PREDICTIVE MODELING OF METAL ADSORPTION ONTO BACTERIAL SURFACES IN GEOLOGIC SETTINGS. A Dissertation. Submitted to the Graduate School

Transcription

1 PREDICTIVE MODELING OF METAL ADSORPTION ONTO BACTERIAL SURFACES IN GEOLOGIC SETTINGS A Dissertation Submitted to the Graduate School of the University of Notre Dame in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy by David M. Borrok, B.S., M.S. Jeremy B. Fein, Director Graduate Program in Geological Sciences Notre Dame, Indiana April 2005

2 PREDICTIVE MODELING OF METAL ADSORPTION ONTO BACTERIAL SURFACES IN GEOLOGIC SETTINGS Abstract by David M. Borrok Bacterial surfaces are capable of adsorbing large quantities of metals, and are thought to partly control the distribution, fate, and bioavialability of metals in near-surface geologic systems. Geochemical models have been employed that are capable of predicting the extents of metal adsorption onto specific bacterial species under laboratory conditions. However, our ability to extrapolate these models to predict the distribution and fate of metals in realistic geologic systems is limited. This dissertation presents the work of a number of closely linked, but individual studies that attempt to quantitatively describe the adsorption reactions on bacterial surfaces so that we can predict the extent and importance of these reactions in geologic systems. This dissertation is the synthesis of more than 300 individual experiments (batch adsorption experiments, potentiometric titrations, chemotaxis experiments, etc.) and corresponding surface complexation models and modeling parameters that test the following questions: (Ch. 2) Are modeling parameters developed from laboratory experiments conducted using bacteria treated with acid similar to those for

3 David Borrok bacteria in natural (non-acid treated) systems? (Ch. 3 & 4) Do consortia of bacteria from natural and contaminated systems exhibit universal adsorption behavior? (Ch. 5) How will salt concentration affect the adsorption behavior of bacteria over the ionic strength ranges found in natural systems? (Ch. 6) Can adsorption models be used to predict bacterial chemotaxis in complex multicomponent systems? The results from these studies demonstrate that (Ch. 2) acidic solutions can damage the bacterial surface by displacing structurally bound Mg and Ca, (Ch. 3 & 4) consortia of bacteria from uncontaminated environments exhibit similar extents of Cd adsorption, while consortia of bacteria from contaminated environments adsorb Cd to much greater extents, (Ch. 5) ionic strength has a negligible impact on the adsorption of protons, Cd, and Pb onto bacterial surfaces, and (Ch. 6) adsorption reactions can control bacterial chemotactic responses and chemical equilibrium models can be used to predict these responses in multicomponent systems. These studies are successful in bringing us closer than ever before to predicting the true extent of bacterial surface adsorption reactions in real systems. ii

4 CONTENTS FIGURES.v TABLES...viii ACKNOWLEDGMENTS...x CHAPTER 1: INTRODUCTION...1 CHAPTER 2: THE EFFECT OF ACIDIC SOLUTIONS AND GROWTH CONDITONS ON THE ADSORPTIVE PROPERTIES OF BACTERIAL SURFACES Introduction Materials and Methods Growth Condition Metal Adsorption Experiments Acidification and Dissolved Organic Carbon (DOC) Experiments Results Discussion Effect of Acid on the Adsorptive Properties of Bacterial Surfaces Effect of Growth Conditions on the Binding Properties of Bacterial Surfaces Modeling of Bacteria-Metal Adsorption Reactions Implications for Describing Bacteria-Metal Adsorption in Nature Summary and Conclusions.37 CHAPTER 3: PROTON AND Cd ADSORPTION ONTO NATURAL BACTERIAL CONSORTIA: TESTING UNIVERSAL ADSORPTION BEHAVIOR Introduction Materials and Methods Sampling and Growth of Bacteria Freezing of Bacteria Potentiometric and Cd Adsorption Experiments DGGE Analysis.46 ii

5 3.3 Results and Discussion Surface Complexation Modeling Conclusions.60 CHAPTER 4: Cd AND PROTON ADSORPTION ONTO BACTERIAL CONSORTIA GROWN FROM INDUSTRIAL WASTES AND CONTAMINATED GEOLOGIC SETTINGS Introduction Materials and Methods Sample Descriptions Sampling and Growth of Bacteria Freezing of Bacteria Potentiometric Titration and Cd Adsorption Experiments Gram-Staining and DGGE Analysis Results and Discussion Gram-staining DGGE Results Potentiometric Titrations Cd Adsorption Experiments Modeling of Data The Range of Proton and Cd Adsorption by Bacterial Consortia Difference in Adsorption Among Bacterial Consortia Development of a Universal Bacteria-Metal Adsorption Model..89 CHAPTER 5: THE IMPACT OF IONIC STRENGTH ON THE ADSORPTION OF PROTONS, Pb, Cd, AND Sr ONTO GRAM NEGATIVE BACTERIAL SURFACES: TESTING NON-ELECTROSTATIC, DIFFUSE LAYER, AND TRIPLE LAYER MODELS Introduction Materials and Methods Bacteria and Growth Conditions Potentiometric Titration Experiments Cd and Pb Adsorption and Kinetics Experiments Ionic Strength Isotherm Experiments with Sr, Cd, and Pb Modeling Approach Discrete Site Surface Complexation Model Model Calculations Bacterial Surface Electric Field Models Experimental and Modeling Results Potentiometric Titration Experiments Cd Adsorption/Desorption Kinetics Cd and Pb Adsorption Experiments Cd, Pb, and Sr Isotherm Experiments Discussion Proton Adsorption Behavior and Ionic Strength iii

6 5.5.2 Metal Adsorption/Desorption Kinetics Cd, Pb, and Sr Adsorption and Ionic Strength Effects Summary and Conclusions CHAPTER 6: ENVIRONMENTAL CONTROL AND PREDICTION OF BACTERIAL CHEMOTACTIC RESPONSES Introduction Materials and Methods Culture Methods D Motion Analysis Potentiometric Titration and Nickel Adsorption Experiments Results and Discussion Ni 2+ Chemotaxis Experiments E. coli Responses to Leucine and Maltose Modeling Potentiometric Titration and Ni 2+ Adsorption Experiments Relating Adsorption and Chemotaxis Chemotaxis in Real Environments Prediction of Chemotaxis in Natural Environments 159 CHAPTER 7: CONCLUSIONS 161 REFERENCES 166 iv

7 FIGURES 2.1 Scanning Electron Microscope (SEM) images of acid-washed B. subtilis (A) and P. mendocina (B) cells Cd adsorption data for Acid Wash, NPS, and APS experiments with B. subtilis at 1 g/l Cd adsorption data for Acid Wash, NPS, and APS experiments with P. mendocina at 0.3 g/l Co adsorption data for Acid Wash and NPS experiments with B. subtilis at 1 g/l Pb adsorption data for Acid Wash and NPS experiments with B. subtilis at 0.05 g/l and Pb at 10 ppm (48 mol/l) Cd adsorption data for Acid Wash and NPS experiments with B. subtilis at 0.05 g/l and Cd at 5.4 ppm (48 mol/l) Concentration data and best-fit curves for Mg (short dashes), Ca (solid), and DOC (long dashes) as a function of ph, with B. subtilis at 1 g/l Concentration data and best-fit curves for Mg (short dashes), Ca (solid), and DOC (long dashes) as a function of ph, with P. mendocina at 1 g/l Co adsorption data for experiments with P. fluorescens (1 g/l) grown in full LB medium and 10% LB medium Co adsorption data for experiments with S. oneidensis MR-1 (1 g/l) grown under aerobic and anaerobic conditions Cd adsorption data for Acid Wash and NPS experiments with B. subtilis, B. cereus, P. mendocina, or P. aeruginosa at 1 g/l Example DGGE gel with numbered lanes Potentiometric titration data (solid circles), and best-fit model using 4 surface sites (curve), for the forest soil #2 consortium.52 v

8 3.3 Cd adsorption onto bacterial consortia cultured from soil and aquatic environments Comparison of the natural adsorption edge (solid curve) to Cd adsorption data from experiments using individual, pure strains of bacteria DGGE gels Representative potentiometric titration data reported as protons consumed per gram of consortium Cd adsorption onto bacterial consortia grown from industrial wastes and contaminated soils and groundwater Total functional group site densities for bacterial consortia isolated from industrial wastes and contaminated soils and groundwater Individual functional group site densities for each bacterial consortium Potentiometric titration data for P. putida reported as H + consumed per gram of bacteria with average best-fit curves for the non-electrostatic model (solid curve and squares), DLM (solid curve and triangles), and TLM (solid curve and circles) Potentiometric titration data for P. mendocina reported as H + consumed per gram of bacteria with average best-fit curves for the non-electrostatic model (solid curve and squares), DLM (solid curve and triangles), and TLM (solid curve and circles) Apparent proton binding stability constants (pk a ) for P. putida and P. mendocina at ionic strengths of 0.01 M, 0.1 M, and 0.5 M Apparent total functional group site densities for P. putida (gray bars) and P. mendocina (clear bars) at ionic strengths of 0.01 M, 0.1 M, and 0.5 M Results from adsorption and desorption kinetics experiments conducted using P. putida. Closed circles () represent desorption results for ph Results from adsorption and desorption kinetics experiments conducted using P. mendocina Cd adsorption data for 10 g/l (ο = 0.01 m ionic strength, = 0.1 m ionic strength, = 0.5 m ionic strength) and 3 g/l ( ) P. putida experiments Cd adsorption data for 10 g/l (ο = 0.01 m ionic strength, = 0.1 m ionic strength, = 0.5 m ionic strength) and 3 g/l ( * ) P. mendocina experiments 119 vi

9 5.9 Pb adsorption data for 1 g/l (ο = 0.01 m ionic strength, = 0.1 m ionic strength, = 0.5 m ionic strength) and 3 g/l ( * = 0.1 m ionic strength) P. putida experiments Pb adsorption data for 0.5 g/l (ο = 0.01 m ionic strength, = 0.1 m ionic strength, = 0.5 m ionic strength) and 3 g/l ( * = 0.1 m ionic strength) P. mendocina experiments Cd adsorption isotherm data for 3 g/l P. putida () and 3 g/l P. mendocina (ο) at constant ph (5.9 ± 0.2), constant total Cd concentration (10 ppm), and variable ionic strength (0.001 to 0.6 M) Pb adsorption isotherm data for 1 g/l P. putida () and 1 g/l P. mendocina (ο) at constant ph (5.5 ± 0.2), constant total Pb concentration (10 ppm), and variable ionic strength (0.001 to 0.6 M) Sr adsorption isotherm data for 6 g/l P. putida () and 6 g/l P. mendocina (ο) at constant ph (6.3 ± 0.2), constant total Sr concentration (10 ppm), and variable ionic strength (0.001 to 0.6 M) A Negative chemotaxis of E. coli cells in response to Ni 2+. Angular velocity measured as a function of ph at constant (100 µm) aqueous Ni 2+ concentration () or with no Ni 2+ present () B Negative chemotaxis of E. coli cells in response to Ni 2+. Angular velocity measured as a function of aqueous EDTA concentration at constant (100 µm) Ni 2+ concentration () and control measurement with no Ni 2+ present () Chemotaxis of E. coli cells in response to leucine and maltose Raw data for triplicate potentiometric titrations of E. coli cells Ni 2+ adsorption onto E. coli cells as a function of ph Negative chemotaxis as a function of the concentration of Ni 2+ adsorbed to the surface of E. coli 158 vii

10 TABLES 2.1 ADSORPTION CONSTANT VALUES FOR BEST-FIT SURFACE COMPLEXATION MODELS SAMPLE GROWTH INFORMATION RELATIVE BAND INTENSITIES IN INDIVIDUAL LANES OF DGGE GELS SITE CONCENTRATIONS AND PROTON BINDING CONSTANTS (K a ) with 1 UNCERTAINTIES, FOR THE FOUR FUNCTIONAL GROUP SITES IDENTIFIED THROUGH TRIPLICATE POTENTIOMETRIC TITRATIONS Cd METAL BINDING CONSTANTS (K) FOR BEST-FIT ADSORPTION MODELS SAMPLE LOCATION AND GROWTH INFORMATION RELATIVE BAND INTENSITIES WITHIN INDIVIDUAL LANES OF DGGE GELS AVERAGE PROTON BINDING CONSTANTS (K a ) AND SITE CONCENTRATIONS WITH 1 ERRORS BASED ON BEST-FIT MODELS OF TITRATION DATA Cd METAL BINDING CONSTANTS (K) FOR BEST-FIT ADSORPTION MODELS. FUNCTIONAL GROUP SITES ARE THOSE IDENTIFIED THROUGH POTENTIOMETRIC TITRATIONS AVERAGE NON-ELECTROSTATIC, DLM, AND TLM PROTON BINDING STABILITY CONSTANTS AND FUNCTIONAL GROUP SITE CONCENTRATIONS FOR TRIPLICATE P. putida TITRATIONS AVERAGE NON-ELECTROSTATIC, DLM, AND TLM PROTON BINDING STABILITY CONSTANTS AND FUNCTIONAL GROUP SITE CONCENTRATIONS FOR TRIPLICATE P. mendocina TITRATIONS LOG APPARENT Cd BINDING STABILITY CONSTANTS FOR BEST-FIT NON-ELECTROSTATIC ADSORPTION MODELS FOR 10g/L P. putida AND P. mendocina EXPERIMENTS viii

11 5.4 LOG APPARENT Pb BINDING STABILITY CONSTANTS FOR BEST-FIT NON-ELECTROSTATIC ADSORPTION MODELS FOR 1g/L P. putida AND 0.5 P. mendocina EXPERIMENTS PROTON AND Ni 2+ BINDING STABILITY CONSTANTS AND FUNCTIONAL GROUP SITE CONCENTRATIONS FOR E. coli ix

