ENGINEERING APPROACHES TO SHEWANELLA TAXIS STUDIES RUI LI

Transcription

1 ENGINEERING APPROACHES TO SHEWANELLA TAXIS STUDIES BY RUI LI A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Chemical Engineering 2012

2 ABSTRACT ENGINEERING APPROACHES TO SHEWANELLA TAXIS STUDIES By Rui Li Shewanella species are famous for their broad range of terminal electron acceptors and the ability to perform taxis towards both soluble and insoluble electron acceptors, which may help explain their nearly ubiquitous presence in widely disparate environmental niches around the world. Studies of Shewanella s tactic properties are important to understand their competitiveness and roles in elemental (e.g. nitrogen, sulfur, iron, manganese and others) cycling and bioremediation (e.g., precipitation of soluble uranium oxides). Population-level microbial taxis involves a complex interplay of cellular process (e.g., growth, metabolism, chemotaxis, and random motility) and molecular processes (e.g., diffusion of electron donors and acceptors that serve as attractants). This dissertation describes the use of multiple approaches including engineering tools, biological assays and mathematical models to study population-level growth and taxis of Shewanella in response to applied and cell-generated gradients of soluble electron acceptors. The model was able to reproduce key trs of the observed cell growth and migration patterns in either diffusion gradient chamber (DGC) or motility assays, which validate the use of our approaches to measure and simulate Shewanella s taxis in response to electron acceptor gradients. New hypotheses relevant to Shewanella s taxis were investigated to help understand how ecological niches containing various electron acceptor gradients influence the distribution of Shewanella baltica strains with different genotypes and hence different chemotactic behaviors.

3 We studied the impact of opposing gradients of nitrate and fumarate on the chemotactic behaviors of S. oneidensis MR-1 fumarate reductase and nitrate reductase mutants, where the mixture of mutant strains could be partially separated into two populations via formation of two separate chemotactic bands, one moving toward each electron acceptor source. We also studied cell behavior in the presence of insoluble electron acceptor manganese dioxide (MnO 2 ). A novel mechanism, mediated energy taxis, we proposed by which shewanellae use self-secreted riboflavin as both an electron shuttle and an attractant to direct cell movement toward local sources of insoluble electron acceptors. To test this hypothesis, taxis measurements were conducted in the presence of in various insoluble electron acceptors and with chemotaxis mutants. The results strongly supported the hypothesis, and mathematical models based on these mediated energy taxis mechanisms were able to predict experimental trs.

4 ACKNOWLEDGEMENTS I would like to thank my advisors Dr. Mark Worden and Dr. James Tiedje, for their guidance, support, and encouragement during my graduate at Michigan State University. I am also grateful to them for recruiting me into the Shewanella project, and offering me a very unique opportunity to conduct research across the fields of engineering, biology and mathematics. It has been a pleasure and my honor to work with them. They did not only offer me professional advice on my major, but also showed me how to work as an indepent researcher, and how to collaborate with people from different areas. I would like to thank my committee members Dr. Scott Barton, Dr. Christina Chan, and Dr. Gemma Reguera, for agreeing to mentor me and for their input and advice. I would like to thank members of both Tiedje and Worden groups, for their help and friship: Jennifer Auchtung, Jie Deng, Christina Harzman, Tamara Cole, Shoko Iwai, Erick Cardenas, Adina Howe, Ederson Jesus, Timothy Johnson, James Kremer, Marius Vital, Maryam Thompson, Jiarong Gao in Tiedje group, and Chloe Liu, Bhushan Awate, Ying Liu, Lee Alexander, Alexander Negoda in Worden group. My projects have been collaborative throughout my graduate study, and I would like to acknowledge my collaborators. I would especially like to thank Jennifer Auchtung, who worked with me in my first two years at MSU. Her deep and comprehensive understanding of microbiology impressed me. Her kind help was invaluable to me when I started my PhD research. iv

5 I would like to thank Dr. Chichia Chiu and Dr. Paul Satoh, for their insight and suggestion on my research. I would like to thank Dr. Daniel Jones and Dr. Ramin Vismeh for their help on mass spectrometry. I would like to thank Dr. Scott Barton and Hanzi Li, for their helpful discussion with me about electrochemistry. In my life I have been most influenced by my parents. They are both chemical engineers, so my understanding of chemical engineering started when I was very young. This dissertation would not have been possible without their love, understanding and support. v

6 TABLE OF CONTENTS LIST OF TABLES... ix LIST OF FIGURES... x Chapter 1 INTRODUCTION Significance Project Objectives Background Electron acceptor taxis in Shewanella Metal reduction in Shewanella Dissertation overview Shewanella taxis in a diffusion gradient chamber and modeling Separating Shewanella cells with different genetic background by opposing electron acceptors gradients Microbial reduction of insoluble electron acceptors via mediated energy taxis NADH oxidation by activated glassy carbon electrode... 7 Chapter 2 SHEWANELLA TAXIS TOWARDS ELECTRON ACCEPTORS IN A DIFFUSION GRADIENT CHAMBER AND MODELING Introduction Materials and methods Strains and growth conditions Motility assays DGC experiments Mathematical model Estimation of model parameters Results and discussion Parameter valuation Response of Shewanella to fumarate gradient in DGC CHAPTER 3 SEPARATION OF S. ONEIDENSIS MR-1 MUTANTS BY OPPOSING GRADIENTS OF ELECTRON ACCEPTORS vi

7 3.1 Introduction Method and materials Strains and growth conditions DGC experiments and modeling Results and discussion CHAPTER 4 ENERGY TAXIS TOWARD INSOLUBLE ELECTRON ACCEPTORS MEDIATED BY SOLUBLE ELECTRON SHUTTLES Introduction Materials and methods Strains and growth conditions Taxis assays Riboflavin concentration analysis Results and discussion Conceptual model Taxis toward riboflavin and MnO Mathematical model of mediated energy taxis by S. oneidensis MR CHAPTER 5 NADH OXIDATION BY ACTIVATED GLASSY CARBON ELECTRODE Introduction Materials and Methods Electrode activation by cyclic voltammetry and azines immobilization Electrochemical characterization Electrode activation with constant potentials NAD + immobilization Results and discussion NADH electrocatalysis Electrode activation with constant potentials NAD + immobilization Conclusion APPENDIX A vii

8 APPENDIX B APPENDIX C APPENDIX D REFERENCE viii

9 LIST OF TABLES Table 2.1 Constants for simulations in Chapter 2 33 Table 3.1 Strains and plasmids used in Chapter Table 4.1 Constants for simulations in Chapter Table 5.1 Methylene green immobilization on the constant potential activated electrodes.. 87 Table 5.2 Correlation coefficient of variables in Table ix

10 LIST OF FIGURES Figure 2.1 Schematic diagram of the diffusion gradient chamber. 25 Figure 2.2 Experimental and model-predicted fumarate concentration profiles in the DGC Figure 2.3 Batch culture and continuous culture of S. oneidensis MR Figure 2.4 Chemotaxis sensitivity curve for taxis toward fumarate..28 Figure 2.5 Experimental and modeling results for DGC Run 1 in Chapter 2 29 Figure 2.6 Experimental and modeling results for DGC Run 2 in Chapter Figure 2.7 Regression based on the experimental data and simulation data.31 Figure 3.1 Experimental and modeling results in Chapter 3.40 Figure 3.2 Results of flow cytometry and simulation 41 Figure 4.1 Schematic diagram illustrating microbial reduction of insoluble electron acceptors via riboflavin mediated energy axis. 59 Figure 4.2 Motility assays for wild-type S. oneidensis MR-1, its SO2240 SO3282 doubledeletion mutant and chea-3 mutant with riboflavin, FMN, and MnO 2.60 Figure 4.3 Radial movement of the reduction zone boundary...61 Figure 4.4 Motility assays for S. oneidensis MR-1, its SO2240 SO3282 double-deletion mutant a n d c h e A - 3 m u t a n t w i t h 1 0 m M Fe(OH) Figure 4.5 Motility assays results depicting S. oneidensis MR-1 taxis to oxidized riboflavin and MnO Figure 4.6 Rate of growth of reduction zone in agar containing either riboflavin or MnO 2.64 Figure 4.7 Linearity test for yellow color intensity recorded by digital camera and riboflavin concentration..65 xi

11 Figure 4.8 Simulation results depicting S. oneidensis MR-1 taxis to 5 mm MnO 2 36 h after inoculation..66 Figure 4.9 Model-predicted profiles for cell density, MnO 2 concentration, Mn(II) concentration, reduced riboflavin concentration, and oxidized riboflavin concentration.67 Figure 5.1 Schematic diagram of example strategy to fabricate bioelectronic containing azine electrocatalyst and NAD + on activated electrode.. 80 Figure 5.2 Activation of glassy carbon (GC) electrode Figure 5.3 Activity of bare glassy carbon, activated glassy carbon, and carbon nanotube electrodes...82 Figure 5.4 The NADH concentration study of absorbed methylene green and oxazine 170 on activated glassy carbon electrodes.83 Figure 5.5 Activity of glassy carbon electrode with absorbed methylene green, glassy carbon electrode with electropolymerized methylene green, and activated glassy carbon electrode with absorbed methylene green.. 84 Figure 5.6 Characterization of electrodes activated by constant potentials xii

12 Chapter 1 INTRODUCTION 1.1 Significance of the problem The genus Shewanella is a group of facultative anaerobic bacteria with worldwide distribution. To date, over 20 different organic and inorganic compounds in soluble or insoluble form have been found to be terminal electron acceptors of Shewanella, enabling the genus to inhabit diverse ecosystems [1]. In the presence of electron acceptors gradients, Shewanellae are able to migrate along the gradients, which is known as electron acceptor taxis [2]. Electron acceptor taxis directs cells into more favorable environments with higher electron-acceptor concentrations, which may provide a competitive advantage, increasing their chances of survival. For example, within anaerobic media Shewanella cells can track motile algae producing oxygen possibly by aerotaxis [3]. The ability to understand and harness taxis toward electron acceptors may also contribute to understanding Shewanella s important ecological roles in biogeochemical cycling of several elements, as well as facilitating Shewanella s use in bioremediation processes [4], such as toxic metal oxide stabilization and denitrification [5]. Researchers have been studying Shewanella taxis for over 15 years [2, 6-8]. This study focuses on the Shewanella taxis at population level. Population-level microbial taxis is complex, involving interplay of cellular processes (e.g., growth, metabolism, tactic migration, and random movement) and molecular processes (e.g., diffusion, consumption and possibly regeneration of carbon sources, energy sources, and attractants). To elucidate the mechanism of the interplay, a multidisciplinary platform including biological assays, engineering tools, as well as mathematical models is useful. 1

13 1.2 Project Objectives The first objective of this project was to adapt existing experimental and mathematical modeling methods and develop a multidisciplinary research platform for studying Shewanella taxis. The diffusion gradient chamber (DGC) was developed as an experimental tool to study the simultaneous transport and reaction processes listed above under controlled conditions [9]. The DGC has several advantages over other methods of studying microbial taxis, including (1) multiple gradients can be set up simultaneously; (2) a variety of gradient may be tested, including linear, steady-state gradients or nonlinear, transient gradients [10]. Experimental studies in the DGC have been successfully used to characterize aerobic microbial chemotaxis and document interesting cellular phenomena resulting from chemotaxis [9, 11-13]. Mathematical models of the DGC were developed to provide a quantitative description of the simultaneous cellular and molecular processes that underpin population-level cell dynamics and interactions with tactic attractants [10]. These models were experimentally validated and then used for interpreting experimental results, as well as exploring possible patterns of populationlevel tactic behavior under conditions that would be difficult or time-consuming to test experimentally [10, 12]. DGC experiments to study Shewanella taxis in response to electron acceptors other than oxygen require an anaerobic condition, because oxygen the preferential electron acceptor and preferential chemoattractant. The second objective is to use the adapted experimental and modeling methods to test hypotheses about Shewanella s taxis in response to electron acceptors. Ziemke et al. proposed that ecological niches in Baltic Sea constructed by various electron acceptor gradients might be 2

14 responsible for the distribution of Shewanella baltica strains with different genotypes [14, 15]. The DGC s proven ability to separate populations of bacterial cells based on their different chemotactic responses to electron-acceptor gradients [11] is useful to validate this hypothesis. Our study presents evidence that S. oneidensis MR-1 cells migrate chemotactically into regions of higher manganese dioxide concentration by following a spatial gradient of a secreted electron shuttle (riboflavin). A mathematical model adapted from DGC model is used to validate the proposed mechanism by which chemotaxis enhances the metal-reduction rate of S. oneidensis MR Background Electron acceptor taxis in Shewanella Electron acceptor taxis in Shewanella was initially reported as metabolism-indepent [2]. However, recent research has discovered that functional anaerobic respiratory systems (e.g., cytochrome c and reductase) are necessary for taxis towards electron acceptors [6, 16]; and disruption of ΔpH component of the proton motive force (PMF) abolishes the taxis ability of Shewanella cells [6]. Hence, together with previous study [7], in Shewanella, taxis towards electron acceptors is governed by an energy taxis mechanism, responding by altered motility patterns to changes in internal energetic conditions [17]. Compared with classical chemotaxis, energy taxis is not mediated by a response to chemical per se, but response to changes in the redox state of the respiratory chain or in the PMF, resulting from the metabolism of the chemical [6]. 3

