Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Characterization of a novel bacterial transducer based on genetically engineered bioluminescence Blouin, Kim 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-0136.pdf [ 6.16MB ]
JSON: 831-1.0103804.json
JSON-LD: 831-1.0103804-ld.json
RDF/XML (Pretty): 831-1.0103804-rdf.xml
RDF/JSON: 831-1.0103804-rdf.json
Turtle: 831-1.0103804-turtle.txt
N-Triples: 831-1.0103804-rdf-ntriples.txt
Original Record: 831-1.0103804-source.json
Full Text

Full Text

CHARACTERIZATION OF A NOVEL BACTERIAL TRANSDUCER BASED ON GENETICALLY ENGINEERED BIOLUMINESCENCE by KIM BLOUIN B.Sc.A. Universite Laval, Quebec, 1990 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES, (Department of Electrical Engineering, Biotechnology Laboratory) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA February 1994 Kim Blouin, 1994 In presentin g thi s thesi s i n partia l fulfilmen t o f th e requirement s fo r a n advance d degree a t th e Universit y o f Britis h Columbia , I  agre e tha t th e Librar y shal l mak e i t freely availabl e fo r referenc e an d study . I  furthe r agre e tha t permissio n fo r extensiv e copying o f thi s thesi s fo r scholarl y purpose s ma y b e grante d b y th e hea d o f m y department o r b y hi s o r he r representatives . I t i s understoo d tha t copyin g o r publication o f thi s thesi s fo r financia l gai n shal l no t b e allowe d withou t m y writte n permission. Department o f H~LEC7%  it4/ £uGr,-tJ£&R<'(06 The Universit y o f Britis h Columbi a Vancouver, Canad a Date B fe&  ./99>V DE-6 (2/88 ) Abstract The research described in this thesis is aimed at helping with the development of better sensors for instrumentation involved in the monitoring of otherwise difficult-to-detect chemical compounds and their toxic effects. The proposed transducer is based on genetically engineered bioluminescence in the biofilm-forming bacterium Caulobacter crescentus. We studied the biochemical mechanisms involved in the bioluminescence process which revealed a complex multi-phase kinetic behaviour. The bioluminescence profiles were characterized in terms of relevant protocol-related parameters such as substrate concentration, environment, order of mixing, etc.  A coupled-enzyme model of the bioluminescence mechanism is proposed in order to explain and interpret the complex profiles. The interpretation is supported by an in vitro  analysis isolating the two enzymes directly involved in in  vivo  bioluminescence. This study demonstrates the importance of precise control of protocol-related parameters required to obtain reproducible and meaningful results. A rudimentary toxicity protocol was developed based on the knowledge gained during the characterization studies. The results obtained were comparable to previously published data and hence demonstrated the considerable potential of the bacterial transducer for use as a toxicity sensor. The data generated also supported the validity of the model proposed during the characterization of the bioluminescence mechanism. The bacterial transducer was thus found to be appropriate for future application in a flow-through toxicity assessing instrumentation system. The first stage of this next phase of the project was accomplished by designing and building a prototype of a flow-through system. Some preliminary flow-through data were generated which demonstrated that the prototype (and the flow-through concept) are indeed advantageous and worthy of further development. Some suggestions for specific directions for such further development are outlined and discussed, as well as some other applications for the flow-through instrumentation system. ii Table of Content s page Abstract ii List of Figures v Acknowledgments vii CHAPTER 1. INTRODUCTION 1 1.1. Literature review 4 1.1.1. Conventional technology for measuring toxicity 4 1.1.2. Microtox® as a non-animal alternative to evaluate toxicity 7 1.1.3. Biosensors for toxicity measurement based on engineered bacteria 9 1.2. Objectives 11 1.3. Overview 12 CHAPTER 2. BACKGROUND BIOCHEMISTRY AND MICROBIOLOGY 14 2.1. Definitions and descriptions of important biological terms 14 2.1.1. Biochemical aspects 14 2.1.2. Genetic aspects 17 2.2. The bioluminescence mechanism 20 2.3. The Caulobacter life cycle 25 CHAPTERS. CHARACTERIZATION OF THE IN  VIVO BIOLUMINESCENCE OF THE BACTERIAL TRANSDUCER 27 3.1. Materials and methods 27 3.1.1. Measurements of in vivo bioluminescence 27 3.1.2. Measurements of in vitro bioluminescence 31 3.2. Results and discussion 32 3.2.1. Characterization of in vivo bioluminescence 34 in 3.2.2. Characterization of in vitro bioluminescence 44 3.2.3. Gas chromatography results 50 3.3. Conclusions 54 CHAPTER 4. TOXICITY ASSAY PRELIMINARY RESULTS 56 4.1. Measuring toxicity 56 4.2. Materials and methods 58 4.3. Results and discussion 60 4.4. Conclusions 68 CHAPTER 5. FLOW-THROUGH SYSTEM DESIGN 70 5.1. Motivations for the development of a flow-through system 70 5.2. Design and development of a flow-through prototype instrument 72 5.2.1. Flow-cell design 73 5.2.2. Fluid handling system design 76 5.2.3. Optical and electronic system design 78 5.3. Preliminary tests of the flow-through system 88 5.3.1. Materials and methods 90 5.3.2. Results and discussion 90 5.4. Conclusions 97 CHAPTER 6. CONCLUSIONS 98 6.1. Summary of main conclusions and scientific contributions 98 6.2. Future work 102 7. References 104 8. Appendix A 115 9. Appendix B 119 IV List of Figure s page Figure 2.1. Chemical structure of flavin mononucleotide 15 Figure 2.2. Chemical structure of n-decyl aldehyde (decanal) 16 Figure 2.3. Structure of nicotinamide adenine dinucleotide 17 Figure 2.4. Expression of genetic information from DNA to the production of a protein 18 Figure 2.5. Schematic diagram of the biochemistry involved in bioluminescence ...23 Figure 2.6. The life cycle of the Caulobacter crescenhis 26 Figure 3.1. Typical bioluminescence intensity profile of the recombinant bacterial transducer 33 Figure 3.2. Effect of varying the decanal concentration on the secondary phase of the luminescence response profile 35 Figure 3.3. Comparison of the effect of different diluent and incubation time on the in vivo bioluminescence signal 37 Figure 3.4. Effect of variations in decanal concentration on in vitro bioluminescence 45 Figure 3.5. Comparison of the different kinetics obtained by controlling the formation of the luciferase-FMNHOOH complex prior to the addition of decanal 47 Figure 3.6. Effect of variations in the decanal concentration on the intensity of the plateau region of the in vitro kinetics 48 Figure 3.7. Standard decanal concentration curve obtained by GC analysis 51 Figure 3.8. Measured out-gassing of decanal and its effect on bioluminescence 53 Figure 4.1. Phenol dose-response characteristic 61 Figure 4.2. Sodium hypochlorite dose-response characteristic 62 v Figure 4.3. Mercuric chloride dose-response characteristic 63 Figure 4.4. Effect of incubation time on the sodium hypochlorite dose-response characteristic 64 Figure 5.1. Block diagram of the flow-through system 73 Figure 5.2. Schematic diagram of the flow-cell fixture for supporting the biofilm and exposing it to the flow stream 75 Figure 5.3. Diagram of the filter/amplifier 86 Figure 5.4. Photodetector coupling to the flow-cell 88 Figure 5.5. Diagram of the overall flow-through system 89 Figure 5.6. Biofilm bioluminescence monitoring 92 Figure 5.7 Photomicrograph of a section of a typical biofilm used in the flow-through system 96 Figure 8.1. Schematic of the photolithographic layout for the flow-cell 116 Figure 8.2. Scale drawing (side view) of the flow-cell 117 Figure 9.1 Block diagram of amplifier/filter 120 Figure 9.2 Instrumentation amplifier 122 Figure 9.3 Gain amplifier 122 Figure 9.4 Clipper detector 123 Figure 9.5 Filter 124 Figure 9.6 Limiter 124 Figure 9.7 Summer 125 Figure 9.8 Power supply 125 Figure 9.9 Top view of the box 126 Figure 9.10 Front view of the box 127 Figure 9.11 Rear view of box 127 Figure 9.12 Filter test results 128 Figure 9.13 limiter test results 130 vi Acknowledgments I would like to offer my thanks to my supervisor Dr. Robin F.B. Turner for his guidance and support provided throughout my degree, and especially for the extra efforts that he invested when it was required. I would like to acknowledge The Biotechnology Laboratory staff and facility and especially the expert technical assistance of Chris Sherwood and Gary Lesnicky, and also The department of Electrical Engineering for their interest and involvement in interdisciplinary work which allowed me to enter the Biotechnology field of research. I also acknowledge the collaboration of Dr. John Smit who gave us the opportunity of getting involved in this project and to Stephen G. Walker who made the genetic work on the bacteria and always was a good resource person, and Dr. Ramey and his technicians for making available their assistance and gas chromatography apparatus for our analysis. I would like to acknowledge the financial support provided by les Fonds pour la Formation de Chercheurs et I'Aide a la Recherche (Fonds FCAR), the Natural Sciences and Engineering Research Council of Canada (NSERC), the department of Electrical Engineering and the Kashmir Singh Manhas Fund which made possible the realization of this project. Je voudrais specialement remercier Paul-Eugene et Madeleine Blouin ainsi que les autres membres de ma famille pour leur support moral et financier qui fut inconditionnel au cours des nombreuses annees d'etudes qui ont mene a l'accomplissement de cette these. I would also like to thank the Stacey familly for their support and especially Norma who assisted in the editing of this work. vn CHAPTER 1 . INTRODUCTIO N Applications in science and engineering often require the measurement of a variety of physical, chemical and/or biological variables in order to monitor or characterize some defined system. Such measurements are usually accomplished using a transducer, or sensor, that converts the measurand signal into an electrical signal. Transducers capable of measuring physical variables (e.g.  pressure, temperature, mass, speed, acceleration, etc.) have been commonplace for many years since these measurands can easily be isolated and quantified with simple and, more importantly, stable devices. Chemical and biological variables, however, are not so straightforward to measure, often requiring far more complex transducers and associated instrumentation. For instance, the determination of chemical concentrations especially in complex media may require a number of sample pre-treatment steps before the actual analysis which, itself, could involve multiple steps. For a few special cases, however, relatively simple chemical sensors have been devised based primarily on electrochemical or optical detection methods, coupled with some (ideally) in situ mechanism (e.g.  a perm-selective membrane system) for achieving specificity for a particular measurand, or analyte, species. Over the past 20 to 25 years, a large sub-class of chemical sensor has emerged that is based on the exploitation of biochemical or biophysical mechanisms in order to convert certain chemical measurand signals, typically of some biological relevance, into electrical signals. Devices comprising this class are commonly referred to as "biosensors", although the term is now used somewhat loosely in the literature and is often taken to include complete instrumentation systems necessary to obtain an electronic signal representing the desired analyte concentration. The "bio-" component may be as simple as a single enzyme that produces a detectable product, or as complex as whole living cells. Several difficult problems have slowed the commercialization of biosensors, but potential applications are many and diverse and include research applications in biology, medical diagnostics, clinical 1 (i.e. patient) monitoring, bioprocess monitoring and environmental monitoring, to name only a few areas. The application area of particular interest in the present work is that of environmental monitoring. For example, the detection and control of heavy metal and toxic organic pollutants in industrial waste water effluents and marine sediments requires the development of new approaches for monitoring these sites. Ideally, monitoring would include rapid or continuous measurements for key indicators of pollution at (or at least near) the sites being monitored. At the present time, standard chemical assays are not readily accomplished in the field and there are no reliable continuous monitoring systems available. However, the development of biosensor devices, attuned to specific pollutants, represents a promising approach, and bacteria have emerged as the best biological vehicle for this type of biosensor development. This is primarily due to recent advances in the understanding of several examples of genetically-derived capabilities to sense pollutants, along with newly developed technology for coupling this sensing capability to appropriate reporter systems such as genes that allow an organism to emit light (i.e.  bioluminescence). Such microbial biosensors can, in principle, be designed to function either as specific toxin sensors (or detectors), or as assay systems for assessing the relative toxicity of a range of toxic agents. Genetic engineering has allowed the development of specialized (recombinant) organisms that offer great potential for use as the biological transducer component of environmental biosensors. However, the engineered biochemical mechanisms involved in generating the detectable signals are extremely complicated and are interconnected with indigenous biochemical pathways that further