Environ. Sci. Technol. 2003, 37, 728-734
Impact of Metal Sorption and Internalization on Nitrification Inhibition Z H I Q I A N G H U , † K A R T I K C H A N D R A N , †,‡ DOMENICO GRASSO,§ AND B A R T H F . S M E T S * ,† Environmental Engineering Program, University of Connecticut, Storrs, Connecticut 06269-2037, and Picker Engineering Program, Smith College, Northampton, Massachusetts 01063
The goal of this study was to explore the relationship between metal extracellular sorption, intracellular accumulation, and nitrification inhibition. Metal sorption on nitrifying biomass was rapid and could be described by linear partitioning with partition coefficients (Kp) of 20.3 ( 0.1, 0.4 ( 0.0, 0.1 ( 0.0, and 0.2 ( 0.0 L/g biomass chemical oxygen demand for Cu, Zn, Ni, and Cd, respectively. On the other hand, intracellular Zn, Ni, and Cd concentrations continued to increase with time beyond 12 h after metal addition, whereas intracellular Cu attained equilibrium after 4 h. Metal internalization kinetics could be described by an intraparticle diffusion model, with characteristic diffusion time constants (td) of 9.4, 64.6, 80.5, and 66.1 h for Cu, Zn, Ni, and Cd, respectively. Ultimate internalized percentages of the total cell-associated metal were 1.4 ( 0.0, 4.3 ( 0.5, 7.6 ( 1.0, and 2.7 ( 0.2% for Cu, Zn, Ni, and Cd, respectively. Nitrification inhibition was not a function of the sorbed metal fraction but correlated well with intracellular Zn, Ni, or Cd fractions. An intraparticle diffusion model coupled with a saturation-type biological toxicity model fit the inhibition data for varying initial Cd concentrations and exposure periods. In contrast, no relationship between intracellular or sorbed Cu concentrations and nitrification inhibition was observed. In the presence of 1 mM Cu, less than 13.3 ( 10.5% cells remained viable as compared to 72.8 ( 7.5, 104.8 ( 1.7, and 84.7 ( 7.0% (assumed 100% viable cells in metal-free control) in the presence of 1 mM Zn, Ni, and Cd, respectively. Hence, the observations that inhibition by metals such as Zn, Ni, and Cd is related to their intracellular fraction and the slow kinetics of metal internalization indicate that metal inhibition can easily be underpredicted from short-term batch assays. Furthermore, the inhibitory mechanism of Cu was very different from Zn, Ni, and Cd and may involve rapid loss of membrane integrity.
Introduction At certain concentrations, metals become toxic to most microorganisms and can have a significant effect on microbial * Corresponding author phone: (860)486-2270; e-mail:
[email protected]. † University of Connecticut. ‡ Present address: Metcalf & Eddy, 60 East 42 St., New York, NY 10165. § Smith College. 728
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 4, 2003
activity in wastewater treatment systems. It is thus hardly surprising that shock loads of metals often result in reduced process efficiency or even failure (1, 2). Nitrification, involving the sequential conversion of ammonium to nitrite and nitrate primarily by nitrifying autotrophic bacteria, is a critical step in biological nitrogen removal processes. Because of their slow intrinsic growth rates and their sensitivity to a number of environmental conditions such as pH, dissolved oxygen concentration, and temperature (3-6), nitrifying bacteria are susceptible to inhibition. Consequently, nitrification inhibition by metals has attracted considerable attention (7-9). Metal partitioning (uptake) including extracellular sorption, transmembrane transport, and intracellular accumulation occurs in many microorganisms (10, 11). Transmembrane transport may be passive, active, or both depending on the metal concentrations and viability of the biomass (12, 13). According to some, metal uptake is primarily a physicochemical rather than a biological transport process (14). Thus, active metabolism may be relatively unimportant in the uptake process and uptake not dependent on cell viability and growth (15). Sorption plays an important role in metal uptake (2, 16, 17) because extracellular polymers and bacterial cell surfaces contain a variety of binding sites including amino, carboxylic, hydroxyl, and phosphate functional groups because of their (phospho)lipid, protein, and polysaccharide moieties (18-20). Transport of metal across the bacterial cell membrane has been studied extensively (21). It is believed that metals first diffuse across the outer wall through porins and then transport in different ways across the cytoplasmic membrane (13, 22). When starvation for essential metals (e.g., Mn2+, Zn2+, Ni2+, Fe2+) occurs, specific metal transport can be induced via various transporter families such as those driven by ATP (13, 21). However, when metals are present in excess, they can accumulate by fast and relatively nonspecific metal transport systems, which are constitutively expressed (13, 21). Because of structural similarities, nonessential metals may accidentally enter the cell through essential metal transport systems (e.g., Cd may enter with Mg through a nonspecific metal transport system in Alcaligenes eutrophus) (13). Once inside the cell, metals can interact with functional groups and degrade protein structure and function (13). Nonessential metals may also displace essential metals from their metabolic sites (e.g., Cd2+ with Zn2+ or Ca2+; Ni2+ with Fe2+; Zn2+ with Mg2+) and inhibit the function of various