Environ. Sci. Technol. 2001, 35, 3646-3652
Isolation and Partial Characterization of Dissolved Copper-Complexing Ligands in Streamwaters F E N G C H A N G W U * ,† A N D EIICHIRO TANOUE‡ State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences, Guiyang, 550002 People’s Republic of China, and Division of Earth and Environmental Sciences, Graduate School of Environmental Studies, Nagoya University, Nagoya 464-8601, Japan
We have separated two groups of copper-complexing ligands (the weak and strong ligands) from streamwaters in the Lake Biwa watershed by modified immobilized metal ion affinity chromatography (IMAC). The weak ligands were about 0.54-1.21% of the total dissolved organic matter (DOM), as determined by UV absorbance, and the strong ligands were about 0.06-0.21%. The results show that the stronger ligands were retained longer on the IMAC column, eluted later, and were accompanied by shorter wavelength UV absorbers, fluorescence maxima patterns with shorter wavelength excitation, and relatively “fresher” ′ values organic matter. The weak ligands with logKCuL of 6.6-7.7 had predominant humic-like fluorescence and may have been considerably degraded, while stronger ligands ′ values of 8.9-9.3 had only protein-like with logKCuL fluorescence and were relatively newly produced, labile material, as indicated from their amino acid composition. The protein-like fluorescence was mainly due to aromatic tryptophan probably bound to proteins or peptides. The results presented here have significant implications regarding the possible sources and biogeochemical role of organic ligands in aquatic environments.
Introduction Natural organic ligands have been investigated for many years in terms of metal speciation in aquatic environments, and their concentrations and binding strength have been reported and reviewed (1-3). Copper is highly complexed by organic ligands in natural waters; this complexation is reported to control its biogeochemical cycling, bioavailability, and toxicity in these systems (1, 2, 4-6). In oceanic waters, it has been reported that two ligand classes with log K values of 12-13 (L1) and 9-10 (L2) are responsible for copper speciation (2, 7, 8). In freshwaters, there exist at least three ligand classes with log K values ranging from 15 (strong ligand) to 8.6 (weak ligand) (6). In all natural environments, the relatively weak ligand class is suggested to represent Cu binding by humic materials, while the stronger ligand class may be related to * Corresponding author phone: (705)748-1011 ext 1370; fax: (705)748-1569; e-mail:
[email protected]. Present address: Environmental and Resource Studies, Trent University, Peterborough, Ontario K9J 7B8, Canada. † Chinese Academy of Sciences. ‡ Nagoya University. 3646
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 18, 2001
primary productivity (2, 4, 9). The nature and geochemical characteristics of organic ligands, however, remain poorly understood largely because most relevant studies have been carried out without the isolation of organic ligands (10-13). Recently, immobilized metal ion affinity chromatography (IMAC) has been used successfully to isolate coppercomplexing organic ligands from oceanic, coastal, and estuarine waters, and their physical and chemical properties such as conditional stability constants, total primary amines, carbohydrate contents, and molecular mass distributions have been reported (10-14). However, to our knowledge, the chemical characterization of organic ligands in freshwaters has not been reported. In this study, the IMAC approach was improved and was applied to the isolation of organic ligands from streamwaters in the Lake Biwa watershed, Japan. The chemical nature of the organic ligands was characterized by absorbance spectra, three-dimensional excitation emission matrix spectroscopy (3DEEM), fluorescence quenching titration, and amino acid composition.