12 ACKNOWLEDGMENTS I would like to extend my appreciation and thanks to Dr. Jeremy Fein for his guidance during my Ph.D. work. It would be difficult to find a better advisor and mentor. I would also like to thank my committee members, Dr. Peter Burns, Dr. Patricia Maurice, Dr. Clive Neal, and Dr. Ken Kemner for their time, consideration, and helpful suggestions, as well as other professors and colleagues who helped me achieve success in this endeavor, including Dr. Charles Kulpa, Dr. Ben Turner, Dr. Jinesh Jain, Kathryn Docherty, and Jennifer Schaefer. Aqueous metal and DOC analyses were conducted at the Center for Environmental Science and Technology (CEST) at the University of Notre Dame. Other equipment and laboratory supplies were provided through the Environmental Molecular Science Institute (EMSI) at the University of Notre Dame. Research funding was provided by the National Science Foundation through grants EAR , EAR , and EAR I would like to thank the Arthur J. Schmitt Foundation for their generous financial support, which made it possible for me to pursue this degree. x

13 CHAPTER 1 INTRODUCTION Bacteria are Earth s smallest living creatures and have existed and evolved on our planet for billions of years, efficiently harvesting the energy from geochemical reactions. The bacterial cell wall acts as interface with the environment, and is capable of adsorbing solutes and sensing the presence and abundance of chemical stimulants. Thus, bacterial surfaces are important in mediating many important geological processes, including the transport and cycling of metals and the precipitation and dissolution of minerals. This dissertation presents the work of a number of closely linked, but individual studies that attempt to quantitatively describe the adsorption reactions on bacterial surfaces so that we can predict the extent and importance of these reactions in geologic settings. Bacterial surfaces contain carboxyl, phosphoryl, hydroxyl, and amino functional groups (Beveridge and Murray, 1976; Beveridge and Murray 1980). These functional groups deprotonate as a function of increasing ph, giving the bacterial surface a net negative charge above a ph of approximately 2.0. Negatively charged functional group sites are capable of adsorbing large quantities of metals (Beveridge and Murray 1980; Beveridge, 1989) and some hydrophobic and polar organic compounds (Daughney and Fein, 1998; Fein et al., 1999). 1

14 Bacteria are present in a wide range of near-surface environments. Reported bacterial abundances range from about 1 x 10 4 to 1 x 10 7 cells/ml in fresh surface waters (Fisher et al., 1998; Almeida et al., 2001; Kisand et al., 2001). Larger concentrations (up to 1 x 10 9 cells/g) of bacteria have been reported for aquifer and lake sediments (Balkwill and Ghionese, 1985; Zheng and Kellogg, 1994; Barnes and Nierzwicki-Bauer, 1997; Alfreider et al., 1997; Martino et al., 1998). Concentrations as high as 1 x 10 9 cells/g have been reported for organic-rich soils (Barnes and Nierzwicki-Bauer, 1997), while bacterial concentrations in seawater (5 x 10 5 to 3 x 10 6 cells/ml) are generally intermediate in value (Zweifel and Hangstrom, 1995). Because of their affinity for binding metals, their abundance in near-surface geologic systems, and their extremely high surface to volume ratios, bacteria can partly control the speciation and distribution of metals in many environments (Ledin et al., 1996; Barnes and Nierzwicki-Bauer, 1997; Ledin et al., 1999). Hence, accurate and quantitative models that describe bacteria-metal adsorption are critical to understanding the availability and cycling of metals (Tornabene and Edwards, 1972; Tortell et al., 1999), for predicting the behavior of heavy metal contaminants (Volesky, 2001), and in the development of a number of contaminant remediation strategies (Macaskie and Basnakova et al., 1998; Espisito et al., 2001). A number of geochemical models, each with their own adjustable parameters, have been developed that are capable of describing the adsorption of protons, metals, and other chemicals onto individual bacterial species in simplistic laboratory systems (Xue et al., 1988; Plette et al., 1995; Plette et al., 1996; Fein et al., 1997; He and Tebo, 1998; Cox et al., 1999; Pagnanelli et al., 2000; Fein et al., 2004). However, it 2

15 is unclear whether any of these models and modeling parameters can be accurately applied to realistic geologic and industrial settings. A number of proposed models utilize empirically-derived bulk partitioning constants based on some variation of Langmuir or Freundlich isotherms (e.g., Plette et al., 1995, 1996). Although these models are useful for specific situations, they cannot be applied to ph, ionic strength, and component concentrations not directly studied in the laboratory (Bethke and Brady, 2000; Koretsky, 2000). As a potential solution to this problem, surface complexation models (SCMs) similar to those previously used for adsorption of metals onto mineral surfaces (Koretsky et al., 1998) have been proposed for modeling the adsorption of protons and metals onto bacterial surfaces (e.g., Fein et al., 1997). In this approach, the stability of the adsorbed metal-bacterial surface complex can be quantified with a unique equilibrium constant. Experimental data is used to constrain a system of mass action and mass balance equations, which can be solved explicitly to determine the equilibrium constants for the adsorption reactions (and/or concentrations of individual chemical species). The equilibrium constants developed using the SCM approach are valid for systems not directly studied in the laboratory (i.e., different ph, ionic strength, and component concentrations). This flexibility is what makes the SCM approach so attractive for describing complex geologic settings. Using the SCM approach, functional groups present on the bacterial surface can be represented using a number of discrete monoprotic acids, each of which undergoes the following deprotonation reaction: R-A i H R-A i - + H + 3

16 where R is the bacterium to which the functional group type, A, is attached. The acidity constant, K a, for reaction (1) can be expressed as: K a = [R - A - i [R - A i ]a + H H ] where [R-A i - ] and [R-A i H ] represent the concentration of deprotonated and protonated sites, respectively, and a + H represents the activity of protons in the bulk solution. Metal complexation with the deprotonated forms of the monoprotic acids can be expressed as: M m+ + R-A - (m -1) i R-A i (M) where M m+ is the aqueous metal cation of interest. The equilibrium constant, K, for this reaction is given by: K = [R - A a M i (M) m 1 [ R A m+ i ] ] where [R-A i (M) m-1 ] is the concentration of the metal-functinal group complex of interest and a m+ is the aqueous activity of the metal cation. M Ideally, each monoprotic acid used in the SCM would represent a unique chemical functional group (i.e., carboxyl, phosphorayl, hydroxyl, amino) present on the bacterial surface. Although some recent spectroscopic studies (Kelly et al., 2002; Boyanov et al., 2004) have provided broad constraints on the bacteria-metal surface 4

17 complexes of interest, our understanding of the distribution of these functional groups on the bacterial surface and the specific complexes they form with metals is still rudimentary. Hence, the SCM approach presented here is an approximation of the mechanisms involved in what is likely an extremely complex and heterogeneous chemical system, and may not be an exact representation of the chemical speciation of the system. This dissertation explores what additional simplifying assumptions may be necessary to apply surface complexation models to geologic systems. Chapter 2 of this dissertation examines the question of whether exposure to acidic solutions affects the adsorption capacity of bacterial surfaces. The ph of soil and water in the vast majority of geologic systems ranges from about 5 to 8. Hence, bacteria in these systems grow and interact with their environments without ever experiencing the shock of acidic conditions. However, to properly constrain key modeling parameters such as binding site densities and metal-binding stability constants, experimental adsorption measurements must be collected over a broad ph range (about 2 to 10). Moreover it is common practice to subject bacteria to acid washes to remove metals acquired from growth media prior to experimentation (Fein et al., 1997, Yee and Fein, 2001, Ngwenya et al., 2003). Some evidence suggests that exposure to acidic solutions can cause structural damage to bacteria (Beveridge, 1989) and in some cases may even increase the metal binding capacity of the bacterial surface (Wong et al., 1993; Chang et al., 1997). We test the affect of acidic solutions by performing Cd, -Co, and -Pb batch adsorption experiments with starting solutions covering a range of ph values, utilizing acid-washed and non-acid-washed Gramnegative (P. mendocina and P. aeruginosa) and Gram-positive (Bacillus subtilis, 5

18 B. cereus) bacterial species. Bacterial acidification experiments were also conducted with several of the same species, and the dissolved organic carbon (DOC) content was measured for a number of the acidification experiment samples to assess structural damage to bacterial cell walls. We quantify these effects using the SCM approach (described above) to determine whether existing modeling parameters must be modified before they can be extrapolated to realistic geologic systems. This study also documents preliminary testing on the effects of nutrient and oxygen levels on the adsorptive capacity of individual bacterial species. Chapter 3 of this dissertation addresses the question of whether bacterial species from natural environments adsorb metals to universal extents, or whether bacterial surface adsorption properties are unique for different bacterial species. Geologic systems potentially contain dozens of bacterial species. If the cell wall functional group sites of each bacterial species exhibit unique adsorption properties, it would be necessary to determine site densities and binding constants for each site on each bacterial species of interest. This would be an impractical task because experimentation on dozens of different bacterial species (many of which cannot even be isolated in the laboratory) would be necessary. However, if bacterial species exhibit similar, or universal, adsorptive behavior, existing models and relatively few modeling parameters could be used to describe real systems. Recent observations suggest that a number of individual bacterial species (Daughney et al., 1998; Small et al., 1999; Yee and Fein, 2001; Kulczycki et al., 2002; Ngwenya et al., 2003) and artificial mixtures of individual species (Yee and Fein, 2003) exhibit similar extents of metal adsorption (and can be described with similar proton and metal binding 6

19 constants). However, it is not known whether these individual species are representative of the bacteria found in realistic (non-laboratory) systems. In this study, we examine the properties of consortia of bacterial species, originating from a wide range of natural environments. I test whether these consortia adsorb metals to universal extents and to the same extents previously observed in experiments using individual, laboratory-cultivated bacterial species. One objective of this study is to use a generalized SCM approach to quantify the adsorption behavior of all the consortia (as a single unit) using a single set of thermodynamic surface stability constants rather than individual stability constants for each consortium. Chapter 4 of this dissertation examines whether bacterial consortia from highly-perturbed, contaminated environments exhibit universal adsorptive properties. Previous studies involving bacteria isolated from industrial wastes suggest that some bacterial species are indeed capable of enhanced metal adsorption, and that these bacteria can develop a specificity for adsorbing a given metal. For example, Wong et al. (1993) and Malekzadeh et al. (2001) show that specific bacteria isolated from electroplating effluents are capable of adsorbing elevated quantities of Cu and U, respectively. Chang et al. (1997) describe a bacterium isolated from hospital sewage that is capable of adsorbing large quantities of Hg and other metals, while Esposito et al. (2001) identify a bacterium isolated from a water purification plant that shows, good performance in heavy metal removal. These studies suggest that bacteria from contaminated sites may have evolved the capacity to adsorb unusually high metal concentrations, and may exhibit site concentrations and/or binding constants 7

20 that are significantly higher than for bacterial consortia from uncontaminated environments. Chapter 5 of this dissertation examines the ionic strength dependence of proton and metal adsorption onto bacterial surfaces. I conduct proton, Cd(II), Pb(II), and Sr(II) adsorption experiments over the ionic strength range of M to 0.6 M, using the Gram negative bacteria, Pseudomonas mendocina and Pseudomonas putida. The ionic strength of aqeuous geologic systems ranges from very low in some fresh surface and ground waters ( 0.01 m) to very high ( 0.5 m) in marine waters and basinal brines. Ionic strength has been show to greatly affect the extents of the adsorption of protons and metals onto mineral surfaces (Dyer et al., 2004). However, existing datasets are too sparse to definitively constrain these effects for bacterial surfaces or to determine the best way to account for them using thermodynamic models. In this study, we first determine the magnitude of bacterial surface electrostatic effects, and then attempt to determine the type of electrostatic corrections that must be incorporated into existing SCMs to describe these reactions in systems with differing salt concentrations. Chapter 6 of this dissertation breaks from the convention established in the previous chapters by examining the role of the bacterial surface in controlling important cellular processes. In this study, I specifically test the hypothesis that the adsorption of chemoeffectors to the bacterial cell wall influences bacterial chemotactic responses. Most bacteria are motile at some stage in their development, enabling them to reach optimal living conditions (Fenchel, 2002). Motile bacteria are often capable of utilizing concentration gradients of chemicals (i.e., oxygen, metals, 8

21 ph) in their environments to move away from or toward different stimuli through a process call chemotaxis. Metals like Fe(II), Ni(II), Cu(II), and Mn(II) have been show to act as chemoattractants or chemorepellents for some bacteria (Jerez, 2000; De Sanchez and Schiffrin, 1996; Childers et al., 2002; Seymour and Doetsch, 1973; Eitinger et al., 2000). Chemotactic responses to metals have important implications for carbon and nutrient cycles, bioremediation (Childers et al., 2002), and may play an important role in the attachment of bacteria to specific mineral surfaces (Sanchez and Schiffrin, 1996). The negatively charged functional group sites present on bacterial surfaces may play a critical role in triggering chemotactic responses to metals by greatly increasing the concentration of metals at the bacteria/solution interface. If such a model is valid, the chemotactic response of bacteria in geologic systems will vary substantially from controlled laboratory experiments, based on environmental factors such as ph, ionic strength, and competing binding agents. In this study, we measure the chemotactic response of Escherichia coli to Ni 2+ in experimental systems where the total concentration of Ni 2+ in solution is held constant, but the ph and the concentration of a competing binding agent (ethylenediaminetetraacetic acid, EDTA) are varied. Each of the first 4 studies (presented in Chapters 2-5) test important questions that must be addressed before a comprehensive modeling approach (with accurate modeling parameters) for bacterial surface-metal adsorption reactions in geologic systems can be developed. The final study (Chapter 6) illustrates the importance of being able to make these modeling predictions in geologic systems, and elucidates the role of adsorption in the important bacterial cellular processes of chemotaxis. As a 9