physiological cations (13, 23). Furthermore, redox-active metals, such as copper, can catalyze the production of hydroxyl radicals and promote stress through redox-cycling activity, resulting in impairment of membrane function (24, 25). To date, the relationship between metal partitioning and microbial toxicity has rarely been addressed (2). Quite often, metal partitioning or uptake (14-16, 26, 27) and metal toxicity to microorganisms (1, 28-31) have been examined separately. In addition, few studies have been devoted to the process of cellular metal internalization (27, 32, 33). Except for a few studies that indicate the high sensitivity of nitrifying activity to metals (34, 35), very little is known about metal interaction with nitrifying bacteria. A better understanding of the mechanism by which metal interacts with nitrifying bacteria appears important and useful for operation of wastewater treatment plants as it may help devise remedial approaches to prevent failure due to shock loads of these substances. The objectives of this research were therefore to (i) evaluate the relationship between metal partitioning and 10.1021/es025977d CCC: $25.00
2003 American Chemical Society Published on Web 01/14/2003
TABLE 1. Comparison of Different Extracting Solutionsa To Determine Intracellular and Extracellular Copper Concentrationsb 1 2 3 4 5 6 7 8 9 10
totalc solubled extracellular 1e extracellular2 f extracellular 3g intracellularh intracellular backgroundi truly internalizedj sum of 2-6 recovery (%)k
1 mM EDTA
50 mM EDTA
50 mM DTPA
0.0237 ( 0.0023 0.9068 ( 0.0387 0.1392 ( 0.0110 0.0092 ( 0.0011 0.0183 ( 0.0003 0.0111 ( 0.0003 0.0072 ( 0.0006 1.0969 ( 0.0269 98 ( 2
1.1194 ( 0.0020 0.0333 ( 0.0050 0.9068 ( 0.0434 0.0287 ( 0.0010 0.0024 ( 0.0003 0.0107 ( 0.0010 0.0067 ( 0.0000 0.0039 ( 0.0010 0.9821 ( 0.0482 88 ( 4
0.0298 ( 0.0006 0.9862 ( 0.0094 0.0337 ( 0.0033 0.0003 ( 0.0007 0.0118 ( 0.0001 0.0077 ( 0.0000 0.0041 ( 0.0001 1.0618 ( 0.0126 95 ( 1
a All solutions contain 0.1 M NaCl. The nitrifying biomass (1062 ( 7 mg L-1 COD) was pulse-spiked with approximately 1 mM Cu. b In mM Cu g-1 XCOD biomass ( 1 SD, n ) 2. c Both added and background Cu were included. d Residual Cu in supernatant after Cu addition and centrifugation. e Supernatant Cu after first-time washing with different solutions. f Supernatant Cu after second-time washing. g Supernatant Cu after third-time washing. h Cu in digested cell pellet after three times washing. i Cu in digested cell pellet that was not spiked with Cu after three times washing. j Difference in cell pellet Cu with and without Cu spike. k Sum of all fractions (row 9)/total metal × 100 (row 1).
metal inhibition for a mixed nitrifying consortium; (ii) determine metal internalization kinetics after transient exposure of metals; and (iii) develop a mathematical model to describe nitrification inhibition that captures both metal transport and biological toxicity effects. Four test metals (copper, zinc, nickel, and cadmium) were selected for this work because of their widespread industrial use and documented toxicity to microorganisms.
Materials and Methods Nitrifying Bioreactor and Reagents. Nitrifying biomass was cultivated in two continuously stirred tank reactors (10 L each) operated at solids retention time (SRT) of 20 d and hydraulic retention time (HRT) of 1 d. Reactors were fed an inorganic medium devoid of organic carbon, with ammonium (300 mg of N/L, (NH4)2SO4) as the sole energy source with requisite macro- and micronutrients described elsewhere (35). Sodium carbonate (1 M) was automatically added to maintain reactor pH at 7.4 ( 0.1 and fulfilled both carbon and alkalinity requirements. Filtered laboratory air was provided to ensure adequate mixing and aeration. Reactor performance was monitored via reactor and effluent chemical oxygen demand (COD), effluent NH4+-N, NO2--N, and NO3-N concentration measurements. Upon attainment of steady state (experiments run for more than 60 d (3 SRTs) after reactor startup with effluent NH4+-N and NO2--N concentrations less than 0.1 mg/L, NO3--N concentration 300 ( 30 mg/L), mixed liquor was withdrawn from the reactor for use in batch studies. Inorganic compounds (CuCl2‚2H2O, ZnSO4‚7H2O, NiSO4‚ 6H2O, and CdCl2‚2H2O, all from Fisher Scientific, Fair Lawn, NJ, certified ACS grade) were used to prepare stock solutions for metal partitioning and inhibition studies. The stock solutions and their dilutions were prepared with Milli-Q purified water. Metal Sorption and Internalization Kinetics. Aliquots of biomass were removed from the continuous-flow reactor, supplemented with 20 mM MOPS (3-(N-morpholino)propanesulfonic acid) as a buffer, pH adjusted to 7.0 ( 0.05, and spiked with individual metals to yield initial analytical concentrations of approximately1 mM. MOPS (20 mM) had no inhibitory effect on nitrifying activity as reported earlier (35). For internalization experiments with Cd, a range of initial analytical concentrations (0.2, 0.35, 0.5, and 1mM) was tested. Cell suspensions were kept at room temperature (25 °C), continuously mixed, and aerated by magnetic stirring (100 rpm). At predetermined intervals (0.5, 1, 2, 4, 6, 8, 12, and 24 h), aliquots were withdrawn, and soluble and intracellular metal concentrations were determined. The amount of biomass-partitioned metal was calculated from the difference