complicate the resulting signals. These organisms must therefore be thoroughly characterized in order to properly interpret the signals and obtain repeatable, quantitative (if needed) results. This characterization must include the elucidation of an appropriate assay protocol and optimization of the conditions under which the organisms are maintained during use. It also remains to develop an 2 appropriate instrumentation system, including fluid handling systems, that allows for an effective means of presenting aqueous environmental samples to the micro-organisms, and efficient detection of the resulting signals. The work described in this thesis is directed toward exactly this kind of system design and characterization of a new microbial transducer (biosensor) for environmental monitoring. The "transducing organism" utilized in this work is a freshwater strain of Caulobacter crescentus  which has been genetically engineered to emit light (i.e. bioluminescence signals) under certain defined test conditions. Briefly, the intensity profile of light emission under these conditions is either constant or follows a predictable time course that is perturbed by the presence of toxic substances in the test medium. This particular organism is designed to function in assays of relative toxicity of a variety of environmental toxins, although additional organisms based on similar genetic constructs are currently under development that would allow the determination of specific toxins in aqueous environmental samples. The work presented here provides a basis for the rapid characterization of these future strains as they become available. This work is of fundamental importance to the development of a range of useful microbial biosensors based on genetically engineered Caulobacters. Biosensor instruments of this type could ultimately be used, for example, by government agencies to track pollutant substances in ecosystems, to assess the relative toxicity of specific compounds, to monitor ecosystem behavior/changes, to measure environmental quality, and to assess environmental impact of modifications to the environment. Industries could also benefit from such technology, applying it to establishing pollution targets, conservation strategies and developing preventative technologies. Consumer and public interest groups could also utilize this type of environmental monitoring technology in order to obtain accurate information contributing to informed public participation. The work described in this thesis represents an essential 3 step toward the development of needed instrumentation to serve these and other applications. 1.1. Literatur e revie w In the field of toxicity assessment, the toxicity of a compound under investigation can be characterized in a number of ways. For example, for pharmacological applications, the compounds (e.g.  drugs) have to be assessed in terms of human toxicity. It is not always possible or practical to obtain human toxicity data, even for drugs or other pharmaceuticals. The accepted, albeit non-ideal, alternative is to select an appropriate animal model or other organism to obtain experimental dose-response data [37]. What is tacitly assumed here is that such data are related, qualitatively at least, to data that would "ideally" be obtained from humans or other organisms of relevance in a particular study. This is true, not only for drugs, cosmetics, etc.,  but also for environmental toxins, where it is necessary to determine the effects of some agent (or mixture of agents) on the general flora and fauna that could potentially become exposed. It is important to distinguish between the quantitative detection of a toxicant and its relative toxicity. There is a fundamental difference between the two types of analysis. The former consists of the measurement of the amount of the toxicant present in a sample, as opposed to the latter which consists of an evaluation of the effect of the toxicant on living organisms. The methods used to perform either analysis are also, in general, very different. Even though the development of C.  crescentiis could lead to transducers capable of quantitative detection of specific compounds, the work accomplished thus far involves a transducer capable only of general toxicity assessment that leads to relative toxicity data. Therefore, we will focus the remainder of this review on this topic. 1.1.1. Conventiona l technolog y for measuring toxicity The most commonly used parameters to express the toxicity of a compound are the LD50 and the closely related LC50, which stand for the "median lethal dose" (per unit 4 body weight) or "concentration", respectively. In other words, it is the dose of a compound that causes 50% mortality in a population [28, 73], Other toxicity parameters, such as ED50 and EC50, are also used and refers to the "median effective dose" (per unit body weight) or "concentration", respectively. Both of these parameters refer to the dose of a substance that can be expected to cause a particular effect to occur in 50% of the population [28, 37]. The effective dose parameters are not necessarily based on a lethal effect, but on some other observable effect on the organism used for the study. The effect used for EC50 can, for example, be eye irritancy, skin rash or other similar physiologic effects. The toxicity parameters may be accompanied by a specified exposure period, especially in cases where the study is based on exposure rather than a single dose effect. It is usually assumed to be necessary and essential to obtain a precise assessment of the lethal effects of highly toxic substances, since a small difference in exposure can distinguish a safe from a lethal situation. A precise LD50 is not always necessary for many less toxic materials (e.g.  household products). In fact, the quantitative significance of LD50 values is a controversial subject and has been widely discussed in the literature [7, 49, 58, 69, 84, 99]. The consensus is that the LD50 is an inherently imprecise value; it is not a biological constant and should be de-emphasized as a quantitative figure of merit. Therefore, qualitative or relative values of LD50 are considered sufficient for all practical purposes. This consensus for the LD50 parameter can also be extended, using similar arguments, the other mentioned parameters. Conventional methods used to obtain toxicity data for different compounds are all based on the same basic principle. First, a suitable type of organism has to be chosen for the study. A wide range of types of organisms are routinely used for toxicity testing of different compounds in different environments. For example, fish (e.g.  rainbow trout, fathead minnows) or daphnid (e.g.  Daphnia  magna,  a small crustacean) are used for studying the effects of aqueous toxic compounds [55, 74], and rodents (e.g.  rabbits, rats and mice), or larger mammals (e.g. dogs and cats) are more commonly used for acute (i.e. 5 single dose) oral toxicity testing [73] of non-aqueous ingestable compounds. Once the organism is chosen, the population is separated into groups that are independently exposed to various concentrations of the toxic compound under study. The exposure to the toxic agent can be done in different ways, depending on which parameter is sought. For example, oral administration to a rodent population leads to the LD50 of the compound, and the exposure of a fish population to a certain concentration of the compound in solution for a predetermined period of time leads to the LC50. Therefore, the toxicity testing of a compound can be summarized as involving three steps: 1) choice of a suitable organism, 2) exposure of the organism to various concentrations or doses of the compound, and 3) determination of the appropriate toxicity parameter, relative to suitable control groups. The advantages and disadvantages of these toxicity tests are very varied. Toxicity testing using specific target organisms can yield the absolute toxic level of a compound only for that organism. To be able to predict the effect of a compound on specific inhabitants of an ecosystem is often of major importance for environmental impact studies. However, the absolute toxic level of any compound is clearly not experimentally obtainable for humans, and therefore the evaluation of a safe range has to be based on the results from other organisms. This evaluation is very complex and inherently not very precise, hence it is always done in a very conservative way [37]. These toxicity tests are also very time consuming and involve large costs. For example, the required time to perform exposure assays (such as the fish assay) range from 24 to 96 hours, and usually involves costs of several hundred dollars per test [13]. In addition, the effort, facilities and time required to maintain the population between and during testing is considerable. On the ethical side, the public interest in animal welfare has instigated many debates over the experimental use of animals. This public concern has stimulated the effort to find non-animal "alternatives". These include use of tissue culture, excised animal organs, bacterial systems, and others [27]. Even though such indirect toxicity tests have a lot of 6 undesirable and impractical qualities, they have an important role to play in toxicity testing. 1.1.2. Microtox ® as a non-animal alternative to evaluate toxicity As the interest in non-animal alternatives for toxicity testing grew, new techniques were developed using bacteria. Many types of bacteria have been used to assess toxicity using a method similar to that described for conventional toxicity testing. But a (naturally) bioluminescent type of bacteria has recently lead to the development of a system that has become very popular, especially for industrial testing (e.g.  self-monitoring of plant effluents). The system was developed in the late 1970's by a group of researchers lead by Bulich, and was eventually given the commercial name of Microtox®. The Microtox® assay is based on the same basic principle as the more conventional methods. The organism used is the bioluminescent Photobacterium phosphoreum (NRRL B-11177); these bacteria have evolved to emit light as a by-product of a genetically conferred, natural biochemical pathway (to be discussed further in Chapter 2). The bacteria are available from the manufacturer in the form of a lyophilized (i.e. freeze dried) powder, and they are treated simply as one of three reagents required to perform the assay. The testing of a compound is performed as follows [13]: 1) reconstitution of the freeze-dried bacteria with the diluent solution, 2) initial light emission measurements are made for a predetermined number of cuvettes containing diluted bacteria, 3) one of various dilutions of the compound under test is added to each cuvette, 4) final light measurements are made after a fixed amount of incubation time with the toxic compound (typically: 5, 15 and/or 30 minutes). The data are then compiled and an EC50 for the chosen incubation time of the compound is obtained. In this case, the EC50 refers to the effective concentration necessary to cause 50% decrease of the luminescence. There are several hundred papers in the literature describing and evaluating applications involving this system, including many reviews [11, 13, 17, 44, 55, 74]. 7 One of the main features of the Microtox® system that gives it a definite advantage over the more conventional toxicity assays is the speed at which it can analyze toxic compounds. The Microtox® assay leading to the 5-minute EC50 of a compound can be performed in less than an hour [13]. This speed of the assay comes mainly from the sensitivity of the bacteria to toxic compounds. This sensitivity allows observable effects on the monitored parameter to be measured in a much shorter exposure time (5 to 30 minutes to get 50% decrease in luminescence) with about the same concentration of toxic compound than, for example, with the trout assay (24 to 96 hours to observe 50% death in the population) [17, 74]. The Microtox® system maintains this speed advantage over other single-celled organisms {e.g.  Spirillum volutans  [74]), even though they are also much faster than the conventional methods, similarly due to their sensitivity to the toxic effects of most compounds tested. In this case, the speed advantage is related to the ease of counting the portion of the population that is affected by the toxic compound (i.e.  measurement of light emitted). For the conventional methods, such enumeration is a lengthy manual process. The Microtox® system takes advantage of the luminescence phenotype of the P. phosphoreum to relate the toxic effect to the population number. Many studies have been done to evaluate the quality of the correlation of toxic effects to the luminescence intensity on which Microtox® is based. These studies concluded that the correlation is satisfactory, although not perfect [11, 55, 74]. Good correlation between cell viability and the luminescence intensity has also been demonstrated for recombinant bacteria having the luminescence phenotype [41]. The luminescence intensity measured in lieu of counting the population of cells affected appears to be acceptable, and allows considerable simplification of the toxicity assay. The measurement of the emitted light intensity, which is proportional to the viable fraction of the population, is typically performed using a luminometer capable of low-intensity light measurement. 