15 In a typical Escherichia coli aerotaxis (a common form of energy taxis), the electrons are finally transferred to oxygen during aerobic respiration, in which accumulation of OH in the cytoplasm and H + in the periplasm generates a proton gradient across the membrane, or PMF. A membrane-bound ATP synthase directs re-entry of the protons into the cytoplasm, which dissipates the PMF. The energy released from this process is captured and used for ATP generation. In the meanwhile, PMF signal is detected by methyl-accepting chemotaxis proteins (MCP, also termed receptors). Receptors are usually transmembrane proteins with a sensory domain located towards the N terminus, and the signaling domain towards the C terminus. The chemotactic signals are transduced to a histidine protein kinase, CheA, which is connected to the receptors via CheW. CheA regulates other proteins interacting with the flagellar motor switch apparatus to regulate the direction of flagella rotation, and hence cell migration. For a review on the mechanism of energy taxis, see [17, 18]. Shewanella has CheA protein highly resembles that of E. coli CheA. The genome analysis of S. oneidensis MR-1 indicated that this organism contains three CheA genes, i.e., chea-1, chea-2 and chea-3; however, only chea-3 has an essential role in chemotaxis to anaerobic electron acceptors [8]. On the other hand, mechanism of Shewanella taxis is different from well-known mechanism of energy taxis in E. coli on several ways. First, although CheA genes are important regulators for aerotaxis in E. coli [19], none of three CheA genes is essential for aerotaxis in Shewanella [16]. Second, in E. coli taxis, the PAS domain is responsible for sensing energy-related parameters, such as oxygen, light and redox; hence energy-sensing receptors containing PAS domain (e.g., Aer) are essential for energy taxis [20]. Shewanella does 4

16 not require MCP containing PAS domain [6, 16], but a receptor containing a Cache domain [6]. However, to date, energy taxis mechanism in Shewanella is still not well known Metal reduction in Shewanella As a dissimilatory metal-reducing bacterium (DMRB), Shewanella is challenged by transferring electrons across cell membrane, from the cytosol to extracellular insoluble terminal electron acceptors, such as iron and manganese oxides. The current understanding of electron transfer indicates that the electron conduit is composed of three multiheme cytochromes (i.e., CymA, MtrA, and MtrC) and an integral outer-membrane protein (i.e., MtrB). Electrons come from menaquinone pool in the cytoplasmic membrane, and are transferred into the periplasm by CymA. MtrA in the periplasm then transfer electrons to MtrC or OmcA at cell surface, where electrons are further transferred to an oxidized substrate [21]. The outer membrane cytochromes can transfer electrons to insoluble terminal acceptors by one or more of following mechanisms: (1) direct contact between cytochromes and the insoluble substrate [22]; (2) conductive protein nanowires [23, 24]; (3) metal chelators/siderophores which solubilize the oxidized mineral and allow it to diffuse to the bacterial cell surface [25, 26]; (4) shuttle molecules produced by the cell cycling between the cell and extracellular substrate [27, 28]. 1.4 Dissertation overview Shewanella taxis in a diffusion gradient chamber and modeling 5

17 To obtain a systems-level understanding of Shewanella biology and ecology, including the influence of electron acceptor availability on Shewanella s growth, metabolism, and transport needs to be elucidated, the response of populations of S. oneidensis MR-1 cells to an applied gradients of the electron acceptor fumarate were studied in the DGC. Mathematical models capable of predicting cellular growth and taxis under the influence of gradients were used to interpret the results. Formation and migration of a tactic band of MR-1 cells up the fumarate gradient was observed in the DGC and reproduced by the model. The predicted velocity of the chemotactic cell band advancing toward the attractant source (0.139 cm/h, R2=0.996) closely matched that measured in the DGC (0.134 cm/h, R2=0.997). These results validate the use of the DGC system including its mathematical model to measure and simulate Shewanella chemotaxis in response to electron acceptor gradients and establish it as a research tool to help elucidate Shewanella s role in environmental processes Separating Shewanella cells with different genetic background by opposing electron acceptors gradients Difference in the ability of Shewanella species to respond to and use available electron acceptors is thought to play an important role in their ecology [14, 15]. Ziemke et al. proposed a hypothesis that the distribution of S. baltica strains with different genotypes might result from various electron acceptor gradients in the Baltic Sea [15]. To validate this hypothesis, the DGC approach was exted to including opposing gradients of two electron acceptors (i.e. nitrate and fumarate) and two mutants in fumarate and nitrate reductase genes. The DGC was able separate the two populations of S. oneidensis MR-1 mutants based on their different tactic responses to 6

18 electron-acceptor gradients, which provides insight into the role of taxis in the microbial ecology of Shewanella species Microbial reduction of insoluble electron acceptors via mediated energy taxis To explore the role of microbial taxis in manganese dioxide (MnO2) reduction, we conducted motility assays with S. oneidensis MR-1 in dilute agar in which MnO2 particles were dispersed. An expanding dark-to-clear transition zone was observed, indicating the migration of cells and reducing MnO 2 as they moved. To explain this behavior, we proposed that riboflavin secreted by the cells acts as both an electron shuttle to reduce MnO2 and as a tactic attractant. Reduced riboflavin diffuses away from the cells, and is oxidized by MnO2. The oxidized riboflavin then diffuses back to the cells, establishing a spatial gradient that triggers microbial energy taxis toward the MnO2. The resulting simultaneous riboflavin-mediated chemotaxis and redox cycling enables MnO2 to be reduced without direct contact with the migrating cells and allows the cells to migrate toward the insoluble MnO2. The hypothesis was validated by both experimental results and mathematical simulations NADH oxidation by activated glassy carbon electrode Glassy carbon electrodes were activated by cyclic voltammetry (CV). The activation increased the surface area, as indicated by an increase in capacitance ( µf cm -2 ), and also increased the current density (0.35 ma cm -2 ) for NADH oxidation at 0.05 V versus 7

19 Ag AgCl in 20 mm NADH solution (ph 7.4). Multiple azines could be easily absorbed onto the activated electrodes. The activated electrode with absorbed methylene green (MG) achieved a current density at 0.7 ma cm -2 for NADH oxidation under 0.05 V vs. Ag AgCl in 20 mm NADH solution (ph 7.4), a 100-fold increase compared to electropolymerized methylene green electrode. When constant potentials were used to activate electrode, the optimal activation potential was found to be around 2.5 V. The activated carbon surface could also be used for NAD + /NADH immobilization, where the immobilized NAD + /NADH was active as a cofactor of alcohol dehydrogenase. 8

20 Chapter 2 SHEWANELLA TAXIS TOWARDS ELECTRON ACCEPTORS IN A DIFFUSION GRADIENT CHAMBER AND MODELING 2.1 Introduction Shewanella species are known for their broad range of terminal electron acceptors and their ability to grow in a wide range of environments [5]. Taxis towards electron acceptors was first considered as metabolism-indepent [2], but later was proposed to be an energy taxis [6, 7]. Shewanella s ability to migrate in response to spatial gradients of electron acceptors, coupled with its wide range of potential electron acceptors, may help explain its nearly ubiquitous presence in widely disparate environmental niches around the world. The ability to understand and harness electron acceptors taxis in Shewanella could facilitate studies on its ecological roles and its use in bioremediation processes. The DGC was developed as an experimental tool to study the population-level microbial taxis under controlled conditions [9]. Compared with other methods to study microbial taxis, the DGC is convenient for establishing well characterized multiple steady-state (or unsteady-state) gradients in multiple directions. In addition, with a removable lid, samples may be withdrawn without sacrificing the experiment. Many different inoculation protocols may be used, including uniform inoculation across the gel, in a line, or at a point [10]. The DGC has been successfully applied to studies of bacterial cells including E. coli [9-11], Pseudomonas nautica [29] and Pseudomonas fluorescens [9]. As part of these studies, the DGC was used to characterize the role of microbial taxis and growth in a variety of interesting population-level cellular phenomena. A mathematical model of the DGC was developed to provide a quantitative description of the 9

21 simultaneous cellular and molecular processes, including cell growth, random motility, and chemotaxis; metabolism of carbon sources and electron acceptors; and diffusion of carbon sources and electron acceptors [10]. This model was able to reproduce bacterial behavior at population-level behavior under the influence of applied gradients of chemoattractants [10-12]. This chapter describes use of the DGC to characterizegrowth and taxis of S. oneidensis MR-1 under the influence of a spatial gradient of the electron acceptor fumarate. The mathematical model originally developed for the DGC was adapted to describe MR-1 cells growing with fumarate as the electron acceptor. The model was used to interpret the experimental results and reproduce the migration patterns observed in the DGC, thereby providing insights into the complex interplay of transport and reaction phenomena governing cellular growth and migration of S. oneidensis MR Materials and methods Strains and growth conditions S. oneidensis MR-1 and its mutants were grown either aerobically on LB medium (10 g/l tryptone, 5 g/l yeast extract and 10 g/l NaCl) at 30 C or anaerobically in M1 medium [1] containing 40 mm sodium lactate (electron donor and carbon source) for all the described experiments and electron acceptors at concentrations indicated below. To inoculate either DGC or soft agar plates, Shewanella cells were grown in 10 ml aerobic LB overnight to an OD 600nm ~ 1.1, pelleted by centrifugation, and re-susped in 100 µl M1 medium. Ten µl cell suspension ws loaded into sterile pipette tip. This tip was stabbed vertically into the gel at the 10

22 midpoint of the DGC s arena or petri-dish, and then the inoculum was gradually expelled as the pipette was withdrawn, leaving a vertical column of cell suspension within the gel. Fumarate was chosen as the electron acceptor because of its high solubility and one-step reduction Motility assays In motility assays, Shewanella cells are injected into soft agar and generate a gradient of the tactic attractant by reducing electron acceptors during anaerobic respiration. The cells then move up the self-generated gradients. The protocol of the assay has been outlined in Bencharit and Ward s publication [7]. Briefly, 20 ml soft 0.25% anaerobic M1 agar containing sodium lactate and an electron acceptor at various concentrations was poured into each petri dish. In all cases, the carbon source was present in excess, and the electron acceptor was the limiting nutrient. After inoculation, a digital camera was used to record the tactic migration patterns while the motility was mounted on a transilluminator box in an anaerobic glove box with ~4% H 2 (the balance in N 2 ). Two 30-cm long, 8 W fluorescent lights in the box provided diffuse lighting from a 45 angle beneath the DGC. The bottom of the box was covered by flat black paper. Using this illumination method, cells growing in the agar diffracted the fluorescent light upward, so that regions of high cell density appeared as bright regions against a dark background [10] DGC experiments Details of the DGC system (Figure 2.1) have been published previously [9]. The DGC consists of a rectangular arena (5 cm 5 cm 1 cm) that is filled with dilute agar and is bounded on each side by a liquid reservoir. Each reservoir is separated from the agar by either a 11

23 polycarbonate membrane, which allows solute exchange between the gel and the reservoir, or an impermeable rubber sheet that blocks solute exchange. Different concentrations of the tactic attractant(s) are maintained in the reservoirs. Diffusion from the higher concentration reservoir (the source) through the membrane and across the gel to the lower concentration reservoir (the sink) results in the development of a continuous gradient across the gel. Experiments in the DGC were carried out following protocols described previously [9], except that strict anaerobic condition were maintained by operating the DGC in the glove box with ~4% H 2 (the balance in N 2 ). Briefly, the DGC was filled by 45 ml M1 medium containing sodium lactate and 0.15% agar. Lactate is not a tactic attractant for Shewaenlla cells [2] and was present in excess, making the electron acceptor be the limiting nutrient. Fresh M1 medium containing the electron acceptor at a specified concentration plus lactate was continuously pumped through the reservoirs at a flow rate of approximately 2.5 ml/h to keep each reservoir s concentration constant. The gradient was established for a specified amount of time, and then the chamber was inoculated. After inoculation, the DGC was stationed on a transilluminator box located in the anaerobic glove box as described above. Pictures were taken by a digital camera. When grown in the presence of lactate and fumarate, Shewanella MR-1 oxidizes the lactate to acetate and reduces fumarate to succinate. Motility experiments showed that lactate a not tactic attractant for S. oneidensis MR-1 when the electron acceptor is limiting nutrient, consistent with previous results [2]. The metabolic by-products, acetate and succinate, were also tested in motility assays, did not significantly affect either cellular growth or migration (data not shown). 12