between the initial and the measured soluble metal concentrations after correction for the background values. Determination of Intracellular Metal Concentrations. Intracellular metal concentrations were measured with a modified EDTA washing procedure (27, 33, 36, 37). Briefly, suspensions of the nitrifying enrichment culture with and without a prior 1mM dose of metal were centrifuged at 1600g for 5 min. Microbial cell pellets retained after centrifugation were resuspended in 30 mL of washing solution (1 mM EDTA, pH 7.0, and 0.1 M NaCl to prevent osmotic shock) and agitated at 150 rpm for 30 min to remove surface-bound metal, followed by further centrifugation twice for 5 min at 1600g. Between each centrifugation, supernatant was removed, and pellets were resuspended in washing solution for 10 min. Washing procedures that employed 50 mM EDTA and 50 mM DTPA (all containing 0.1 M NaCl) were also tested, and results were compared with the employed 1 mM EDTA for metal recovery. Concentrated nitric acid (Fisher Scientific, trace metal grade) was added to the pellets, and the contents were quantitatively transferred to a glass reaction tube. The suspension containing 4 M nitric acid was digested at 100 °C for 24 h. The cooled digest was filtered (0.45 µm), and the metal concentration in the filtrate was measured by atomic absorption (AA) spectrometry. EDTA up to 3 mM is efficient in removing sorbed metal fractions permitting a measure of the intracellular metal concentrations (32, 33, 38, 39). Because Cu is the strongest biomass-sorbed metal among the metals we tested (14, 15, 40), the optimum method determined for Cu was used for determination of all intracellular metal concentrations. The intracellular Cu concentration inferred from 1 mM EDTA extraction was slightly higher than from 50 mM EDTA or DTPA extraction (t-test, p < 0.05) (Table 1), which could be due to a less efficient removal of surface bound metals at 1mM concentration. However, the 1 mM EDTA washing procedure had the highest recovery of all conditions tested (98% vs 88% for 50 mM EDTA or 95% for 50 mM DTPA). EDTA concentrations higher than 1 mM were previously found to inhibit ammonium oxidation (35), which may be an indication of membrane damage resulting in release of intracellular metals. Indeed, at 2 mM EDTA, less than 87.5 ( 11.1% cells were found viable (41). Consequently, the 1mM EDTA solution was selected for further experiments. Metal Partitioning. MOPS was added to a final concentration of 20 mM in the nitrifying biomass samples, and the pH was adjusted to 7.0. Aliquots of stock metal solution were added to the biomass samples to achieve metal doses from 0 to 2.0 mmol/g biomass COD. For Cu, pH readjustment was necessary after Cu addition by adding 0.1 M NaOH. The mixtures were placed on a mechanical shaker and constantly agitated at 25 ( 1 °C and 150 rpm for 1 h. Biomass was VOL. 37, NO. 4, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
729
separated by filtration through a 0.45-µm membrane (MillexHA, Millipore), and soluble metal fractions were measured by AA spectroscopy. Metal-biomass partition coefficients (with units of L/g COD biomass), reflecting the distribution of a metal between solid (biomass) and aqueous phase, were calculated from data pairs on biomass associated metal concentration (mg/g COD biomass) and aqueous metal concentration (mg/L). Nitrifier Substrate Oxidation Activity. Maximum specific substrate oxidation rates were measured in duplicate using a batch respirometric assay described elsewhere (42). Briefly, biomass aliquots (50 mL) were collected from the continuousflow reactor. Assays were performed at a final pH of 7.0 ( 0.05 in 50 mL water-jacketed glass vessels, maintained at 25 ( 0.5 °C. Biomass suspensions were aerated with pure oxygen prior to substrate (NH4+-N or NO2--N) injection. Initial concentrations of 5 mg/L of NH4+-N and 10 mg/L of NO2--N permitted measurement of substrate-independent specific substrate oxidation rates (data not shown). Dissolved oxygen (DO) concentrations in the vessel were measured with a DO probe (YSI model 5331, Yellow Springs, OH) and continuously recorded at 4 Hz by a personal computer interfaced to a DO monitor (YSI model 5300, Yellow Springs, OH). The effects of Cu, Zn, Ni, and Cd on nitrification kinetics were investigated individually in batch assays. Aliquots prepared from stock solutions containing Cu, Zn, Ni, or Cd were spiked into batch vessels, and the standard procedure for the substrate oxidation assay was followed. To elucidate the effect of metal exposure time on inhibition, biomass samples in the absence and presence of 1mM analytical concentration of the test metals (for Cd, a range of concentrations of 0.2, 0.35, and 0.5 mM was tested) were preincubated at 25 °C and aerated by magnetic stirring at 100 rpm. At predetermined intervals, aliquots were withdrawn, and nitrification kinetics were determined. Inhibition of biological activity was inferred from differences of measured maximum specific oxygen uptake rates (SOUR) between metal spiked biomass samples and metal-free controls (35). Bacterial Viability. In this work, we utilize the terms “ activity” and “viability” to indicate cellular metabolism and intact membranes, respectively. The numbers of nitrifying bacteria with intact and damaged membranes were measured using a LIVE/DEAD Baclight bacterial viability kit (Molecular Probes, Eugene, OR) and a FACS Calibur flow cytometer (Becton Dickinson, Mountain