8 The combined speed and convenience of an instrumental method for counting the bacteria, give rise to another important factor making the Microtox® system attractive, its relatively low operating cost. The costs related to the physical facilities and maintenance of the population for conventional toxicity testing are far greater than those associated with the Microtox® system. Since the bacteria required for the Microtox® assay can be stored as a lyophilized powder, there are practically no population maintenance costs. Since the Microtox® assay can be performed with small amounts of the reagents and compound to be tested, the overall cost is much smaller than, for example, the fish assay which requires much larger volumes of everything necessary. One economic analysis of the Microtox® system concluded that the cost of this assay is about 5% of the estimated cost associated with an acute fish assay [17]. Even though the Microtox® system offers some significant advantages, it still possesses some disadvantages. The assay suffers from an extreme sensitivity to the assay protocol that leads to weak reproducibility, and hence comparability, of results obtained from different laboratories. Even though the correlation between the Microtox® system and other bioassays has been shown to be fairly good in many cases, the low reproducibility of the toxicity data obtained for the same compound by different laboratories is a definite weakness [17, 55, 74]. For example, the 5-minute EC50 for phenol, a common biocide compound, is reported in many publications and exhibits a variance of more than a factor of two in the reported values [13, 17, 44, 55, 74]. Interestingly, even though the system offers this low precision, many users that perform routine toxicity tests have adopted the Microtox® system. It is apparent that the compromise in precision is worthwhile and acceptable to a majority of users for economical reasons. 1.1.3. Biosensor s for toxicity measurement base d on engineered bacteri a Early bacteria-based biosensors were developed for the detection of metabolized substances such as C02, ethanol, sugars, etc.  [45, 46]. These biosensors usually use an 9 oxygen or pH probe [9, 45] onto which the bacteria are immobilized or trapped. This setup limits the type of substance that can be detected and it is not practical to adapt these devices to detect toxic or non-metabolized agents. But with advances in genetic engineering, it has become possible to improve or adapt these types of biosensors so that some of the limitations can be overcome. Bioluminescence is one useful characteristic that can now be engineered in bacteria. This phenotype is of special interest for biosensor development because of the high sensitivity with which we can detect its manifestation. Modern light detection technology allows us to detect single photons which translate, in bioluminescence terms, to being able to monitor the occurrence of single biochemical reactions. The greater understanding of cloned lux  genes (responsible for light production) has led to the development of bioluminescence reporters for the study of genes promoters [4, 24, 25, 30, 47, 77, 81, 86, 93, 96]. These reporters are very important tools that aid in the study of gene expression which, in turn, basically control everything that happens in a cell. The understanding of gene expression and its regulation has been a fundamental goal in biology for many years and advancements in this field have yielded much technology that can now be exploited for the development of improved biosensors. The characterization of an increasing number of cloned genes and the development of more sensitive solid state microelectronic devices capable of low-light-level detection has already opened the door to new directions and applications for bioluminescent bacteria-based biosensors. Being of particular interest for us, some biosensors have recently been developed that demonstrate the feasibility of this concept for toxicity sensors [41, 50, 57, 90, 94]. The light-producing bacteria currently used in these biosensors are Escherichia coli (engineered bioluminescence), Pseudomonas  or Photobacterium  strains (natural bioluminescence) [4, 92]. Most of the work done to date with these bacteria involves simple addition of the bacteria to samples of effluent or soil to be tested. These setups have inherent sensitivity limitations since only the bacteria in close proximity to the light 10 detector produce a useful signal. Some attempts have been made to immobilize E. coli  in a matrix {e.g.  agar) adjacent to the light detector, but the stability of the cells in this unnatural environment, combined with uncertainties about rates of movement in and out of the matrix of the nutritional substances for the bacteria and the substances being detected, makes this setup a non-ideal choice. The bacterium Caulobacter  crescentus,  that is the subject of this thesis, offers many advantages over previously used species. This species of bacteria can naturally adhere to almost any material and form a dense monolayer biofilm. As opposed to other adherent bacteria, the Caulobacters do not produce a polysaccharide "slime" that could cause similar problems as with the agar matrix for E. coli.  Furthermore, the Caulobacters are natural inhabitants of most environments where toxicity monitoring is required [60, 70, 71]. Caulobacters are known to be robust bacteria and capable of growth under minimal nutrient conditions [4, 72], A very important factor is that several strains of  Caulobacter crescentus have been used in molecular genetic research for many years, and it is now straightforward to carry out all the necessary bacterial genetic manipulations [6, 20, 21, 22, 23]. These characteristics make the Caulobacters an excellent choice to address many of the practical problems encountered with other bacteria-based biosensors. 1.2. Objective s Some problems still remain in the way of the development of a commercially viable line of biosensors based on bacterial bioluminescence. First, the bioluminescence signals emitted by recombinant bacteria need to be properly characterized in order to specify an appropriate assay protocol and instrumentation system. Second, the nature of toxicity monitoring requires the development of biosensors that can be used in a wide variety of environments, a quality lacking in most bacteria-based biosensors. Third, the methods used for coupling the light emitting bacteria to the photon detection system has to become more efficient in order to achieve the necessary sensitivity and reproducibility. Finally, the 11 complexity of most of the bacteria-based biosensor systems limit their use to laboratory conditions with specially trained technicians. It is in the interest of most potential users of a toxicity monitoring system to develop a portable instrument capable of field operation without the skills of a microbiologist. The main goals of this work are oriented toward the development of a biosensor system, based on an engineered strain of Caulobacter crescentus, that will address some of these problems encountered by other biosensors. The specific objectives of this thesis are treated separately in each of three chapters. The objective of Chapter 3 is to characterize the engineered bioluminescence signals from the C.  crescentus transducer on which the proposed toxicity biosensor is based; this will allow a better understanding of the processes involved in generating the bioluminescence and will provide the necessary knowledge to design a future optimized system. The objective of Chapter 4 is to demonstrate the potential and feasibility of a toxicity testing system based on the C crescentus  transducer by obtaining sample toxicity data and comparing it to other toxicity assays. The objective of Chapter 5 is to investigate the feasibility and advantages of a prototype system based on a flow-cell configuration that would take advantage of the inherent ability of the C.  crescentus to self-immobilize in a monolayer biofilm. 1.3. Overvie w Since the project described in this thesis is highly interdisciplinary, Chapter 2 aims to introduce readers coming from other fields to the biological and biochemical aspects of this work. Chapter 2 will also present the reader with essential definitions and technical terms used in cell biology in describing the biochemical processes involved in the production of bioluminescence, and in describing the genetic engineering required to produce the construct used as a transducer in this thesis. A description of the important physiological characteristics of the Caulobacter crescentus strains will also be presented. 12 Chapter 3 will detail the experimental characterization of the "transducer" element, C. crescentus, used in the remainder of the thesis. The effects of different protocol factors such as the reagent concentrations and control of the formation of intermediate species will be linked to specific changes in the luminescence signal obtained. The correlation between in  vitro  and in  vivo  results will be used to interpret and better understand the complex luminescence signals from the transducer. The results of Chapter 3 enabled the development of a practical toxicity assay that will be presented in Chapter 4 to demonstrate the potential and feasibility of a toxicity testing system based on the characterized transducer. Toxicity data for several different toxic compounds will be presented and compared to those obtained using other toxicity assay systems. Chapter 5 will discuss the design of a prototype flow-cell system that is intended to eventually be used for the development of a toxicity testing instrument based on the C.  crescentus transducer. Finally the main conclusions derived from this body of work will be summarized in Chapter 6, along with some comments and suggestions for further work. 13 CHAPTER 2. BACKGROUN D BIOCHEMISTR Y AND MICROBIOLOG Y To build up a base of knowledge that will allow the reader to more easily understand the biological subjects applied in this thesis, some background material will be presented and a number of terms will be introduced and defined. The section will be separated into three main parts: The first part will contain descriptions of material related to the biochemical aspects of bioluminescence, followed by material that is relevant to the genetic aspects. The second part will be an introduction to the bioluminescence mechanism, and the last part will provide a brief description of the life cycle of Caulobacter crescentus. 2.1. Definition s an d descriptions o f importan t biologica l term s 2.1.1. Biochemica l aspects The enzyme involved in the biochemical reactions leading to the emission of light in bioluminescence is called luciferase. An enzyme is a polypeptide or protein having a catalytic effect on biochemical reactions. The bacterial luciferase is formed of 2 polypeptide subunits, named a  and |3, encoded by the hixA and hixB genes respectively; its ensemble molecular weight is about 79 kD (where one Dalton, D, is defined as the mass of one |H atom). It is found in several naturally bioluminescent bacterial strains such as Vibrio harveyi, Vibrio  fisheri, Photobacteiium phosphor-eum,  to name only a few. Each strain has a specific type of luciferase, although they all have similar structures and the same basic function. One of the substrates for the luciferase reaction is reduced flavin mononucleotide (FMNH2). This molecule binds to the luciferase to start the sequence of reactions leading to bioluminescence. The FMNH2 molecule is not very stable, being vulnerable to auto oxidation, leading to the formation of FMN. In cell biochemistry, both forms of flavin mononucleotide are involved in the electron transport chain as a coenzyme for the flavoproteins located in the membrane. The complexity of this process makes its 14 description beyond the scope of this thesis. However, we can say that the electron transport chain is responsible for the formation of adenosine tri-phosphate (ATP) molecules, which are the energy "currency" units of the cell. The FMN and FMNH2 are therefore members of a set of molecules that are directly related to the well being of the cell. Thus, if a toxic substance enters the cell and interferes with any of the processes leading to the availability of FMN and/or FMNH2, then the dependence of the luciferase reaction on FMNH2 provides a potential mechanism for assessing the toxic effect on the cell. The structures of FMN and FMNH2 are shown in Figure 2.1. O HN' 0 N ^ ^ N ^ ^C < H -H -H -CH2 - C —  OH - C — O H - C — O H 0 H 2 ~ O -0 - p _ O" - C H 3 - C H 3 0 0 HN" o H II H -H -H -H II 1 I i CH2 - C — O H ~C—OH - C ~ O H O 1 I I 1 I I C H 2 — 0 ~ P -o-- C H - C K - 0 " Figure 2.1. Chemical structure of flavin mononucleotide. The oxydized form (FMN) is shown on the left and the reduced form (FMNH2) on the right. The second substrate molecule involved in bioluminescence is decanal or n-decyl aldehyde. This molecule reacts with a luciferase-substrate intermediate to increase the photon emission efficiency of the light emitting reaction. It can either be generated by the bioluminescent bacteria (if the required genes are present) or can be added exogenously for recombinant constructs (if the required genes are lacking). The availability or amount 15 of decanal is not known to be biochemically linked to any other important reactions in the cell as in the case of FMN. Decanal is a 10-carbon aliphatic aldehyde with the structure shown in Figure 2.2. H H H H H H H H H Q H — C —C — C—C—C — C — C — C— C— C H H H H H H H H H H Figure 2.2. Chemical structure of n-decyl aldehyde (decanal). The luciferase enzyme requires the reduced form of FMN (i.e. FMNH2) to catalyze bioluminescence. Since FMNH2 is known to be unstable, it is relevant to introduce the enzyme responsible for maintaining an adequate level of FMNH2, derived from the stable FMN; this enzyme is NADH:FMN oxidoreductase. This enzyme reduces FMN by oxidizing a reduced molecule of nicotinamide adenine dinucleotide (/3-NADH). The /? -NAD+ molecule is also an electron carrier, like FMN. In addition to its role in reducing FMN, it is also involved in the respiratory chain, or electron transport chain and other oxidative processes occurring in the cell. The structures of /3-NAD+ and /3-NADH are shown in Figure 2.3. 