24 In Run 1, all four reservoirs of the DGC were sealed with rubber sheets, and the agar poured into the arena contained both lactate and 5 mm fumarate. Hence, there was no preestablished fumarate gradient across the agar. In Run 2, the south reservoir served as the fumarate source (15 mm fumarate), and the north reservoir served as the fumarate sink (0 mm fumarate). The other two reservoirs (east and west) were sealed with rubber sheets. Fresh solution containing the specified fumarate concentrations plus lactate was continuously pumped through the reservoirs. Before inoculating the agar, an unsteady-state fumarate gradient was established by allowing fumarate to diffuse from the source reservoir, through the agar, to the sink reservoir for 10 h. The DGC system was inoculated as described above Mathematical model Derivation of model for energy taxis velocity The derivation of an energy-taxis model parallels that used for the RTBL chemotaxis model [30]. The 1-D approach accounts for cell movement in + and x directions (x). Each cell is assumed to swim in a straight line and then tumble before running in another direction. In chemotaxis, the tumbling frequency varies with changes in the concentration of receptor proteins bound to chemoattractant molecules. By analogy, in energy taxis, the tumbling frequency is assumed to vary with changes in some molecular indicator of energy status (E). In energy taxis, the swimming speed (v) can vary with the concentration of a limiting electron acceptor (S) [31]. However, this effect has been reported to be insignificant in modeling energy taxis [32], so v is assumed here to be constant. The tactic velocity (V) deps on probabilities of cells tumbling and switching directions [30]: 13

25 V = v + p p + p + p (2-1) where p is the probability per unit time that a cell moving in the + direction will switch to a cell moving in the direction, and p is the probability per unit time that a -moving cell will switch to + -moving cell. These switching probabilities are given by p (1 2 + / + / ψ ) = p t (2-2) where p t is the tumbling probability for cells moving in the + or direction, and ψ is the directional persistence. The depence of p t on bound receptor concentration has been reported during E. coli chemotaxis by Berg and Brown [33]. A similar depence of p t on E is assumed for energy taxis: ln 1 p t 1 de = ln + σ p 0 dt (2-3) where p 0 is the tumbling probability when E is constant, and σ is the tactic sensitivity. This equation indicates that, if the cell s energy status increases during a run (de/dt > 0), its tumbling probability decreases. Substituting Eq. (3) into Eq. (2) gives p + / + / ψ ) = p0 exp de σ dt (1 2 (2-4) 14

26 Under the conditions used in this study, the electron donor was present in excess, and, because Shewanella strains are nonfermenters [5], energy production was limited by the electron acceptor concentration (S). Under such conditions, respiration has been shown to follow hyperbolic saturation kinetics with respect to S [34], suggesting a hyperbolic depence of E on S: E = ke S K D + S (2-5) where k E is the maximum specific reaction rate, and K D is the saturation constant. The rate of change in E due to changes in S deps on the degree of saturation: + / de dt = de ds + / DS Dt (2-6) where de ds K = k D E (2-7). 2 ( K D + S) The material derivative of S is given by + / DS Dt S = t S ± v x (2-8) Making the usual assumption that the temporal gradient has a small contribution compared with the spatial gradient [30, 35], + / DS Dt S = ± v x (2-9) 15

27 Substituting Eqs. (6) through (9) into Eq. (4) gives p exp S σv k x (1 2 + / ψ ) = p D 0 E 2 ( K D + S) K (2-10) This equation indicates that cells swimming up a spatial gradient of S experience an increase in E and reduce their tumbling probability, thereby exting their run lengths. For large cell populations, the net effect is bulk energy taxis up the spatial gradient. Substituting Eq. (10) into Eq. (1), gives V = v tanh σ vk E ( K K S 2 ) x D + S D (2-11) For a shallow gradient, V is approximated by V K = χ D 0 2 ( K D + S) S x (2-12) 2 0 E where χ = v σk (2-13) Eqs. (12) and (13) have the same mathematical form as equations previously used to model chemotaxis by E. coli [10, 11] and were used in this study to model population-level energy taxis by S. oneidensis in response to spatial gradients of oxidized electron shuttle. Frymier et al. and others have discussed the need for a correction factor to compensate for differences in a cell s swimming speed in one and three dimensions and thereby adjusted the one-dimensional chemotaxis coefficient ( χ 1D 0 ) to obtain a value suitable for use in three- 16

28 dimensional models ( χ 3D 0 )[36, 37]. Theoretical values for the correction factor have been reported for one and three dimensional systems [36, 38]. A similar correction factor could be used to convert a 3D χ value into a 0 2D χ 0 value for use with the two-dimensional DGC model. However, our approach of fitting the two-dimensional DGC model to the x-y projection of 3-D patterns captured by the camera gives an effective χ 0 value that includes the correction factor, 3D 2D eliminating the need to evaluate a correction factor between χ 0 and χ 0 values The mathematical model of the DGC The mathematical model of the DGC consists of coupled, unsteady-state, differential mass balance equations for cells and the electron acceptor (i.e., fumarate). The initial carbon source concentration was at least 4 times that of fumarate. Oxidation of one lactate molecule to acetate generates four electrons, while reduction of one fumarate molecule to succinate consumes two electrons. Thus, the change in lactate concentration would be expected to be only half that of fumarate. Under these conditions, fumarate would be expected to be the rate-limiting nutrient, and the toxic effect of acetate could be neglected in the DGC model. The cell mass balance equation is u t = J u + qu (2-14) 17

29 where u is the cell density, t is the time, J u is the cell flux, and q is the specific growth rate. The cell growth rate is assumed to be limited by the electron acceptor concentration, as described by the Monod model: S q =ν CS + S (2-15) where ν is the maximum specific growth rate, and C S is the half-saturation constant for the ratelimiting electron acceptor consumption. The cell flux is expressed as a sum of random movement and chemotactic movement: Ju = µ u + V u u (2-16) Substituting Eq. (2-12), (2-15) and (2-16) into Eq. (2-14) gives u t 2 K = µ u - χ D 0 ( K D + S) 2 S u S + uν CS + S (2-17) The fumarate balance is S t 2 uν S = DS S Y CS + S (2-18) where Y is a yield coefficient (mass of electron acceptor consumed/mass of cells produced), and D S is the fumarate diffusion coefficient within the agar. A zero-flux boundary condition is applied for cells on all the DGC boundaries (Ω), because cells cannot cross the membranes or rubber sheets that separate the agar from the reservoirs: 18

30 K µ u - χ D 0 u S = 0 2 ( K D + S) Ω (2-19) A Neumann boundary condition (Eq. (2-20) and (2-21)) is used where a membrane allows fumarate exchange between the agar and a reservoir, and a zero-flux boundary condition (Eq. (2-22) and (2-23)) is used where a rubber sheet seals off the reservoirs: S ks = y y= 0 DS S ks = y y= 2R DS S = 0 x x= 0 S = 0 x x= 2R ( S Sres, S ) y= 0 (2-22) ( S Sres, N ) y= 0 (2-23) (2-20) (2-21) where S res,s and S res,n are the fumarate concentrations in the south and north reservoirs, respectively, and k S is the mass transfer coefficient for fumarate transport through the membrane. The model was solved numerically using an alternating direction implicit (ADI) algorithm as described above. The program was written in FORTRAN and MATLAB, and the results were plotted using MATLAB Estimation of model parameters The mass transfer coefficient for fumarate transport across the membrane was determined as described before [10]. Briefly, the fumarate solution was pumped through a reservoir, and an 19

31 identical solution without fumarate was placed in the arena. The arena solution was mixed using a magnetic stir bar, and the fumarate concentration was measured over time by HPLC analysis. The following mass balance describes how S would be expected to change with time as fumarate diffuses across the membrane: S0 ln( ) = S0 S ks Aarena Varena t (2-24) where S 0 is fumarate concentration in the arena, A arena is the membrane area, and V arena is the liquid volume in the arena. The experimental data were plotted in the linear form suggested by Eq. (2-24), and the resulting slope was used to calculate a k S value. A fumarate gradient was established in DGC without inoculation. The Ds value was then adjusted to optimize agreement between the model s predictions and the experimental results. The ν value was determined as the slope of the ln(cell density) versus time plot during exponential growth under anaerobic conditions. The yield coefficient (Y) was also determined from these experiments by applying the following relationship during the exponential growth phase: Y = St ut Si ui (2-25) where S t and S i are the fumarate concentrations at two different times, and u t and u i are the corresponding cell densities. The C S value was estimated from chemostat experiments that were carried out in a glove box under fumarate-limiting conditions. A peristaltic pump was used to 20

32 deliver the growth medium (LM medium with 40 mm lactate and 10 mm fumarate) to the magnetically stirred, 72 ml chemostat tank. At each dilution rate, the system was allowed to run for a minimum of ten residence times, and then steady-state was confirmed by a constant OD 600. The experimentally measured steady-state cell concentrations were plotted as a function of the dilution rate (D), and the half-saturation constant for the fumarate (C S ) was determined from non-linear regression using the following mass balance on the rate-limiting substrate (i.e., fumarate) [39]: DCS u = Y S0 ν D (2-26) A simplified capillary assay was performed to estimate cellular motility parameters. [40]. Briefly, 1 ml syringes with 22G, 1-inch, blunt-ed, stainless-steel cannulae containing a solution of attractant were placed into a suspension of motile bacteria, and the number of bacteria that accumulate inside the capillary was determined by plate counting. The µ values were estimated using following equation: 2 π 4 2 N µ = RM t πr bc (2-27) where N RM is the number of cells accumulating in the cannula of radius (r) without attractant after time (t), and b c is the initial cell number concentration in the chamber. Sensitivity curves were obtained by performing the capillary assay with a set of capillaries containing fumarate at 21

33 different concentrations. The K D value was estimated by fitting Eq. (3-14) to the data using nonlinear regression [41]. N K S = C D 2 ( K D + S) (2-28) where N is the number of cells accumulating in the cannula with attractant (i.e. fumarate), and C is a constant. The χ 0 value was estimated by Eq. (2-29) [42]: 2 ( 1+ S 0 ) N 1 (2-29) χ 0 = µ DS S S 0 N RM where S 0 and S are the dimensionless attractant concentrations initially present at the mouth of the capillary, and far into the capillary, respectively, scaled by K D. Over ten indepent replicates of the capillary assay were used to estimate both µ and K D. 2.3 Results and discussion Parameter valuation The k S and D S values are valuated as described above to be of cm/h and cm 2 /h, respectively. These values are able to predict the fumarate concentration profiles in the DGC without inoculation (Figure 2.2). 22

34 Batch cultures grown on 5 to 40 mm fumarate exhibited similar growth rate. Analysis of these data gave a Y value of g dry cell per mol fumarate ( cell density OD600 per mm fumarate (R 2 = 0.98), and a ν value of (R 2 = 0.99) (Figure 2.3A). Regression based on data from continuous cultures gave a C S of 2.92 mm (Figure 2.3A). Capillary assay experiments conducted in the absence of fumarate gave a µ value of cm 2 /h with standard deviation of Nonlinear regression was used to fit Eq. (2-28) to the data, resulting in an optimal K D value of 20 mm (Figure 2.4). The value for χ 0 was calculated by fitting Eq. (2-29). These values for random motility and chemotaxis coefficient are larger than those typically reported for other bacteria [41, 43, 44]. Additional capillary assays were conducted using traditional capillary tubes, but similar results were obtained. When these constants were used in the DGC model, the model significantly over-predicted cell migration speed. A trial-and-error approach was then used to obtain cell-transport parameters that were similar in magnitude to published values for other bacteria and provided reasonable agreement between the model s predictions and the experimentally observed behavior: µ = cm 2 /h, χ 0 = 0.09 cm 2 /h, and K D = mm. Other researchers have also reported problems using the capillary assay for Shewanella [2]. Li and coworkers attempted to characterize chemotactic responses of Shewanella to electron acceptors using capillary assays and cell enumeration via counting under a microscope. They found that fumarate elicited a response at 2 mm and that saturation occurred at electron acceptor concentrations above 100 mm [4]. These results are similar to our results (Figure 2.4) obtained using cell enumeration via colony forming units. The reason(s) for the apparent discrepancy between S. oneidensis MR1 s chemotactic behavior in the 23