View, CA) equipped with an argon-ion laser emitting at 488 nm. Viable and dead cells are detected by differential staining with a mixture of a green fluorescent nucleic acid stain, SYTO 9 (stains all cells, live or dead), and a red fluorescent nucleic acid stain, propidium iodide (PI, stains only bacteria with damaged membranes). In addition, a reduction in the SYTO 9 fluorescent emission results when both dyes are present in the cell. Equal volumes of PI and SYTO 9 stock solutions were combined to prepare a fresh dye mixture. A 3 µl aliquot of the dye mixture was added per mL of bacterial suspension. Enumeration of stained cells was facilitated by excitation at 488 nm and detection at 585 (red) and 530 nm (green) for propidium iodide and SYTO 9, respectively. As a result, viable and dead cells can be distinguished by appropriate gating as they occupy distinct regions in the fluorescence cytogram. For all FACS analyses, samples were analyzed until at least 10 000 observations were made. Known mixtures (100:0, 75:25, 50:50, 25:75, and 0:100) of live and dead cells were prepared by mixing aliquots of cells directly retrieved from the reactor (resuspended in 20 mM MOPS, pH 7) and aliquots of cells subject to 2% formaldehyde killing (in 20 mM MOPS, pH 7 for 3 h). The percentage of cells exhibiting the viable signature in the fluorescence cytogram was plotted against the fraction of viable cells in the mixture analyzed, resulting in a linear (R 2 ) 0.92) calibration curve for analysis of experimental samples. 730
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 4, 2003
Nitrifying cultures (20 mM MOPS, pH 7.0 ( 0.05) were spiked with individual metals to yield initial analytical concentrations of 1 mM with 15 min incubation. Microbial cell pellets retained after centrifugation (1600g) were resuspended in the same volume of 0.1 mM NaCl, followed by centrifugation twice for 5 min at 1600g. Between each centrifugation, supernatants were decanted and pellets resuspended in 0.1 mM NaCl. Samples were prefiltered through 25-µm pore size filters (CALBIOCHEM, La Jolla, CA) to prevent clogging of the flow cytometer nozzle. Filtered samples were subsequently subject to Live/Dead staining and examined via flow cytometry. The percentage of viable cells after the various metal spikes was then quantified using the constructed calibration curve. Analytical Procedures. Metal concentrations were measured according to standard methods (43) by flame atomic absorption spectroscopy (model 5100, Perkin-Elmer Co., Norwalk, CT) with method detection limits of 0.3, 0.2, 1.7, and 0.1 µM for Cu, Zn, Ni, and Cd, respectively. Biomass concentrations were measured as COD using commercially available reagents (HACH COD vials, 0-1500 mg/L). The concentrations of Cu2+and Cd2+ were measured directly by ion-selective electrodes (ISE, Orion model 94-29, 94-48) with respective MDLs of 0.1 and 5 µM. The concentrations of Ni2+and Zn2+ were calculated with the MINEQL+ (V 4.5) chemical equilibrium speciation algorithm as reported earlier (35, 44). NH4+-N was analyzed using an ammonia gas-sensing electrode (HNU Systems) with an operating range between 10-2.2 and 104.1 mg/L NH4+-N. Concentrations of NO2--N and NO3-N were measured according to standard methods (43). Glassware and plastic ware, when appropriate, were soaked in 1 M HNO3 overnight and rinsed with 5 vol of Milli-Q purified water before use. Modeling. The kinetics of metal internalization were described by a standard intraparticle diffusion model (45, 46). Basic assumptions to the model include the following: (i) a well-stirred solution of limited volume, V (L), (ii) a spherical sorbent (i.e., the biomass) with mean diameter of d, (iii) biomass initially free from sorbate (i.e., the metal), and (iv) transport controlled by intraparticle mass transfer. The total mass of metal Mt (µmol) in the biomass particle after time t, expressed as a fraction of the corresponding mass (M∞) after infinite time is
Mt M∞
)1-
∞
6R(R + 1) exp(-Dqn2t/d2)
n)1
9 + 9R + qn2R2
∑
(1)
where D is the intraparticle diffusion coefficient (m2/s) and qn are nonzero solutions to the equation
tan qn )
3qn 3 + Rqn2
(2)
and the parameter R relates the fractional uptake (the ratio of final metal mass in the biomass particle to the initial total metal mass) as follows:
M∞ 1 ) VC0 1 + R
(3)
where C0 is the initial solute concentration in the external phase (µmol/L) of volume V. In this study, a characteristic diffusion time td () 0.5d2/D), which represents the efficiency of diffusive mass transport (inversely related to diffusion
FIGURE 2. Internalization kinetics of metals with nitrifying biomass (968 ( 169 mg/L COD, pH 7.0). Lines are best fits to an intraparticle diffusion model. Cu (b, - -); Zn (O, - - -); Ni (9, - - -); Cd (0, s). Error bars indicate one standard deviation. FIGURE 1. Sorption kinetics of metals on nitrifying biomass (968 ( 169 mg/L COD, pH 7.0). Insert: Metal sorption at shorter time. Cu (b); Zn (O); Ni (9); Cd (0). Error bars indicate one standard deviation. coefficient), was used to describe overall intraparticle diffusion (46, 47). Equation 1 is then modified to
( ) qn2t
Mt M∞
∞
)1-
∑
n)1
6R(R + 1) exp -
2td
9 + 9R + qn2R2
(4)
The terms were summed until further addition had no more effect (n ) 1-40). Solutions to eq 4 were developed by determining best-fit td via least-squares error (LSE) analysis using the SOLVER routine in Microsoft Excel.