Materials and Methods Sampling. River samples were collected in the headwater streams of River Kiryu, Daido, Azusa, and Egadani in the Lake Biwa watershed in April and June 1999. Water samples were filtered through GF/F glass-fiber filters (Whatman, Maidstone, U.K.) after sampling and were kept frozen until analysis. The IMAC Isolation Procedure. In the IMAC principle of ligand isolation, a water sample is passed through an agarose gel to which iminodiacetate (IDA) moieties are covalently bonded via a hydrophilic spacer arm. The spacer arm facilitates the interaction of the immobilized metal ions with relatively inaccessible metal ligand sites in large molecules. The chelator IDA ligands on the gel are charged with copper, which binds to the IDA groups, but one of copper’s coordination sites is left free to bind to copper-complexing ligands in the water sample. The organic ligands are specifically immobilized by copper on the gel and can be eluted by competing ligands or decreased pH (14-16). The IMAC isolation in this study initially followed the procedure reported in a previous study (12), which included five steps: column packing and equilibrium, immobilization of copper ion, loading of samples, elution of organic ligands, and column regeneration. Briefly, 20 mL of Chelating Sepharose Fast Flow gel (Pharmacia, Uppsala, Sweden) was used in a column (20 cm × 16 mm i.d., Pharmacia) as the solid matrix for IMAC. The column was charged with 340 µmol copper ions (0.02 M copper sulfate). After copper ions were loaded, the column was rinsed with 0.1 M borate buffer (pH ) 8.2, 0.1 M NaCl). Samples were then loaded at a flow rate of 90 mL‚cm-2‚h-1. The column was subsequently rinsed with the borate buffer. The retained organic ligands on the solid matrix, complexed to copper on the column, were then eluted by a mobile phase at a flow rate of 20 mL‚cm-2‚h-1. In this study, two types of mobile phase, pH ) 2 and 4 HCl solution (both with 0.1 M NaCl), were tested comparatively. Absorption and Fluorescence Spectroscopy. Absorption spectra were obtained between 230 and 500 nm at 0.5 nm intervals using a spectrophotometer (Shimadzu, MPS-2400, UV-vis multipurpose) equipped with matching 2-cm quartz cells. Each sample was scanned three times, and the resulting spectra were smoothed. The spectral slope was calculated from a linear-least-squares regression of the plot of ln(absorbance) vs wavelength for the interval between 230 and 400 nm. The slope was used to determine changes in the 10.1021/es0019023 CCC: $20.00
2001 American Chemical Society Published on Web 08/10/2001
optical properties of the ligand fractions and was compared with 3DEEM fluorescence patterns with excitation wavelength ranging from 230 to 400 nm. Fluorescence was measured with three-dimensional excitation/emission matrix spectroscopy (3DEEM) by a fluorescence spectrophotometer (Hitachi, Model F-4500). The wavelengths ranged from 230 to 400 nm for excitation (5 nm bandwidth) and from 250 to 600 nm for emission (2 nm bandwidth). A system procedural blank was prepared by passing a mobile phase through a blank copper-loaded column; the blank 3DEEM was subtracted to eliminate possible column contamination. Matlab program was used to obtain the surface and contour plots of 3DEEM, in which excitation/emission (Ex/Em) maxima can be identified. Binding Characteristics of Organic Ligands. Fluorescence quenching titration was performed to characterize the binding properties of the moieties associated with the major fluorescence maxima of the organic ligands. This method was developed by Weber and colleagues and was applied to copper binding with humic substances and DOM (17, 18). An underlying assumption of the method is that there are no binding sites that are not associated with a fluorescence signature. Standardized copper(II) nitrate solution was added incrementally to a final concentration of CCu ) around 60 µM in 0.8 mL of ligand fractions under the conditions of an ionic strength of 0.1 M, pH ) 8.15, and 25 °C. For 1:1 stoichiometry for copper(II) complexation with organic ligands, the complexing reactions that fit the quenching titration data can be described by a linear regression program in the following equation (13):
RCu
) C ‚(1-X) + (1-X X ) K′
CCu‚
L
(1)
CuL
where CCu was the total copper concentration; CL was the ligand concentration; X ) (FLinit - FL)/(FLinit - FLend) ) [CuL]/ CL, which was the fraction of the ligand bound to copper (II), CuL, expressed in terms of the measured fluorescence intensity, FL. FLinit, and FLend were the limiting intensities before and after copper(II) titration and corresponded to those when all ligands were completely free and occupied, respectively. K′CuL was a conditional stability constant; RCu was the inorganic side-reaction coefficient for copper(II) and was determined to be RCu ) 11 in the similar condition (19). For 1:1 copper complexes with one ligand, one linear curve may be observed in the plotting of CCu‚(1 - X/X) vs (1 - X). If nonlinearity of the diagram was observed, two 1:1 complexation models of two discrete ligand classes with different stability constants were applied (20). Amino Acid Analysis. Ligand fractions were hydrolyzed for 22 h at 110 °C and 6 N HCl before undergoing HPLC analysis for amino acids except for tryptophan. The ampule was opened, and the hydrolysate was evaporated to dryness in a rotary evaporator at a bath temperature less than 60 °C and was redissolved in Milli-Q water for the HPLC analysis. Since tryptophan is labile in the presence of acid and oxygen commonly used in acid hydrolysis, for tryptophan analysis, ligand fractions were hydrolyzed for 16 h at 110 °C and 4.2 N NaOH in N2 atmosphere, and ascorbic acid was added as an antioxidant (21). The alkaline hydrolysate was neutralized by HCl prior to HPLC analysis. All amino acids were determined by precolumn orthophthaldialdehyde (OPA) fluorescent derivatization, HPLC separation of derivatized amino acids, and fluorescence detection as originally proposed by Lindroth and Mopper (22). Liquid chromatography was carried out using a Waters HPLC system to separate individual amino acids. The analytical precision expressed as standard deviation from multiple standard injections of 25 µL was less than 0.8% for valine, methionine, isoleucine, phenylalanine, and leucine, 1.2-1.9% for serine, histidine,
FIGURE 1. (a) IMAC chromatographs of two mobile phases. The sample was headwater from Kiryu River in June 1999 in the Lake Biwa watershed, Japan. (b) The fluorescence intensity of individual Ex/Em maxima in the ligand fractions as shown in Figure 1a when pH ) 4 mobile solution was used. Peaks A, B, C, and D are Ex/Em maxima measured as 3DEEM in the fractions are shown in Figure 2 and Table 2. and glycine, 2.0-4.7% for glutamic acid, threonine, alanine, arginine, tryptophan, and tyrosine, and 9.9% for aspartic acid.