22 whole, this dissertation highlights the importance of bacterial surface adsorption reactions, and presents a framework for quantifying these reactions in real systems. 10

23 CHAPTER 2 THE EFFECT OF ACIDIC SOLUTIONS AND GROWTH CONDITONS ON THE ADSORPTIVE PROPERTIES OF BACTERIAL SURFACES 2.1 Introduction The surfaces of bacteria contain organic acid functional groups that adsorb metal cations over a wide ph range (e.g., Beveridge and Murray, 1976; Beveridge and Murray 1980). Because bacteria are abundant in virtually all near-surface geologic systems, bacterial adsorption can contribute to the overall fate and transport of metals (Tornabene and Edwards, 1972; Warren and Ferris, 1998; Ledin et al., 1999; Tortell et al., 1999). Models that are capable of quantifying bacteria-metal adsorption are critical for predicting the mobility of metals under conditions not directly studied in the laboratory. Recently, surface complexation models (SCMs) have been invoked to describe the adsorption of metals onto individual functional group sites on the bacterial cell wall (Xue et al., 1988; Plette et al., 1995; Plette et al., 1996; Fein et al., 1997; He and Tebo, 1998; Fein, 2000; Haas et al., 2001). This modeling approach requires identification of the bacterial surface complexes, determination of the thermodynamic stability constants of those complexes, and determination of the concentrations of binding sites on the bacterial surface. 11

24 To constrain these parameters, adsorption measurements must be collected over a wide ph range. However, some evidence suggests that exposure to relatively acidic conditions can alter the metal binding properties of bacteria. Chang et al. (1997) and Wong et al. (1993) observed an increase in adsorptive capacity in response to repeated acidification of the gram-negative bacterial species Pseudomonas aeruginosa and P. putida, respectively. In their studies, adsorbed metals (Cu, Cd, or Pb) were driven from the bacterial surfaces by using aliquots of 0.1 M HCl to decrease the ph of the bacteria-metal suspensions to about 2.0. An increase in adsorptive capacity was observed when bacteria were resuspended in stock solutions rich in Cd (in the Chang et al. [1997] study) or Cu (in the Wong et al. [1993] study). The authors attributed the increase in metal adsorption to structural changes caused by the addition of HCl. Other investigators have suggested that the concentrations of available nutrients during growth can affect the binding properties of bacterial surfaces (Cox et al., 1999). When nutrients are limited during growth, specific acidic functional group sites on bacterial surfaces may not develop. Nutrient levels can also induce changes in the size and shape of bacteria, but whether these changes affect the density of functional group sites on bacterial surfaces is unknown. Binding properties of facultative bacteria may be affected similarly by changing from an oxygen-rich to an anoxic environment during growth. The effect of acid on bacterial surfaces is important, because most experiments examining the sorption of metals by bacteria involve exposure of bacteria to acidic conditions. For example, Yee and Fein (2001) observed that a 12

25 number of gram-positive and gram-negative bacterial species exhibit similar extents of Cd adsorption. On the basis of this evidence they proposed that a universal set of thermodynamic stability constants could be used in an SCM to describe bacteriametal adsorption in natural systems, where the determination of individual species abundances is difficult. However, in their study gram-positive bacteria were washed with acid (ph 1.5), and gram-negative bacteria were exposed to acidic parent solutions (ph < 2.8) prior to experimentation. Changes in the extent of bacteria-metal adsorption due to acid treatment may also have important implications for modeling bacterial adsorption in acidic natural environments and for industrial biosorption applications. In addition, bacteria-metal adsorption models can be applied in natural settings only if differences in growth conditions typical of those between laboratory and natural systems do not affect the binding properties of the bacteria. Here, we report studies of bacteria-cd, -Co, and -Pb adsorption. Starting solutions covered a range of ph values, and bacteria were acid-washed and non-acidwashed gram-negative (P. mendocina and P. aeruginosa) and gram-positive (Bacillus subtilis, B. cereus) species. We also conducted bacterial acidification experiments with several of the same species, and measured dissolved organic carbon (DOC) for a number of the acidification experiment samples to assess structural damage to bacterial cell walls. Additional bacteria-co adsorption experiments were conducted with bacteria grown in nutrient-rich versus nutrient-limited solutions (P. fluorescens) and aerobic versus anaerobic growth conditions (Shewanella oneidensis MR-1). The objectives of this study were (1) to determine if and how exposure to acidic solutions affects the concentrations of functional groups on bacteria surfaces and (2) to 13

26 determine whether changes in nutrient and oxygen levels during growth of the selected bacteria affect their metal binding properties. 2.2 Materials and methods Growth Conditions Bacteria used in the experiments to test the effect of acid were initially cultured for 24 h at 32 C in 3 ml of trypticase soy broth (TSB) with 0.5% yeast extract, then transferred to 1 L of broth of the same composition and cultured for another 24 h at 32 C. The cells were removed from the nutrient medium by centrifugation and rinsed twice with 0.1 M NaClO 4 (the electrolyte used in the experiments). Acidwashed cells were prepared by suspension in 0.03 M HNO 3 (ph 1.5) for 30 minutes. The purpose of the acid wash step is to remove contaminant cations from the bacterial surfaces. Figures 2.1A and B show that B. subtilis and P. mendocina cells, respectively, remain intact after the acid-washing step. Three additional rinses with 0.1 M NaClO 4 were then completed (for both acid-washed and non-acid-washed bacteria). After the final wash, cells were transferred to weighed test tubes and centrifuged for 1 h, with 3 pauses to decant the remaining electrolyte solution. The final moist/wet weight of each bacterial pellet was calculated and then converted to a dry weight, which is reported here. Previous work has shown that the wet:dry weight ratio is approximately 5:1 (Borrok et al., 2004a), and that 1g/L dry weight corresponds to about cells/ml (Borrok and Fein, 2004). The effect of nutrient concentration during growth was tested by using P. fluorescens, cultured and washed as described above except that Luria broth (LB) and diluted (10%) LB replaced TSB 14

27 and the bacteria were grown at room temperature. Shewanella oneidensis MR-1 was grown aerobically in TSB broth (as described above) and also in an anaerobic chamber (in the presence of nitrogen) to test the effect on cellular adsorptive properties of oxygen content during growth. Cells grown in the anaerobic chamber were harvested after reaching early stationary growth phase (about 7 days). Both aerobic and anaerobic bacteria were washed 5 times with 0.1 M NaClO 4. A B Figure 2.1. Scanning Electron Microscope (SEM) images of acid-washed B. subtilis (A) and P. mendocina (B) cells. Photographs show clumps of intact cells that are similar in size and shape to cells that have not been acid-washed. The cells were prepared for imaging by suspension in 0.1 m NaClO 4 electrolyte. Minute aliquots of the suspension were allowed to dry on metallic stubs with no additional treatment. Images A and B were taken under ultra-high vacuum at 20 kv at 12,400 x and 10,280 x magnification, respectively Metal adsorption experiments All metal adsorption experiments were conducted in batch reaction vessels. Cadmium adsorption experiments were conducted with B. subtilis, B. cereus, P. aeruginosa, or P. mendocina at bacterial concentrations of 1.0 g/l. Additional Cd experiments were conducted with B. subtilis at 0.05 g/l and P. mendocina at 0.3 g/l. Cobalt adsorption 15

28 experiments were conducted with B. subtilis, P. fluorescens (LB and 10% LB), and S. oneidensis MR-1 (aerobic and anaerobic), all at 1.0 g/l. The Pb adsorption experiment used B. subtilis at 0.05 g/l. All experiments were conducted with the chosen metal at 10 ppm (diluted from a 1,000-ppm atomic absorption standard), except that the experiment with Cd and B. subtilis at 0.05 g/l used Cd at 5.4 ppm. The chosen concentrations of metal and bacteria are high to improve accuracy in our experiments; however, the bacteria/metal ratios were are similar to those found in some natural and industrial settings. Two types of stock solutions were made: (1) those in which the bacteria were added prior to ph adjustment (termed here acidic parent solutions or APS) and (2) those adjusted to ph 6 8 prior to the addition of the bacteria (termed here neutral parent solution or NPS). The ph of the APS stock solutions ranged from 2.6 to 2.8 after the addition of bacteria. Three types of experiments were conducted using B. subtilis and P. mendocina: (1) bacteria were washed with acid and suspended in an acidic parent solution ( Acid Wash ), (2) bacteria were not washed with acid but were suspended in an acidic parent solution ( APS ), and (3) bacteria were not washed with acid and were suspended in a neutral parent solution ( NPS ). Acid Wash and NPS (but not APS) experiments were performed with P. aeruginosa and B. cereus. The Co adsorption experiments using P. fluorescens and S. oneidensis MR-1 were conducted according to the APS procedure. Known masses of bacteria were suspended in the chosen parent solution and stirred gently until the distribution was homogeneous. Aliquots (approximately 10 ml) of the bacteria-metal-electrolyte suspensions were transferred into individual 16

29 batch reaction vessels. The ph of each batch experiment was adjusted to the desired value by using a small volume of concentrated NaOH or HNO 3. Each batch experiment was then allowed to equilibrate on a rotating rack for 2 h, and the final (equilibrium) ph was measured. Previous kinetics experiments showed that most bacterial adsorption reactions equilibrate within 30 min (Fein et al., 1997; Yee and Fein, 2001). Each sample was then centrifuged, filtered through a 0.45-µm nylon syringe filter membrane, acidified to prevent precipitation, and analyzed for the dissolved metal by using an inductively coupled plasma-atomic emission spectroscopy (ICP-AES) technique. Calibration standards were made with the same electrolyte used in the experiments. Analytical uncertainty based on comparison to standards was less than approximately ±2%. The decrease in aqueous metal concentration during the experiment was due to metal adsorption onto the bacterial cell wall. Control experiments in our laboratory have demonstrated that metal adsorption onto the experimental apparatus is negligible (data not shown) Acidification and dissolved organic carbon experiments Acidification experiments measured the concentration ranges of exchangeable metals associated with the bacterial surfaces after completion of the incubation and washing procedure. In these experiments, we suspended 1.0 g/l of non-acid-washed bacteria in a 250-mL Teflon reactor vessel with 0.1 M NaClO 4 electrolyte (and no added metals). The ph of the electrolyte-bacteria mixture was adjusted downward from its initial nearly neutral value by using small amounts of concentrated HNO 3. The suspension was stirred vigorously with a floating stir bar to ensure homogeneity. 17

30 After each downward desorption step, 10-mL aliquots of suspension were transferred to a separate reaction vessel. After equilibration for 2 h on a rotating rack, the final ph was measured, and the samples were filtered and analyzed for a range of dissolved metals. Initial analysis of the solutions was conducted by using an inductively coupled plasma-mass spectrometer (ICP-MS). Analytical uncertainty was less than approximately ±4% (from standards). On the basis of these results, additional analyses for Mg and Ca were performed using the ICP-AES technique described above. On several of the samples collected during the acidification experiments, DOC analyses were performed as a rough estimate of the extent of structural damage to the bacterial cell walls. Analyses were performed with a Shimadzu 5000 DOC analyzer, and calibration standards were made by diluting a potassium hydrogen phathalate standard containing 1,000 ppm carbon with the electrolyte from the experiments. Experimental error was ±3% (from standards). 2.3 Results The results of the Acid Wash, APS, and NPS experiments with B. subtilis and P. mendocina are presented in Figures 2.2 and 2.3, respectively. All three types of experiments for each bacterial species exhibit different adsorption edges. For each bacterial species, the adsorption edges corresponding to the Acid Wash experiments occur at the lowest ph values, while the adsorption edges associated with the NPS experiments occur at the highest ph values. For example, in the B. subtilis experiments, 50% Cd adsorption occurs at ph 4.0 for acid-washed bacteria, but for the APS and NPS experiments, 50% Cd adsorption occurs at ph 4.2 and ph 5.4, 18

31 respectively (Figure 2.2). The ph values at 50% Cd adsorption for the P. mendocina experiments are 4.2, 5.0, and 6.2, for the Acid Wash, APS, and NPS, experiments, respectively (Figure 2.3). However, with increasing ph, the extent of adsorption measured in the P. mendocina APS experiment approached the NPS results, and at the highest ph conditions studied, the NPS experiment yielded more adsorption than the APS experiment. This pattern was not observed in the B. subtilis experiments. The extent of Co adsorption onto B. subtilis in the Acid Wash and NPS experiments is presented in Figure 2.4. As observed in the Cd experiments, the Co adsorption edge generated with acid-washed bacteria occurred at a lower ph than the NPS adsorption edge. However, there was little difference between the Acid Wash and NPS results for Pb and Cd sorption onto B. subtilis at 0.05 g/l (Figures 2.5 and 2.6, respectively) Cd Adsorbed (%) APS Data NPS Data Acid Wash Data ph Figure 2.2. Cd adsorption data for Acid Wash, NPS, and APS experiments with B. subtilis at 1 g/l. Corresponding surface complexation model fits for Acid Wash and NPS data are shown with solid and dashed lines, respectively. 19