Results and Discussion Metal Sorption and Internalization Kinetics. The sorption of Cu, Zn, and Ni to the nitrifying biomass was rapid and attained equilibrium within 1 h (Figure 1), consistent with previous results on metal sorption by microbial biomass and activated sludge (14, 15). In contrast, the sorption of Cd to the nitrifying biomass was much slower and did not attain equilibrium within 12 h. Slow kinetics of Cd sorption by bacteria have been documented elsewhere (48, 49). In contrast with the generally fast sorption kinetics, intracellular metal transport was slow: intracellular Zn, Ni, and Cd concentrations continued to increase with time throughout the experiment (up to 12 h for Zn and 24 h for Ni and Cd; Figure 2). A notable exception was Cu, which rapidly internalized and attained equilibrium after 4 h (Figure 2), consistent with earlier observations using yeast and algae (33, 50). To evaluate whether the internalization kinetics could be described by a diffusion process, an intraparticle diffusion model (45) was applied. Figure 2 provides results using this model with correlation coefficient (R 2) of 0.84, 0.87, 0.92, and 0.96 for Cu, Zn, Ni, and Cd, respectively. Characteristic diffusion times (td) were 9.4, 64.6, 80.5, and 66.1 h for Cu, Zn, Ni, and Cd, respectively. Experiments were repeated for a range of analytical Cd concentrations from 0.2 to 0.5 mM, and similar slow Cd internalization rates were observed (data not shown), which were adequately fit by the intraparticle diffusion model (R 2 from 0.87 to 0.92). Few reports on metal internalization kinetics in bacteria exist to date. In a recent study (27) on Zn uptake (dosed at 5 µM) by the Gram-positive bacterium Rhodoccocus opacus, maximum accumulation of internalized Zn was observed
FIGURE 3. Metal partitioning isotherms on nitrifying biomass (752 ( 25 mg/L COD, pH 7.0, 1 h incubation). Cu (b); Zn (O); Ni (9); Cd (0). Error bars indicate one standard deviation. after 20 min followed by a significant decrease of the internalized concentration, which was attributed to homeostatic controls. Such controls might be irrelevant in our study since our metal concentrations were at much higher (mM) levels. Our results were however consistent with those of Bates et al. (32), who observed slow internalization of Zn, at a concentration of 15 µM, by algal cells over a 6-h incubation period. Yet, it remains unclear why Cu transport into cells was faster than for the other metals (Zn, Ni, and Cd). It has been suggested that Cu uptake may be energy-dependent (21, 51, 52). Furthermore, relatively rapid Cu internalization may be related to affinity of Cu to biomass and its unique mode of action (see below). Metal Partitioning. Within the examined narrow ranges of metal concentrations ( Cd > Ni, in qualitative agreement with earlier reports on metal partitioning to activated sludge biomass (14, 15, 40). From the metal partitioning (Figure 3) and internalization (Figure 2) data, maximally attained intracellular fractions of the total biomass associated metal were calculated as 1.4 ( 0.0, 4.3 ( 0.5, 7.6 ( 1.0, and 2.7 ( 0.2% for Cu, Zn, Ni, and Cd, respectively. Therefore, the internalized metal represented only a small fraction of the biomass associated metal mass. Metal Dose-Response Relationship. In short-term batch assays (about 1 h), the specific ammonium oxidation rate (SOURNH4) decreased as the applied metal dose to nitrifying biomass increased (Figure 4). Previously reported observations have demonstrated that free-metal cation concentration VOL. 37, NO. 4, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
731
FIGURE 4. Short-term inhibition of nitrifying activity as function of free metal cation concentration (biomass at 955 ( 67 mg/L COD, pH 7.0, no additional exposure time except ca. 15 min elapsed between metal spike and ammonium spike). Cu2+ (b); Zn2+ (O); Ni2+ (9); Cd2+ (0). Inhibition is expressed with reference to the SOUR of a culture incubated without metal for same duration after removal from parent reactor. Error bars indicate one standard deviation. correlated with inhibition (34, 35). It has been suggested that inhibition is related to metal complexation with intracellular functional groups (e.g., -SH) (13). The molar inhibitory effect toward ammonium oxidation was cation specific and followed: Cu2+ Zn2+ > Cd2+ > Ni2+. Hence, surprisingly, the inhibitory character of the metals tested in our study was not perfectly congruous with their sulfide complexation potential series, which follows Cu2+ > Cd2+ > Ni2+ > Zn2+ (metal sulfide stability constants: 1036.1, 1027.0, 1026.6, and 1024.7 for Cu, Cd, Ni, and Zn, respectively) (53). Effect of Exposure Time on Nitrification Inhibition. To quantify the effect of exposure time on nitrification inhibition, a fixed metal dose was applied in the batch reactor, and the extent of inhibition was measured as a function of time. It was observed that nitrification inhibition by metals increased with incubation time (Figure 5). Depending on the initial metal dose, the inhibition at internalization equilibrium (taken at 24 h) increased by factors of 1.9 ( 0.5, 2.5 ( 0.7, and 2.3 ( 0.3 for Zn, Ni, and Cd, respectively, as compared with short-term (ca. 15 min elapsed