16 ' V C —  CONH , CH3 ^ N K O 0 = P — O 0 H O O H 0 = = p — O O C H -O HO O H NH2 / 9  Y c I I I \ M ^ c . c O H H \ / C —  CONH , C C CH, \ N / (X. ^ 0 = p — o 0 H O O H 0 = p — O O £ HO O H NH2 -C, / 9  y CH2 V " C \ N ^ C Figure 2.3. Structure of nicotinamide adenine dinucleotide. The oxidized form (j6-NAD+) is shown on the left and the reduced form (/3-NADH) on the right. 2.1.2. Geneti c aspects. Since we work with a genetically engineered strain of bacteria, it is important to become acquainted with a few terms relating to the genetic mechanisms involved in bioluminescence. In what follows, we will introduce and explain some of the terms required to understand the mechanisms involved in genetic engineering, in general, and in the expression of genetic information leading to bioluminescence, in particular. A gene is a segment of DNA (deoxyribonucleic acid), the structure of which encodes the information necessary to synthesize a particular protein or polypeptide chain. The genes essential for cell growth are grouped together to form the chromosomal DNA of a cell. DNA can also take the form of so-called plasmids, which are smaller, circularized DNA molecules not linked to the chromosome(s). In bacteria, natural plasmids usually contain genes that convey antibiotic resistance. Plasmids are stable, easily manipulated/modified and can be transferred into (or between) cells using well developed techniques, hence plasmids can also be used as a vehicle (or vector) for introducing some foreign DNA into the cell. 17 The expression of genetic information leading to the production of a protein starts with transcription. This is the process during which a messenger ribonucleic acid (m-RNA) molecule containing the complementary genetic information of a certain DNA segment {i.e.  one or more genes) is synthesized. The mRNA strand will then connect with a ribosome where the process of translation takes place. Translation is the process during which the genetic information contained in the mRNA structure is translated into polypeptide sequences. The polypeptide chains undergo a complex folding process leading to a 3-dimensional configuration (or conformation) that gives the protein its biological activity.. Some proteins may be complete at this stage, however some polypeptides may subsequently undergo further maturation that involves addition of smaller molecules. The processes described are illustrated in Figure 2.4. DNA mmmmm Transcription Y Protein mRNA Translation Figure 2.4. Expression of genetic information from DNA to the production of a protein. Some proteins may also undergo some additional (post-translational) modifications (these are usually very simple in bacteria). Some needs of the cell require the activity of several enzymes and/or proteins. The genes encoding the enzymes and/or proteins involved in a specific function are often A/V 18 linked together, along with DNA sequences required for their regulation, in a section of DNA called an operon. During transcription, the synthesized mRNA will contain the information of the whole operon. Therefore, during translation, the same mRNA molecule can produce multiple copies of several different polypeptides encoded in its structure. The control sites of the operon include a promoter region and an operator region. These two regions regulate the transcriptional mechanisms. The promoter is the site where an RNA polymerase molecule (an enzyme catalyzing mRNA synthesis) binds to start the transcription process. Different promoter sequences can influence the rate of initiation of transcription by a factor of more than 1000 [95]. Next to the promoter is a region called the operator which, in some simple regulatory modes, is the binding site of a repressor molecule. The repressor can physically block mRNA synthesis by binding to the operator site. There are many other regulation mechanisms involved in gene expression, and the overall processes are much more complicated than we need to discuss here, but suffice it to say that gene expression can not only be repressed but also induced. The repression and induction processes may involve several regulatory genes that, in turn, involve many proteins which can interact with molecules that enter the cell from the environment. Gene expression might be induced in a simple way by a specific molecule (which, for example, interferes with the binding of the repressor) or in a more complicated way involving physical or environmental signals such as heat. Recombinant DNA techniques can be exploited to create genetically engineered strains of bacteria that express (i.e.  produce and use) proteins that are not present in the natural (i.e.  wild type) bacteria. The details of this approach are presented in many popular texts [80, 95]. Briefly, the two main processes involved are cloning and selection. Cloning involves submitting a population of bacteria to treatments resulting in the "implantation" of desired foreign genes (typically on a plasmid vector) into some fraction of the population. Selection involves separating this fraction from the non-implanted bacteria and ensuring that the new population retains the genetically engineered 19 characteristics. The type of plasmid is chosen according to the strain of bacteria and selection mechanism desired. Selection is usually based on plasmid-conferred resistance to an antibiotic. The desired genes can be inserted into the plasmid along with specific regulatory genes and control sites, thereby allowing their expression under specific, desired conditions. The plasmid also contains DNA sequences that allow the plasmid to be replicated in the host bacterium and thereby passed onto its progeny. The engineered plasmid construct with which we did the work described in this thesis uses a constitutive expression system where the lux  genes (responsible for bioluminescence) are expressed constitutively (i.e.  at a constant rate that does not involve the action of an inducer). 2.2. Th e bioluminescence mechanis m In naturally bioluminescent bacteria, there can be up to seven genes in the hoc operon [64], These genes encode four different enzymes: the luciferase (hixAB)  and three other enzymes (hixC,  hixD,  luxE)  involved in the generation of the aldehyde substrate. The luciferase also requires FMNH2, which involves other enzymes not encoded in the lux operon; these other enzymes are naturally present in all cells. The involvement of these many enzymes makes very difficult the interpretation of the resulting temporal behaviour of the luminescence signal, since each enzyme has a different kinetic behaviour. However, luminescence can be obtained from a simpler gene construct containing only luxAB,  if the long chain aliphatic aldehyde substrate is added exogenously (i.e.  instead of being synthesized by the cell). The use of the hixAB  genes simplifies the kinetic modeling as it excludes the enzymes related to the production and recycling of the decanal, and also allows experimental control of the decanal concentration. Even though our knowledge of the bacterial bioluminescence system has progressed sufficiently to allow its use as a genetic reporter, not all aspects of the mechanism are yet folly understood. There is some disagreement amongst research groups concerning the model of the molecular biochemistry [56], however it can be 20 adequately described (for our purposes) using the following set of 5 reactions [1, 35, 36, 43, 100]: E,+FMNH2 > E rFMNH2 (2.1) E rFMNH2 + 0 2 > ErFMNHOOH (2.2) ErFMNHOOH + R-CHO > [E,-FMNHOH]*+R-COOH (2.3) [ErFMNHOH]* > E,-FMNHOH + hv (light) (2.4) Ej-FMNHOH > E, + FMN + H20 (2.5) where the luciferase is represented by El; the reduced form of decanal is represented by R-CHO, and the oxidized form of the decanal is represented by R-COOH. The exact mechanism of light emission in this model, and the identity of the emitter, is still the subject of much discussion and research [1, 52, 59, 61]. However, the difference amongst the available models has very little effect on our discussion. The bioluminescence cycle starts with the binding of the reduced FMN to the luciferase (Reaction 2.1). The intermediate complex is then oxidized in the presence of oxygen (Reaction 2.2). The decanal is required at the next step to obtain maximum efficiency of the bioluminescence. The decanal is not essential to get bioluminescence, but the intensity of the resulting luminescence is so dim that the reaction without aldehyde was thought to be dark for many years [32, 33]. The decanal therefore becomes oxidized by the luciferase complex that is then left in an excited state (Reaction 2.3). The excited intermediate then releases a photon as it relaxes, giving rise to the bioluminescence (Reaction 2.4). The relaxed complex then releases the remaining substrate, and the luciferase becomes free to restart the cycle (Reaction 2.5). With the aldehyde added exogenously, and the 0 2 present in the broth, FMNH2 is the only other substrate required for luminescence. FMNH2 has a relatively short life since it is susceptible to auto-oxidation [33, 40, 100]. The principle enzyme responsible for the 21 reduction of FMN to FMNH2 for the luminescence [65] is NADH:FMN oxidoreductase. Therefore a coupled-enzyme set of reactions containing the NADH:FMN oxidoreductase and the luciferase is appropriate to model in  vivo  bioluminescence in cases where the aldehyde is added exogenously, such as in the case of the C.  crescentus transducer described in this thesis. The kinetics of such a coupled-enzyme system more accurately describes the complex kinetics of the in  vivo  luxAB  system than does a single enzyme system. In the coupled model, the five biochemical reactions, (1) through (5), are linked explicitly to the function of the luciferase. In order to complete the biochemical model, we append the following reaction, linking the function of the NADH:FMN oxidoreductase (symbolized by E2) required for the production of the necessary substrate FMNH2: FMN + NADH + H+ + E2 > NAD+ + FMNH2 + E2 (2.6) This two-enzyme model, involving these six reactions, will also be used to analyze results from in  vitro  studies using a crude extract of bioluminescent bacteria containing the two enzymes. The inter-relationships between the six reactions are illustrated schematically in Figure 2.5. 22 o, FMNH, -> E rFMNH2 P-NAD (3-NADH + H FMN 4 ErFMNHOOH R-CHO H20 R-COOH ErFMNHOH <- [ErFMNHOH]* hv Figure 2.5. Schematic diagram of the biochemistry involved in bioluminescence. The coupling between reactions involving the bacterial luciferase (Ej) and the NADH:FMN oxidoreductase (E2) with their respective substrates is illustrated. Not surprisingly, many factors can influence the kinetics of the bioluminescence. FMN concentration, often ignored in the conventional luciferase assays, causes inhibition of the luciferase when used at concentrations higher than a few [xM  [2, 8, 40]. For example, the protocol for light production provided by the supplier of the commercial luciferase suggests using FMN at an inhibiting level of 40 fiM  [87]. The aldehyde chain length, as well as its concentration, also influences the luciferase kinetics and can cause inhibition of the luciferase [34, 39, 56, 65]. As suggested by others [65], it may be possible to optimize the aldehyde chain length for in  vivo work as a tradeoff between light intensity, ease of mass transport, and solubility. Reagents and conditions directly related to the concentration of substrates are important for optimization of bioluminescence assays. Since these effects are not yet well characterized, each researcher tends to use 23 different amounts and different procedures, and it is therefore difficult to compare results. Other constituents of the reaction mix also influence bioluminescence, such as buffers and stabilizers. For example, the presence of inert protein such as bovine serum albumin (BSA) has been shown to increase the in  vitro  luminescence intensity and shift the optimum concentration of decanal required [62], Gelatin has been reported to have a similar effect on in vivo bioluminescence assays using recombinant E.  coli  [51]. The same group also reported effects on luminescence intensity resulting from using Hank's and N-2-hydroxyethypiperazine-N'-2-ethanesulfonic acid (HEPES) buffers as opposed to phosphate buffered saline (PBS) and Tris buffers. Therefore, the characterization and measurement of the bioluminescence must be done in a very controlled way since so many factors have an influence on it. One of the main contributions to this field resulting from our work is, in fact, our classification and resolution of these protocol-related issues. Bioluminescence a s a  toxicity reporter . The foregoing discussions allow us to understand the basic mechanisms involved in the production of light in C. crescentus, but the use of these bacteria as transducers for the measurement of toxicity requires further explanation. The relationship between the bioluminescence and toxicity is complex and not well characterized, however the principles underlying it can be expressed in fairly simple terms. As previously stated, the construct that we use is constitutively expressed. This means that there is no specific inducer; if the cells are in a medium (at an appropriate temperature) that supports growth, then the luciferase should be produced at a constant rate in each cell. A toxic substance can influence the bioluminescence in many ways: it can directly impair the expression of the lux  genes, which would eventually affect the light intensity, or it can indirectly affect other parts of the cell, causing metabolic stresses or even cell death. As mentioned previously, the bioluminescence reactions require two reagents (FMN and NADH) that are involved in many key natural biochemical processes within the cell. Any substance affecting the viability of the cell (i.e.  having a toxic affect) does so by affecting, either directly or indirectly, the main biochemical reactions occurring 24 in the cell. By this action, it will also affect the availability of the substrates involved in the bioluminescence reactions. This briefly summarizes how the presence of toxic substances can modulate the bioluminescence signal. 