32 The results of the acidification and DOC experiments with B. subtilis and P. mendocina are presented in Figures 2.7 and 2.8, respectively. As ph conditions became acidic, Mg and Ca were liberated from both bacterial species. The DOC concentration in solution increased only slightly over the entire ph range for B. subtilis, but it increased dramatically with decreasing ph for P. mendocina Cd Adsorbed (%) APS Data NPS Data Acid Wash Data ph Figure 2.3. Cd adsorption data for Acid Wash, NPS, and APS experiments with P. mendocina at 0.3 g/l. Corresponding surface complexation model fits for Acid Wash and NPS data are shown with solid and dashed lines, respectively. The Co adsorption data for P. fluorescens grown in diluted LB medium match (within experimental uncertainty) the Co adsorption data for P. fluorescens grown in regular LB medium (Figure 2.9), suggesting that nutrient levels do not have a 20

33 significant effect on Co adsorption by this microbe. In addition, the Co adsorption data for anaerobically and aerobically grown S. oneidensis MR-1 demonstrate that oxygen content during growth does not significantly affect the extent of Co adsorption onto S. oneidensis MR-1 under the experimental conditions (Figure 2.10) Co Adsorbed (%) NPS Data Acid Wash Data ph Figure 2.4. Co adsorption data for Acid Wash and NPS experiments with B. subtilis at 1 g/l. Corresponding surface complexation model fits for Acid Wash and NPS data are shown with solid and dashed lines, respectively. Results from the additional Acid Wash and NPS experiments with Cd and B. subtilis, B. cereus, P. aeruginosa, and P. mendocina are presented in Figure In all cases the adsorption edges generated in the Acid Wash experiments occur at lower ph than do the adsorption edges for the NPS experiments. However, the difference is greatest for the gram-positive bacterium B. cereus. In the B. cereus 21

34 experiments, 50% Cd adsorption occurs at ph 3.7 in the Acid Wash experiment, but at ph 5.7 in the NPS experiment. At the highest ph studied, the results from the Acid Wash experiments with the gram-negative species P. aeruginosa and P. mendocina fail to reach the maximum extents of adsorption observed in the NPS experiments with these same species Pb Adsorbed (%) Acid Wash Data NPS Data ph Figure 2.5. Pb adsorption data for Acid Wash and NPS experiments with B. subtilis at 0.05 g/l and Pb at 10 ppm (48 mol/l). 22

35 NPS Data Acid Wash Data Cd Adsorbed (%) ph Figure 2.6. Cd adsorption data for Acid Wash and NPS experiments with B. subtilis at 0.05 g/l and Cd at 5.4 ppm (48 mol/l) Mg, Ca (ppm) Mg Ca DOC DOC (ppm) ph Figure 2.7. Concentration data and best-fit curves for Mg (short dashes), Ca (solid), and DOC (long dashes) as a function of ph, with B. subtilis at 1 g/l. 23

36 Mg, Ca (ppm) Mg Ca DOC DOC (ppm) ph Figure 2.8. Concentration data and best-fit curves for Mg (short dashes), Ca (solid), and DOC (long dashes) as a function of ph, with P. mendocina at 1 g/l Cd Adsorbed (%) LB Media 10% LB Media ph Figure 2.9. Co adsorption data for experiments with P. fluorescens (1 g/l) grown in full LB medium and 10% LB medium. 24

37 Co Adsorbed (%) Aerobic Anaerobic ph Figure Co adsorption data for experiments with S. oneidensis MR-1 (1 g/l) grown under aerobic and anaerobic conditions. 2.4 Discussion Effect of acid on the adsorptive properties of bacterial surfaces Bacillus subtilis and P. mendocina were subjected to acidic conditions in both the Acid Wash and APS experiments. The purpose of the Acid Wash treatment is to remove any adsorbed metals from the cell wall prior to introduction of the experimental metal of interest. Bacteria in the APS experiments were not washed with acid, but were subjected to acidic (ph 2.8) stock solutions. Hence, the difference between the Acid Wash and the APS experiments was that contaminant 25

38 cations removed through acid washing would be retained in the APS experimental systems. Differences between the APS and the Acid Wash experimental results suggest that the electrolyte rinses were insufficient to remove all of the metals adsorbed from the growth media. Competition for binding sites between the growth medium metals and the experimental metal (Cd) probably resulted in slightly less adsorption of Cd in the APS experiments than in the Acid Wash experiments (Figures 2.2 and 2.3). The acidification experiments demonstrate which metals (other than the experimental metal) were present in solution during the APS experiments. An initial ICP-MS multi-element scan indicated that only Mg and Ca were present at concentrations greater than 1 ppm. Subsequent acidification experiments suggested that maximum levels of approximately 3 ppm for Mg and 9 ppm for Ca were present in solution during the B. subtilis APS experiment (Figure 2.7), while maximum levels of approximately 6 ppm for Mg and 10 ppm for Ca were present in solution during the P. mendocina APS experiment (Figure 2.8). During the acidification experiments, DOC was monitored to determine whether the release of Mg and Ca was due to desorption of metal (not structurally bound) from the bacterial surfaces or to the possible structural breakdown of cells. Calcium and Mg are structural components within the cell walls of gram-positive bacteria and are the major constituent metals in the outer membranes of gramnegative species (Beveridge and Koval, 1981; Beveridge, 1989). The DOC level remained relatively constant over the entire ph range of the experiment with B. subtilis (Figure 2.6); however, DOC increased with decreasing ph in the 26

39 experiment with P. mendocina (Figure 2.8), suggesting that the acidic conditions caused some damage to the P. mendocina cells. Figures 2.1A and B show that B. subtilis and P. mendocina cells remain intact after exposure to acidic solutions, suggesting that cell damage did not result in cell lyses. However, damage to P. mendocina cells may have included the loss of proteins or lipopolysaccharides in the cells outer membrane brought about by the displacement of structurally-bound Mg and Ca (Beveridge and Koval, 1981). Under acidic conditions, the driving force of protonation appears to be great enough to displace structurally bound Mg and Ca from some bacteria. Unlike the adsorption edge generated by the B. subtilis APS experiment, the APS adsorption edge generated with P. mendocina crosses the NPS data at high ph. We attribute part of this Cd adsorption behavior to the added DOC in solution. DOC values increased by more than 200% in the acidification experiment with P. mendocina (Figure 2.8), and this DOC might compete with the bacterial cell wall in binding Cd. Hence, aqueous complexation of Cd with DOC decreases the extent of Cd adsorption onto the non-dissolved bacterial fraction. The effect of this competition is most noticeable at higher ph, where the DOC would be more negatively charged. This effect was observed only in experiments with gram-negative bacteria, suggesting that destruction of the outer membrane material is responsible for the release of DOC. This conclusion is consistent with the results of Beveridge (1989), who described the release of blebs of outer membrane material from gramnegative bacteria in response to displacement of structurally bound Mg and Ca. 27

40 With both B. subtilis and P. mendocina, the adsorption edges occurred at higher ph values (by 1-2 units) in the NPS experiments than in the APS experiments (Figures 2.2 and 2.3). The total concentrations of Mg + Ca available for competition with Cd were identical in NPS and APS experiments (because the Mg and Ca were not released and rinsed away prior to either experiment). Therefore, competition of Mg + Ca for binding sites cannot account for this shift. A similar shift occurs for the NPS adsorption edge relative to the Acid Wash adsorption edge for Co adsorption onto B. subtilis (Figure 2.4). Subjecting bacteria to acidic solutions appears to significantly increase the number of cell wall functional group sites that are available for metal binding. We further hypothesize that during the NPS experiments, Mg and Ca were structurally and irreversibly bound to specific functional groups, rendering these sites unable to adsorb Cd and shifting the adsorption edge to higher ph values. However, once the NPS solution was subjected to acidic conditions (for the experiments conducted at ph values less than approximately 4), Mg and Ca broke free from these functional groups, allowing the sites to compete freely to adsorb Cd. This explanation accounts for the confluence of the NPS, APS, and Acid Wash experimental data below approximately ph 4. These observations contradict the assumption that all metal adsorption reactions involving bacteria are fully reversible. However, adsorption reactions do appear to be fully reversible once structurally bound metals have broken free during acid washing or interaction with acidic solutions (Fowle and Fein, 2000). In other words, it is the initial breaking free of Mg + Ca that is an irreversible process. It is also possible that the loss of divalent cations from within the bacterial surface has an appreciable effect on the overall 28

41 electronegativity of the bacteria. A larger negative charge could result in increased affinity for metal binding, which would also be consistent with our results. In contrast to the results presented above, the Cd and Pb adsorption edges from the Acid Wash and NPS experiments with only B. subtilis at 0.05 g/l are not significantly different (Figures 2.5 and 2.6). We hypothesize that the metals in these experiments exerted the same effect on the bacterial cell wall as did protons in the previous experiments. In both experiments, an extremely high molar ratio of metal to bacteria was used (about 1 mmol/g). The surface sites available for binding metals became saturated at these high metal:bacteria ratios. This saturation apparently created conditions for enhanced adsorption, even in the absence of acidification, most likely by liberating Mg and Ca from the bacterial surface and thereby creating additional sites for metal uptake. However, spectroscopic evidence is clearly necessary to verify this proposed mechanism Effect of growth conditions on binding properties of bacterial surfaces The same bacterial species appear to adsorb similar extents of metals, regardless of the richness of their growth media (Figure 2.9) or (in the case of facultative bacteria) the availability of oxygen (Figure 2.10). Scanning electron microscope (SEM) images of P. fluorescens bacteria grown under nutrient-limited conditions (10% LB medium) showed that they retained their rod-shaped character, but were only a fraction of the length of P. fluorescens cells grown in full-strength LB. Despite this size difference, there was no significant difference in the extent of Co adsorption onto each bacterial population (Figure 2.9). Similar size differences were observed for S. 29

42 oneidensis MR-1 grown in aerobic vs. anaerobic conditions. However, despite these changes, both anaerobic and aerobic S. ondeidensis MR-1 adsorbed Co to similar extents (Figure 2.10). These preliminary results suggest that changes in growth conditions have little effect on the metal adsorption properties of bacterial surfaces Modeling of bacteria-metal adsorption reactions For comparison with previously published results and to serve as a basis for comparison to the NPS adsorption edges in this study, the Acid Wash experimental results were modeled by using an SCM approach. The NPS experimental results were modeled to develop a set of thermodynamic constants that reflect realistic (circumneutral ph) geologic conditions, as well as to quantify the changes in functional group site concentrations and adsorption equilibrium constants that result from acid exposure. The APS experimental results were not modeled, because the structurally bound cations released into solution during these experiments compete with the experimental metal for adsorption to functional group sites. The results from the two experiments conducted at high metal:bacteria ratios also were not modeled, because the Acid Wash and NPS results were nearly identical. The SCM approach uses mass action equations that describe the dominant chemical reactions taking place on the bacterial cell wall, in conjunction with mass balance constraints on the system, to define the concentrations of individual surface and aqueous species. We represent the deprotonation of a specific type of organic functional group on the bacterial surface with the following reaction: R - AH R - A - + H +, (1) 30

43 where R is the bacterium to which functional group type A is attached. The number of discrete surface binding sites A, the concentration of each of these sites, and their proton adsorption constants have previously been determined through modeling of potentiometric titration data. Fein et al. (2004) developed a four-site non-electrostatic model for proton binding onto B. subtilis with protonation constant (pk a ) values of 3.3, 4.8, 6.8, and 9.1. Borrok and Fein (2005) developed a similar four-site nonelectrostatic model for proton binding onto P. mendocina with pk a values of 3.4, 4.7, 6.5, and 9.3. To develop our model, we describe metal adsorption onto the bacterial surface as an interaction between deprotonated surface sites and aqueous metal cations, M m+, as follows: M m+ + R - A (-1) R - A(M) (m-1). (2) The corresponding mass action equation is K = [R - A(M) [a M m+ (m-1) ][R A ] ] (3) where K is the equilibrium constant for Reaction 2, the brackets represent the concentrations of functional group sites in mol/l, and a m+ is the activity of the aqueous metal cation. Values for the acidity constants (equilibrium constants for Reaction 1) and site concentrations for the bacterial surfaces used in these experiments are known (Borrok and Fein, 2005; Fein et al., 2004); therefore, if we measure the absolute concentrations of adsorbed and total metal at a particular ph, we can use the experimental data to constrain the value of the equilibrium constant for Reaction 2. M 31

44 Calculations were performed with the program FITEQL 2.0 (Westall, 1982), using a non-electrostatic model. A number of previous studies have employed constant capacitance electrostatic models to quantify the electrostatic surface effects (e.g., Fein et al., 1997; Daughney et al., 1998, Yee and Fein, 2001). However, as shown by Fein et al. (2004), the high capacitance value (8.0 F/m 2 ) used in these studies is nearly equivalent to using a non-electrostatic model. Hence, the modeling parameters developed in this study are comparable to previous studies. The best-fit model for the experimental data was determined by testing model fits for different combinations of bacterial surface complexes (R - A(M) (m-1) in Equation 2). The relative goodness of fit of each tested model was quantified by using the residual function V(Y) from the FITEQL 2.0 output (Westall, 1982). Because Acid Wash and NPS data points converge under acidic conditions (below about ph 4.0), we neglect data points in this range in our models to facilitate a valid comparison of the remaining data. Adsorption of metals onto two separate functional group sites provided a best fit to the B. subtilis and P. mendocina Acid Wash data. Models that excluded either one of these sites did not fit the data over the entire ph range. Because low ph data points were neglected, metal binding onto the lowest pk a site for each bacterium was not necessary. The binding stoichiometries for these models are compiled in Table 2.1, while the model fits are presented with the data in Figures 2.2, 2.3, and 2.4. The calculated metal adsorption constants for Cd (log K = 4.2) and Co (log K = 3.7) adsorption onto site 2 (pk a = 4.8) of B. subtilis compare favorably with those reported previously for similar experimental conditions (for Cd, log K = 4.0 [Yee and Fein, 2001]; for Co, log K = 3.5 [Fein et al., 2001]). The adsorption constant for site 2 (pk a 32