Results and Discussion Improved IMAC and Isolation of Organic Ligands. In previous studies in oceanic waters, 0.5 M NaCl + pH ) 2 HCl solution was used as a mobile phase, the eluted fractions were directly monitored by UV absorbance, and one major peak was observed in the IMAC chromatograph (12, 13). Since pH may have an effect on UV absorbance and fluorescence properties of organic ligands, as reported in previous studies on organic ligands and DOM (13, 17, 23, 24), all eluted fractions were adjusted to pH ) 8.15 before being monitored for UV absorbance and fluorescence. To test the effect of pH on the elution pattern, we independently used pH ) 4 and pH ) 2 HCl mobile solutions. Figure 1a shows the IMAC chromatographs of these two mobile phases. It is shown that one major peak, which occurred at an elution volume of about 35 mL, was observed in both cases, but another small peak occurring later at the elution volume of about 62 mL was seen only when the pH ) 4 mobile phase was used. It is probably because the organic ligands eluted later could have been destroyed by strong acids in the pH ) 2 situation. Similar results were also reported by Gordon et al. (25). Thus two organic ligand groups (group 1 and 2) corresponding to the two major peaks in the IMAC chromatograph were isolated by choosing the pH ) 4 mobile solution, group 1 occurred at the elution volume of 35 mL, and group 2 occurred at 62 mL. These groups were subjected to further characterization, as will be discussed in later sections. The modified IMAC was applied to all studied headwaters in the Lake Biwa watershed. Two similar organic ligand groups were isolated in all samples. Table 1 shows the absorbance of the two groups and their contribution to the bulk DOM. It is shown that the absorbance values of group 1 and 2 were in the range of 1.21-16.8 × 10-5 cm-1 and 0.07-4.03 × 10-5 cm-1, respectively, in the headwaters investigated, accounting for 0.54-1.21% and 0.06-0.21% of the bulk DOM, respectively. VOL. 35, NO. 18, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3647
TABLE 1. Absorbance of Organic Ligands, and Their Contribution to the Bulk Dissolved Organic Matter (DOM), as Determined by UV Absorbance
TABLE 3. Concentrations and Conditional Stability Constants of Organic Ligand Fractions Calculated by Fluorescence Quenching Titrationd
absorbance at 254 nm (10-5 cm-1)
L1
organic ligandsa samples
bulk DOM
group 1
group 2
group 1/ group 2
Kiryu Daido
1100 1900
April 1999 7.9 (0.72%) 16.8 (0.88%)
1.84 (0.17%) 4.03 (0.21%)
4.3 4.2
Azusa Egadani Kiryu Daido
100 600 1800 1900
June 1999 1.21 (1.21%) 3.51 (0.59%) 9.80 (0.54%) 10.90 (0.57%)
0.07 (0.07%) 0.34 (0.06%) 1.98 (0.11%) 3.27 (0.17%)
17.3 10.3 4.9 3.3
a The data in parentheses represent the contribution of organic ligands to bulk DOM.
TABLE 2. Characteristic Data of Organic Ligand Fractionsd log K′HLb
Ex/Em maximum (nm)a original water 320/444 (peak A) 235/434 (peak B) 260/310 (peak C) fraction 1 fraction 2 fraction 3 fraction 4
fraction 5 fraction 6
-
Group 1 315/424 (peak A) 235/440 (peak B) 265/330 (peak C) 315/390 (peak A) 235/426 (peak B) 265/332 (peak C) 235/434 (peak B) 255/328 (peak C) 235/310 (peak D) 235/430 (peak B) 255/324 (peak C) 235/316 (peak D)
+ 2.40 (peak B) + 3.23 (peak A) 2.51 (peak B) 2.17 (peak C) 2.28 (peak B) 1.99 (peak C) 2.08 (peak D) -
Group 2 250/350 (peak C) 250/340 (peak C)
2.06 (peak C) 2.05 (peak C)
spectral slopec 0.0163
0.0164 0.0177 0.0258 -
0.023 0.024
a The data represent the Ex/Em fluorescence maxima, as shown in Figure 2. b logK′HLwas calculated independently based on the fluorescence - pH relationship, and the data in parentheses indicate the fluorescence Ex/Em maxima shown in Figure 2, which were used to estimate the logK′HL. c Spectral slope was calculated based on 230-400 nm spectrum. + data cannot be accurately estimated. - data not available. d The organic ligand fractions (group 1 and group 2) are the same as shown in Figure 1a.