35 capillary assay and that in the DGC are unclear. A single set of parameter values, listed in Table 2.1, was used for all simulations Response of Shewanella to fumarate gradient in DGC. Shewanella s growth and migration patterns in response to symmetrical, cell-generated gradients (Run 1) are shown in Figure 2.5. The experimentally observed patterns (left) are reasonably well predicted by the model (middle). The predicted velocity of the chemotactic band (0.285 cm/h, R 2 =0.999) closely matched the experimental value DGC (0.278 cm/h, R 2 =0.998) (Figure 2.7 upper plot). Predicted fumarate gradients due to Shewanella s metabolism are shown on the right. Figure 2.6, shows the comparable cell and fumarate patterns that result when an unsteady-state fumarate gradient is established in the DGC prior to inoculation (Run 2). In both experimentally observed patterns and the model prediction, an arc-shaped, chemotactic band developed and migrated up the applied fumarate gradient. The velocity of the chemotactic band measured in the DGC (0.086 cm/h, R 2 =0.997) was 11% slower than that predicted by the model (0.097 cm/h, R 2 =0.996) (Figure 2.7 lower plot). The asymmetric cell pattern observed in response to the applied gradient is strikingly different from the symmetrical, circular chemotactic band observed in Figure 2.5. Cellular uptake of fumarate south of the inoculation point generates a steep concentration gradient that triggers chemotactic band formation; however, north of the inoculation point, the fumarate gradient remains too shallow to trigger a chemotactic band. A 24

36 comparison of the experimental curves in Figures 2.5 and 2.6 indicates a slight flattening of the chemotactic band as the cells migrate into higher fumarate concentrations. A similar behavior was observed for E. coli growing into an advancing gradient of the chemoattractant aspartate in the DGC [9, 10]. In both cases, the degree of flattening observed experimentally was underpredicted by the DGC model. This minor discrepancy may arise from inaccuracies inherent in the use of a relatively simple chemotaxis model. This study validates the DGC system s ability to experimentally characterize, and mathematically predict, Shewanella s growth and tactic response under the influence of electronacceptor gradients. The DGC could also be used to investigate Shewanella s response to multiple electron acceptors, and competition with other microbial species. It could also be integrated with genomic, transcriptomic, or proteomic analyses to develop a systems-level understanding of how Shewanella s versatile redox capabilities contribute to its role in environmental. 25

37 Figure 2.1 Schematic diagram of the diffusion gradient chamber 26

38 Figure 2.2 Experimental and model-predicted fumarate concentration profiles in the DGC at times of 48 h, 72 h, 96 h, and 120 h. The fumarate concentration in the source reservoir (distance of 0 cm) was 5.7 mm, and that in the sink reservoir (distance of 5 cm) was 0 mm. Symbols show experimentally measured concentrations, and solid lines show model predictions. 27

39 Figure 2.3 Batch culture (A) and continuous culture (B) of S. oneidensis MR-1 on M1 medium with lactate concentration of 20 mm and fumarate concentrations of 1 mm (squares) and 10 mm (circles). The initial cell density was OD600nm. Both experiments were carried out under anaerobic condition at room temperature. The data represent the averages from three indepent cultures. For (B) the circles represent the experimental data with 10 mm fumarate feed, and the solid line represents the best-fit curve obtained by non-linear regression of Eq. (3-12). 28

40 Figure 2.4 Sensitivity curve for taxis toward fumarate. The circles represent cells in the capillary tubes, and the solid line represents the best fit curve obtained by non-linear regression of Eq. (3-14, 3-15). LM medium was used as buffer solution. The cell suspensions in the capillaries were serially diluted and plated onto LB plates, and the resulting colonies were counted. The capillary assays were carried out for 1 h in an anaerobic glove box at room temperature. 29

41 Figure 2.5 Experimental (left) and modeling (middle and right) results for DGC Run 1. Brighter areas correspond to higher cell densities. In the three dimensional graphs of simulated fumarate gradient profiles (right), the z axis represents the fumarate concentration (mm), and the x and y axes correspond to the dimensions of the DGC (mm). The time shown in the upper left-hand corner of experimental results corresponds to the time after inoculation. 30

42 Figure 2.6. Experimental and modeling results for DGC Run 2. The brighter areas correspond to higher cell densities. The time shown in the upper left-hand corner of experimental results corresponds to the time after inoculation. 31

43 6 No pre-established gradient 5.5 Traveling distance (cm) Experimental data regression Experimental data 2.5 Simulation regression Simulation Time (hours) Traveling distance (cm) Pre-established gradient Experimental data regression Experimental Simulation Simulation regression Time (hours) Figure 2.7 Regression based on the experimental data and simulation data. The slop of the regression lines represent the average migration speed of chemotactic band without preestablished gradient (Run 1, upper plot) and with pre-established unsteady state gradient (Run 2, 32

44 low plot). For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of this dissertation. 33

45 Table 2.1 Constants used in simulations. Parameter Unit Estimated value D S cm 2 h k S cm h Y (g dry cell) (mol fumarate) C S mm 2.92 ν h µ cm 2 h χ 0 cm 2 h K D mm

46 CHAPTER 3 SEPARATION OF S. ONEIDENSIS MR-1 MUTANTS BY OPPOSING GRADIENTS OF ELECTRON ACCEPTORS 3.1 Introduction Ziemke et al. used Randomly Amplified Polymorphic DNA (RAPD) genotyping and water column of Gotland Deep to study the genetic structure of S. balitca population, the dominant culturable denitrifier in the low oxygen and anoxic water of central Baltic Sea [15]. They proposed a hypothesis that distribution of the S. baltica strains with different RAPD genotypes in the water column was largely based on ecological niching, which assumes that a specific adaptation of a given RAPD genotype to a specific environment. They believed that taxis towards specific electron acceptors could be important in this process [15]. The DGC has been shown to be able to separate bacteria on the basis of their motility and chemotactic responses to gradients of attractants [9]. A wild type strain and mutant strain of E. coli could be physically sorted with the DGC according to different physiological profiles [11]. Hence, to verify Ziemke et al. s hypothesis, we exted the DGC approach described above to include gradients of nitrate and fumarate and to include mutants in fumarate and nitrate reductase genes. Our experiments demonstrated the DGC s ability to separate populations of S. oneidensis MR-1 strains based on their different chemotactic responses to electron-acceptor gradients, providing insight into the role of chemotaxis in the microbial ecology of Shewanella species. 3.2 Method and materials 35

47 3.2.1 Strains and growth conditions Two mutants of S. oneidensis MR-1, one unable to reduce fumarate and the other unable to reduce nitrate, were created by deletions of the fumarate reductase or nitrate reductase gene ( fcca or napa), respectively, (The fcca mutant also contained the green fluorescent protein gene gfpmut3 stably integrated into the chromosome in the Tn7 attachment site, allowing us to distinguish the cell populations based upon fluorescence). The green fluorescent protein (GFP) reporter gene (gfpmut3*) was inserted into the chromosome of a fumarate mutant as described by Ciche et al. [45]. Briefly, donor strain E. coli WM3064 harboring the plasmid purr25 (encoding minitn7ksgfp) and E. coli BW29427 harboring plasmid pux-bf13 (encoding Tn7 transposase) were grown and mated with the S. oneidensis MR-1 cells by triparental mating on LB agar containing DAP. Following mating, GFP-labeled exconjugant cells were selected in the absence of DAP and presence of kanamycin and streptomycin and tested for GFP fluorescence by using a SpectraMax M2 Multi-Mode microplate reader (Molecular Devices, Sunnyvale, USA) set at an excitation wavelength of 475 nm and emission detection at 515 nm. All the strains studied are listed in Table 3.1. The cells were grown on LB medium was amed with 30 µg of kanamycin/ml, 20 µg of streptomycin/ml, and 300 µg of diaminopimelic acid (DAP)/ml when required DGC experiments and modeling A nitrate gradient (0 to 2 mm) was established in the opposite direction to the fumarate gradient (12 to 0 mm). Both the fluorescent fumarate reductase mutant (JMA1023), which responded chemotactically to fumarate but not nitrate; and the non-fluorescent nitrate reductase 36

48 mutant (GCC01), which responded chemotactically to nitrate but not fumarate [6], were inoculated 12 h after gradient initiation. Samples were taken from the chemotactic moving band 14 h after inoculation using a 1 ml syringe. A LSR II Flow Cytometer (BD, Sparks, USA) was used to test cell fluorescence with blue laser at 488 nm, FITC detector, 505 LP longpass dichroic mirror and 530/30 bandpass filter. After inoculation, the DGC was stationed on a transilluminator box located in the anaerobic chamber as described above. The model used for single gradient and single culture in the DGC was used to describe the fumarate gradient and nitrate reductase mutant. Additional balances were added to describe the nitrate gradient and fumarate reductase mutant. Briefly, the balances for both strains had the same random motility term. However, the balance for JMA1023 contained chemotaxis and consumption terms deping on nitrate concentration, and the balance for GCC01 contained chemotaxis and consumption terms deping on fumarate concentration. The balance for both electron acceptors had same diffusion term but different consumption terms for the two strains. The model was solved numerically using an alternating direction implicit (ADI) algorithm as described above. 3.3 Results and discussion When a mixture of the two mutants, one unable to reduce fumarate and the other unable to reduce nitrate were placed in opposing gradients of fumarate and nitrate, two chemotactic bands developed, one moving toward each chemoattractant source (Figure 3.1). Flow cytometry for cells collected from the leading edge of each chemotactic band indicated that 92.1% ± 5.0% 37

49 of cells moving toward the fumarate source were non-fluorescent (i.e., nitrate reductase mutants), while 87.5% ± 5.2% of cells moving towards the nitrate source were fluorescent (i.e. fumarate reductase mutants). The percentage of cells having the fluorescence trait expected for that chemotactic band may have been less than 100% due to background noise from agar and dead cells; our positive control containing only fluorescent cells under same conditions also showed a similar fluorescent percentage (89.4% ± 6.1%) (Figure 3.2). However, we cannot rule out the possibility that a small amount of nitrate reductase mutants were attracted by the nitrite produced during nitrate reduction of the other strain. These results indicate that Shewanella populations arising from mutant strains that differ in only two genes can be physically separated via differential chemotaxis traits. Cruz-García et al. reported that during S. onediensis MR-1 nitrate reduction in liquid culture, the intermediate product nitrite accumulates to levels that strongly inhibit cell growth [46]. Our batch growth experiments in liquid culture confirmed that the fumarate reductase mutant cells cannot grow with 2 mm nitrite in the liquid LM media (data not shown). However, in separate control experiments these cells grew and migrated tactically in LM agar containing 2 mm nitrite (data not shown), and no evidence of inhibition was observed in the DGC experiment. This result indicates cell growth, metabolism and tolerance to toxic substances may be different in liquid culture and agar. The concentration of nitrite accumulating in the agar was very low (1 ppm, about mm) around cell migration rings shown in Figure 3.1 (tested as described by Cruz-García et al. (11)). Based on these observations, we did not include nitrite accumulation, inhibition, or chemotaxis towards nitrite in the DGC model. Cell-transport constants for the modified DGC model were adjusted to reasonably match the experimental results (Figure 3.1). The simulation results predicted that near the nitrate source, 100% of cells are fumarate reductase mutants, and close to fumarate source, 100% of 38

50 cells are nitrate reductase mutants (Figure 3.2). Thus the model was able to reproduce the observed, chemotaxis-driven physical separation of two MR-1 mutants based on differences in their ability to metabolize, and thus create spatial gradients of, electron acceptors. This finding is consistent with Ziemke et al. s hypothesis that ecological niches constructed by various electron acceptor gradients might be responsible for the distribution of S. baltica strains with different genotypes [15]. While the vertical gradients in average nitrate concentration the Baltic Sea should be too shallow to induce chemotaxis, the average nitrate concentration provides a background in which S. baltica populations can generate local nitrate gradients steep enough to induce chemotaxis when carbon sources become available. Strains whose genotype is ideally suited for a given background nitrate concentration would migrate most effectively into nutrientrich areas. Over many generations, even subtle competitive advantages between strains could result in a genetic stratification with depth in the Baltic Sea in response to stratification in nitrate concentration. The DGC s ability to generate well-controlled electron-acceptor concentration gradients enables the impact of differences in chemotactic properties on competition to be investigated. While our study demonstrated competition between strains having extreme differences in ability to use nitrate (wild-type vs. nitrate-reductase mutant), a similar approach could be also used with strains having more subtle differences. 39

51 Table 3.1 Strains and plasmids Strain or plasmid Description Reference Strains Shewanella oneidensis MR -1 Wild type [1] S. oneidensis MR-1 mutants CCG01 napa [46] SO0970 fcca [47] JMA1023 fcca minitn7-gfpmut3* This study E. coli WM3064 BW29427 Plasmids purr25 pux-bf13 Donor strain for conjugation, dap auxotroph Donor strain for conjugation, dap auxotroph, tra pir Mini Tn7KsGFP, GFP driven by P lac (P A1/04/03 ) promoter, mobilizable orit IncP α, suicide orir R6k γ; Ap r (bla) Tn7 transposase genes tnsabcde, mobilizable orit IncP α, suicide orir R6k γ; Ap r (bla) [48] [45] [48] [49] 40