between metal spike and ammonium spike) observed inhibition. This timedependent behavior prompted us to investigate the correlation between metal partitioning and inhibition. Inhibition Related to Sorbed Metal Concentration. Poor correlation (R 2 ) 0-0.65) was noticed between observed inhibition and sorbed metal concentrations using a linear inhibition model (Figure 6). The sorption of Cu, Zn, and Ni to the nitrifying biomass was rapid and achieved equilibrium within 1 h while nitrification inhibition due to metal exposure was time dependent up to 24 h. Therefore, sorbed metal concentrations were not good predictors of the metals’ effect on nitrification kinetics. Inhibition Related to Intracellular Metal Concentration. Internalized metals are generally thought to interact with cellular functional groups, destroy protein structure and function, and interfere with physiologically important ions (13). Hence, inhibition may be correlated to the intracellular metal concentrations. Indeed, the extent of inhibition of NH4+-N oxidation kinetics correlated well with the intracellular Zn, Ni, and Cd concentrations (Figure 7), suggesting that intracellular Zn, Ni, and Cd were directly responsible for the observed inhibitory effects on nitrifying activity. The full collection of metal inhibition results was welldescribed by a saturation-type (rectangular hyperbola) biological toxicity model (54, 55) from the characteristic shape of the inhibition versus intracellular concentrations (R 2 ) 0.82-0.94 as compared to R 2 ) 0.60-0.78 for linear model) 732
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 4, 2003
FIGURE 5. Effect of exposure time on nitrification inhibition by (a) Cu(b), Zn(O), and Ni(9) at 1 mM total analytical concentrations and (b) Cd (in mM) at 0.2 (b), 0.35 (O), and 0.5 (0) in the batch assays. Metals were added at time 0.
FIGURE 6. Inhibition of NH4+-N oxidation as a function of sorbed metal concentration. Cu (b); Zn (O); Ni (9); Cd (0). (Figure 7). It reads
I)
Imax[M] KM + [M]
(5)
where Imax is the maximum degree of inhibition (I) (100%); [M] is the intracellular total metal concentration, [µmol g-1 XCOD]; and KM is a half-maximum response coefficient, [µmol g-1 XCOD]. The KM value, representing the intracellular metal concentration causing 50% inhibition, was estimated at 0.50, 3.24, and 0.80 µmol g-1 XCOD for Zn, Ni, and Cd, respectively. Inhibition Modeling. As inhibition increased at longer exposure times, the short-term inhibition tests are of limited applicability and need to be supplemented with a transportbased model describing kinetics of metal biomass interactions. Combining the intraparticle diffusion model and the
TABLE 2. Characteristics of Metal Internalization, Partitioning, and Complexation metal
td (h)a
Kp (L/g XCOD)b
Kc
Cu Zn Ni Cd
9.4 64.6 80.5 66.1
20.3 ( 0.1 0.4 ( 0.0 0.1 ( 0.0 0.2 ( 0.0
1036.1 1024.7 1026.6 1027.0
a t , characteristic metal diffusion time in nitrifying biomass. b K , d p metal partition coefficient on nitrifying biomass. c K, stability constants of metal sulfide complexes (53).
FIGURE 7. Inhibition of NH4+-N oxidation as a function of intracellular metal concentration. Continuous lines are results of a saturation-type biological toxicity model. Cu (b); Zn (O, - - -); Ni (9, - - -); Cd (0, s).
FIGURE 8. Comparison of calculated inhibition (using the intraparticle diffusion model and biological toxicity model) and measured inhibition at various initial Cd doses (0.2, 0.35, and 0.5 mM) and exposure times (1, 6, 12, and 24 h). Model line prediction is shown in bold. Upper and lower 95% confidence lines are also shown. biological toxicity model, inhibition could be predicted. This was done for Cd as a representative metal. The full collection of Cd inhibition results at 0.2, 0.35, and 0.5 mM initial Cd concentrations and various exposure time (1, 6, 12, 24 h) was well described by the combined diffusion and toxicity models (Figure 8). Inhibition by Copper: Mode of Action. In comparison to Zn, Ni, and Cd, copper was internalized rapidly (ca. 4 h) while inhibition due to Cu exposure continued to increase with time beyond 24 h. Furthermore, there was no correlation between nitrification inhibition and intracellular Cu concentration (R 2 ) 0.03, Figure 7). This prompted us to investigate whether the suggested mechanism of copper toxicity that involves disruption of the cytoplasmic membrane (50, 56) also held for the nitrifying enrichment cultures. In the presence of 1 mM Cu, cells had a much lower specific fluorescence signal (30 ( 1 total events/s) than in the presence of same concentrations of Zn (143 ( 22), Ni (119 ( 8), Cd (144 ( 7), and the metal-free control (138 ( 13 total events/ s). Furthermore, on the basis of LIVE/DEAD Baclight staining, less than 13.3 ( 10.5% cells were viable after 15 min exposure to 1 mM Cu as compared to 72.8 ( 7.5, 104.8 ( 1.7, and 84.7 ( 7.0% viable cells in the presence of same concentrations of Zn, Ni, and Cd. The biological response of Cu, therefore, is unique among the metals studied, may involve cytoplasmic membrane rupture, and cannot be explained by the operative mechanism for the other metals that interact with cellular functional groups and degrade protein structure and function (13).