Optical and Fluorescence Characteristics of Organic Ligands. In the following chemical and optical characterization of organic ligands, we chose the ligand fractions (fractions 1, 2, 3, 4, 5, and 6) isolated from Kiryu River shown in Figure 1a as a case study. Our approach was to get characteristic information about the copper-complexing ligands by comparing differences between the IMAC-isolated ligand fractions and original water and between various fractions themselves. Some optical and fluorescence characteristics of the ligand fractions and original water are listed in Table 2. The spectral slopes of the ligand fractions were higher than that of the original water (Table 2), indicating that the IMAC-isolated ligands were strong UV absorbers. In group 1, the spectral slope increased from 0.0164 in fraction 1 to 0.0258 in fraction 3, suggesting an increase in absorbance at the shorter wavelength of the spectrum. The spectral slope was 0.023-0.024 in the group 2 fractions eluted last. Figure 2 shows 3DEEM patterns of the IMAC-isolated ligand fractions and original water. Four major Ex/Em 3648
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 18, 2001
fraction 2a fraction 4b fraction 5c
L2
CL1 (µM)
log K′CuL1
CL2 (µM)
log K′CuL2
12.03 4.08 3.45
7.71 8.85 9.25
22.73 4.33 10.36
6.81 6.64 6.69
a-c The data were calculated based on fluorescence peak A at Ex/Em 315/390 nm, peak C at Ex/Em 255/324 nm, and peak C at Ex/Em 250/350 nm, respectively, as shown in Figure 2. d The organic ligand fractions are the same as shown in Figure 1a.
maxima can be discerned in the ligand fractions (Figures 2 and 3); peak A resulted from excitation at around 310-320 nm and emission in the visible region, peak B from excitation in the UV region at around 230-250 nm and emission in the visible region, and peaks C and D from both excitation and emission in the UV region. The patterns of fluorescence maxima and intensities changed with the fractions eluted, as shown clearly in Figures 1b, 2, and 3. There were three major Ex/Em peaks in the original water, peak A at Ex/Em 320/444 nm, peak B at 235/434 nm, and peak C at 260/310 nm. In the ligand fractions 1 and 2, the fluorescence patterns were similar to those of the original water: peak A at Ex/Em 315/390-424 nm, peak B at 235/426-440 nm, and peak C at 265/330-332 nm. In fractions 3 and 4, peak A disappeared; however, there were still three remaining fluorescence maxima: peak B at Ex/Em 235/430-434 nm, peak C at 255/ 324-328 nm, and peak D at 235/310-316 nm. It seems that the fluorescence patterns shifted to the shorter excitation wavelength as ligands were constantly eluted. This pattern is in agreement with the increase in the spectral slope mentioned earlier, indicating the similarity between absorbing and fluorescing components of the organic ligands. It is interesting to note that there was only one fluorescence peak at Ex/Em 250/340-350 nm in fractions 5 and 6 eluted last. The different fluorescence patterns observed between the ligand fractions are considered to reflect their different chemical properties. In terms of Ex/Em maxima, peaks A and B in the ligand fractions are similar to humic-like fluorescence in previous studies on DOM (26-29). While peaks C and D are similar to the protein-like fluorescence in previous reports (28-31), they are also within the Ex/Em maxima ranges of tryptophan and tyrosine standard (Figures 2 and 3). The binding properties of the ligand fractions can be calculated based on their copper quenching titration and eq 1. The results of the calculated CL and logK′CuL1 of major fluorescence maxima in the ligand fractions are listed in Table 3. Two ligand classes were detected in all fractions. The logK′CuL of the weaker ligand class (L2) ranged from 6.64 to 6.81 in all ligand fractions tested, while the logK′CuL2 of the stronger ligand class ranged from 7.71 to 9.25 and increased with the fractions eluted later (Table 3). This result is consistent with previous reports that the stronger ligands are retained longer on the IMAC column and eluted later (10, 12). Thus group 1 and 2 ligand fractions (Figure 1a and Table 1) can be referred to as the weak and strong ligand fractions, respectively. The ligand classes (L1 and L2) in this study are comparable to those in similar studies in oceanic waters (12, 13). Compared to previous reports in both freshwater (6) and marine waters (2, 7, 8), where the bulk of this kind of work has been done, the strong ligand class in this study is like the weak ligand. Humic substances were reported to be one of the most important classes of ligands for copper in natural waters (2, 6, 32). The fluorescence properties in group 1
FIGURE 2. 3DEEM fluorescence patterns of organic ligand fractions. The ligand fractions are the same as shown in Figure 1a. Fluorescence units are arbitrary. fractions provide evidence that supports the contribution of humic substances to copper complexation. Most significantly,
our results imply that proteins or peptides, as indicated from the protein-like fluorescence, particularly in group 2 fractions, VOL. 35, NO. 18, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3649