52 Figure 3.1 Experimental (left) and modeling (right) results. Brightness and contrast of experimental images were adjusted for clarity. Fumarate concentration was 12 mm in the south reservoir and 0 mm in the north reservoir. Nitrate concentration was 2 mm in the north reservoir and 0 mm in the north reservoir. The west and east reservoirs were sealed off. A mixture of S. oneidensis MR-1 fumarate-reductase mutants and nitrate-reductase mutants was inoculated 12 h after gradient initiation. Brighter areas correspond to higher cell densities. The time shown in the upper left-hand corner of experimental results corresponds to the time after inoculation. 41

53 100.00% 80.00% A. Non-fluorescent population (nitrate reductase mutants) 60.00% 40.00% 20.00% Experimental data Simulation 0.00% Nitrate source Fumarate source % 80.00% B. Fluorescent population (fumarate reductase mutants) 60.00% 40.00% 20.00% Experimental data Simulation 0.00% Fumarate source Nitrate source Figure 3.2 Results of flow cytometry and simulation for non-fluorescent population (A) and fluorescent population (B). Error bars indicate standard deviation of three replicates. 42

54 CHAPTER 4 ENERGY TAXIS TOWARD INSOLUBLE ELECTRON ACCEPTORS MEDIATED BY SOLUBLE ELECTRON SHUTTLES 4.1 Introduction Microbial reduction of insoluble electron acceptors is important in microbial physiology, biogeochemical cycles of metals [1], and microbial fuel cells [5]. Shewanella species reduce a wide range of terminal electron acceptors, including iron oxides (Fe(OH) 3, Fe 2 O 3 ) [50] and manganese dioxide (MnO 2 ) [51]. Insoluble electron acceptors or their reduction products also play a role in bioremediation of uranium [52, 53]. The Mtr respiratory pathway has been proposed for metal reduction by Shewanella [54], in which the terminal reductase MtrC either directly reduces an insoluble terminal acceptors [22], or reduces a self-secreted electron shuttle molecule [55] that transfers the electrons to the terminal acceptors [27, 28, 56]. Biogenic, nonreductive chelation of metal oxides into soluble organic-metal complexes may also help cells transfer electrons through the cell membrane to insoluble electron acceptors [25, 26]. When the availability of soluble electron acceptors becomes rate-limiting, many bacteria migrate tactically up a concentration gradient into a region having a higher electron-acceptor concentration. Bacterial taxis is a biased random walk consisting of straight runs interspersed with tumbles. During runs, cells continuously monitor either an extracellular chemoattractant concentration or an intracellular indicator of energy production. When an increase is detected, the run is exted [57], resulting in chemotaxis or energy taxis, respectively. 43

55 Shewanella cells exhibit the unusual ability to migrate tactically through dilute suspensions of insoluble electron acceptors in the absence of soluble electron acceptors. Under these conditions, there would presumably be no continuous electron-acceptor gradient on the length scale of a cell s run, suggesting that a tactic mechanism other than conventional chemotaxis or energy taxis may be involved. Childers et al. suggested that Geobacter, another metal-reducing bacterium, might use chemotaxis toward soluble reduced metal species, such as Mn(II) and Fe(II) [58]. Bencharit and Ward found that S. oneidensis MR-1 cells exhibited chemotaxis towards soluble reduction products like Mn(II) and Fe(II), and might use this ability to locate sources the insoluble electron acceptors Mn(III/IV) and Fe(III) oxides for dissimilatory purposes [7]. They also suggested that the humic acid analog anthraquinone-2,6-disulfonate (AQDS) was another attractant of S. oneidensis MR-1 [7]. Because AQDS is an exogenous electron shuttle for Shewanella fuel cells [59], we hypothesized that Shewanella uses the oxidized form of a self-secreted electron shuttle as a tactic attractant to migrate toward nearby insoluble electron acceptors. Flavins, including FMN and its hydrolysis product riboflavin, have been shown to be secreted by various Shewanella species, even in the presence of oxygen [27, 28, 60], suggesting that Shewanella might use one or more flavin(s) as an electron-shuttling attractant(s). In this study, we conducted experimental and modeling studies to establish a novel mechanism (mediated energy taxis) by which Shewanella simultaneously enhances respiration and achieves taxis toward insoluble electron acceptors. In this mechanism, Shewanella secretes a reduced electron shuttle (e.g., riboflavin [60]) that is oxidized by nearby insoluble electron acceptors. The resulting oxidized shuttle then serves as an attractant to direct energy taxis of Shewanella cells toward the insoluble electron acceptors. Motility assay experiments were used 44

56 to demonstrate that S. oneidensis MR-1 is tactic toward oxidized riboflavin and exhibits tactic behavior in the presence of dilute suspensions of insoluble electron acceptors (e.g., MnO 2 particles). The migration velocity and cell density of the tactic bands were shown to vary with concentrations of both riboflavin and MnO 2 particles. A mathematical model based on the new mechanism was developed and used to reproduce experimentally observed trs in S. oneidensis MR-1 s growth, riboflavin-mediated taxis, and MnO 2 reduction. 4.2 Materials and methods Strains and growth conditions Wild-type S. oneidensis MR-1, its in-frame chea-3 deletion mutant [8], its SO2240 SO3282 double-deletion mutant (ΔSO2240ΔSO3282) [6], S. putrefaciens CN-32, and S. sp. W were tested in this study. The ΔSO2240ΔSO3282 mutant does not express a major and a minor methyl-accepting chemotaxis protein (MCP) involved in energy taxis in S. oneidensis [6], and has partially abolished energy taxis capability. Cells were grown either aerobically on LB medium at 30 C or anaerobically on M1 basal medium [1] at room temperature containing sodium lactate and electron acceptors at concentrations indicated below Taxis assays 45

57 Motility assays [7] were conducted anaerobically to test S. oneidensis MR-1 s tactic responses. Twenty ml soft (0.25%) M1 agar containing 40 mm of the non-tactic carbon source, sodium lactate [2, 61], and an electron acceptor (riboflavin and/or insoluble electron acceptors) at various concentrations was poured into each petri dish. In all cases, the carbon source was present in excess, and the electron acceptor was the limiting nutrient. Freshly prepared amorphous δmno 2 or amorphous Fe(OH) 3 particles [1] were grounded into powder by mortar and pestle. Crystalline MnO 2 and crystalline Fe 2 O 3 were purchased from Sigma-Aldrich. These metal oxides dispersed readily and remained susped in the agar. Cells for inoculation were grown aerobically in 5 ml LB medium at 30 C overnight to an OD 600 nm of 1.1, washed by centrifugation (3220 g) and re-susped in 100 µl 30 mm HEPES buffer. Riboflavin was added to the cell suspension to a final concentration 200 nm unless otherwise indicated, and then a sterile pipette tip was used to inject 8 µl of the cell suspension into the swarm agar. Adding a trace amount of riboflavin into the cell suspension prior to inoculation was found to increase repeatability of the experiments. However, variation of riboflavin concentrations in the inoculum from 100 nm to 1000 nm did not lead to obvious differences in results for both experiments and mathematical simulations. Inoculated plates were incubated in an anaerobic chamber containing 4% H 2 (the balance in N 2 ) at room temperature. A digital camera was used to record the tactic migration patterns while the motility assay plate was mounted on a transilluminator box [10] Riboflavin concentration analysis 46

58 Liquid chromatography-mass spectrometry (LC/MS/MS) analysis was performed to verify riboflavin secretion in experiments conducted without riboflavin added to either the agar or the inoculum. Samples were centrifuged at 3220 g for 10 min to settle the agar, MnO 2 and most cells. The supernatants were then passed through a 2 um pore size filter. A Q TRAP 3200 LC/MS/MS system (Applied Biosystems) was used in the positive ion mode. The LC column was a 50 mm 2.1 mm Ascentis Express C18 (Supelco) (2.7 μm particle size). 4.3 Results and discussion Conceptual model The conceptual model underlying mediated energy taxis (Figure 4.1) is based on following premises: 1: Shewanella cells eliminate electrons produced during respiration by secreting reduced riboflavin, which diffuses through the medium, shuttling electrons from the cells to nearby insoluble electron acceptors; 2: the insoluble electron acceptors oxidize the reduced riboflavin in a reaction that consumes the insoluble electron acceptors and generates oxidized riboflavin; 3: the oxidized riboflavin serves as an attractant for Shewanella energy taxis; 4: the insoluble electron acceptors do not diffuse but are small enough and well enough dispersed that a continuum model based on an insoluble electron acceptors concentration is sufficient to predict population-level cell behavior. A possible variant of the conceptual model was also considered: that the species driving taxis is a soluble product of the redox reaction. However, in this scenario, respiratory activity by a tactic band moving in the positive x direction would generate a positive gradient of oxidized 47

59 substrate and a negative gradient of reduced product. If the reduced product were driving taxis in the direction opposite to its gradient, the product would have to be a chemorepellent. This situation would be inconsistent with the observations that the reduction products of MnO 2 reduction, Mn(II) and Mn(III), are chemoattractants for Shewanella [2, 7]. However, inorganic Mn(III) is unstable without chelates due to fast disproportionation [62] and can be reduced by HEPES [2], the buffer used in this study. Hence, Mn(III) was considered unlikely to play a prominent role under our experimental conditions Taxis toward riboflavin and MnO 2 Because oxidized riboflavin is yellow and reduced riboflavin is colorless, tactic migration of S. oneidensis MR-1 populations created a sharp, circular yellow-to-clear swarm boundary that separated the outer (yellow) region containing oxidized riboflavin from the inner (clear) reduction zone (Figure 4.2 A). Riboflavin s demarcation of the swarm boundary was fortuitous, because the S. oneidensis MR-1 cell density in the tactic cell bands was typically too low to be visible [7]. A tactic response was not observed for the non-tactic chea-3 mutant (Figure 4.2 A), indicating that chea-3 is essential for S. oneidensis MR-1 taxis [8] under the conditions tested. The ΔSO2240ΔSO3282 double mutant exhibited a swarm boundary that migrated more slowly than that for the wild-type strain (Figure 4.2 A), consistent with results for other soluble electron acceptors [6, 7]. In motility assay plates containing flavin mononucleotide (FMN), another yellow-colored redox cofactor secreted by Shewanella species [27, 28], formation and migration of a yellow-to-clear swarm boundary was also observed (Figure 4.2 B). 48

60 In motility assay plate experiments conducted without riboflavin but with amorphous MnO 2 (δmno 2 ) particles uniformly dispersed throughout the agar, an expanding visible swarm boundary was again observed (Figure 4.2 C), similar to that seen with soluble electron acceptors. As before, the chea-3 mutant did not show a tactic response, and the ΔSO2240ΔSO3282 doubledeletion mutant (Figure 4.2 C) showed slower tactic migration than the wild-type. Over time, a few black particles appeared in the gel (Figure 4.2 C) which were not reduced by the tactic cell band. These particles were presumably larger MnO 2 particles that grew via Ostwald ripening [63]. We did not attempt to develop a mutant unable to secrete flavins, because flavins serve as essential enzyme cofactors [60]. Experiments were conducted to determine whether secreted riboflavin may have been involved in mediating δmno 2 particle reduction and cellular taxis in motility assays in which no riboflavin was added to either the inoculum or the agar containing the δmno 2 particles. Samples were taken before and after inoculation, both inside and outside the swarm boundary. LC/MS/MS analysis detected no riboflavin for samples taken before inoculation and outside the migration band. However, riboflavin was detected in samples taken inside the tactic band at an average concentration of about 100 nm, consistent with previous reports [27, 64]. The MS signal attributed to riboflavin had a mass-to-charge ratio (m/z) of 377.2, and secondary MS analysis of the peak yielded an ion with a m/z ratio of 243.1, identical to values reported for riboflavin [27]. Supplementing the δmno 2 -containing agar with low concentrations of riboflavin prior to inoculation increased the swarm boundary s average migration rate. During 72 h experiments 49