45 = 4.7) for Cd adsorption onto P. mendocina (log K = 4.4) is comparable to that for Cd binding onto B. subtilis. We used two approaches for modeling the NPS data. In one approach, we assumed that the values of the Cd adsorption constants were identical to those derived from the Acid Wash models, and we allowed the concentrations of functional group sites to vary to achieve the best fit to the data. This approach was used to determine the maximum number of functional group sites that become available for metal binding after structurally bound Mg and Ca are displaced by protons. Alternatively, the NPS data were also modeled by using the same reaction stoichiometries as in the Acid Wash model and the same site concentrations, but then solving for new Cd adsorption constants. This approach determined the hypothetical change in metal binding affinity caused by acid washing, under the assumption that no new functional group sites were created. This type of modeling cannot probe the exact mechanisms involved that lead to increased binding capacity in response to acid. Hence we cannot rule out either of the proposed mechanisms (more sites vs. increased binding affinity). Model fits for the cases where Cd and Co adsorption constants, as determined in the Acid Wash experiments, were held constant and site concentrations were allowed to vary are presented in Figures 2.2, 2.3, and 2.4. The concentrations of functional group sites increased in response to acid washing by 4 times and 5 times for Cd adsorption onto P. mendocina and B. subtilis, respectively. For Co adsorption onto B. subtilis, the increase in functional group site concentration was about 3 times. The concentrations of additional acid-activated sites (normalized to the mass of bacteria) created among these three examples averaged about mol/g. This 33

46 number is about 3 times the maximum concentrations of Mg and Ca released during acid washing (Figures 2.7 and 2.8). This finding may suggest that the structural metals are bound in multi-dentate fashion, with a number of functional groups attached to each divalent cation. Alternately, displacement of Mg and Ca by protons may not only increase the concentration of binding sites but also alter their overall affinity for binding. Clearly, spectroscopic measurements are required to distinguish between these possible end members. X-ray adsorption fine structure (XAFS) analysis of acid-washed and non-acid washed bacterial surface-metal complexes could be used to determine the nature of the acid-activated binding sites. If the acidactivated binding sites are chemically different than the original functional group binding sites, it would suggest they likely have different affinities for binding metals. TABLE 2.1. ADSORPTION CONSTANT VALUES FOR BEST-FIT SURFACE COMPLEXATION MODELS Bacterium and Experiment Metal Adsorption Constant (log K) B. subtilis-cd Site 1 (pk a 3.2) Site 2 (pk a 4.8) Site 3 (pk a 6.8) Site 4 (pk a 9.1) V(Y) Value Acid Wash NA NA 3.23 NPS NA NA 15.3 B. subtilis-co Acid Wash NA NA 30.5 NPS NA NA 2.54 P. mendocina-cd Site 1 (pk a 3.4) Site 2 (pk a 4.7) Site 3 (pk a 6.5) Site 4 (pk a 9.4) V(Y) Value Acid Wash NA NA 0.98 NPS NA NA 22.3 All adsorption constants are for monodentate metal adsorption reactions onto deprotonated functional group sites (reaction 2). NA = Not applicable for best-fit model. 34

47 When functional group site concentrations were held constant in the modeling, we obtained equally good fits to the data as when the functional group site concentrations were allowed to vary. As shown in Table 2.1, the calculated bestfitting Cd and Co adsorption constants in this modeling approach are significantly lower than those obtained though modeling of the corresponding Acid Wash data Implications for describing bacteria-metal adsorption in nature Potentiometric titration data demonstrate that most, if not all, bacterial species exhibit significant buffering capacity, even at very low ph (Borrok et al., 2004a; Fein et al., 2004; Martinez et al., 2002; Plette et al., 1995). Therefore, proton and metal adsorption data are needed at acidic ph to constrain the binding behavior of the functional groups that are proton-active at low ph. However, our study suggests that exposure to acidic ph conditions irreversibly alters the bacterial cell wall, increasing proton and metal binding capacity by displacing structurally bound Mg and Ca. Hence, experiments with bacteria in acidic conditions probably overestimate the extent of adsorption in natural systems, where bacteria grow at circumneutral ph. Experiments like those conducted by Yee and Fein (2001) to define the universal adsorption edge need to be revised if the results are to reflect the extent of metal adsorption under near-neutral conditions. Figure 2.11 depicts the results of NPS and Acid Wash experiments on a number of gram-positive and gram-negative bacteria, in conjunction with the universal adsorption edge determined under Acid Wash and APS experimental conditions by Yee and Fein (2001). Contrary to the universal adsorption behavior documented by Yee and Fein (2001) for acid-activated 35

48 bacteria, we found significant differences in the extent of adsorption for gramnegative and gram-positive bacteria under natural ph (NPS) conditions. However, the gram-positive bacteria appear to have similar maximum, or acid-activated, functional group site concentrations. As in our study, Mullen et al. (1989) showed that the gram-negative bacteria Escherichia coli and P. aeruginosa adsorbed more Cd than the gram-positive species B. subtilis and B. cereus. Our NPS results show that the gram-negative species tested here adsorb about 30% more Cd than do the grampositive species at ph 6.0. The differences in adsorptive capacity are probably due to fundamental differences in the structures of the bacterial cell walls (Beveridge and Koval, 1981). This gap in adsorption was obscured in the Yee and Fein (2001) study, because acid washing of the gram-positive bacteria increased their adsorptive capacity to a level resembling that of the gram-negative bacteria. At high ph, the acid-washed gram-negative bacteria in Figure 2.11 exhibited a loss of adsorptive capacity, in comparison to the NPS results. Apparently, the functional groups responsible for binding at high ph (perhaps the outer membrane material) were partly washed away during acid treatment, resulting in lower overall adsorption. The development of a true universal model that can describe bacteria-metal adsorption over the range of ph conditions in the environment will require a more mechanistic understanding of cell wall structural changes that occur in response to exposure to acidic conditions. For now, bacteria-metal adsorption in circumneutral aqueous settings seems to be estimated best by subtracting approximately 1 log unit from previously published Cd adsorption constant values (Yee and Fein, 2001) and approximately 0.5 log units from previously published Co adsorption constant values 36

49 (Fein et al., 2001). Metals with adsorption constants similar to those of Cd and Co would probably require similar corrections; however, whether adsorption constants for other metals with higher affinities for adsorption (e.g., Pb, Cu, Al) are affected similarly is not known. At extremely high metal:bacteria ratios the effect of acid appears to diminish (Figures 2.5 and 2.6), negating the need for correction Cd Adsorbed (%) ph P. mendocina NPS B. subtilis NPS B. cereus NPS P. aeruginosa NPS P. aeruginosa Acid Wash P. mendocina Acid Wash B. cereus Acid Wash B. subtilis Acid Wash Figure Cd adsorption data for Acid Wash and NPS experiments with B. subtilis, B. cereus, P. mendocina, or P. aeruginosa at 1 g/l. Data are compared to the universal adsorption edge (solid line) defined by Yee and Fein (2001). 2.5 Summary and conclusions When bacteria are exposed to acidic solutions, structurally bound Mg and Ca are displaced by protons, resulting in increased metal binding capacity for the bacteria 37

50 (Figures 2.2, 2.3, and 2.4). Modeling results demonstrate that acid washing of bacteria, inducing the release of Mg and Ca, may cause a 3 to 5 fold increase in the concentration of effective functional group sites. The cell wall of the gram-negative species P. mendocina appears to be disrupted by the displacement of Mg and Ca during acidification, resulting in an increase in the release of DOC. Displacement of Mg and Ca in the outer membrane has previously been shown to induce sloughing off of organic material (Beveridge, 1989). This organic material can compete with bacterial cell walls for the adsorption of metals. Minor changes in the size of P. fluorescens and S. oneidensis MR-1 were observed in response to changes in the concentrations of nutrients and oxygen, respectively, in the growth medium; however, metal binding properties of the bacteria appeared to be unaffected by these changes. Our study demonstrates that adjustments to bacteria-metal adsorption data generated under acidic conditions may be necessary if the results are to be applied to natural settings. Our results also suggest that treating bacteria with acid before they are used in remediation scenarios might significantly improve their performance as adsorbents for some metals. Preliminary work also indicates that the concentration of nutrients during growth of bacteria and the presence or absence of oxygen during growth of facultative bacteria have little effect on their metal binding properties. 38

51 CHAPTER 3 PROTON AND Cd ADSORPTION ONTO NATURAL BACTERIAL CONSORTIA: TESTING UNIVERSAL ADSORPTION BEHAVIOR 3.1 Introduction Bacteria are present in a wide range of environments, and the adsorption of metals onto bacterial surfaces can control the speciation and distribution of metals in many aquatic and near-surface systems (Ledin et al., 1996; Barnes and Nierzwicki-Bauer, 1997; Ledin et al., 1999). Accurate and quantitative models that describe bacteriametal adsorption are critical to understanding the availability and cycling of metals (Tornabene and Edwards, 1972; Tortell et al., 1999), for predicting the behavior of heavy metal contaminants (Volesky, 2001), and in the development of a number of contaminant remediation strategies (Macaskie and Basnakova et al., 1998; Espisito et al., 2001). Surface complexation models (SCMs), originally developed to describe the adsorption of metal ions onto mineral surfaces, have recently been applied to describe the adsorption of metals onto bacterial surfaces (Xue et al., 1988; Plette et al., 1995; Plette et al., 1996; Fein et al., 1997; He and Tebo, 1998; Fein, 2000). These models use a thermodynamic approach to describe metal adsorption onto bacterial surface functional groups through a series of mass action and mass balance equations. 39

52 One of the most problematic obstacles to the application of surface complexation modeling to realistic systems is that a given bacteria-bearing natural system can contain dozens of different bacterial species, and the number of species of environmental interest is huge and undetermined. If bacterial surfaces are unique and if each species exhibits unique adsorption properties, then it would be a Herculean task to determine the binding site concentrations and binding constants for each bacterial species of environmental interest. Experimentation on dozens of different bacterial species would be necessary just to describe the metal behavior in a single environment. A potential solution to this dilemma arises from the recent observations that a number of bacterial species exhibit similar extents of metal adsorption (and similar proton and metal binding constants), as determined in laboratory experiments using individual pure strains of bacteria (Daughney et al., 1998; Small et al., 1999; Yee and Fein, 2001; Kulczycki et al., 2002; Ngwenya et al., 2003), and artificial mixtures of pure strains of bacteria (Yee and Fein, 2003). In this study, we test the hypothesis that consortia of bacteria, originating from a wide range of soil and aquatic environments, adsorb metals to similar extents and that the adsorption behavior of all the consortia can be described with a single set of universal thermodynamic stability constants rather than individual stability constants for each species present. 3.2 Materials and Methods Sampling and Growth of Bacteria 40

53 Soil and/or water samples were collected from Northern Indiana and Southern Michigan. Sample locations included a forest, a natural wetland, a river, and a wastewater treatment facility. All materials used for sample collection, including glass jars, lids, and digging scoops were sterilized and sealed in plastic bags prior to use. Soil samples were collected by removing the top 3 to 5 inches of topsoil and debris and then directly scooping a soil specimen using a glass sample jar. Water samples were collected by dipping the sample jar directly into the water. In the case of the wetland water sample, water was collected from a near-surface piezometer using disposable plastic tubing. Immediately after collection, sample lids were loosely placed over the jars to allow for aerobic conditions while preventing contamination. Control growth experiments were completed by briefly opening jars of sterilized growth media near sample locations and exposing them to the atmosphere. Lack of bacterial growth in the control samples suggests that the bacterial consortia we produced were most likely not contaminated during handling with species not present in the original sample. Approximately 10 grams of soil or 10 ml of water from the samples were used to inoculate the chosen broth solution. Bacteria were grown using either trypticase soy broth (TSB) with 0.5% yeast extract or soil broth (SB). SB was made with 250 ml soil extract, 2 g glucose, 1 g yeast extract, and de-ionized (DI) water (to 1.0 L). Soil extract was made by autoclaving 500 g of soil with 1 L of DI water for 1 hr at 15 psi/121 C and decanting the fluid product using Whatman No. 2 filter paper. Inoculated broth solutions were gently shaken at room temperature until they were harvested or a portion was used for re-inoculation. Approximately 10 ml of the initial 41

54 bacteria-broth suspension was used to inoculate larger quantities of identical broth solutions to dilute out the solid fraction present from soil samples. The number of reinoculations prior to harvest, the broth mixture used, and the growth duration for each consortium are presented in Table 3.1. Matrix Location Date Collected Soil St. Mary s Forest #1 N. Indiana Water Wastewater Treatment Plant (Final Clarifier) N. Indiana Soil St. Mary s Forest #2 N. Indiana Water St. Joseph River N. Indiana Water Wetland S. Michigan Soil Wetland S. Michigan TSB = Trypticase Soy Broth. SB = Soil Broth TABLE 3.1 SAMPLE GROWTH INFORMATION Growth Media Number of Inoculations Duration of Growth for Each Inoculation (Days) 1/8/03 SB 3 7 2/5/03 TSB SB 1 8 2/4/03 SB 2 6 2/16/03 SB 1 8 3/8/03 TSB 1 3 SB 1 7 3/8/03 SB Because a particular set of growth conditions will support growth of only particular types of bacterial species, the bacteria grown for our experiments probably included only a subset of the total bacterial population present in each environment sampled. Experiments have shown that many of the individual bacterial species within natural consortia cannot survive repeated inoculations in laboratory growth media (Kaeberlein et al., 2002). However, by growing the bacteria directly in broth solutions and limiting the number of re-inoculations, we hoped to achieve a range of diverse consortia that is at least representative of the diversity that exists in nature. 42