2.3. Th e Caulobacter lif e cycl e To complete this introduction, we will describe one of the physiological characteristics making the C.  crescentus  an ideal choice for bacteria-based biosensors. This characteristic involves the natural life cycle of C. crescentus  leading to the formation of a self-immobilizing biofilm. The cell undergoes several steps starting with a so-called swarmer state, which is a highly motile cell that "swims" using a flagellum, similar to that possessed by the intestinal bacterium E. coli.  As the cell matures, it loses its flagellum and pili. This step is followed by elongation of the stalk and, during this stage, the cell can attach to a surface, thereby contributing to the enlargement of the biofilm. The stalk cells can also attach to each other to form rosettes of various sizes. Once they are attached, as part of a biofilm or a rosette, they can form a predivisional cell that consists of a stalk cell, which becomes elongated to form the beginning of a swarmer cell. Cell division occurs after the formation of a cross band between the two daughter cells. Note that this type of division does not end with two identical daughter cells as in most cases involving mitotic division; in C. crescentus,  the process of division ends with a swarmer cell and a stalk cell. A diagram illustrating the different stages in the life cycle of a typical C.  crescentus bacterium is shown in Figure 2.6. 25 B D Figure 2.6. The life cycle of the Caulobacter  crescentus,  showing the different stages of A) division, B) the swarmer cell, C) the predivisional cell, D) stalk elongation. C. crescentus cells can stick to a surface to form a biofilm (left), or to each other to form a rosette (right). 26 CHAPTER 3. CHARACTERIZATIO N O F THE IN VIVO  BIOLUMINESCENCE OF THE BACTERIAL TRANSDUCE R Despite the widespread interest in using lux genes as a reporter system for sensors or other applications, the characterization of the temporal behaviour of constructs with engineered bioluminescence have not been reported in the literature. This has resulted in poor repeatability, and difficulties in interpreting and comparing previous results published in the literature. Under controlled conditions, however, we have shown that the luminescence intensity profile produced by the expression of the InxAB  genes exhibits a well defined and repeatable temporal response characteristic. In this chapter, this temporal characteristic is investigated and analyzed in terms of the kinetic behaviour of the underlying biochemical processes involved in the InxAB  mechanism. We will propose an interpretation of the kinetics based on data obtained both from in  vivo  and in  vitro experiments. This study provides insight into adapting the luciferase system for use as a gene activity reporter for applications in toxicity sensing using Canlobacter crescentus and Escherichia coli,  and potentially for a broad range of other bacteria. Our discussion of the in  vivo  bioluminescence will improve the immediate utility of current and previous results, and significantly improve the quality of future assays. It will also facilitate the development of better tools for scientific applications which use the hixAB genes. 3.1. Material s and method s 3.1.1. Measurement s o f in vivo bioluminescence Bacterial strain s an d growt h conditions . C.  crescentus  CB2A [89] was grown at 30°C in a peptone-yeast extract medium (PYE) [70] supplemented with CaCl2 to 0.01% and MgS04 to 0.02%. E.  coli  DHa5 (Bethesda Research Laboratories, Gaithersburg, Md.) was grown at 37°C in L medium [68] consisting of 10 g tryptone, 5 g yeast extract and 5 g NaCl per liter. Agar was added to 1.6% (wt/vol) for solid media. 27 The selection agents used for the different constructs were tetracyline hydrochloride (12.5 /ig/ml final concentration) or streptomycin sulfate (50 /xg/ml final concentration). DNA manipulation s an d geneti c constructions . The bioluminescence obtained from the C.  crescentus  and E. coli  strains was a result of the expression of the plasmids RTB7 and pSW23 respectively. Isolation of plasmid DNA, restriction digests, blunting reactions and ligations were done according to standard methods [80] and were carried out by personnel working in the laboratory of J. Smit in the department of Microbiology and Immunology at UBC. For completeness, these methods are briefly summarized in the following two paragraphs. RTB7 was constructed by J. Ajioka or M. Devan (unpublished construct) and is a fusion of R300B(H) and pTB7 [4]. R300B(H) was constructed by the ligation of Hindlll linkers and RSF1010 [83] that was digested with Hpal (J. Ajioka - unpublished) to form a plasmid containing a unique Hindlll site. The plasmid pTB7, containing the luxAB  gene from Vibrio  harveyi regulated by the anti-tetracyline promoter, was partially digested with Hindlll to open the plasmid at the Hindlll site adjacent to the ampicilin resistance gene, and ligated to Hindlll-digested R300B(H) to form the RTB7. The plasmid RTB7 includes a gene conferring resistance to streptomycin for selection. The plasmid pSW23 was constructed from three existing plasmids: pT7-3, pUC18 and pRK767. The luxA and hixB  genes of Xenorhabdus luminescens  were removed from pT7-3 [93] by digestion with SstI and BamHI. The fragments containing the lax  genes were isolated, purified and the ends were made blunt. The fragment was ligated into the plasmid pUC18 previously digested with the blunt cutting Smal [98]. Two different plasmids were then produced and separated. The recombinant plasmid containing the lux genes oriented in the opposite direction to the lac  promoter was selected and digested with EcoRI and BamHI to cut out the lux  containing fragment. The lux  genes were then ligated into pRK767 downstream of the lac  promoter, also digested with EcoRI and 28 BamHI [48], to form pSW23. The plasmid pSW23 confers resistance to tetracyline for selection. Plasmids were introduced by electroporation [19, 29]. Typical in vivo assays . The luminescence was measured with a luminometer Model 1250 from LKB Wallac® (now BioOrbit®). For all luminometer assays, standard 4 ml cuvettes (polypropylene) were used. Bacteria used for the samples were grown in fresh media and stored on ice. In Vivo Assay #1 was used to study the effect of added decanal concentration. A mixture of 950 /zl of media and 50 fj.1  of bacteria broth (OD60o ~  0.6 to 0.8 absolute units) was placed in the cuvette and incubated at room temperature for 5 minutes. The bioluminescence was initiated by the injection of 20 /zl of various dilutions of decanal prepared from a concentrated stock of 5.31 M. The decanal was diluted in 50% methanol and 50% water. A hand pipeter, Pipetman p20, was used for the injections. Immediately after injection, the cuvette was quickly placed in the luminometer and exposed to the photomultiplier tube (PMT). A tube for air injection (sparging) was inserted into the cuvette in order to ensure proper mixing and oxygenation of the sample during measurement. The sparging system used is not a standard luminometer feature. This apparatus was customized from a piece of stainless steel tubing (OD= 1.5 mm, ID= 1.0 mm, with a flattened end to cause smaller bubbles) inserted into the cuvette through an injection port located in the top cover plate of the luminometer. Air, filtered through a 0.2 tim filter, was used for sparging. A pressure regulator was installed on the building air line to control the sparging rate. In Vivo Assay # 2 was used to study the effect of substitution of the PYE with other defined media. This assay was thus used to study the effect of the presence of the different substrates involved in bioluminescence. For this assay, we designed an injection system that enabled injections through the luminometer head, into the cuvette while it was exposed to the PMT. The original LKB injection system was modified as follows: The 29 injection needle was custom made from stainless steel tubing (OD= 1.5 mm, ID= 1.0 mm). A 10 ml reagent reservoir was also added to the system. The reservoir has a closure system enabling airtightness to minimize the evaporation of volatile reagents. The 3-way valve and syringe holder of the LKB injection system were used unmodified. We used silicone rubber tubing to connect the needle, the 3-way valve and the reagent reservoir together. A mixture of 450 /xl of PYE or HEPES buffer (10 mM, pH 7.2) and 50 /xl of bacteria broth was introduced into the cuvette. Three specific mixtures were used: 1) pure HEPES, 2) HEPES buffer with either 40 /xM of FMN (stock 0.2 mg/ml in water) or 1.3 mM of /3-NADH (stock 10 mg/ml in HEPES buffer, kept on ice and dark), 3) all three reagents, HEPES, 40 itM of FMN and 1.3 mM of /3-NADH together. The mixtures prepared in the cuvette were incubated for 5 minutes at room temperature prior to measuring. The 10 ml injection reagent reservoir was filled with 250 /xl of diluted decanal (0.01% v/v in 2% methanol), mixed with 100 /xl of methanol and 9.75 ml of HEPES buffer (10 mM, pH: 7.2). The luminescence was initiated by injecting 500 /xl of this injection reagent into the cuvette. The mixing was ensured by the injection itself and no sparging was performed. Gas Chromatographi c assa y o f decanal . An isotherm protocol was developed to measure decanal with a laboratory gas chromatography (GC) apparatus. The instrument used was a Perkin-Elmer® Model Sigma 3B. The detection device was a flame integration detector (FID). Helium was used as the carrier gas and hydrogen as the combustion gas. The column used was a DBWAX® Model P/N J125-7012, 15 m long with an inner diameter of 0.33 mm and a C18 coating film of 1 /xm thick. The injector and detector temperatures were set at 190 °C. The column temperature was set at 75 °C. The standard used for internal calibration was linalool (C10H18OH), which was added to the samples to a final concentration of 135 /xM. The test calibration samples consisted of various defined concentrations of decanal in PYE or water and 1% methanol. 30 The injection of 2.5 /xl of sample was performed using a 10 /xl Hamilton® syringe. The decanal concentration of the unknown test samples were determined by comparing the ratio of the decanal peak area to the linalool peak area, with a set of ratio values obtained from standard samples containing known amounts of decanal. 3.1.2. Measurement s o f in vitro bioluminescence Two slightly different in  vitro  assays were developed. All the reagents were obtained from Sigma Chemical Company and used as received, without further purification. The assays were based on the Sigma assay [87] and carried out using the LKB 1250 luminometer. The same injection system described in the previous section and standard 4 ml polypropylene cuvettes were used for both assays. In Vitro Assay #1. A mixture of 25 /xl of decanal (in concentrations varying from 0.0001 to 20% diluted in methanol), 50 /xl of 2-mercaptoethanol (0.01% in water), 350 /xl of HEPES buffer (pH 7.2, 50 mM), and 100 /xl of 0-NADH (10 g/ml in 0.6% BSA and 10 mM HEPES, kept on ice and dark) was placed in a cuvette. The 10 ml injection reagent reservoir was filled with 40 /xl of bacterial luciferase (extract of Vibrio  harveyi from Sigma 10 g/ml in 0.3% BSA and 10 mM HEPES), 2 ml of FMN (0.2 g/ml in water), 7.96 ml of 0.6% BSA in 10 mM HEPES. The luminescence was initiated by injecting 500 /xl of the injection reagent into the cuvette. The injection ensured proper mixing and no sparging was performed. In Vitro  Assay #2. A mixture of 25 /xl of decanal (in concentrations varying from 0.0001 to 20% diluted in methanol), 50 /xl of 2-mercaptoethanol (0.01% in H20), and 450 /xl of HEPES buffer (pH 7.2, 10 mM) was placed in the cuvette. The 10 ml injection reagent reservoir was filled with 40 /xl of bacterial luciferase (10 g/ml in 0.3% BSA and 10 mM HEPES, pH: 7.2), 2 ml of FMN (0.2 g/ml in distilled water), 2 ml of /3-NADH (10 g/ml of 0.6% BSA and 10 mM HEPES, kept on ice and dark), 5.96 ml of 0.6% BSA in 10 mM HEPES. The luminescence was initiated by injecting 500 /xl of the injection reagent 31 in the cuvette. Again, the injection ensured proper mixing and no sparging was performed. 