Discussion. The intrinsic slow growth of nitrifying bacteria and their high sensitivity to environmental perturbations often render nitrification the rate-determining step in biological nitrogen removal during wastewater treatment operations. Metals can cause nitrification inhibition, especially when applied as shock loads. This work sought to provide insights in the mechanism of nitrification inhibition by metals. We examined the correlation between extracellular metal sorption, intracellular metal accumulation, and nitrification inhibition. We observed that inhibition by Zn, Ni, and Cd correlated well with the intracellular, but not the sorbed fractions (Figure 7). Further, the intracellular fractions increased slowly with time. By employing an intraparticle diffusion model coupled to a saturation-type biological toxicity model, the observed inhibitions at varying Cd concentrations were successfully described. Our results indicate no direct correlation between intracellular or sorbed Cu concentrations and nitrification inhibition. Copper is characterized by high complexation potential (i.e., with sulfide), high degree of partitioning to nitrifying biomass, and fast internalization kinetics (Table 2). Such difference in physicochemical behavior as compared to other metals may affect its biological response. The bacterial LIVE/ DEAD test indicated a fast and high degree of cell viability loss unique to Cu exposure; the exact mode of action of Cu on bacterial activity is yet to be elucidated. The key finding of this work is that, because of the slow metal internalization kinetics, exposure time may dictate remedial approaches. Using results from short-term batch inhibition assays without correction to infer responses in continuous flow reactors, which may experience longer metal exposures, may be erroneous. The actual manifestation of the kinetically limiting inhibition phenomena on performance of continuous-flow reactors subject to shock-loads of metals is yet to be examined. In summary, the relationship between cellular metal partitioning and nitrification activity was evaluated employing batch experiments with a suite of metals and nitrifying cultures. Major findings were as follows: (i) Sorption of Cu, Zn, Ni was fast and attained equilibrium within 1 h, while the sorption of Cd was much slower. Internalization of Zn, Ni, and Cd, on the other hand, was slow and increased throughout the experimental time, while internalization of Cu was relatively fast. Internalization kinetics were well-described by an intraparticle diffusion model. (ii) Inhibition by Zn, Ni, and Cd correlated well with the measured intracellular fractions but not with the sorbed fractions. Inhibition was well-described by a saturation-type biological toxicity model based on intracellular Zn, Ni, or Cd concentrations. (iii) Short-term batch assays may not reflect ultimate inhibition by metals because of their slow internalization kinetics. (iv) No correlation was observed between the intracellular or sorbed fractions of Cu and inhibitory effect on nitrifying activity. Bacterial viability dropped dramatically after exVOL. 37, NO. 4, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
733
posure to Cu but not to Zn, Ni, or Cd, indicating that inhibition by Cu follows a different mechanism that may involve cytoplasmic membrane disintegration.
Acknowledgments We are grateful to Dr. Michele Barber in the Department of Molecular and Cell Biology at the University of Connecticut for assistance with flow cytometry. This study was funded, in part, by the Connecticut Department of Environmental Protection through a U.S. EPA Long Island Sound Study grant under Section 119 of the Clean Water Act.
Literature Cited (1) Neufeld, R. D. J. Water Pollut. Control Fed. 1976, 48, 1940. (2) Battistoni, P.; Fava, G.; Ruello, M. L. Water Res. 1993, 27, 821. (3) Dangcong, P.; Bernet, N.; Delgenes, J.; Moletta, R. Water Environ. Res. 2000, 72, 195. (4) Ford, D. L.; Churchwell, R. L.; Kachtick, J. W. J. Water Pollut. Control Fed. 1980, 52, 2726. (5) Shammas, N. K. J. Water Pollut. Control Fed. 1986, 58, 52. (6) Painter, H. A. Water Res. 1970, 4, 393. (7) Davis, J. A.; Jacknow, J. J. Water Pollut. Control Fed. 1975, 47, 2292. (8) Stensel, H. D.; McDowell, C. S.; Ritter, E. D. J. Water Pollut. Control Fed. 1976, 48, 2343. (9) Gruttner, H.; Winther-Nielsen, M.; Jorgensen, L.; Bogebjerg, P.; Sinkjer, O. Water Sci. Technol. 1994, 29, 69. (10) Chang, D.; Fukushi, K.; Ghosh, S. Water Environ. Res. 1995, 67, 822. (11) White, C.; Wilkinson, S. C.; Gadd, G. M. Int. Biodeterior. Biodegrad. 1995, 35, 17. (12) Blackwell, K. J.; Singleton, I.; Tobin, J. M. Appl. Microbiol. Biotechnol. 1995, 43, 579. (13) Nies, D. H. Appl. Microbiol. Biotechnol. 1999, 51, 730. (14) Nelson, P. O.; Chung, A. K.; Hudson, M. C. J. Water Pollut. Control Fed. 1981, 53, 1323. (15) Wang, J.; Huang, C. P.; Allen, H. E.; Poesponegoro, I.; Poesponegoro, H.; Takiyama, L. R. Water Environ. Res. 1999, 71, 139. (16) Brown, M. J.; Lester, J. N. Water Res. 1979, 13, 817. (17) Brown, M. J.; Lester, J. N. Water Res. 1982, 16, 1539. (18) Beveridge, T. J.; Murray, R. G. E. J. Bacteriol. 1980, 141, 876. (19) Cox, J. S.; Smith, D. S.; Warren, L. A.; Ferris, F. G. Environ. Sci. Technol. 1999, 33, 4514. (20) Texier, A.; Andres, Y.; Illemassene, M.; Cloirec, P. L. Environ. Sci. Technol. 2000, 34, 610. (21) Rensing, C.; Rosen, B. P. In Molecular Biology and Toxicology of Metals; Zalupa, R. K., Koropatnick, J., Eds.; Taylor & Francis: New York, 2000; p 129. (22) Achouak, W.; Heulin, T.; Pages, J.-M. FEMS Microbiol. Lett. 2001, 199, 1. (23) Amor, L.; Kennes, C.; Veiga, M. C. Bioresour. Technol. 2001, 78, 181. (24) Howlett, N. G.; Avery, S. V. Appl. Environ. Microbiol. 1997, 63, 2971. (25) Stillman, M. J.; Presta, A. In Molecular Biology and Toxicology of Metals; Zalupa, R. K., Koropatnick, J., Eds.; Taylor & Francis: New York, 2000; p 1. (26) Lopez-Sanchez, J. F.; Rubio, R.; Samitier, C.; Rauret, G. Water Res. 1996, 30, 153.