FIGURE 3. Fluorescence Ex/Em maxima distribution of organic ligand fractions and original water. The fractions are the same as shown in Figure 1a. The 3DEE plots are shown in Figure 2. may also play a major role in stronger complexation with copper. Since previous studies have shown that organic ligands (logK′CuL ) 7-9.5) play a significant role in trace metal speciation, the mobilization and transport of nutrients and pollutants (As, Zn, Cd, Pb, and Ce), and soil weathering in terrestrial ecosystems (33-36), the IMAC-isolated ligands should represent significant organic ligands that biogeochemically occur in natural terrestrial ecosystems. Previous studies have reported that the humic-like fluorescence similar to peak A in this study is mainly associated with the presence of various functional groups (carboxyl, carbonyl, hydroxy groups) (26, 37, 38). There are only a few reports on fluorescence similar to peak B in freshwaters, and consequently the responsible chemical structure is unknown. To further characterize the functional groups responsible for all fluorescence peaks, the relationship between fluorescence intensity and pH was investigated. Humic substances derived from freshwaters and soils reportedly show a decreasing intensity of fluorescence emission with decreasing pH in the absence of metal ions (13, 23, 26). The gradual decrease of fluorescence intensity with decreasing pH in the low pH region can be used to calculate the protonation constant of possible functional groups by log[(FLH-pH - FL)/(FL - FLL-pH)] ) logK′HL - pH (13), where FLH-pH is the fluorescence intensity at the highest pH, and FLL-pH is the fluorescence intensity at the lowest pH in the low pH region. The fluorescence intensities of peaks A, B, C, and D in the ligand fractions decreased with decreasing pH in the region below pH ) 5 (Figure 4), indicating that acidic functional groups may play a role. The calculated logK′HL values of individual fluorescence peaks are listed in Table 2. The logK′HL values of peaks A and B, ranging 2.28-3.23, were close to logK′H2L (2.5-4.3 except for maleic acid) of dicarboxylates of authentic materials (3941). The existing literature on authentic reagents (39-41) suggests that the protonation constant can decrease to as low as 2.2 if molecules are combined with a few carboxyl moieties. Therefore it is possible that both peaks A and B were associated with carboxyl groups, and peak B may be associated with the presence of more carboxyl moieties since the logK′HL values of peak B were slightly lower than those of peak A. The literature survey also indicates that the logK′H2L can reach below 2 when other donor groups, such as carbonyl (e.g. pyruvate, logK′H2L ) 2.2), amino, or imino moieties are bound in the vicinity of a carboxyl moiety in a molecular structure. For instance, ethylenediaminetetraacetic acid and nitriloacetic acid have a logK′HL value of 2.0, and values of amino acids are in the range of 2.4 (glutamic acid etc) to 1.9 (histidine). Our observations that logK′HL values of peaks C and D fluorescence (1.99-2.1) were much lower than those of peaks A and B may be due to the involvement of other donor groups such as amino acids. The aromatic 3650
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 18, 2001
FIGURE 4. Relationship between fluorescence intensities of Ex/Em maxima and pH in the ligand fractions. The ligand fractions are the same as shown in Figure 1a.
FIGURE 5. Relative distribution of amino acids in ligand fractions. The ligand fractions are the same as shown in Figure 1a. Asp: aspartic acid; Glu: glutamic acid; Ser: serine; His: histidine; Gly: glycine; Thr: threonine; Ala: alanine; Arg: arginine; Tyr: tyrosine; Val: valine; Meth: methionine; Iso: isoleucine; Phe: phenylalanine; Leu: leucine; Trp: tryptophan. amino acids may be the most likely donor groups, as will be discussed in later sections. Little information is available about the relationships between physical, chemical structure, and binding properties of organic ligands in natural environments. IMAC provides a unique approach to examine these relationships. It is interesting to note that the stronger organic ligands for copper were related to shorter wavelength absorbers, fluorescence maxima with shorter Ex wavelength, and lower protonation constants (Tables 2 and 3), reflecting the simultaneous changes in physical and chemical structures of organic ligands. Structural and compositional changes that result in shorter-wavelength excitation maxima may include an increase in the number of aromatic rings, of conjugated bonds in the chain structure (26). These changes can be also responsible for the increasing affinity of the organic ligands. Chemical Nature of Organic Ligands for Copper. The molar percent abundance of amino acids in the total hydrolyzed amino acids (THAA) in the ligand fractions was analyzed to identify differences in composition among them (Figures 5 and 6, Table 4). The most abundant amino acids in the original water were aspartic acid, arginine, glycine, glutamic acid, and alanine, accounting for 66%, of the THAA. The relative contributions of amino acids in fractions 1 and
FIGURE 6. Relative distribution of amino acid groups in ligand fractions. The ligand fractions are the same as shown in Figure 1a.