61 containing 10 mm δmno 2 particles dispersed in the agar, the migration rate increased from cm/h (R 2 = 0.96) without riboflavin to cm/h with mm riboflavin in the agar (R 2 = 1) (Figure 4.3 A and B). The increase in migration rate with increasing riboflavin concentration suggests that the riboflavin concentration secreted by Shewanella cells (~100 nm) is enough to mediate reduction of insoluble electron acceptors and energy taxis but not enough to saturate the cells respiratory capacity. Riboflavin supplementation also resulted in formation of a white, insoluble reduction product, possibly Mn(II) carbonate [2] (Figure 4.3 B and F). In motility assays conducted using crystalline MnO 2 without added riboflavin, no swarm boundary was observed within 48 h (Figure 4.3 C). However, supplementing the crystalline MnO 2 with riboflavin did yield tactic behavior (Figure 4.3 D). In some cases, tactic migration of Shewanella cells through agar pre-loaded with riboflavin and MnO 2 resulted in concentric, stationary circles (Figure 4.3 E and F) that were stable after exposure to oxygen for 1 week or longer. The rings are presumably composed of insoluble Mn(II) compounds. S. putrefaciens CN-32 and S. sp. W were also tested for their responses to riboflavin and MnO 2. These strains have the highest reported coulombic efficiency among Shewanella strains in biofuel cells [65], indicating they effectively donate electrons to insoluble electron acceptors (i.e., electrodes). In motility plates with riboflavin and MnO 2, these strains exhibited tactic behavior similar to S. oneidensis MR-1. In motility assays containing amorphous Fe(OH) 3, reduction zones were observed for the wild-type strain but not for the chea-3 mutant. However, the double-deletion mutant gave results 50

62 similar to the wild-type strain (Figure 4.4 A). As was observed for δmno 2, supplementing Fe(OH) 3 -containing agar with riboflavin prior to inoculation increased the swarm boundary s migration rate. During 72h experiments containing 10 mm Fe(OH) 3 particles dispersed in the agar, the migration rate increased from cm/h without added riboflavin (R 2 = 0.97) to cm/h with mm riboflavin in the agar, R 2 = 1) (Figure 4.4 B and C), presumably due to a riboflavin-mediated increase in Fe(OH) 3 reduction rate of Fe(OH) 3 with increased riboflavin concentration [28]. To further confirm that the color change in the reduction zone is not due to some abiotic process, 2 µl samples were taken from plates with 0.2 mm riboflavin, 0.2 mm FMN, 10 mm δmno 2, and 10 mm amorphous Fe(OH) 3 by using capillary tubes, and the cell numbers were counted. Within the reduction zones, average cell densities were 663±93 cells/µl on the riboflavin plates, 688±83 cells/µl on the FMN plates, 7112±1124 cells/µl on δmno 2 plates, and 3071±1812 cells/µl on amorphous Fe(OH) 3 plates. No cells were found in samples taken outside the reduction zones Mathematical model of mediated energy taxis by S. oneidensis MR-1 The mathematical formulation is based on one developed by Widman et al. to describe Escherichia coli chemotaxis in a diffusion gradient chamber [10-12], and later modified by Li et al. to describe S. oneidensis MR-1 taxis towards soluble electron acceptors [61]. The model 51

63 consists of coupled, unsteady-state mass balance equations for cells, oxidized riboflavin, reduced riboflavin, and the insoluble electron acceptor. The cell mass balance is u t 2 K = µ u - χ0 D 2 ( K D + S) S u S + νu CS + S k d u (4-1) where u is the cell density, t is time, μ is the random motility coefficient, χ 0 is the taxis coefficient, S is the concentration of oxidized riboflavin, K D is the taxis saturation constant, ν is the maximum specific growth rate, C S is the half-saturation constant for consumption of the ratelimiting electron acceptor, and k d is the death rate constant. The first term on the right-hand side of the equation accounts for random motility, and the second term, which has a mathematical form reminiscent of the RTBL chemotaxis model [30], has been used to describe populationlevel energy taxis [37, 61]. The last two terms describe cell growth and death, respectively. The electron acceptor was the limiting nutrient under the experimental conditions, so the growth is assumed to be limited by the oxidized riboflavin concentration. The death rate is assumed to be proportional to the cell concentration. The mass balance for extracellular oxidized riboflavin is S t 2 u( ν + km ) S = DS S + krqm Y CS + S (4-2) 52

64 where D S is the diffusion coefficient for oxidized riboflavin in the agar, Y is a yield coefficient (mass of cells produced/ amount of oxidized riboflavin consumed), k m is the maintenance energy rate constant, k R is the rate coefficient for oxidation of reduced riboflavin by MnO 2, Q is the reduced riboflavin concentration, and M is the MnO 2 concentration. The second term describes consumption of oxidized riboflavin due to cell growth and maintenance. The last term accounts for generation of oxidized riboflavin as reduced riboflavin is oxidized by MnO 2. Since a detailed mechanism for the reaction between riboflavin and MnO 2 particles is not yet available, we assumed the reaction rate was proportional to the concentrations of both reduced riboflavin and MnO 2. The mass balance for extracellular reduced riboflavin is Q t 2 u( ν + km ) S = DS Q + Y CS + S + ksu krqm (4-3) The second term describes reduction of oxidized riboflavin by cellular respiration for growth and maintenance. In the third term, which accounts for riboflavin secretion, k S is the riboflavin secretion rate constant. The diffusion coefficients for the oxidized and reduced forms of riboflavin are assumed to be identical. Based on the assumption that cells would not secrete additional riboflavin once the extracellular concentration reached a saturating level, k S is set to zero if S + Q exceeded mm. 53

65 The MnO 2 balance, which describes consumption via reaction with reduced riboflavin, is M t = k R QM (4-4) Zero-flux boundary conditions are applied for all components on all boundaries (Ω): µ u - χ K ( K D 0 u S = 2 D + S) Ω 0 (4-5) S Ω = 0 (4-6) Q Ω = 0 (4-7) M Ω = 0 (4-8) Because of the inherent complexity of this system, an exhaustive effort was not made to indepently evaluate all constants. D S was evaluated experimentally using a diffusion gradient chamber [61]. Values for μ and χ 0 for S. oneidensis MR-1 were obtained from the literature [61]. K D and C S were estimated to be around mm, the reported concentration of Shewanellasecreted riboflavin [27, 64]. A trial-and-error approach was used to determine values for the other parameters that gave reasonable agreement between the model and experimental results. A single set of parameter values, listed in Table 4.1, was used for all simulations. Initial conditions for Q and M were Q(x,y) = 0 and M(x,y) = M 0 for all x and y. Initial conditions for u and S were u(x,y) = u 0 and S(x,y) = S 0 for x, y within the inoculation zone, and u(x,y) = 0 and S(x,y) = 0 for 54

66 x, y outside the inoculation zone. The values of M 0, u 0 and S 0 were experimentally determined. An alternating direction implicit (ADI) algorithm described previously [10] was used to numerically integrate the system of coupled, partial differential equations. Experimental results and model predictions depicting energy taxis by S. oneidensis MR-1 in response to oxidized riboflavin are compared for different times and initial riboflavin concentrations in Figure 4.5 (top) and Figure 4.6 (top). Two trs observed experimentally were reproduced by the model. First, a tactic cell band (swarm boundary marked by a yellow-to-clear transition) forms and migrates away from the inoculation point, leaving a circular reduction zone in its wake. Second, a higher initial riboflavin concentration results in a slower migration rate of the swarm boundary. The leading edge of the boundary appears sharper in the experimental results than in the simulations. While there was some minor nonlinearity between the yellow color intensity recorded by the camera and riboflavin concentration (Figure 4.7), the difference in sharpness is likely due to the use of the relatively simple RTBL model to describe complex cellular behavior governing taxis. Previous published models have also underestimated the sharpness of chemotactic bands [10, 61]. Experimental results and model predictions depicting energy taxis by S. oneidensis MR-1 in response to the insoluble electron acceptor (MnO 2 ) are also compared for different times and initial MnO 2 concentrations in Figure 4.5 (bottom) and Figure 4.6 (bottom). The same two trs described above for riboflavin were also observed both experimentally and in the model predictions. However, the experimental migration rate decreased with time, likely due to an aging process Mn oxides undergo in aqueous suspension that makes them less reactive [1]. 55

67 Simulations were also performed using model parameters that eliminated chemotaxis, random movement, or flavin secretion to explore the influence of these processes on predicted cell growth and migration patterns. Eliminating chemotaxis or random movement significantly reduced the predicted the migration rates (Figure 4.8 A, B and C), consistent with experimental data. Eliminating flavin secretion significantly decreased both migration rate and MnO 2 reduction rate, even when riboflavin initial concentration at the inoculation point was set to 100 nm (Figure 4.8 D). The mathematical model s predictions showed reasonable quantitative agreement with the experimental results across the wide range of experimental conditions tested. Following validation, the model was used to predict variables that are difficult to measure in order to elucidate mechanisms underlying tactic band formation via mediated energy taxis. Figure 4.9 shows predicted concentration profiles of cells, MnO 2, Mn(II), reduced riboflavin, and oxidized riboflavin across a Shewanella tactic band, with variables scaled so that all five profiles can be viewed on one plot. The shapes of these profiles are intuitively reasonable and illustrate several aspects of the assumed mechanism. The chemotactic band of cells migrates outward from the inoculation point into MnO 2 -rich regions. Reduced riboflavin secreted by the cell band is rapidly oxidized by MnO 2 outside the cell band but accumulates inside the cell band, where the MnO 2 has been depleted, resulting in band of reduced riboflavin that tracks the cell band. Reduction of MnO 2 in the vicinity of the cell band results in a symmetrical patterns indicating Mn(II) production at the expense of MnO 2. The sharp gradient of oxidized riboflavin within the cell 56

68 band generates the concentration gradient that drives energy taxis of cells into the cell band, generating the steep cell band. Collectively, the experimental and modeling results presented here establish how mediated energy taxis enables Shewanella to migrate tactically in environments containing insoluble electron acceptors as the sole electron acceptor, even when neither the insoluble electron acceptors nor its reduction product is a chemoattractant. The observation that the tactic wave solubilizes MnO 2 indicates a link between taxis and respiration. Riboflavin s ability to cycle between a reduced form that shuttles electrons from the cells to MnO 2 particles and an oxidized form that serves as a tactic attractant enables it to establish that link. Energy taxis differs from metabolism-indepent chemotaxis, in that chemotaxis requires a sensor specific for the chemoattractant, whereas energy taxis uses a generic sensor for some energy-related property [17, 66]. For example, receptor SO2204, which plays a major role in S. oneidensis MR-1 s energy taxis towards electron acceptors, is believed to participate in detecting a transmembrane ph [6]. This feature of energy taxis avoids a potential problem inherent in classical chemotaxis, in which chemicals having a structure similar to a metabolically important chemoattractant trigger futile chemotaxis. In addition, this feature allows a single sensor system to direct taxis toward multiple insoluble electron acceptors, consistent with Baraquet et al. s observation that among 27 MCP-encoding genes of S. oneidensis MR-1, only five influenced taxis [6]. The novel mediated energy taxis mechanism presented here is supported by several observations in the literature. Harris et al. reported that S. oneidensis MR-1 cells exhibited an increased run length in the presence of insoluble electron acceptors [31]. Such behavior is 57

69 expected, because the insoluble electron acceptors would oxidize riboflavin secreted by the cells, generating oxidized riboflavin that would activate energy taxis and thereby increase run length. Harris et al. reported that ΔcheA-3 mutant was nearly non-chemotactic in the presence of MnO 2 [31], consistent with our findings. This lack of a tactic response suggests that the chea protein s activity is downstream of the energy sensor in the chemotaxis signaling pathway. Harris et al. [31] reported an increase in Shewanella swimming speed in the presence of insoluble electron acceptors (positively poised electrodes and MnO 2 ). They suggested the effect was indepent of electron shuttles because the shuttles were present throughout the solution, while the response was only observed near the insoluble electron acceptors. Instead, they attributed the behavior to a previously unreported phenomenon electrokinesis, which they defined as enhanced motility near insoluble electron acceptors in the absence of shuttle activity. However, shuttle activity cannot be ruled out, because, even though the the electron shuttle may be present throughout the solution, the oxidized form is likely to be present only near the insoluble electron acceptors. Thus, oxidized riboflavin could represent a previously unrecognized electron sink present only near the insoluble electron acceptors that could enhance energy production via respiration and explain the increase in swimming speed. Our results indicate that the riboflavin-mediated energy taxis can be significant even at very low riboflavin concentrations, making its presence and influence easy to overlook. The proposed mechanism is also reasonable from the standpoint of thermodynamics. Riboflavin has a redox potential of 0.21 V vs. SHE [27]. Substances having redox potentials more positive than this value would be suitable insoluble electron acceptors for riboflavinmediated energy taxis. Moreover, the larger the potential difference between the insoluble 58