3.2. Result s and discussio n A typical luminescence response profile of our recombinant Caulobacter crescentvs transducer is shown in Figure 3.1 (A) (obtained using In Vivo  Assay #1). The initial phase of the response profile features an immediate, high intensity peak (referred to here as the first or primary phase) that decays rapidly and is followed by an extended, lower intensity phase referred to here as the secondary phase. These qualitative features of the typical bioluminescence profile produced by our assay are highly reproducible and well behaved. Figure 3.1 (A) and 3.1 (B) show prototypical response profiles obtained when the assay is carried out in media that (A) supports growth or (B) does not support growth; both types of assay were used in the characterization reported here. The maximum intensities of the primary phases are indicated by It and I/, and the maximum intensities of the second phases are indicated by I2 and I2'. The time of occurrence of the secondary phases, following the injection of decanal, are indicated by T and T'. We will use these parameters to show the effects of other factors influencing the luminescence profiles. 32 CO c CD CD O c CD O CO CD c 'E Z5 1 ---1 • 1  •  1 - T  H • i \ • i  •  i A . . _ I, \ • • r~i —-r - 1 T - l 1  1  1  1  1 ; A " /  \ - /  \ ; \ , i  , ' I  ,  I 1 1  '  1 B _ . -V r V ' . i '_ -7— 1 0 5  1 0 1 5 2 0 2 5 0  5  1 0 1 5 2 0 2 5 Time (Min.) Tim e (Min.) Figure 3.1. Typical bioluminescence intensity profile of the recombinant bacterial transducer. (A) shows the complex in vivo kinetics of the hixAB system when the assay is carried out in PYE medium. (B) shows the effect of the substitution of PYE by HEPES buffer without addition of any exogenous substrates other than decanal. I1? I/, I2and I2' indicate the maximum intensities of the primary phase and secondary phase responses respectively. T indicates the time of occurrence of the secondary peak. Note that a subsequent injection of an identical dose of decanal, after completion of the luminescence response resulting from the initial injection, results in the reproduction of a virtually identical two-phase luminescence profile. We hypothesize that this repeatability is evidence that the luminescence profile (and therefore the secondary phase decay) is due to decanal limitation. We will verify this hypothesis using the results obtained during the characterization studies. We will also bring to light several protocol-related factors, often neglected (by others), that significantly affect the kinetics. These neglected factors make difficult the interpretation and comparison of the different results 33 previously reported in the literature. Figure 3.1 (B) illustrates the impact that some factors can have on the luminescence profile. We will further discuss these factors later in this chapter. 3.2.1. Characterizatio n o f in vivo bioluminescence Decanal dependence . Since exogenous decanal is required to produce luminescence in our assays, and its concentration must be specified in the protocol, the study of the effect of variations in the decanal concentration was an important first step in the overall characterization. These experiments were further motivated by the fact that many assays described in the literature are performed with poor experimental control [14, 15, 75, 82, 97] and are often misinterpreted due to various effects deriving from such a lack of control. The following set of results will demonstrate the importance of a well controlled assay. The compiled data relating the effect of variations in the decanal concentration to the maximum intensity and time of occurrence of the maximum intensity of the second phase are shown in Figure 3.2. These results illustrate that a sufficiently low decanal concentration will cause a complete merging of the secondary phase with the primary phase. In this case, the absence of the secondary phase can be explained by the complete depletion of the decanal during the production of light forming the primary phase. This would leave no decanal to sustain luminescence during the second phase. Increasing decanal concentration delays, proportionally, the time of occurrence of the secondary peak, while broadening the profile of the entire secondary phase. It also has the effect of drastically decreasing the intensity of the second peak above a concentration of a few /xM. Further increase will eventually lead to the total inhibition of the second phase of the response. At levels of decanal higher than the value causing total inhibition of the second phase, a subsequent injection does not result in the reproduction of similar kinetics (as is the case for lower concentrations). This behaviour is attributed to a combined effect of the luciferase enzyme decanal inhibition and (potentially lethal) damage to the cells. This 34 hypothesis is further supported by other results (not shown) illustrating that the cell number is proportional to the luminescence intensity of the secondary peak. Therefore, a decreasing luminescence intensity is consistent with a decrease in cell population {i.e.  cell death). ^-v ^ •°" 1 "1 ' o > & 2.0 -*-—-C/) c (D •S 1.5 -^ 05 & £. 1.0 -Seconda o 1 0.0-—r i •  i  •  i + ^"~^~~~-^- \ * " \ / \ ^ \ / -// \/^ ^ \ \ / / \ /% \ ^ \ 1 . . . . . . . i  i  i  i  • -• -. " =H 100 200 300 Decanal concentration ((. i M) 20 15 o c 10 3 Z3 O CD 5 ^ . 13 400 Figure 3.2. Effect of varying the decanal concentration on the secondary phase of the luminescence response profile. The data shows the high and low of duplicate experiments. The curve fit passes through the average of the duplicate points. Note that the relation between bioluminescence and decanal concentration can be further influenced by the fact that the decanal is volatile and not very soluble at high concentrations. This may influence the mass transport of decanal through the cell membrane and its distribution inside the cell, as well as affecting the chemical kinetics through (for example) enzyme inhibition or substrate limitation. This factor almost certainly influences the observed behaviour of the intensity of the secondary phase in the 35 case of high decanal concentration. Further experimentation to differentiate between these various influences could be carried out, but the results showing qualitative inhibition at high concentrations are satisfactory for our purposes here. A detailed characterization of the transducer at high decanal concentration has no immediate significance, since it is not practical to operate the transducer in the inhibited regime (high concentrations of decanal). The intensity of the primary phase of the response is not observably inhibited by low or high decanal concentrations. Due to the very short duration of this phase, coupled with limitations in our instrumentation system, we could not detect any conclusive trend regarding changes in the peak intensity of the primary phase as a function of the decanal concentration. Media dependence . It is also important to study the effects of the media composition on the response characteristics. These experiments involved substituting the PYE medium used in the previous study with a series of defined media that permit elucidation of the specific effects contributed by the presence of each of the principle components. In  Vivo  Assay #2 was used to perform this set of experiments. The substitutions were defined by the following cases: 1)HEPES buffer only. 2) HEPES with /3-NADH and FMN. 3) HEPES and FMN. 4) HEPES and /3-NADH. 5) PYE with no incubation time. 6) PYE with 5 minutes incubation time. The resulting kinetic behaviour observed in each case were analyzed in order to obtain an understanding of the significance and the biochemical origins of the two phases of the luminescence response. 36 1.2 ^? O > 1-0 . 'S 0. 8 -9 c 8 °-6 c 0 o eg 0.4 c '£ 3 0.2. 0.0. c o <n 0 Q . 0 » ¥ SO Q to 0) Q . 0 Ill lea m ma. 90 to Q . 0) First phase Second phase SSffii' sail! mi i X Q Q . m §§§ II Won 10) SiiE • a (l) -p> 03 .Q 13 O c .— c o c LU >-Q_ fM -M-3 S:-(D .to # 3 | 0 SE! £ S r " ^ . * E ' SO IsC". iiii' m *Q3 T3 CD XJ 13 O _c 111 >-Q_ .fl> P2-£ Bacteria diluent Figure 3.3. Comparison of the effect of different diluent and incubation time on the in vivo bioluminescence signal. The histogram gives the variation in peak intensity of the first and second phase. When bacteria are diluted using HEPES buffer only, a very low initial luminescence level is observed, followed by a broad secondary phase response as shown in Figure 3.1 (B). This behaviour corresponds to the previously described kinetics (Figure 3.1(A)) with the primary phase absent, and suggests that a necessary substrate for the primary phase is missing in the buffer as opposed to the PYE medium. We therefore experimented with the addition of the two essential substrates (which are jS-NADH and FMN according to the biochemical model described in Chapter 2) to the buffer. When the bacteria were diluted in HEPES buffer with both /3-NADH and FMN, the luminescence profile recovered the primary phase that was absent in the previous case. This result supports the hypothesis that at least one of the substrates, /3-NADH or FMN, is linked to the presence of the primary peak. The next two cases (HEPES plus FMN, and HEPES 37 plus /3-NADH) were performed to identify which of the two substrates was responsible for the occurrence of the primary phase. HEPES plus FMN allowed the primary phase to form, but HEPES plus (S-NADH did not. The conclusion, therefore, was that the primary phase is dependent on the presence of FMN in the media. Note that PYE is a complex media that normally contains FMN. An additional experiment was performed to examine the effect of the incubation time of the bacteria in the presence of exogenous FMN on the primary peak. Using the PYE medium, we performed the same in vivo assay with various incubation times. It was found that a minimum incubation time of 5 minutes is necessary for the primary phase to completely form. Zero incubation time prior to the decanal addition results in an initial luminescence level comparable to the level obtained when no external FMN is present. Hence, incubation is necessary to allow the primary phase to form. Interpretation o f the primary phase . The requirement of a minimum incubation time provided an important clue to the understanding of the kinetics involved in the formation of the primary phase. Understanding the origin and kinetic behaviour of the two phases is essential for interpreting the luminescence signals, and hence for sensor development. Recall that a model using a coupled enzyme system involving an FMN reductase and the luciferase was discussed in Chapter 2. We shall derive our interpretation of the results presented here from this previously described model of the biochemistry (Reactions 2.1-2.6 introduced in the Chapter 2), along with other results reported in the literature. Based on the fact that luciferase-substrate complexes are believed to lead to the formation of long-lived intermediates [34, 39], and by looking at the biochemistry, it is reasonable to hypothesize that Reaction 2.1, as well as Reaction 2.2, would reach equilibrium in the presence of oxygen prior to the decanal addition. If this is indeed the case, then, preceding the decanal addition, a certain proportion of the luciferase is present as a luciferase-FMNHOOH complex. The injection of decanal then allows the fast 38 reactions leading to luminescence to proceed, as shown in Reactions 2.3 and 2.4 [39]. Consequently, the initial luminescence intensity should be proportional to the amount of pre-formed luciferase complex. This hypothesis is supported by the following analysis. The experiment with buffer dilution (Figure 3.3) shows that the initial level of luminescence is related to the presence of exogenous FMN and the incubation time. The low peak intensity of the first phase obtained from simple buffer dilution (no FMN added) corresponds to the actual in  vivo equilibrium concentration of the luciferase-FMNHOOH complex. By adding FMN exogenously, we allow more complex to form therefore leading to a higher intensity initial peak. This implies that the in  vivo  equilibrium concentration of complex is less than the concentration present when exogenous FMN is added. Furthermore, if the amplitude of the primary peak is directly related to the amount of luciferase-FMNHOOH complex, then the concentration of the complex ought to take some time before reaching equilibrium when exogenous FMN is added to the sample. Therefore, if insufficient incubation time is allowed for the complex to form before injecting the decanal, the primary peak should not be amplified. This situation is represented in cases 5 and 6 (Figure 3.3.) using PYE as the diluent, where we can see the expected effect of incubation time on the primary phase. In summary, the low intensity of the primary phase in both cases 1 and 5 corresponds to the in  vivo equilibrium level of the enzymatic complex. Thus, the primary phase requires the presence of exogenous FMN and some incubation time to allow the formation of the complex. This supports the hypothesis regarding the direct relationship between the presence of luciferase-FMNHOOH complex and the initial intensity of the first phase. Interpretation o f the secondary phase . The experiment examining the effect of variations in the decanal concentration also supports a relationship between the primary peak and the presence of pre-formed luciferase complex. In contrast, however, to the observed modulation of the secondary peak with decanal concentration, we observed that 39 the primary peak was not inhibited by high decanal concentration. This absence of inhibition of the primary peak is explained by in vitro  results from other published studies [38, 66,  67], in which it has been recognized that luciferase is more resistant to decanal inhibition when in a complexed state. One group reports that there are actually two binding sites for the aldehyde on the luciferase enzyme, one catalytic and one inhibitory [38]. Evidence suggests that the inhibitory binding site of the aldehyde is made inactive or inaccessible by the binding of the flavin substrate (FMNH2). As expected from