55 Multiple experiments using the same samples, but grown from different media (with different growth durations) were completed to avoid some of the inherent bias in culturing the bacteria (Table 3.1), however all growth conditions were aerobic so all anaerobes were eliminated through the growth procedures. As a preliminary indication of diversity, Gram staining was conducted on most of the consortia used in the experiments. Diversity was more precisely determined through denaturing gradient gel electrophoresis (DGGE) Freezing of Bacteria Prior to each re-inoculation and/or harvest for experimentation, bacterial cells suspended in their broth solutions were frozen for future DNA extraction and/or reuse. Cells frozen for re-use were suspended in a mixture of 10% sterilized glycerol and flash frozen with liquid nitrogen prior to storage in a -80 C freezer. Cells frozen for DNA extraction were not mixed with additional agents and were placed directly into the -80 C freezer Potentiometric Titration and Cd Adsorption Experiments Bacteria were harvested from the growth media by centrifugation, transferred to test tubes, and washed five times in 0.1M NaClO 4. Sodium perchlorate was chosen as the experimental electrolyte because perchlorate does not form complexes to an appreciable extent with protons or Cd under the experimental conditions. During each wash, the bacteria were suspended in fresh electrolyte solution using a vortex machine and stir rod. Bacteria were centrifuged for 5 minutes at 8000 rpm (7150 g) 43

56 to form a pellet at the base of the test tube and the electrolyte was discarded. After the final wash, the bacteria were placed in weighed test tubes and centrifuged (7000 rpm [5500 g] at 25 C) for 1 hour, stopping three times to decant all supernatant. After 1 hour, the weight of the moist bacterial pellet was determined. The weight recorded at this stage and reported here and throughout this paper is the wet weight. Previous work in our laboratory has shown that this wet weight is a reproducible quantity, but changes over a modest range depending upon the bacterial species weighed. The wet weight to dry weight ratio ranged from 3.1:1 to 6.5:1 and averaged 5:1 in individual experiments using Pseudomonas fluorescens, Shewenella oneidensis, Staphyloccocs aureus, Bacillus subtilis and Bacillus cereus. The bacteria pellet was immediately used in potentiometric titrations or in Cd adsorption experiments. Although the bacterial cells remain viable after this treatment (Fein et al., 1997), they are not expected to be metabolizing during experiments because of the lack of nutrients and short (<3 hr) experimentation times. In each metal adsorption experiment, 1g/L (dry weight) of a bacterial consortium was suspended in a ph-neutralized stock solution of 0.1 M NaClO 4 and 10 ppm Cd. After an initial 10-minute equilibration time, the bacterial stock solution was divided into individual reaction vessels. The ph of each reaction vessel was adjusted by adding minute aliquots ( 0.05 ml) of 1.0 M or 0.1 M HNO 3 or NaOH. After adjustment of the ph, and an additional 2 hours of reaction time on a rotating rack, the final (equilibrium) ph of each vessel was measured. Previous studies in our laboratory have demonstrated that equilibrium of the adsorption reaction generally occurs in less than 1 hour and that the adsorption reaction is fully reversible (see 44

57 Fowle and Fein, 2000). The individual vessels were then centrifuged and the resultant supernatant filtered (0.45 µm). The filtered supernatant was analyzed for Cd using an inductively coupled plasma atomic emission spectroscopy technique. Standards were made using the same electrolyte as in the experiments, and uncertainties were within ± 2% of external standard measurements. The concentration of metal adsorbed to bacteria in each vessel was calculated by subtracting the concentration of metal that remained in solution (supernatant) from the original 10 ppm in the stock solution. Cells used in titration experiments were unfrozen from the frozen stock and grown in the same broth solution used for the metal adsorption experiments. Titrations were not completed on all consortia for all growth conditions, but were completed for a single consortium from each environment using one growth medium. Cells frozen from the next to final inoculation used in the metal adsorption experiments were unfrozen and inoculated/cultured for use in the titration experiments. Hence the final growth times and number of inoculations remained the same for each consortia for each type of experiment. Several duplicate Cd adsorption experiments (data not shown) were performed using bacterial consortia grown from the frozen stock to test if the freezing process affected the observed adsorption behavior. The results using the frozen cells were indistinguishable from those using cells prior to the freezing step, indicating that the freezing process had no measurable effect on the experimental results. Minor variations in the diversity of the bacterial consortia in response to the freezing process cannot be ruled out. However, for this study, we determined that potential minor variations in diversity caused by deep- 45

58 freezing were preferable to the large changes in diversity expected from the alternatives of either growing the consortia for longer periods of time or re-sampling and re-growth of the consortia prior to each type of experiment conducted here. The cells were harvested from the growth media and washed as described above; however, for the titrations, they were suspended in approximately 10 ml of 0.1 M NaClO 4 that had been purged of CO 2 by N 2 bubbling for 60 min. The suspension was immediately placed into a sealed titration vessel maintained under a positive pressure of N 2. Titrations were conducted using an automated burette assembly with aliquots of N NaOH and/or N HNO DGGE Analysis DNA was extracted directly from the cell suspensions in the growth media using a MoBio Laboratories, Inc. Ultraclean soil DNA kit. Extracted DNA was frozen at - 20 C prior to amplification. A custom-made universal bacterial primer set (EUB 341 and EUB 534, 200 base pairs in length with a GC-clamp) was used during the polymerase chain reaction (PCR) process to ensure that only the DNA extracted from bacteria would be amplified. However, this approach cannot guarantee that DNA from every bacterial species in the sample becomes amplified. DGGE was carried out using a Dcode universal mutation detection system (Bio-Rad). The PCR product was loaded into a gel with a 30% to 60% gradient of a chemical denaturant (urea and formamide). The DNA was forced to travel through the denaturant in response to an electrical potential. Gels were run at 60 C and a potential of 60 V for 14 hours. The gel was then stained using ethidium bromide, fluoresced, and photographed using a 46

59 Kodak EDAS 290 photographic system. Because of the differences in DNA base pair sequences among different bacterial species, the DNA from a specific species will denature at a specific point in the gel, forming a characteristic band. The intensity of these bands is directly related to the concentration of DNA present in the sample. Hence, analysis of the band positions and intensities can determine the minimum number of bacterial species present, and their relative abundances. The position and relative intensities of these bands were analyzed using Kodak ID 3.6 software. Although this is a sufficient test for diversity in our study, additional sequencing of the bands would be necessary to determine the identity of each band and to verify that each band represents only one species. 3.3 Results and Discussion Gram-staining revealed both Gram-positive and Gram-negative bacterial species in all consortia examined. DGGE results indicated that most of the consortia displayed three to six bands (species) and that there was little or no similarity of bands from one sampled environment to the next (Table 3.2, Fig. 3.1). Two of the consortia displayed only two bands, while one forest soil consortium contained only one band. One of the wetland water samples displayed only one very faint band, indicating either that something other than bacteria dominated this consortium, or, more likely, that the DNA extracted from this sample was not successfully amplified using the PCR technique and chosen primers. Gram staining of this consortium identified numerous rod-shaped Gram-negative bacteria and much fewer oval-shaped Grampositive bacteria. 47

60 Figure 3.1. Example DGGE gel with numbered lanes. 1 Control, E. Coli; 2-4 Wetland Soil, SB; 5-7 Wetland Water, SB; 8-10 Wastewater, SB; Forest Soil #1, SB. TSB = Trypticase Soy Broth. SB = Soil Broth. TABLE 3.2 RELATIVE BAND INTENSITIES IN INDIVIDUAL LANES OF DGGE GELS Number of Bands 3 St. Mary s Forest #2, SB 1 Wetland Soil, SB 1 Wastewater Effluent, SB 2 Wastewater Effluent, TSB 2 River Water, SB 2 Wetland Water, TSB 1 11% 5% ± 1.0% 2% ± 0.3% 31% ± 5.5% 3% ± 0.4% 72% ± 4.5% 2 26% 2% ± 0.6% 27% ± 0.2% 69% ± 5.5% 15% ± 0.3% 28% ± 4.5% 3 18% 15% ± 1.8% 52% ± 1.7% 18% ± 2.0% 4 31% 11% ± 0.4% 19% ± 1.9% 12% ± 1.5% 5 14% 22% ± 0.4% 18% ± 1.0% 6 45% ± 3.1% 34% ± 0.5% The band intensity is a proxy for the abundance of a specific bacterial species within each consortia. Statistics are based upon 1 3 replicates (1 uncertainties), 2 2 replicates (maximum difference), or 3 no replicates were used. The bands for each consortium are unique and are not comparable to bands in other consortia. Samples displaying a single band are not shown. TSB = Trypticase Soy Broth. SB = Soil Broth 48

61 All the consortia exhibited significant buffering capacity over the entire ph range studied (2.5 to 9.7). Representative titration data from forest soil consortium #2 are presented in Fig The titration curves for all of the consortia were similar in shape to each other and to those determined previously for a wide range of individual bacterial species (Fein et al., 1997; Yee and Fein, 2001). The extent of Cd adsorption onto each bacterial consortium studied here increases with increasing ph (Fig. 3.3), defining a ph adsorption edge. Data collected from experiments using consortia of bacteria from each tested site form a relatively narrow band, demonstrating that the experimental consortia adsorb metals to similar extents (Fig. 3.3). Changing the growth media and duration of growth for individual samples had no discernable effect on the relative positions of the adsorption edges, despite the fact that the bacterial speciation within the consortia appeared to change (Table 3.2) Surface Complexation Modeling The similar adsorption behaviors that we observed for each of the consortia in this study suggest that it may be possible to model metal adsorption onto bacterial consortia with a single universal set of thermodynamic properties. We model the observed adsorption behavior using a discrete site SCM approach. This model of cell wall adsorption is a simplification of the mechanisms involved in what is likely an extremely complex and heterogeneous chemical system. However, discrete site surface complexation models offer a number of advantages. For example, they are relatively simple, yet they can describe adsorption behavior over a wide range of chemical conditions, and they are relatively easy to adapt for use in many 49

62 contaminant transport codes. We use the SCM approach to explicitly account for cell wall adsorption reactions, and we use the experimental data to constrain a system of mass action and mass balance equations to solve for equilibrium constants and/or concentrations of individual chemical species (Fein et al., 1997). We represent bacterial surface functional groups using a limited number of discrete site types, each of which undergoes deprotonation according to the following reaction: (1) R-AH R-A - + H + Where R is the bacterium to which the functional group type, A, is attached. The acidity constant, K a, for reaction (1) can be expressed as: (2) K a = - [R - A ] a H [R - AH ] + Where [R-A - ] and [R-AH ] represent the concentration of deprotonated and protonated sites, respectively, and a + represents the activity of protons in the bulk H solution. The titration data were modeled to determine functional group site concentrations and acidity constant values. These parameters were calculated using the chemical equilibrium program FITEQL 2.0 (Westall, 1982). We chose to neglect the effects of the surface electric field on the adsorption reactions (non-electrostatic model) for two reasons. Because all the experiments were conducted at a single ionic 50

63 strength value (0.1 m) there is no way to uniquely constrain a type of surface electric field model that best accounts for the experimental data. Also, in order to implement an electrostatic model, the surface area of the bacteria used in the experiments must be calculated. Each consortium used in these experiments contained a variety of bacterial species of different shapes and sizes. Therefore, the overall surface area could not be calculated with any certainty. In our modeling approach, we determine the minimum number of discrete functional group types that are required to account for the observed buffering capacity of each consortium by sequentially testing models with one through five protonactive sites. In each case, the four site models yield significantly better fits than models with fewer sites. The five site models in each case do not converge, indicating that the data cannot constrain a model with 5 discrete functional group types. An example model fit for the titration data generated from the forest sample #2 consortium is presented in Fig. 3.2, and the average functional group site concentrations and acidity constant values are compiled in Table 3.3. We refer to the functional group sites with pk a values of 3.12, 4.70, 6.57, and 8.99 as Sites 1 through 4, respectively. The limited chemical (Beveridge and Murray, 1980; Beveridge and Fyfe, 1985) and spectroscopic information (Boyanov et al., 2002; Kelly et al., 2002) available for pure strains of bacteria suggests that the majority of these sites are likely phosphoryl and carboxyl. However, additional spectroscopic information is necessary to chemically identify the sites present in each consortium. The calculated site concentrations and acidity constants are similar for all the consortia studied, despite the diversity represented in the different consortia. The 1 51

64 Total Protons Added (m) ph Figure 3.2. Potentiometric titration data (solid circles), and best-fit model using 4 surface sites (curve), for the forest soil #2 consortium. Values are positive for net acid added; negative for net base added. 52

65 TABLE 3.3 SITE CONCENTRATIONS AND PROTON BINDING CONSTANTS (K a ) with 1 UNCERTAINTIES, FOR THE FOUR FUNCTIONAL GROUP SITES IDENTIFIED Consortia Wetland Soil, TSB Wastewater Effluent, TSB Forest Soil # 2, SB River, SB a Wetland Water, TSB Average 3.12 ± 0.13 THROUGH TRIPLICATE POTENTIOMETRIC TITRATIONS Proton Binding Constants (-log K a ) Site Concentrations (x 10-5 moles/g wet weight) Site 1 Site 2 Site 3 Site 4 Site 1 Site 2 Site 3 Site ± 4.62± 6.62 ± 9.18± 7.06± 5.70 ± 2.69 ± 3.90 ± ± 4.74 ± 6.46 ± 8.94 ± 8.06 ± 9.79 ± 4.96± 4.39 ± ± 4.74 ± 6.65 ± 9.11 ± 7.61 ± 6.94 ± 4.62 ± 6.56 ± ± 4.74 ± 6.67 ± 8.74 ± 3.34 ± 3.32 ± 1.45 ± 1.70 ± ± ± ± ± ± ± 1.71 a Only one of the three titrations was performed to a low enough ph to constrain fit. TSB = Tripticase Soy Broth. SB = Soil Broth. Constants are based on equation 2, K a = [R - A - ]a + H [R - AH ] 4.47 ± 2.19 uncertainties associated with the average negative logarithm of the acidity constant (pk a ) value for each functional group site (collected over all titrations) range from 0.21 (Site 4) to 0.11 (Site 2). Analogous 1 uncertainties for functional group acidity constant values that were determined from similar experiments using individual bacterial species range from 0.45 to 0.20 (Yee and Fein, 2001), suggesting that the uncertainties determined in this study may arise more due to experimental variations rather than from real systematic variations between the different consortia tested. The 1 uncertainties (0.17 to 0.27 on a log scale) for each of the four functional group site concentrations are slightly greater, suggesting there may be more variability in site concentrations than acidity constants for bacteria in nature. To a good first 53