TABLE 4. Molar Percent Ratios of Amino Acids in the Total Hydrolyzable Amino Acids (THAA) in Ligand Fractionsa molar %
original water
frac 1
group 1 frac 2 frac 3
frac 4
group 2 frac 5 frac 6
Asp Glu Ser His Gly Thr Ala Arg Tyr Val Meth Iso Phe Leu Trp
14.68 12.37 6.50 2.04 13.31 10.79 11.49 14.52 0.75 5.21 1.61 3.20 1.08 1.91 0.54
21.50 0.73 0.10 4.99 7.84 5.19 19.88 26.52 0.04 4.36 0.06 2.90 2.49 3.28 0.12
21.96 0.74 0.22 4.80 7.32 5.45 21.60 25.62 0.20 3.66 0.18 2.56 2.38 3.15 0.16
20.29 28.20 0.72 2.04 4.31 1.40 10.83 7.16 0.80 6.42 0.26 5.16 3.23 7.45 1.74
14.43 17.47 9.75 5.05 8.38 1.46 9.75 4.15 0.56 7.98 0.03 3.20 9.72 5.25 2.82
13.12 19.00 10.06 1.92 11.85 9.27 8.28 7.41 0.27 4.24 1.97 3.51 4.21 4.23 0.66
15.44 21.25 7.86 3.60 7.25 2.54 11.05 2.62 1.48 7.54 0.55 3.51 5.66 7.82 1.84
a The organic ligand fractions (group 1 and group 2) are the same as shown in Figure 1a. The notation of amino acid species is the same as those in Figure 5.
2 were similar to that of the original water, the predominant amino acids were arginine, aspartic acid, alanine, and glycine, accounting for about 76% of the total. While in fractions 3-6 with protein-like fluorescence, the most abundant amino acids were glutamic acid, aspartic acid, alanine, serine, or leucine, accounting for 54-67% of the total. Differences in amino acid groups between the fractions were also evident. Basic amino acids were more abundant in fractions 1 and 2, while acidic and aromatic species were more abundant in fractions 3-6. Neutral species, however, were similar in all fractions (Figure 6). In terms of individual amino acids, all individual aromatic species increased from fractions 1-4 to fractions 5 and 6, while alanine and arginine decreased from fractions 1- 4 to fractions 5 and 6 (Figure 5). Previous studies on amino acid biogeochemistry have demonstrated that amino acid composition is a sensitive indicator for the degradation of organic matter in sediments and particulates, and it has been reported that acidic species and aromatic species tyrosine and phenylalanine decrease, while basic species increase during organic matter degradation (4246). Similarly, our results indicate that ligand fractions with predominant protein-like fluorescence such as fractions 3-6 were relatively “fresh” and susceptible to degradation, while the weak ligand fractions (fractions 1 and 2) with humic-like fluorescence were “old” and may have been degraded considerably. This conclusion is consistent with their fluorescence properties because protein-like fluorescence was reported to closely relate to biological productivity in previous studies in aquatic environments, and humic-like fluorescence was reported to be refractory humic substances (21, 27-29). The protein-like fluorescence similar to tryptophan or tyrosine fluorescence has been extensively studied in both freshwater and oceanic waters (21, 28-31). Since only three aromatic amino acids (tryptophan, tyrosine, phenylalanine)
FIGURE 7. Relationship between the protein-like fluorescence intensities (Peak C) and concentrations of aromatic amino acids (tyrosine, phenylalanine and tryptophan) in the ligand fractions. Peaks C in the ligand fractions are shown in Figure 2. were reported to fluoresce (47), protein-like fluorescence in aquatic environments is usually considered to be due to those aromatic amino acids. However, confirmation has never been carried out. It is still unclear whether the protein-like fluorescence really comes from the aromatic amino acids, and the relationship between the protein-like fluorescence and aromatic amino acid concentrations is unknown. Our results show that aromatic amino acids in the strong ligand fractions with predominant protein-like fluorescence were abundant relative to those in the weak fractions with humiclike fluorescence (Figure 6). Phenylalanine had the highest concentrations among the three aromatic amino acids, and tyrosine had the lowest (Figure 7). The protein-like fluorescence intensity was found to correlate well with tryptophan concentration (R 2 ) 0.85, n ) 6) but not with phenylalanine and tyrosine (Figure 7). Since the concentration of free dissolved tryptophan was under the detection limit in the ligand fractions, most of the tryptophan in the fractions was bound instead of freely dissolved. This strongly suggests that the protein-like fluorescence in the ligand fractions was related to tryptophan likely bound to proteins or peptides. This study provides the