their model, their study showed that luciferase complexes were less easily inhibited by high aldehyde concentrations. Our data indicates that the luciferase-FMNHOOH complex has a higher resistance to inhibition, which is consistent with the other groups' observations for the luciferase-FMNH2 complex. From these experiments, we conclude that the intensity of the primary phase was directly related to the amount of luciferase-FMNHOOH complex present. Immediately following the decanal addition which gives rise to the initial peak, the previously mentioned equilibrium between Reaction 2.1 and 2.2 in our model is disturbed. A new equilibrium involving all six reactions has to be reached. When the equilibrium is reached (i.e.  the minimum following the initial phase), the secondary phase of the profile is observed. The transition between the two equilibrium states leading to the minimum between the phases will be discussed following the interpretation of the second phase kinetics. The modulation of the intensity of the luminescence due to changes in decanal concentration has been observed in  vitro  during other published luciferase studies [5, 38, 62]. The studies show that the luminescence intensity is maximum at an optimum decanal concentration, above which an effect known as substrate inhibition occurs. These experiments were performed with different decanal concentrations (negligibly changing in time) leading to a pseudo-steady-state level of luminescence. By doing a similar experiment with a constantly changing decanal concentration, one should observe a 40 temporal modulation of the luminescence, where the intensity would correspond to the instantaneous concentration of decanal. The observed bioluminescence should show an inhibiting, optimum, and a limiting regime. Our analysis of the luminescence profile will show that, during the second phase of the luminescence, the intensity is dictated by and reflects, the constantly varying amount of decanal. If the initial amount of decanal is sufficient to cause inhibition, the luminescence will begin with an inhibited intensity. This inhibited intensity corresponds to the minimum between the two peaks. The subsequent increase of the luminescence intensity corresponds to the steady decrease in the decanal concentration due to consumption and out-gassing (due to sparging). A maximum in the second peak occurs when the decanal level reaches an optimum concentration. The decanal concentration, being constantly decreasing, would then cause the final luminescence to decrease corresponding to substrate (decanal) limitation. This is consistent with our previous conclusion that the secondary phase ends by decanal limitation, because subsequent addition of decanal leads to the reproduction of a similar luminescence profile. The intensity profile of the luminescence during the secondary phase basically reflects the decrease in the concentration of decanal with time, and hence an initial addition of higher decanal concentration implies that the decanal requires a longer time to deplete. The time taken for the decanal to decrease to the optimum concentration (corresponding to the secondary maximum) is related to the amount of decanal initially added, and hence the observed delay, T or T', of the secondary phase increases in response to higher decanal concentration (Figure 3.2) as expected. In the case where the initial addition of decanal corresponds to the optimum decanal concentration, there will be no inhibition period between the two phases. This situation corresponds to the lower limit value of the delay that can be practically detected (the two phases are merged). The smallest delay resolved during the experiment corresponded to a concentration of 50 juM. Therefore, we conclude that the in  vivo 41 optimum decanal concentration is somewhat less than 50 ^M because this concentration barely causes any inhibition before the secondary phase maximum is reached. The optimum decanal concentration was also observed (by the same groups studying the decanal inhibition) to depend on several other potential medium components, such as BSA, phosphate ions and probably others. The addition of a stabilizer such as BSA, and/or a buffer such as PBS, usually increases this optimum concentration from values close to 1 IJM to values close to 100 fiM.  We observed the in  vivo  optimum concentration to be between these limits, as expected. Interpretation o f the transient period . The transient period following the initial primary phase maximum exhibits two distinct kinetic behaviours. In the case where we exogenously add FMN (and adequate incubation time is allowed), the initial intensity corresponds to the amount of luciferase-FMNHOOH complex present at equilibrium prior to the decanal addition. As the luminescence proceeds (Reaction 2.4 of the model), the luciferase releases the remaining molecule of FMNHOH (Reaction 2.5) in order to later reform a complex with FMNH2 and restart the cycle. The luciferase, no longer in a complexed state, is no longer protected against decanal inhibition. The decay of the luciferase-FMNHOOH population, combined with the decanal inhibition of the freed luciferase, causes a decrease in the overall luminescence intensity. Hence, a luminescence minimum is observed as a new biochemical equilibrium involving all six reactions is reached (after decanal addition). This minimum corresponds to the inhibited level of luminescence. In the case where no exogenous FMN is added, the amount of luciferase complex is much less than in the former case. As the experiment with buffer dilution shows, the PYE remaining in the broth containing the bacteria is obviously mostly depleted of FMN. The luciferase can not form a large amount of complex, since the substrate (FMN) is limiting. The small amount of complex then present is close to the amount corresponding to the equilibrium intensity following the initial peak (when FMN is 42 added exogenously). Consequently, the luminescence profile does not show any primary maximum and begins at the inhibited level. Consequences relatin g t o toxicit y analysis . Collectively, these results indicates that the intensity of the primary phase is basically a measure of the amount of luciferase complex present, and that only the second phase is meaningful for equilibrium luminescence measurements. This has very important consequences with regard to further development of sensors and scientific tools using the luxAB genes. For example, if we use our recombinant bacteria for biocide assays, then we would not get a very good correlation between the efficiency of the biocide and the maximum intensity of the initial phase, since the primary phase is related to a very stable enzyme complex which is formed before the biocide can take effect. Any noticed effect would be more likely related to luciferase inhibition than to toxic effects on the cell. Measurement of the secondary phase would, on the other hand, give information which is directly related to toxicity, since this phase requires the presence of vital substrates such as /3-NADH and FMN. These substrates are linked to reactions that are essential to the survival of the cell (e.g.  electron transport chain), and anything influencing the availability of these substrates can undoubtedly be considered as having a toxic effect. In addition, the reproducibility of the bioluminescence profile is very good and is a definite advantage for biosensor development. The maximum intensity of the secondary peak of ten profiles (obtained using In Vivo  Assay #1 on independent cuvettes) showed a mean value of 1.507 ALU with a standard deviation of 0.046, corresponding to a coefficient of variation of 4.39%, which is excellent for biological data. We should also mention that profiles obtained using a recombinant Escherichia coli are qualitatively similar to those presented here using C. crescentvs.  The fact that it is possible to obtain similar profiles using a different species of bacteria is important to demonstrates that the processes underlying the bioluminescence, and leading to the profiles observed using C.  cresceritiis,  are not specific to this bacteria. Rather, these 43 processes are linked to the specific genetic manipulations leading to bioluminescence and, therefore, the characterization of the bioluminescence performed using C.  crescentus should be applicable to other recombinant systems resulting from similar genetic modifications. 3.2.2. Characterization of in vitro  bioluminescence The in vitro  experiments were designed to verify and strengthen the credibility of our in vivo model. Two types of in vitro experiment were performed. The modulation of the kinetics in relation to changes in the decanal concentration was studied. The effects on the kinetics of the presence or absence of FMNH2 prior to decanal injection was also studied. Since the in  vitro  results are based on a commercially available two-enzyme system extracted from Vibrio  harveyi, the in vitro results could be interpreted in terms of the same model used to described the in  vivo  reactions. The advantage of the in  vitro system, however, is that the experimental conditions and concentrations of all reactants can be more precisely controlled. The first experiment was intended to verify that the observed modulation of the intensity of the in  vivo  secondary phase is, in fact, linked to the effect of decanal on the luciferase, and not to other biochemical reactions occurring in the cell. The second experiment was intended to verify the link between the formation of a luciferase-FMNHOOH complex and the initial primary phase peak of the in  vivo kinetic profiles. Assay with pre-formed complex. The first set of results were obtained by mixing a sample containing all the substrates necessary for luminescence with the exception of decanal (In  Vitro  Assay #2). Injection of this sample into solutions containing various amounts of decanal then initiated the luminescence. The sample containing the different substrates was given sufficient incubation time to reach equilibrium before contact with decanal. The typical response starts with a sharp and high intensity peak. The peak quickly decreases to a lower intensity plateau-like profile. These data are summarized in Figure 3.4. 44 - I 1  I  I  I  III] 1  1  I  I  1111| f  I  I  I  I 11 If 1  I  I  1  I  l l l| 1  f-TTTTTT ] f ' T T o--' n • —e— u Plateau Peak • • • . n 0 10 "0 0.08 W CD 0) £Z 0.06 ZJ ' D D o o "§ o irq 1  1  I  I  I  111| 1  1  I  I  I  111] 1—I T 100 100 0 1000 0 CD 0.04 ^ § 0.02 0.00 Decanal concentration (u M) Figure 3.4. Effect of variations in decanal concentration on in vitro  bioluminescence. The concentration dependencies of the initial peak intensity (square) and the subsequent plateau intensity (circle) are shown for the case where the luciferase-FMNHOOH complex is allowed to form prior to the decanal addition. Experiments were done in triplicate; the fit is calculated from the average of triplicates at each concentration. The behaviour of the plateau intensity as a function of decanal concentration was as expected and as described in the literature [38, 62]. This is essentially the same behaviour as seen in the in  vivo  case. At low decanal concentrations, the peak intensity increases with decanal concentration for the same reason as does the plateau intensity (i.e. decanal limitation). This behaviour continues up to the optimum decanal concentration. Once the optimum decanal concentration is reached, a decrease in the plateau intensity is observed, corresponding to decanal inhibition. The initial peak 45 intensity is not affected by decanal inhibition, as expected (i.e.  since the luciferase-FMNHOOH complex is resistant to substrate inhibition). The increasing scatter in the peak data at higher decanal concentrations is attributed to the conditions of the experiment. The decanal is not completely soluble at higher concentrations, causing the sample to become turbid and inhomogeneous. However, triplicate experiments adequately confirm that the peak intensity is resistant to inhibition at high decanal concentrations. This behaviour of the initial peak is the same as that observed for the in  vivo  behaviour of the primary phase. Since the complex is not inhibited at high decanal concentration [38], such a correlation between the in  vivo primary phase and the in  vitro  initial peak supports the conclusion that the luciferase-FMNHOOH complex is also responsible for the in  vitro initial peak. As previously mentioned, we could not observe the in  vivo  kinetics of the first phase with sufficient accuracy us to allow quantitative measurements at low decanal concentrations. Therefore, we were not able to directly observe the behaviour of the primary peak at low decanal concentrations and compare them with the in  vitro  results in this range. However, we could observe the absence of inhibition of the in  vivo  primary phase at high decanal concentration, which is consistent with the in vitro results. Assay without pre-formed complex . The second set of results were obtained by mixing a sample containing all of the necessary substrates, with the exception of jS-NADH and decanal (In  Vitro  Assay #1). To initiate luminescence, the sample was injected into the /3-NADH solution with various amount of decanal. In this case, the luciferase-FMNHOOH complex could not form prior to being mixed with decanal. The kinetics observed by this method differed from the previous case only with regard to the primary phase. The initial high intensity peak is absent in this case; the luminescence increases rapidly to reach a plateau-like behaviour. The plateau level of the two different experiments correspond quite well as shown in Figure 3.5. 