66 approximation, a single set of averaged site concentrations and acidity constants can describe the buffering behavior of all the consortia studied here. We use FITEQL 2.0, along with the average acidity constant and site concentration values derived through modeling of the titration data, to determine the cadmium binding constants from the Cd adsorption data. We determine the minimum number of types of binding sites that are required to account for the observed Cd adsorption, and solve for the values of the corresponding equilibrium constants for the adsorption reactions. The speciation of Cd in the system was modeled using the equilibrium constant values from Baes and Mesmer (1976) for aqueous Cd-hydroxide species. We tested a range of adsorption reaction stoichiometries, with Cd adsorption onto the bacterial surfaces modeled as a reaction between Cd +2 and a protonated (x = 1) or deprotonated (x = 0) surface site (Sites 1 to 4, both individually and in combination, were tested in each case) to form a bacterial surface complex, according to the reaction: (3) Cd +2 + R-AH x (x-1) R-AH x (Cd) (x+1) with an equilibrium constant, K, for this reaction given by: (4) K = [R - AH a Cd +2 x (Cd) [R - AH (x+ 1) (x-1) x ] ] The modeling results are compiled in Table 3.4. The best fitting adsorption model for all but one consortium involves mono-dentate adsorption of Cd +2 onto the 54

67 deprotonated form of each of the four functional group sites. Adsorption experiments using the wastewater effluent (soil broth) consortium did not cover a large enough ph range to constrain Cd binding onto all 4 sites. In this case, the adsorption behavior could be accounted for using only 3 of the 4 surface complexes; however, the calculated values of the binding constants for these complexes are similar to those obtained from the experiments that required all four surface complexes. Universal Cd adsorption constants were developed for each functional group site by averaging the binding constants obtained for each site, weighting the values based on the number of data points used to constrain each individual binding constant. The average site concentration and binding constant values yield a universal SCM that predicts an adsorption edge that is in reasonable agreement with the observed Cd adsorption behavior over the entire ph range of the experiments (Fig. 3.3). Clearly, these constants alone are not capable of determining the distribution of Cd among the many important binding ligands (e.g., bacterial surfaces, exopolysaccharide material, mineral surfaces, dissolved organic matter, inorganic ligands, etc.) in natural systems. Additional equilibrium constants for metal complexation with other environmentally relevant ligands are necessary in order for our results to be used to estimate metal distributions in realistic bacteria-bearing systems. 55

68 Cd Adsorbed (%) ph Figure 3.3. Cd adsorption onto bacterial consortia cultured from soil and aquatic environments. All Cd adsorption experiments were conducted using 10 ppm Cd and 10 g/l consortia (wet weight) in a suspension of 0.1 M NaClO 4. The consortia were grown using soil broth (SB) or trypticase soy broth (TSB) with 0.5% yeast extract. = Forest Soil #1, SB, = Wastewater Effluent, TSB, = Wastewater Effluent, SB, = Forest Soil #2, SB, = River Water, SB, = Wetland Water, TSB, = Wetland Soil, SB, + = Wetland Water, SB. The solid line represents the predicted adsorption edge that is calculated using the universal functional group site concentrations and binding constants developed in this study. 56

69 TABLE 3.4 Cd METAL BINDING CONSTANTS (K) FOR BEST-FIT ADSORPTION MODELS Consortia Cd binding Constants (log K) Site 1 (3.12 pk a ) Site 2 (4.70 pk a ) Site 3 (6.57 pk a ) Site 4 (8.99 pk a ) Forest Soil #1, SB Forest Soil #2, SB Wastewater Effluent, SB Wastewater Effluent, TSB Wetland Soil, SB Wetland Water, SB Wetland Water, TSB River, SB Weighted Average 2.83 ± ± ± ± 0.40 Functional group sites are those identified through potentiometric titrations, and weighted averages and 1 errors are based on the number of data points available to constrain each constant. TSB = Tripticase Soy Broth. SB = Soil Broth. Constants are based on equation 4, K = [R - A(Cd) a Cd + 2 [ R - A ] ] The bacterial consortia that we produced in this study are not exact or complete representations of the consortia of bacteria that exist in the locations that we sampled. Although the DGGE results indicate that the consortia were quite diverse and distinct from each other, their constituent species were only a subset of the ones present in the natural systems. However, the similar Cd adsorption behavior that we observed for the bacterial consortia in this study suggests that actual consortia from a wide range of natural systems adsorb metals to a similar extent as well. We hypothesize that the stability constants determined here, in conjunction with other relevant stability constants, can be used to predict the extent of Cd adsorption onto bacterial consortia that are found in a wide range of environments. This hypothesis assumes that the bacteria/metal adsorption behavior is linear over a wide range of 57

70 bacteria/metal ratios. However, the dominant adsorption mechanism may be different under extremely high or low metal loading conditions. If so, the equilibrium constants for these reactions must be determined in order to model the effect of bacterial adsorption on metal speciation in these systems. Although our data indicate that a wide range of bacterial consortia exhibit similar Cd adsorption behaviors, Fig. 3.4 illustrates that the consortia adsorb significantly less metal than do equivalent amounts of individual pure laboratory strains of bacterial species cultured under identical conditions, including Bacillus subtilis, Bacillus cereus, Pseudomonas mendocina, Pseudomonas aeruginosa, and Pseudomonas putida (Borrok et al., 2004b). The reason for the gap in adsorption behavior between pure strains of bacteria and consortia of bacteria cultured from natural environments is uncertain; however, it may be due in part to the number of growth cycles experienced by the bacteria in nutrient-rich laboratory media. The pure strains of bacteria used for the experiments shown in Fig. 3.4 were grown and replated on nutrient-rich media numerous times over a period of months or years, while the consortia of bacteria were grown in nutrient-rich broth solutions for only a few life cycles. The average acidity constants for the consortia from this study are nearly identical to the pk a constants (3.3, 4.8, 6.8, and 9.1 for Sites 1-4, respectively) derived by Fein et al. (2003) to describe proton binding onto the Gram-positive bacterium, B. subtilis, also using a non-electrostatic approach. However, the functional group site concentrations for B. subtilis (8.1 x 10-5, 1.1 x 10-4, 4.4 x 10-5, and 7.4 x 10-5 moles/g wet weight for Sites 1-4, respectively) are higher than those for 58

71 the bacterial consortia for each of the four functional group site types. Overall, the laboratory-cultured pure strain of B. subtilis has about 40% more sites available for metal binding per wet gram of bacteria than the consortia tested in this study. This observation suggests that the proton binding constants for individual bacterial species and for bacterial consortia may be the same. The difference between the adsorption behaviors observed for natural consortia and for individual bacterial species appears to be caused by the lower site density associated with the bacterial consortia rather than by a lower tendency for a proton to bind with each functional group Cd Adsorbed (%) ph Figure 3.4. Comparison of the natural adsorption edge (solid curve) to Cd adsorption data from experiments using individual, pure strains of bacteria. = Pseudomonas mendocina, = Pseudomonas putida, = Bacillus cereus, = Bacillus, subtilis, = Pseudomonas aeruginosa. All Cd adsorption experiments were conducted using 10 ppm Cd and 10 g/l consortia (wet weight) in a suspension of 0.1 M NaClO 4. 59

72 3.4 Conclusions This study demonstrates for the first time that diverse groups of bacteria, cultured from a variety of soil and aquatic environments, adsorb metals to similar extents. This observation could greatly simplify the task of modeling bacteria-metal adsorption in nature. Using the assumption of universality of metal adsorption behavior, along with the equilibrium constants and site concentrations determined in this study (and the relevant constants for competing metals and binding ligands), the effects of bacterial adsorption on the distribution and speciation of Cd in realistic geologic systems can be modeled relatively easily. The behavior of other metals of interest could be predicted based on relatively few laboratory experiments, or on approaches that enable prediction of stability constants for metals onto bacterial surfaces (Fein et al., 2001). Many of the relevant metal-binding constants and site densities for other important binding ligands such as mineral surfaces (Koretsky et al., 1998; Sahali and Sverjensky; 1998) and humic and fulvic acids (Milne et al., 2001; Milne et al., 2003) already exist. Our results suggest that reasonable estimates of the extent of bacterial adsorption in natural environments can be obtained simply by determining the absolute concentration of all bacteria in the system, rather than by having to characterize the adsorption behavior, the distribution, and the concentration of each bacterial species present. Our results also suggest that estimations of the extent of metal adsorption based on pure cultures of individual bacterial species may be higher than the adsorption that occurs onto natural bacterial consortia. The repeated re- 60

73 generation (over months or years) of single strains of bacteria using a high-nutrient growth medium may influence the density of functional group sites on their surfaces. In contrast, our approach of measuring adsorption onto consortia that have been grown under laboratory conditions for only a few generations yields thermodynamic parameters that are likely to better reflect the adsorption behavior of bacterial consortia under natural conditions. 61

74 CHAPTER 4 Cd AND PROTON ADSORPTION ONTO BACTERIAL CONSORTIA GROWN FROM INDUSTRIAL WASTES AND CONTAMINATED GEOLOGIC SETTINGS 4.1 Introduction Determining the speciation of metals in both natural and engineered settings is critical for predicting the mobility of the metals and their impact on the environment. Because metal cations readily adsorb to bacterial surfaces, bacteria have been recognized as important metal-complexing agents (Mullen et al., 1989; Ledin et al., 1996; Barns and Nierzwicki-Bauer, 1997; Ledin et al., 1999), and various models have been proposed to account for metal adsorption onto a number of individual bacterial species (Cox et al., 1995; Plette et al., 1996; Fein et al., 1997; Martinez et al., 2002). It is unclear whether these models can be accurately applied to more realistic settings that can potentially contain dozens of different bacterial species. If the cell wall functional group sites of each bacterial species exhibit unique adsorption properties, it would be necessary to determine site densities and binding constants for each site on each bacterial species of interest. This would be an impractical task because experimentation on dozens of different bacterial species (many of which cannot even be isolated in the laboratory) would be necessary. However, if bacterial 62

75 species exhibit similar, or universal, adsorptive behavior, existing models and modeling parameters could be extrapolated more easily to describe complex systems. Recent observations by a number of researchers working with individual pure strains of bacteria (Daughney et al., 1998; Small et al., 1999; Yee and Fein, 2001; Kulczycki et al., 2002; Ngwenya et al., 2003), and artificial mixtures of pure strains of bacteria (Yee and Fein, 2003), suggest that to a first approximation, bacteria do exhibit universal adsorptive behavior. This universality of adsorptive behavior was tested by Borrok et al. (2004a) using consortia of bacteria grown from a range of uncontaminated soil and water environments. Borrok et al. (2004a) found that these consortia of bacterial species exhibit roughly similar affinities for protons and Cd, although the extents of adsorption were less than what had been measured in some previous studies using individual pure strains of bacteria. Other studies involving bacteria isolated from contaminated industrial wastes suggest that some bacterial species are capable of enhanced metal adsorption, and that these bacteria can develop a specificity for adsorbing a given metal. For example, Wong et al. (1993) and Malekzadeh et al. (2002) show that specific bacteria isolated from metal-rich electroplating effluents are capable of adsorbing elevated quantities of Cu and U, respectively. Chang et al. (1997) describe a bacterium isolated from hospital sewage that is capable of adsorbing large quantities of Hg and other metals, while Esposito et al. (2001) identify a bacterium isolated from a water purification plant that shows, good performance in heavy metal removal. These studies suggest that bacteria from contaminated sites may have evolved the capacity to adsorb unusually high metal concentrations, and may exhibit site concentrations and/or 63

76 binding constants that are significantly higher than for bacterial consortia from uncontaminated environments. In this study, we test whether consortia of bacteria isolated from a variety of industrial wastes and contaminated geologic systems adsorb protons and Cd to similar extents. We use a surface complexation modeling approach to constrain the values of the thermodynamic modeling parameters that best fit data collected in acid-base titrations and Cd adsorption experiments. The raw data and modeling results from this and previous studies are compared to determine whether a single set of modeling parameters are capable of describing all of the experimental data and whether universal bacteria-metal adsorption behavior can be assumed in contaminated environments. 4.2 Material and Methods Sample Descriptions Soil samples were collected from a manufactured gas plant (MGP) site in Iowa, from dredged river sediments in Mississippi, and from a former explosives testing facility in New Jersey. The clay-rich soils from the MGP site were severely impacted by polycyclic aromatic hydrocarbons (PAHs) that had commingled with gasoline range hydrocarbons from a nearby leaky underground storage tank (LUST) site. The contamination is characterized by visible coal tar and concentrations of benzene, toluene, ethylbenzene, and xylene (BTEX), all in excess of remediation criteria. The soil sample collected from river sediment was impacted by dichlorodiphenyltrichloroethane (DDT) and its breakdown products, 1,1-dichloro-2,2-bis(p- 64