direct evidence, for the first time, that the protein-like fluorescence in DOM is linked to the presence of tryptophan. Our observations are also consistent with the low fluorescence efficiency of phenylalanine and agree with findings of Wolfbeis (47) that tryptophan fluoresces, when bound to protein fragments, several times more strongly than tyrosine. At present we have little information about the source of the organic ligands. It is interesting to note the consistency between fluorescence pattern and amino acid compositions in the ligand fractions and that the strong organic ligands for copper(II) were linked to “fresh” organic matter, proteinlike fluorescence, and tryptophan. It has been reported that exudates from certain phytoplanktons and bacteria are two strong sources for protein-like fluorescence and are also strong copper chelators (25, 48, 49). Therefore, the IMACisolated strong ligands with predominant protein-like fluorescence suggest a biogenic source in streamwater rather than refractory. Since our samples were the headwaters, it may indicate that biological activities in the watershed were the possible sources of the strong IMAC ligands in the streamwater. The ratios of the weak to strong ligand fractions (ratios of group 1 to group 2, Table 1) range from 3.3 to 17.3, indicating that the majority of IMAC-isolated ligands were weak ligands and were probably linked to humic substances, which were more refractory and degraded organic matter in the aquatic environment. Most likely, the source of dissolved organic ligands was same, and the changes in various fractions occurred as a result of changes in molecular degradation. Tryptophan, an essential amino acid (50), has an important nutritional value to aquatic organisms due to its relatively VOL. 35, NO. 18, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3651
long aliphatic chain length among the essential amino acids. It may not be synthesized from many other amino acids in heterotrophic metabolism. This study suggests that the protein-like fluorescence in DOM may be directly due to aromatic tryptophan bound to proteins or peptides. Therefore fluorescence provides a method for quick and direct analysis of tryptophan and its bound proteins, with much greater ease of use than more rigorous chemical techniques. This study reconfirms that fluorescence may provide a means to investigate the biogeochemical cycling of some important components (tryptophan and bound proteins or peptides), their biological activities, and their poorly understood role in binding trace metals.
Acknowledgments We wish to thank Drs. Yoshioka, Konohira, Nishida, Ohta, Saino, and Suzuki (Nagoya University) and Hayakawa and Takahashi (Lake Biwa Research Institute) for discussion and help in the lab and field trip. This research was jointly supported by the grant-in-aid for Scientific Research for IGBP, grant-in-aid (11304039, 12800016, and 13878097) from the Ministry of Education, Science and Culture, Japan, and grantin-aid for JSPS fellowship (12098005), Chinese Academy of Sciences, and Lake Biwa Research Institute.
Literature Cited (1) Tanoue, E.; Midorikawa, T. In Biogeochemical Processes and Ocean Flux in the Western Pacific; Sakai, H., Nazaki, Y., Eds.; Terra Scientific Publishing: Tokyo, 1995; p 201. (2) Donat, J. R.; Bruland, K. W. In Trace Elements in Natural Waters; Salbu, B., Steinnes, E., Eds.; CRC Press: Boca Raton, FL, 1995; p 302. (3) Town, R. M.; Filella, M. Aquat. Sci. 2000, 62, 252. (4) Xue, H. B.; Oestreich, A.; Kistler, D.; Sigg, L. Aquat. Sci. 1996, 58, 69. (5) Breault, R.; Colman, J.; Akien, G.; McKnight, D. Environ. Sci. Technol. 1996, 30, 3477. (6) Xue, H. B.; Sunda, W. G. Environ. Sci. Technol. 1997, 31, 1902. (7) Coale, K. H.; Bruland, K. W. Limnol. Oceanogr. 1988, 33, 1084. (8) Sunda, W. G.; Huntsman, S. A. Mar. Chem. 1991, 38, 69. (9) Xue, H. B.; Sigg, L. Limnol. Oceanogr. 1993, 38, 1200. (10) Gordon, A. S. Mar. Chem. 1992, 38, 1. (11) Donat, J.; Kango, R. A.; Gordon, A. S. Mar. Chem. 1997, 57, 1. (12) Midorikawa, T.; Tanoue, E. Mar. Chem. 1996, 52, 157. (13) Midorikawa, T.; Tanoue. E. Mar. Chem. 1998, 62, 219. (14) Gordon, A. S.; Dyer, B. J.; Kango, R. A.; Donat, J. R. Mar. Chem. 1996, 53, 163. (15) Porath, J.; Carlsson, J.; Olsson, I.; Belfrage, G. Nature 1975, 258, 598. (16) Andersson, L. ISI Atlas Biochem. 1988, 1, 318. (17) Saar, R. A.; Weber, J. H. Anal. Chem. 1980, 52, 2095. (18) Ryan, D. K.; Weber, J. H. Environ. Sci. Technol. 1982, 16, 866.