70 electron acceptors and riboflavin, the greater would be (1) the driving force for electron transfer, (2) the oxidized fraction of riboflavin on the insoluble electron acceptors surface, (3) the driving force for diffusion of reduced riboflavin from the cells to the insoluble electron acceptors and (4) the driving force for diffusion of oxidized riboflavin from the insoluble electron acceptors to the cells. These trs are consistent with Harris et al. s observation that electrodes with higher potential and metal oxides with higher redox potential range triggered a stronger response [31]. The first-generation mathematical model presented here was designed to illustrate the mechanism of mediated energy taxis and to reproduce experimental trs with minimal complexity. The effects of riboflavin and FMN were lumped together, because FMN has the same electron carrying capacity as riboflavin [27, 28], can hydrolyze to form riboflavin [60], and has been reported to be undetectable in Shewanella cultures under some conditions [64]. However, additional balances could easily be added to describe multiple shuttles. The derivation of the chemotactic velocity expression assumes a constant swimming velocity. While an increase in swimming speed has been reported as Shewanella cells approach insoluble electron acceptors [31], this effect has been reported to be insignificant in modeling energy taxis [32]. As additional data that relate swimming speed to oxidized shuttle concentration and/or some intracellular measure of energy level become available, more detailed energy-taxis models could be developed. Influences of additional environmental factors on swimming speed could be also incorporated [67-69]. These and other refinements would improve the model s accuracy and facilitate its use in testing hypotheses about mediated energy taxis and designing bioremediation systems for toxic metals that undergo microbial reduction. Finally, this study provides new insight into the relative competitive advantages of Shewanella s mechanism to find and utilize insoluble electron acceptors (mediated energy taxis) 59

71 vs. Geobacter s mechanism (chemotaxis toward soluble products of metal-reduction reactions). Previous discussion of this topic [58] recognized that Shewanella s secretion of shuttles allows it to discard excess electrons but overlooked what may be its primary benefit: enabling taxis toward insoluble electron acceptors. One potential advantage of mediated energy taxis is that it would likely work for virtually any insoluble electron acceptor that can oxidize riboflavin, whereas chemotaxis would only work when the reduced form of the electron acceptor is both soluble and a chemoattractant. A second advantage is that mediated energy taxis would require a smaller inventory of sensor proteins. Shewanella would only need a sensor for the oxidized form of its secreted shuttle(s), whereas Geobacter would typically require a different receptor for each reduced metal species. A third advantage is that Shewanella's use of diffusional shuttles eliminates the need for direct contact between the cells and the insoluble electron acceptor and thus may be better suited than Geobacter s mechanism when insoluble electron acceptors are embedded in microporous matrices. A fourth advantage is that secretion of compounds that serve as attractants can lead to microbial pattern formation and quorum sensing, which can provide additional competitive advantages, including mitigation of damage from hazardous materials [70]. Collectively, these potential advantages of mediated energy taxis help justify the energetic cost of Shewanella s electron-shuttle secretion in environments containing insoluble electron acceptors. 60

72 Figure 4.1 Schematic diagram illustrating microbial reduction of insoluble electron acceptors via riboflavin mediated energy taxis. Oxidation of secreted reduced riboflavin by an insoluble electron acceptor (e.g., a metal surface) generates a gradient of oxidized riboflavin that simultaneously serves as an electron sink and a chemoattractant that directs Shewanella migration toward the insoluble electron acceptor. 61

73 Figure 4.2. Motility assays for wild-type S. oneidensis MR-1 (MR-1), its SO2240 SO3282 double-deletion mutant (Δ2240Δ3282) and chea-3 mutant (Δ chea-3). The soft agar contained 0.2 mm riboflavin (A), 0.1 mm FMN (B) or 10 mm MnO 2 (C). Pictures A and B were taken 24 h after inoculation. Yellow areas indicate oxidized riboflavin or FMN, whereas dark areas around inoculation point indicate reduced electron acceptors. Picture C was taken 36 h after inoculation. Brown areas indicate un-reduced MnO 2, whereas dark areas around inoculation point indicate reduced MnO 2. 62

74 Figure 4.3 Rate of radial movement of the reduction zone boundary was increased in agar plates preloaded with 10 mm MnO 2 (A), 10 mm MnO 2 plus mm riboflavin (B). Shewanella cells could not reduce 10 mm crystalline MnO 2 without added riboflavin (C), but could reduce it if the agar were preloaded with (0.002 mm) (D). Agar plates preloaded with 10 mm MnO 2 and either mm riboflavin (E) or 0.02 mm riboflavin (F) formed concentric, stationary circles. All pictures were taken 48 h after inoculation with S. oneidensis MR-1. Brown areas indicate unreduced MnO 2, whereas dark areas around inoculation point indicate reduced MnO 2, and white particles within dark areas are presumed to be insoluble reduction products. 63

75 Figure 4.4 Motility assays for S. oneidensis MR-1, its SO2240 SO3282 double-deletion mutant and chea-3 mutant with 10 mm Fe(OH) 3 (A). Reduction rate and expanding speed of transition zones were increased in agar plates preloaded with 10 mm Fe(OH) 3 and mm riboflavin (C), compared with no preloaded riboflavin (B). All pictures were taken 24 h after inoculation. Brown areas indicate un-reduced Fe(OH) 3, whereas dark areas around inoculation point indicate reduced Fe(OH) 3. 64

76 Figure 4.5 Motility assays results depicting S. oneidensis MR-1 taxis to 0.2 mm oxidized riboflavin (top two rows) and 5 mm MnO 2 (bottom two rows at various times after inoculation. In both photographs of experiments (rows 1 and 3) and simulations (rows 2 and 4), brighter areas indicate un-reduced electron acceptors, and darker area indicates reduced electron acceptors. The time shown in the upper left-hand corner of experimental results corresponds to the time after inoculation. 65

77 Figure 4.6 Rate of growth of reduction zone in agar containing either riboflavin or MnO 2. Radial positions of reduction zone boundaries are compared; solid circles indicate experimental results, error bars indicate standard deviations, and asterisks indicate simulations. (The boundaries were considered as the locations at which the electron acceptor concentrations dropped to zero.) 66

78 200 Grayscale Value Experimental data Regression line Riboflavin Concentration (mm) Figure 4.7 Linearity test for yellow color intensity recorded by digital camera and riboflavin concentration. Pictures were taken for plates containing various concentrations of riboflavin, and the intensity (grayscale value) was measured by ImageJ (a public domain Java image processing program developed by Wayne Rasband at National Institute of Mental Health). A linear correlation between intensity and riboflavin concentration gave a correlation coefficient R 2 of 0.97 over the riboflavin concentration range used in the experiments. 67

79 Figure 4.8 Simulation results depicting S. oneidensis MR-1 taxis to 5 mm MnO 2 36 h after inoculation. (A) Simulation with chemotaxis, random movement, and flavin secretion (χ 0 = cm 2 h 1, µ = cm 2 h 1, and k S = mm riboflavin (mg cell) 1 h 1 if (S + Q) mm); (B) Simulation with no chemotaxis but random movement and flavin secretion (χ 0 = 0 cm 2 h 1, µ = cm 2 h 1, and k S = mm riboflavin (mg cell) 1 h 1 if (S + Q) mm); (C) Simulation with neither chemotaxis nor random movement but flavin secretion (χ 0 = 0 cm 2 h 1, µ = 0 cm 2 h 1, and k S = mm riboflavin (mg cell) 1 h 1 if (S + Q) mm); and (D) Simulation with chemotaxis and random movement but no flavin secretion (χ 0 = 0 cm 2 h 1, µ = 0 cm 2 h 1, and k S = 0 mm riboflavin (mg cell) 1 h 1 ). Brighter 68

80 areas indicate un-reduced electron acceptors, and darker area indicates reduced electron acceptors. 69

81 Figure 4.9 Model-predicted profiles for cell density, MnO 2 concentration, Mn(II) concentration, reduced riboflavin concentration, and oxidized riboflavin concentration. The concentrations are rescaled for easy comparison. 70

82 Table 4.1 Constants used in simulations Variables Value D S cm 2 h 1 µ cm 2 h 1 χ cm 2 h 1 K D ν C S Y k d k m k R k S mm 0.2 h mm 0.1 mg cell / mmol oxidized riboflavin 0.12 h h mm -1 h mm riboflavin (mg cell) 1 h 1 if (S + Q) mm 0 if (S + Q) > mm 71

83 CHAPTER 5 NADH OXIDATION BY ACTIVATED GLASSY CARBON ELECTRODE 5.1 Introduction Nicotinamide adenine dinucleotide (NADH) acts as an electron-carrying cofactor for a large group of redox enzymes that are widely applicable in biosensor, bioenergy and bioconversion technologies [71]. However, direct electrochemical oxidation of NADH or reduction or NAD + at an electrode surface is kinetically unfavorable, requiring the use of high overpotentials, which may cause cofactor degradation, energy inefficiency, and undesirable byproducts [72]. Electrocatalysts are used to circumvent this problem by accelerating the rate of electron transfer between the electrode and cofactor at moderate voltages. Common electrocatalysts include azines [73-75], quinone derivatives [76], polymerized quinone derivatives [71], metal oxide electrodes such as tin oxide [77] and iron oxide [78], NADHoxidizing enzymes such as diaphorase [79], and high-surface area materials such as carbon nanotubes (CNT) [80] and graphene [81]. However, even using electrocatalysts, most of the reported steady-state current densities for NADH electrocatalysis were still small and far below 1 ma cm 2 under low overpotential (e.g. around 25 ua cm 2 for a poly(toluidine Blue O) modified glassy carbon electrode at 2 mm NADH under V vs Ag AgCl [82], around 57 ua cm 2 for an iron oxide/carbon black composite glassy carbon electrode at 2 mm NADH under 0 V vs Ag AgCl [78], around 170 ua cm 2 for a carbon-nanotube-modified glassy carbon electrodes at 1 mm NADH under 0.3 V vs Ag AgCl ua cm 2 [83]). 72

84 In this study, azines were selected as the electrocatalysts because they have been shown to be effective for bioelectronic applications involving dehydrogenases [73, 84, 85]. Several methods were evaluated for immobilizing azines onto the electrode surface, including adsorption, electropolymerization [74, 75], and covalent binding [73]. Each of these methods has advantages and disadvantages relevant to enzyme applications. The covalent binding method requires the targeting molecule to have reactive groups (e.g., amine group on toluidine blue O) [73], and electropolymerization method cannot be applied to meldola blue and oxazine 170 [86]. We also developed electrochemical methods to activate glassy carbon electrodes that provided three benefits. First, activation allowed azine mediator molecules to be stably immobilized on the electrode via simple adsorption. Second, it increased the capacitance of the electrodes, suggesting an increase in electrode surface area. Third, it increased the rate of direct NADH oxidation in the absence of externally added azines. 5.2 Materials and Methods Electrode activation by cyclic voltammetry and azines immobilization All the electrochemical activation and characterization in this chapter were performed using a potentiostat (CV-50W Voltammetric Analyzer, BASi, West Lafayette, IN). An Ag AgCl 1 mm KCl electrode (CHI111, CH Instruments, Austin, TX) was used as reference electrode and a platinum wire (CHI115, CH Instruments, Austin, TX) as counter electrode. The glassy carbon electrode (3 mm diameter, CHI104, CH Instruments, Austin, TX) was polished on microcloth pads using 0.05 μm alumina powder (Electrode Polishing Kit, CHI120, CH Instruments, Austin, TX) and rinsed with distilled water in an ultrasonic bath for 10 min. All the chemical reagents 73

85 used in this chapter were purchased from Sigma-Aldrich (St. Louis, MO). The electrolyte was purged with nitrogen to exclude oxygen 5 min prior to activation or characterization and throughout the experiment. The glassy carbon electrodes were activated by CV between 2.5 V and -1.5 V with a scan rate at 0.1 V/s for 20 cycles in 100 mm, ph 7.4, phosphate buffer. The activated electrodes were incubated in 1 mm azine solutions for 1 h. Electropolymerization of azines was performed as described before [86]. Briefly, azine monomer solution was prepared by dissolving azine in 10 mm borate buffer, ph 9.1, containing 100 mm NaNO 3. During electropolymerization, CV was performed in 0.4 mm azine monomer solution between -0.5 and 1.5 V with a scan rate at 0.05 V/s for 20 cycles Electrochemical characterization Faradaic redox activity was characterized by peaks in the CV curves obtained at a scan rate at 0.05 V/s in 100 mm phosphate buffer, ph 7.4. NADH oxidation was characterized by chronoamperometry. To obtain NADH concentration profiles, the potential was fixed at 0.05 V, and steady-state current density at each NADH concentration was recorded Electrode activation with constant potentials Cyclic voltammetry is a complex method, involving an applied potential that varies in magnitude over a specified range at a specified sweep rate. To characterize the activation 74