77 dichlorodiphenyl) ethylene (DDE) and 1,1-dichloro-2,2-bis(p-chlorophenyl) ethylene (DDD). Concentrations of the composite material sampled ranged from 100 to 300 µg/kg total DDT+DDE+DDD. Soil from the New Jersey site was heavily impacted by explosive residue, mainly 2,4,6-trinitrophenylmethylnitramine (tetryl). A composite sample of this material was collected from silty soils that ranged from about 1000 to 100,000 mg/kg tetryl. TABLE 4.1 SAMPLE LOCATION AND GROWTH INFORMATION Sample Type Location Broth Type 1 Growth Duration 2 (Days) MGP Soil Northern Iowa SB TSB 7 3 Phosphate Sludge Electroplating Facility - SB TSB 5 2 Indiana Filter Cake Electroplating Facility - SB TSB 8 4 Indiana Coolant Oil Manufacturing Plant Indiana SB TSB 7 3 DDT Soil River Sediment - Mississippi SB TSB 9 3 Tetryl Soil Military Facility - New Jersey SB TSB 7 3 LUST 1 Southern Michigan SB TSB 6 3 LUST 2 Southern Michigan SB TSB TSB = Trypticase Soy Broth, SB = Soil Broth. 2 Growth duration refers to the amount of time a consortium was grown in a particular broth solution prior to harvest and use in experiments. In addition to these soil samples, groundwater samples were collected from shallow (< 25 ft) monitoring wells screened across the water table at LUST sites in two separate towns in Southern Michigan. We refer to these sites as LUST 1 and LUST 2. Gasoline-range hydrocarbons and BTEX compounds were present in the shallow water column at both sites. Benzene concentrations in the groundwater samples from LUST 1 and LUST 2 were 170 µg/l and 300 µg/l, respectively. Sludge materials were collected from the waste coolant oil at a major manufacturing plant and from two different waste streams at an electroplating facility, both in 65

78 Northern Indiana. The coolant oil was vegetable-based and was mixed with waste oil from machines in the plant. The first waste stream sampled at the electroplating facility consisted of filter cake material from a filter press used to remove flocculants from metal-rich fluid baths. The filter cake material was clay-based and rich in metals, including Cr and Zn. The second waste stream sampled was from phosphate sludge material rich in Zn and Mn. This sludge buildup was scraped from the fluid baths several times each month Sampling and Growth of Bacteria All materials used for sample collection, including glass jars, lids, and digging scoops were sterilized and sealed in plastic bags prior to use. Soil and sludge samples were collected by hand, and disposable plastic bailers were used to collect samples from monitoring wells at the LUST sites. Immediately after collection, sample lids were loosely placed over the jars to prevent contamination while maintaining the aerobic conditions thought to be present in each environment. The soil samples contaminated with DDT and tetryl were collected in late 2002, placed in sealed glass jars, and refrigerated prior to their use in this study (late 2003). Control growth experiments were completed by briefly opening jars of sterilized growth media near sample locations and exposing them to the atmosphere. Lack of bacterial growth in the control samples suggests that the bacterial consortia we produced were most likely not contaminated during handling with species not present in the original sample. Approximately 10 grams of soil/sludge or 10 ml of water from the samples were used to inoculate the chosen broth solution of either trypticase soy broth (TSB) with 0.5% 66

79 yeast extract or soil broth (SB). SB was made with 250 ml soil extract, 2 g glucose, 1 g yeast extract, and de-ionized (DI) water (to 1.0 L). Soil extract was made by autoclaving 500 g of soil with 1 L of DI water for 1 hr at 15 psi/121 C and decanting the fluid product using Whatman No. 2 filter paper. A nutrient-rich media (TSB) and a nutrient-limited media (SB) were chosen, because differences in the amounts and types of nutrients often lead to growth of different bacterial consortia from the same environment. The duration of growth for each consortium was determined through visual estimation of the maximum turbidity (cell density). Growth times were short (2 to 4 days) for consortia grown in TSB and longer (5 to 7 days) for consortia grown in SB. Inoculated broth solutions were gently shaken at room temperature until they were harvested or a portion was used for re-inoculation. Approximately 10 ml of the initial bacteria-broth suspension was used to inoculate larger quantities of identical broth solutions in order to dilute out the solid fraction present from soil and sludge samples. Growth conditions were identical to those previously described for consortia of bacteria from uncontaminated environments (Borrok et al., 2004a), and are summarized in Table 4.1. Because many bacterial species are unculturable (Kaeberlein et al., 2002), and because the growth conditions were entirely aerobic, the bacterial consortia grown for our experiments likely included only a subset of the total bacterial population present in each environment sampled. However, by growing the bacteria directly in broth solutions (as opposed to isolating them on plates) and limiting the number of re-inoculations (as opposed to many episodes of re-growth), we hoped to achieve a range of diverse consortia representative of the 67

80 original environments sampled. As a preliminary indication of diversity, Gramstaining was conducted on each consortium. Diversity was more precisely determined through denaturing gradient gel electrophoresis (DGGE) Freezing of Bacteria Prior to each re-inoculation and/or harvest for experimentation, bacterial cells suspended in their broth solutions were frozen for future DNA extraction and/or reuse. Cells frozen for re-use were suspended in a mixture of 10% sterilized glycerol and flash frozen with liquid nitrogen prior to storage in a -80 C freezer. Cells frozen for DNA extraction were not mixed with additional agents and were placed directly into the -80 C freezer Potentiometric titration and Cd Adsorption Experiments Washing procedures and experimental conditions have been described previously (Borrok et al., 2004a), and are briefly summarized below. Bacteria were harvested and washed five times in 0.1M NaClO 4 (the same electrolyte used in the experiments). After the final wash, the bacteria were placed in weighed test tubes and centrifuged (7000 rpm [5500 g] at 25 C) for 1 hour to produce a moist pellet. The weight of the moist pellet recorded during this step is the wet weight we report throughout this study. This wet weight is between 3 and 6.5 times more than the dry weight of the biomass, depending on the bacterial species involved (Borrok et al., 2004a). The bacterial pellet was then used in potentiometric titrations or Cd adsorption experiments. 68

81 In each metal adsorption experiment, 10 g/l of a bacterial consortium was suspended in a ph-neutralized stock solution of 0.1 M NaClO 4 and 10 ppm Cd. The chosen concentrations of metal and bacteria are high to improve accuracy in our experiments; however, the bacteria/metal ratios are similar to those found in some natural and industrial settings. The bacterial stock solution was divided into individual reaction vessels, and the ph of each vessel was adjusted by adding minute aliquots of 1.0 M or 0.1 M HNO 3 or NaOH. After adjustment of the ph, and an additional 2 hours of reaction time on a rotating rack, the final (equilibrium) ph of each vessel was measured. The individual vessels were then centrifuged and the resultant supernatant was filtered through a 0.45 µm nylon membrane. The filtered supernatant was analyzed for Cd using inductively coupled plasma atomic emission spectroscopy. Standards were diluted from 1000 ppm reagent grade stock into a 0.1 M NaClO 4 electrolyte, and analytical error was within ± 2% of external standards. Cells used in titration experiments were unfrozen from the frozen stock and grown in the same broth solution used for the metal adsorption experiments. Titrations were completed on all the consortia used in the Cd adsorption experiments, with three exceptions. LUST 1 SB was discarded after algae were discovered growing in the sample. The LUST 2 - SB sample and the Tetryl Soil SB sample were not frozen with glycerol after the Cd adsorption experiments, and hence could not be used in the titration experiments. A duplicate Cd adsorption experiment was performed using bacterial consortia grown from the frozen stock to test if the freezing process affected the observed adsorption behavior. The results using the frozen cells were 69

82 indistinguishable from those using cells prior to the freezing step. Although minor variations in the diversity of the bacterial consortia in response to the freezing process cannot be ruled out, these were preferable to the large changes in diversity expected from the alternatives of either growing the consortia for longer periods of time or resampling and re-growth of the consortia prior to each type of experiment conducted. After harvesting and washing the cells to be used in the titration experiments, the cells were suspended in approximately 10 ml of 0.1 M NaClO 4 that had been purged of CO 2 by bubbling N 2 for 60 min. The suspension was immediately placed into a sealed titration vessel maintained under a positive pressure of N 2. Titrations were conducted in triplicate using an automated burette assembly. Each suspension was first titrated to the desired starting ph of 2.5, using minute ( ml) aliquots of N HNO 3. Forward titrations were then performed to an approximate final ph of 9.5 through the addition of aliquots of N NaOH. During each titration, the suspension reached a stability of ph units per second prior to the addition of the next aliquot of acid or base. The total volume of acid and base added during each titration ranged from about 2.5 to 3.5 % of the starting volume of the solution Gram-Staining and DGGE Analysis Gram-staining was conducted on each bacterial consortium by first heat-fixing the cells to a glass slide and then staining the cells using a PROTOCOL Gram stain kit from Fisher Scientific. DNA was extracted from the cell suspensions in the growth media using a MoBio Laboratories, Inc. Ultraclean soil DNA kit, and was frozen at - 20 C prior to amplification. A custom-made universal bacterial primer set (EUB

83 and EUB 534, 200 base pairs in length with a GC-clamp) was used during the polymerase chain reaction (PCR) process to amplify a specific 16s rdna sequence (Muyzer et al., 1993). DGGE was carried out using a Dcode universal mutation detection system (Bio-Rad). The PCR product was loaded into a gel with a 30% to 60% gradient of a chemical denaturant (urea and formamide). The DNA was forced to travel through the denaturant in response to an electrical potential. Gels were run at 60 C and a potential of 60 V for 14 hours. The gel was then stained using ethidium bromide, fluoresced, and photographed using a Kodak EDAS 290 photographic system. Because of the differences in DNA base pair sequences among different bacterial species, the DNA from a specific species will denature at a specific point in the gel, forming a characteristic band. The intensity of these bands is directly related to the concentration of DNA present in the sample. Hence, analysis of the band positions and intensities can determine the minimum number of bacterial species present, and their relative abundances. The position and relative intensities of these bands were analyzed using Kodak ID 3.6 software. Although this is a sufficient test for diversity in our study, additional sequencing of the bands would be necessary to determine the identity of each species and to verify that each band represents only one species. 4.3 Results and Discussion Gram-staining Both Gram-positive and Gram-negative bacterial species were present in the MGP samples grown in both TSB and SB broth solutions, the coolant oil samples grown in 71

84 both broth solutions, the DDT soil sample grown in TSB, and the phosphate sludge sample grown in SB. Only Gram-negative species were present in the remaining consortia. The DDT soil sample grown in SB, contained abundant paramecia and the filter cake sample grown in SB contained large quantities of algae, both determined through optical microscopy. Hence, both samples were discarded and not used in experimentation DGGE results Consortia used in the experiments displayed between 3 and 9 separate bands (minimum number of bacterial species). DGGE analysis was not performed on samples DDT TSB and tetryl soil SB, so Gram-staining was the only indication of diversity for these samples. DGGE analysis shows that the band locations and intensities for consortia LUST 1 SB/TSB and LUST 2 SB/TSB are strikingly similar, despite differences in sampling location and growth media. In contrast, a high degree of diversity was achieved among the other consortia. Each of the remaining consortia contained numerous bands and there was little similarity of bands between different consortia or between the same consortia grown using different broth solutions (Table 4.2, Figures 4.1A-C) Potentiometric Titrations All the consortia exhibited significant buffering capacity over the entire ph range studied (2.5 to 9.5). The titration curves for all of the consortia were similar in shape 72

85 and position to each other except for the curves from the MGP and phosphate sludge samples grown in SB. Below approximately ph 6, these samples exhibit significantly TABLE 4.2 RELATIVE BAND INTENSITIES WITHIN INDIVIDUAL LANES OF DGGE GELS. Samples Relative Intensity of Bands for each Sample 1 Filter Cake TSB 26.7± ± ± ± ± ± ± ± ±0.6 Tetryl Soil SB 25.1± ± ± ± ± ± ± ± ±0.1 LUST 1 TSB 50.6± ± ± ±0.3 LUST ± ± ± ±1.3 SB LUST 2 TSB 71.0± ± ±0.6 LUST ± ± ± ±0.2 SB Coolant Oil 43.9± ± ± ± ± ±0.7 TSB Coolant Oil SB 58.6± ± ±1.7 Phosphate 31.2± ± ± ± ± ±7.2 SludgeTSB Phosphate 35.6± ± ± ± ± ±0.6 Sludge SB MGP Soil TSB 44.0± ± ± ±0.1 MGP Soil SB 52.0± ± ± ±1.1 1 The band intensity is a proxy for the abundance of what is assumed to be a specific bacterial species within each consortium. The bands for each consortium are unique and are not comparable to bands in other consortia. Reported errors are based upon 3 replicates (1 uncertainties) or 2 two replicates (maximum difference). A

86 B C Figure 4.1. DGGE gels. Numbers correspond to individual lanes of the same sample analyses were performed in triplicate. A1 = E. coli control, A2 = DDT SB, A3 = Tetryl soil SB, A4 = Phosphate sludge TSB, A5 = Filter cake TSB, A6 = Phosphate sludge SB, B1 = E. coli control, B2 = LUST 2 TSB, B3 = LUST 1 TSB, B4 = LUST 2 SB (only duplicate analyses performed), B5 = LUST 1 SB, B6 = Coolant Oil SB, B7 = Coolant Oil TSB, C1 = E. coli control, C2 = MGP soil TSB, C3 = MGP soil SB. TSB = Trypticase Soy Broth. SB = Soil Broth. 74