3652
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 18, 2001
(19) Midorikawa, T.; Tanoue, E.; Sugimura, Y. Anal. Chem. 1990, 62, 1737. (20) van den Berg, C. M. G. Mar. Chem. 1984, 15, 1. (21) Determann, S.; Lobbes, J. M.; Reuter, R.; Rullkotter, J. Mar. Chem. 1998, 62, 137. (22) Lindroth, P.; Mopper, K. Anal. Chem. 1979, 51, 1667. (23) Hall, K. J.; Lee, F. G. Water Res. 1974, 8, 239. (24) Black, A. P.; Christman, R. F. J. Am. Water Works Assoc. 1963, 55, 753. (25) Gordon, A. S.; Donat, J. R.; Kango, R. A.; Dyer, B. J.; Stuart, L. M. Mar. Chem. 2000, 70, 149. (26) Senesi, N. Anal. Chem. Acta 1990, 232, 77. (27) Mopper, K.; Schultz, C. A. Mar. Chem. 1993, 41, 229. (28) Coble, P. G. Mar. Chem. 1996, 51, 325. (29) Determann, S.; Reuter, R.; Willkomm, R. Deep-Sea Res. 1996, 43, 345. (30) Traganza, E. D. Bull. Mar. Sci. 1969, 19, 897. (31) Mayer, L. M.; Schick, L. L.; Loder III, T. C. Mar. Chem. 1999, 64, 171. (32) Petersson, C.; Bishop, K.; Lee, Y.; Lee, B. Water, Air, Soil Pollut. 1995, 80, 971. (33) Mantoura, R. F. C.; Dickson, A.; Kiley, J. R. Estuar. Coast. Mar. Sci. 1978, 6, 387. (34) Tegen, I.; Dorr, H. Water, Air, Soil Pollut. 1996, 88, 133. (35) Li, Z.; Shuman, L. M. Environ. Pollut. 1997, 95, 219. (36) Kalbitz, K.; Wennrich, R. Sci. Total Environ. 1998, 290, 27. (37) Schnitzer, M.; Khan, S. U. Humic Substances in the Environment; Marcel Dekker: New York, 1972; pp 327. (38) Larson, R. A.; Rockwell, A. L. Arch. Hydrobiol. 1980, 89, 416. (39) Martell, A. E.; Smith, R. M. Critical Stability Constants, Vol. 1, Amino Acids; Plenum: New York, 1974; p 469. (40) Martell, A. E.; Smith, R. M. Critical Stability Constants, Vol. 3, Other Organic Ligands; Plenum: New York, 1977; p 495. (41) Smith, R. M.; Martell, A. E. Critical Stability Constants, Vol. 2, Amines; Plenum: New York, 1975. (42) Steinberg, S. M.; Venkatesan, M. I.; Kaplan, R. Mar. Chem. 1987, 21, 249. (43) Burdige, D. J.; Martens, C. S. Geochim. Cosmochim. Acta 1988, 52, 1571. (44) Gonzalez, J. M.; Grimalt, J.; Albaiges, J. Mar. Chem. 1983, 14, 61. (45) Colombo, J. C.; Silverberg, N.; Gearing, J. N. Org. Geochem. 1998, 29, 933. (46) Dauwe, B.; Middelburg, J. J. Limnol. Oceanogr. 1998, 43, 783. (47) Wolfbeis, O. S. In Molecular Luminescence Spectroscopy, Methods and Applications; Schulman, S. G., Ed.; Wiley: Terscience, New York, 1985; Vol. 77, Part I, p 167. (48) McKnight, D. M.; Morel, F. M. M. Limnol. Oceanogr. 1979, 24, 823. (49) Moffett, J. W.; Zika, R. G.; Brand, L. E. Deep-Sea Res. 1990, 37, 27. (50) Fong, W.; Mann, K. H. Can. J. Fish. Aquat. Sci. 1980, 37, 88.
Received for review November 27, 2000. Revised manuscript received June 12, 2001. Accepted June 18, 2001. ES0019023