86 process, it was desirable to study activation at a constant applied potential. A design of experiments approach was used to investigate the effect of applied potential (+1.5 V, +2.5 V and V) and activation time (15 s, 30 s, and 120 s). The sequence of experiments was randomly generated by Minitab 16 (Minitab Inc., State College, PA) to minimize effect on activation from systematic environmental errors. Minitab s Response Surface function (central composite) was used with 2 factors (i.e., potential and time), 3 center points, 4 cube points, 4 axial points, 2 replicates for each points, resulting in 22 total experiments. After activation, the capacitive surface area and the steady-state current for NADH oxidation at 30 mm 0.5 V was recorded. Capacitive surface area was measured as described before [75]. Briefly, CV was performed in phosphate buffer (100 mm, ph 7.4) in the range of 0.3 to 0.4 V with scan rates of 0.03 to 0.07 V s -1. The capacitance was calculated as the linear regression slope of non-faradaic current against scanning rate. Assuming a conversion factor of 25 μf cm 2 of carbon material, capacitive surface area was obtained. NADH oxidation characterization was performed by chronoamperometry in phosphate buffer (100 mm, ph 7.4, purged with nitrogen) with a stir bar to reduce mass transfer limitation. The current density was calculated by dividing current by capacitive surface area. The azine binding ability of the activated electrodes were also tested by soaking them in 1 mm methylene green solution for 1 h, and measuring the redox peak areas of methylene green in a cyclic voltammogram. The current production in NADH oxidation was measured as described above after methylene green immobilization. 75

87 5.2.4 NAD + immobilization Glassy carbon electrodes were polished and activated as described in section and then incubated in a 5% (v/v) aqueous solution of poly(ethyleneimine) (PEI). We expected that a PEI layer can be immobilized above the activated carbon surface due to possible attraction between positively charged amine groups on PEI molecules and negatively charged groups on electrode surface. A 5 mm 3-carboxyphenyl boronic acid (CBA) solution was activated at room temperature in the presence of 0.02 g/ml ethyl(dimethylaminopropyl) carbodiimide (EDC) and g/ml N-hydroxysuccinimide (NHS) in 100 mm ph7.4 phosphate buffer for 30 min. The NHS-modified CBA was reacted with the PEI functionalized interface at room temperature for 30 min, leading to an amide linkage between the carboxylic acid group of CBA and the amine group of PEI. The PEI-CBA-modified electrodes were reacted with a 5 mm NAD + solution in 100 mm ph 7.4 phosphate buffer for 30 min. The resulting electrodes were rinsed with water, incubated in phosphate buffer for 1 h and rinsed with water again to remove weakly bound NAD +. A schematic diagram of example strategy to fabricate bioelectronic containing azine electrocatalyst and NAD + on activated electrode is given in Figure 5.1. To detect NAD + immobilized on an electrode s surface, two NAD + /NADH assay kits were used (BioAssay Systems, Hayward, CA; and Cayman Chemical, Ann Arbor, MI). The BioAssay Systems kit is based on a lactate dehydrogenase cycling reaction, in which the formed NADH reduces (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reagent. The intensity of the reduced product color (blue), measured by absorbance at 565 nm, is proportional to the NAD + /NADH concentration. The Cayman Chemical kit is based on an 76

88 alcohol dehydrogenase cycling reaction, where alcohol dehydrogenase catalyzes the oxidation of ethanol to acetaldehyde, and the formed NADH reduces 2-(4-iodophenyl)- 3-(4-nitrophenyl)-5- (2,4-disulfophenyl)-2H-tetrazolium (WST-1) to a highly-colored formazan (yellow) which absorbs strongly at 450 nm. The amount of formazan produced is proportional to the amount of NAD + /NADH concentration. The assay solutions were prepared as described in the kit manuals. Either 100 or 50 μl of assay solution was added into a 2 ml centrifuge tube. The NAD + modified electrode was immersed in the solution. Both electrode and tube were sealed by Parafilm (Pechiney Plastic Packaging, Chicago, IL) to avoid evaporation. The reduction products from assays were measured by NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific, Wilmington, DE). The absorbance change in the assay solution was taken as a measure of the NAD + present at the electrode surface that was and able to be taken into the active site of a dehydrogenase, reduced, released from the enzyme, and then reduce a reagent to change the color and OD of assay solution. 5.3 Results and discussion NADH electrocatalysis CV-based activation of glassy carbon electrodes resulted in faradaic peaks that are presumably associated with formation of redox active quinones on the glassy carbon surface (Figure 5.2) [87]. The activated glassy carbon electrodes capacitance ( µf cm -2 ) and current density (0.35 ma cm -2 ) for NADH oxidation at 0.05 V versus Ag AgCl in 20 mm 77

89 NADH solution (ph 7.4). These values are about 200-fold higher than untreated carbon electrodes (Figure 5.3) and comparable to those measured for glassy carbon electrodes onto which 0.21 mg cm -2 CNT had been immobilized (Figure 5.2). All the tested azine molecules, including toludine blue O, methylene blue, methylene green, azur A, meldola blue and oxazine 170, could be stably immobilized onto the activated electrode surface, yielding bioelectronic interfaces exhibiting higher current density for NADH oxidation than when the same azine molecule was electropolymerized onto the glassy carbon electrode. Methylene green had the highest catalytic activity, which was different from the prediction made by Karyakin et al., who expected the highest catalytic activity with oxazine 170 [88] (Figure 5.4). The MG-activated electrode achieved a current density at 0.7 ma cm -2 for NADH oxidation under 0.05 V in 20 mm NADH solution (ph 7.4), a 100-fold increase compared to electropolymerized MG electrode (Figure 5.5) Electrode activation with constant potentials The contour plots of capacitive surface area, current production in NADH oxidation, and current density against activation potentials and durations are given in Figure 5.6. Larger activation potentials and longer activation times increased capacitive surface area (Figure 5.6 A), but decreased current density (Figure 5.6 B). These results suggest that electrooxidation of the glassy carbon can increase electrode surface area, but excess oxidation can degrade the electrocatalytic species (e.g., quinones) on the electrode surface responsible for NADH oxidation. Over the range of potentials and times tested, the constant potential activation method yielded a 78

90 maximum NADH rate of about A (Figure 5.6 C), which is only 3 times larger than produced by electrodes without activation ( ± A)). The azine binding ability and NADH oxidation rates of electrodes activated at a potential of 3.25 V are given in Table 5.1. For activation times longer than 30 s, no redox peak of methylene green was identified, suggesting that methylene green was poorly immobilized on overactivated electrodes. These overactivated electrodes also showed poor NADH oxidation performance. However, longer activation times consistently gave larger capacitive surface areas (Table 5.1). These results suggest that the amount of methylene green immobilized in a way that allows rapid oxidation at the applied potential used for NADH oxidation experiments does not increase in proportion to the capacitive surface area (Table 5.2). Overactivation could reduce the binding density of methylene green and/or it could decrease the efficiency of electron transfer between NADH and the electrode during the NADH oxidation experiments NAD + immobilization During the 30 min incubation of the NAD + -functionalized glassy-carbon electrode in 100 μl BioAssay assay kit solution, the assay solution color changed from yellow to blue. The change in absorbance at 565 nm was 0.364±0.007, which is equivalent to the assay response from 1 μm soluble NAD +. After the first assay, the electrode was rinsed with water, and the same assay was performed again with the same electrode. The absorbance change for the second assay was nearly zero, suggesting that the assay solution removed the interface during the first 79

91 assay. Hence, the color change might have resulted from NAD + dissolved in the solution, and does not necessarily indicate that the NAD + molecules had been bound to the electrode in a biologically active configuration that allowed them to bind to the dehydrogenase active site and reversibly transfer electrons. Assay irreproducibility was not observed with Cayman Chemical assay kit. After the NAD + modified electrodes were immersed in 50 µl assay solution for 5 h, the colorless solution became bright yellow, and the absorbance at 450 nm was 0.063±0.010, which is equivalent to the assay response for 0.25 μm soluble NAD +. The absorbance readings from the controls (i.e., the electrodes without either PEI or CBA layer, or both layers) were 0.001± After water rinsing, a second assay with the same protocol was performed, and the absorbance at 450 nm was 0.060±0.013, which indicated that the immobilized NAD + was stably immobilized and biologically active Conclusion In this chapter, a novel method to electroactivate glassy carbon electrode toward NADH oxidation has been presented and characterized. The method increases capacitive surface area and introduces redox-active functional groups that catalyze direct oxidation of NADH without the addition of external electrocatalysts. Based the location of redox peaks in cyclic voltammograms of activated electrodes, we hypothesize that the functional groups are quinones. The activated electrodes also stably adsorb azine electrocatalysts, thereby offering multiple 80

92 options for cofactor regeneration. The effectiveness of activation at a constant applied potential was explored by using a central composite design of experiments to determine optimal values of activation time and applied potential. At relatively mild activation potential (1.5 V), the activation did not increase capacitive surface area, and at high potential (3.25 V), the activation appeared to over-oxidize the carbon surface, reducing specific NADH-oxidation activity. These trs suggest that mild oxidative potentials generate electrocatalytic species on the electrode s surface, but sustained, strongly oxidative potentials may lead to the degradation of these species. Optimal constant-potential activation was obtained by applying 2.5 V for about 30 s. Capacitive surface area increased consistently with applied potential and increased with activation time up to tens of seconds. However, the azine redox activity measured by CV was not directly proportional to the surface area; after exted periods under strongly oxidizing potentials, electrodes exhibited less azine redox activity. One hypothesis to explain this tr is that functional groups able to adsorb azines are formed on the electrode surface under mild activation conditions (i.e., lower potential or shorter time; e.g., 3.25 V and 15 s), but are subsequently destroyed under harsher activation conditions (e.g., 3.25 V and 2 min). A second hypothesis is that overactivation reduces efficiency of electron transfer between adsorbed azines and the electrode, possibly by establishing an oxidized layer between the glassy carbon and adsorbed azine molecules that acts as an electrical insulator. The relative contributions of these two hypothesized mechanisms could be assessed in future studies The constant-potential experiments allowed the effects of potential and time to be characterized using a design of experiments approach. However, the results indicated that only a three-fold enhancement in direct NADH oxidation rate was possible using constant potential, whereas a 200-fold enhancement was achieved using CV. This result suggests that the surface 81

93 chemistry responsible for the enhancement is facilitated by voltage cycling. In future studies, conditions for CV-based activation can be optimized by a similar design of experiments approach to simultaneously optimize parameters such as the potential range, the sweep rate, and the number of cycles. In Section 5.2.4, we designed a bioelectronic interface to co-immobilize azines, NAD + and dehydrogenases on an activated glassy carbon electron such that continuous cofactor regeneration was achieved via multistep electron transfer between the enzyme and electrode. Our approach combined layer-by-layer deposition of the polyelectrolyte PEI with covalent coupling through amide linkages to bind the azines and NAD +. Initial experiments indicated that immobilized azines can functionally oxidize free-diffusing NADH, and NAD + immobilized as indicated in Figure 1 can be repeatedly taken into the active site of a dehydrogenase and reduced by the enzyme. This interface could be further optimized in future studies. 82

94 Figure 5.1 Schematic diagram of example strategy to fabricate bioelectronic containing azine electrocatalyst and NAD + on activated electrode. 83

95 Figure 5.2 Activation of glassy carbon (GC) electrode. (A) Cyclic voltammograms of glassy carbon electrode before and after activation in 0.1 M phosphate buffer ph 7.45, 30 ºC. (B) CV was performed on glassy carbon electrode, 20 cycles, 0.1 V/s, 0.1 M phosphate buffer, ph 7.45, 30 ºC. 84

96 Capacitance NADH oxidation activity Capacitance (uf/cm 2 ) Current density (ma/cm 2 ) Bare GC Act GC CNT 0 Figure 5.3 Activity of bare glassy carbon (Bare GC), activated glassy carbon (Act GC) and carbon nanotube (CNT) electrodes in 0.1 M phosphate buffer ph 7.4. Current for NADH electrocatalysis was recorded at 50 mv vs. Ag AgCl in 20 mm NADH solution. 85

97 1.2 1 Current density (ma/cm 2 ) Methylene Green Oxazine NADH concentration (mm) Figure 5.4 Effect of NADH concentration on NADH oxidation currents generated by activated glassy carbon electrodes with either methylene green or oxazine 170 adsorbed on the electrode. Experiments were conducted 50 mv vs. Ag AgCl in 100 mm phosphate buffer ph 7.4. The error bars indicate standard deviations of three replicates. 86

98 Redox peak height in CV (ma/cm2) Redox peak height in CV NADH oxidation activity MG-GC PMG-GC MG-Act GC Current density (ma/cm 2 ) Figure 5.5 Relative activities of glassy carbon electrode with absorbed methylene green (MG- GC), glassy carbon electrode with electropolymerized methylene green (PMG-GC), and activated glassy carbon electrode with absorbed methylene green (MG-Act GC). Experiments were conducted in 0.1 M phosphate buffer ph 7.4, at 30 ºC. Steady state current density for NADH electrocatalysis was recorded at 0.05 V vs. Ag AgCl in 20 mm NADH solution. Redox peak height in CV was generated by CV-50W Software (BASi, West Lafayette, IN). 87

99 Figure 5.6 Contour plots showing effect of indepent variables (activation potential and activation time) on depent variables (capacitive surface area (A), current production during NADH oxidation (B), and current density (C)). 88