734
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 4, 2003
(27) Mirimanoff, N.; Wilkinson, K. J. Environ. Sci. Technol. 2000, 34, 616. (28) Bagay, M. M.; Sherrard, J. H. J. Water Pollut. Control Fed. 1981, 53, 1609. (29) Cabrero, A.; Fernandez, S.; Mirada, F.; Garcia, J. Water Res. 1998, 32, 1355. (30) Chua, H.; Yu, P. H.; Sin, S. N.; Cheung, M. W. Chemosphere 1999, 39, 2681. (31) Barber, W. P.; Stuckey, D. C. Water Res. 2000, 34, 2423. (32) Bates, S. S.; Tessier, A.; Campbell, P. G. C.; Buffle, J. J. Phycol. 1982, 18, 521. (33) Vasconcelos, M. T. S. D.; Leal, M. F. C. Environ. Sci. Technol. 2001, 35, 508. (34) Braam, F.; Klapwijk, A. Water Res. 1981, 15, 1093. (35) Hu, Z.; Chandran, K.; Grasso, D.; Smets, B. F. Environ. Sci. Technol. 2002, 36, 3074. (36) Chang, S. I.; Reinfelder, J. R. Environ. Sci. Technol. 2000, 34, 4931. (37) Parent, L.; Twiss, M. R.; Campbell, P. G. C. Environ. Sci. Technol. 1996, 30, 1713. (38) Ferris, F. G.; Beveridge, T. J. Can. J. Microbiol. 1986, 32, 594. (39) Knauer, K.; Behra, R.; Sigg, L. Environ. Toxicol. Chem. 1997, 16, 220. (40) Su, M.-C.; Cha, D. K.; Anderson, P. R. Water Res. 1995, 29, 971. (41) Hu, Z.; Chandran, K.; Grasso, D.; Smets, B. F. Environ. Eng. Sci. (accepted for publication). (42) Chandran, K.; Smets, B. F. Biotechnol. Bioeng. 2000, 68, 396. (43) APHA; AWWA; WEF. Standard methods for the examination of water and wastewater; APHA: Washington, DC, 1998. (44) Schecher, W. D.; McAvoy, D. C. MINEQL+: A chemical equilibrium modeling system; Version 4.5 for Windows; Hallowell, ME, 2001. (45) Crank, J. The mathematics of diffusion; Clarendon Press: Oxford, 1976; p 89. (46) Clark, M. M. Transport modeling for environmental engineers and scientists; John Wiley & Sons: New York, 1996; p 259. (47) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental organic chemistry; John Wiley & Sons: New York, 1993; p 200. (48) Yee, N.; Fein, J. Geochim. Cosmochim. Acta 2001, 65, 2037. (49) Kim, D. W.; Cha, D. K.; Wang, J.; Huang, C. P. Chemosphere 2002, 46, 137. (50) Avery, S. V.; Howlett, N. G.; Radice, S. Appl. Environ. Microbiol. 1996, 62, 3960. (51) Cooksey, D. A. FEMS Microbiol. Rev. 1994, 14, 381. (52) Pena, M. M.; Lee, J.; Thiele, D. J. J. Nutr. 1999, 129, 1251. (53) Morel, F. M. M.; Hering, J. G. Principles of applications of aquatic chemistry; Wiley-Interscience: New York, 1993; p 407. (54) Brown, P. L.; Markich, S. J. Aquat. Toxicol. 2000, 51, 177. (55) Timbrell, J. Principles of biochemical toxicology; Taylor & Francis Ltd.: London, 2000; p 11. (56) Sani, R. K.; Peyton, B. M.; Brown, L. T. Appl. Environ. Microbiol. 2001, 67, 4765.
Received for review July 17, 2002. Revised manuscript received November 26, 2002. Accepted December 4, 2002. ES025977D
