II) and Eu(III) at molecular level. Compared with Cu(II), Eu(III) presents a higher affinity for the fulvic-, humic-like, and microbial humic-like components, but not for protein-like component. Cu(II) and Eu(III) may compete for the same binding sites of the WEOM components.