46 I . U P 0.8 -intensity Q o b) i O 0.4 -c (D g 0.2 -E _ l 0.0-1 •  1  • D ID | D q 1 Q r»'  "§—-D—- D ' !/ 1 f  •  :  >  1 i '  i  •  i  '  i  ' D Comple x present —•— Complex absent • -• » •  •  • i i  •  •  •  i 0.0 0.5 1.0 1.5 2.0 2.5 Time (Min.) 3.0 Figure 3.5. Comparison of the different kinetics obtained by controlling the formation of the luciferase-FMNHOOH complex prior to the addition of decanal. The same decanal concentration is used in both cases (13 juM). The open square/dotted line corresponds to the profile obtained from the first type of in vitro experiment (complex formed). The solid circle/solid line corresponds to the profile of the second type of experiment (no complex formed). The observed modulation the plateau intensity as a function of the decanal concentration in the second case is very similar to the modulation observed in the previous experiment. This similarity in the secondary phase response is expected since, after the primary phase ends, the same biochemistry takes place in both cases. The correlation between the plateau intensities observed in both experiments is shown in Figure 3.6. 47 0.10-f 0.08 -% 0.06 -Plateau inten; o o I.I. • 0.00-o .  • s 1—r o / Q/ • / o 8/ • / a 1111 u | — i —i 11 1 " M • o <?••''' Q • " i — n -• .o 6. 1 P •  11 1  T- T 1 TTTT] 1  1  1  1  1 l l l| 1 -• Comple x absent ° Comple x present -\ • o..o- 6« I.I. o o  o g Q  9 TTTT] 1  1  !  1  1 111 1 • 1  1  I  I  1  1 I I I 1  ' 100 100 0 1000 0 Decanal concentration ((. i M) Figure 3.6. Effect of variations in the decanal concentration on the intensity of the plateau region of the in  vitro  kinetics. The solid squares describes the behaviour when the luciferase-FMNHOOH complex is not allowed to form prior to the decanal addition. The open circles describes the behaviour when the same complex is allowed to form. The only protocol difference between these two experiments was that the /3-NADH was added as a component of the injection mixture in the first case, and as a component of the decanal sample in the second case. The jS-NADH is necessary for the oxidoreductase to reduce FMN to FMNH2 and, without it, the luciferase can not form a complex capable of resisting decanal inhibition. When the decanal and the /3-NADH are added at the same time, there is no time for the complex to accumulate since the decanal reacts with the complex as quickly as it is formed. The absence of the initial peak under these conditions confirms that the luciferase-FMNHOOH complex is linked to the initial sharp peak of both in vivo and in vitro bioluminescence profiles. 48 Interpretation o f the in vitro results. In the experiment using In Vitro  Assay #1, the luciferase had available FMNH2 prior to its contact with decanal. With an adequate incubation time, an equilibrium amount of the luciferase-FMNHOOH complex was formed. Upon addition of decanal, the initial peak was produced, corresponding to the accumulated amount of complex, as in the in  vivo  case. Our model thus explains the behaviour of the in vitro  primary phase and its correlation with the in  vivo primary phase quite well. However, in the in  vitro  system, the secondary phase as described in the in vivo system does not appear to exhibit the same behaviour. A closer analysis of the kinetics and the experimental conditions will show that this behaviour is, in fact, present, but is characterized by a much longer time scale, making it impossible to measure during this experiment. For the in  vitro  system, the optimum decanal concentration is shown to lie between 20 and 40 /xM. Therefore, below this concentration, no secondary peak should be observed. This is analogous to the observed "merging" of the secondary phase with the primary phase in the in vivo system as the decanal concentration approaches the optimum value. As discussed for the in  vivo  case, the secondary phase reflects the changes in concentration of decanal over time. The intensity reaches a maximum when the decanal concentration reaches the optimum. When the initial decanal concentration used in the in vitro experiments is higher than the optimum value, we should expect an initially inhibited luminescence intensity, increasing to a maximum, analogous to the in  vivo secondary peak. A behaviour similar to this was observed for all repeats with an initial decanal concentration of 130 /xM in experiments performed using In Vitro  Assay #2. In the first in  vitro  experiments (at inhibiting levels of decanal concentration), most of the inhibition that precedes the secondary maximum was masked by the large primary peak, which is inhibition resistant. At a decanal concentration of 130 ttM, however, the initial peak decreased quickly, and the secondary phase exhibited the peak behaviour typical of the secondary phase seen in the in vivo  profiles. For even higher decanal concentrations, 49 the fast decrease of the initial peak was also observed, but no secondary maximum could be observed due to the limited recording time. Note that time of occurrence of the secondary peak of the in  vitro  and in  vivo  kinetics are different because the in  vivo samples were sparged, which accelerated the loss of decanal by out-gassing (as verified by the gas chromotography results reported in Section 3.2.3). However, the qualitative behaviours of the second phase of both in vivo and in vitro experiments are comparable to each other and are consistent with the model presented earlier. To obtain reproducible results throughout the in  vitro  experiments, measurements had to be made in the minimum time possible due to the inherent instability of the reagents (especially /3-NADH). Therefore, throughout most of these experiments, slowly occurring effects were not observed. However, one experiment was done (with a decanal concentration of 130 /xM) where the response was recorded until the secondary maximum occurred (approximately 2.5 hours to reach the peak) and could be identified clearly. Following this result, the notion of a "plateau" becomes relative to the length of time of the observation. Therefore, we should refer to the plateau intensity as "the intensity after X elapsed time units" (typically in the order of one minute for the experiments represented in Figures 3.4, 3.5, and 3.6). The intensity of the secondary maximum was slightly lower than the intensity corresponding to the optimum decanal concentration. This difference is attributed to the limitation of substrates other than decanal. The instability of substrates over the two and a half hour period of time required to reach the maximum may result in multiple substrates becoming limiting, making a quantitative analysis impractical. 3.2.3. Ga s chromatography result s In order to verify the loss of decanal by out-gassing due to sparging in the in  vivo experiments, we proceeded to measure the decanal concentration changes by gas chromatography (GC). A standard curve (Figure 3.7) was first obtained for decanal concentrations varying from 6 to 127 ^M. The end points of this range correspond, respectively, to the low-concentration detection limit of the GC and to the decanal 50 concentration initially injected in the in  vivo  assay used to produce the reference in  vivo bioluminescence profile in this experiment. 0.30-.2 0 2 5 " TO „ 0.20 -1 _ < 0.15 -03 0- 0.10 -0.05-i •  i  •  i  >  i  '  i  •  i / / / • ' / y / /* 7 ^ 1 '  1  '  1  ' • / Y = A + B * X Param Vfelu e A 0.0390 7 B 0.0022 2 R =0.9935 8 SD =0.01254, N = 6 1 i  '  i 1 i /m sd 0.00779 0.00013 1 i X ----. -0 20 40 60 80 100 120 140 Decanal Concentration ( M ) Figure 3.7. Standard decanal concentration curve obtained by GC analysis. The peak area ratio is the ratio of the decanal peak area to the linalool peak area. Experiments involving the direct measurement of the decrease in decanal concentration over time due to sparging were performed in parallel with experiments showing its effects on the bioluminescence. The bioluminescence profiles were obtained using In Vivo  Assay #1, with sparging controlled by a constant flow-rate pump, instead of the previously used air line in order to decrease the error due to variations in the air line pressure. The data relating the bioluminescence intensity of the secondary phase to the sparging time were obtained from independent samples measured using In Vivo  Assay #1. By using different sparging rates, we confirmed that sparging was the primary cause of the continuous decrease in decanal concentration by out-gassing. Recall that sparging needs to be used during In Vivo  Assay #1 in order to ensure proper mixing. 51 The variations in decanal concentration caused by uptake and consumption in the bioluminescence reactions are not significant compared to the loss due to out-gassing. Therefore, the decanal evaporation measurement was done with sparged samples that did not contain bacteria in order to simplify the experiment and ensure compatibility with the GC instrument. The samples injected into a GC column have to be as clean and free of debris (such as cells or cell debris from microorganisms) in order to avoid damage to the column. Filters can be used in some cases, but in our case did not yield reproducible results and hence bacteria-free samples were used instead. The relationship between the decanal concentration in the sample and the sparging time is shown in Figure 3.8. Each data point on the curve was obtained from independent samples, all sparged at the same rate as for the data obtained from the bioluminescence assay. The GC analysis was done using the GC assay described in Section 3.1.1. The decanal concentration data is well behaved and fits an exponential decay function as expected for an evaporation/out-gassing process. 52 ^ 120 -^ • ^ 100 -c o ro 8 ° -t _ c 8 6°-c o O 40 -"co ro 20 -o CD Q o -I 28^ D. 'D* 9 M •-•'' \ .Cf' 1 \ d \ . P i i D. P D D \ 1 "E CO • ^ ^ I 1 ' a.. D D 1 1  ' 1 ——J1 1 " _ • -• . 0 5  1 0 1 5 Sparging Time (minutes) 20 50 - 4 0  3 0 - 2 0 - 1 0 3 CD W O CD O CD 5" r—*• CD 13 W > r -c Figure 3.8. Measured out-gassing of decanal and its effect on bioluminescence. The variation of decanal concentration (solid line, solid square) and the bioluminescence intensity (dotted line, open square) are shown as a function of sparging time. The results obtained from the GC analysis support our interpretation of the bioluminescence temporal profile. The results support conclusions coming from both the in vitro and in vivo experiments. In addition to showing that the change in intensity of the bioluminescence is directly related to decanal decrease, it has revealed an approximation of the optimum decanal concentration necessary for maximum peak bioluminescence intensity during the secondary phase. This optimum concentration of 28 /JM  is of major importance for getting the maximum signal out of the C.  crescentus  transducer. Maximizing the signal from the transducer is essential for the development of sensor instrumentation giving the best signal qualities possible (i.e.  high sensitivity, high signal/noise ratio). 53 3.3. Conclusion s The observed bioluminescence due to the expression of recombinant luxAB  genes in C. crescentus  was characterized and interpreted. The observed in  vivo bioluminescence consisted of a complex two-phase intensity profile. To elucidate the processes underlying this complicated response, the behaviour of its different phases has been studied in relation to several factors indicated by the biochemistry describing the luxAB  mechanism. The variations in the concentration of exogenously added decanal had a characteristic effect on the secondary peak. An approximate value was determined for the optimal decanal concentration (i.e.  28 /JM)  corresponding to the maximum intensity of the luminescence observed for the in vivo  case. The peak intensity of the primary phase can be controlled by using hepes buffer (with or without FMN) instead of PYE medium. The primary phase was found to be the appropriate parameter to measure in order to obtain information concerning the amount of luciferase present. However, it is not an appropriate parameter to measure for biocidal studies because of the stability of the complex responsible for its presence. These studies lead to the development of a biochemical model based on a coupled-enzyme system explaining the observed results. The validity of this model was established by the good correlation found between in  vivo,  and in  vitro  experiments (where conditions and reagents could be better isolated and controlled). We also elucidated the non-negligible effects of important protocol-related factors influencing the kinetics of the bioluminescence produced by bacterial luciferase. The adequate control of the different concentrations of substrates, the time allowed for complex formation, the use of buffers and stabilizers, and the method of measurement of the luminescence are important in order to obtain meaningful results. The temperature is another important influencing factor for biochemical studies. All experiments reported here were performed at room temperature unless indicated otherwise. In our experience, however, the slight variations in temperature that may occur during the course of an experiment at room 54 temperature do not produce noticeable effects. Nevertheless, biosensor instrumentation based on the luxAB  mechanism (see Chapter 5) will include provisions for thermostatic control. The understanding and regulation of the mentioned factors are critical for any quantitative measurements involving genetically engineered bioluminescence using the luxAB system. We have presented results indicating that careful control of the solution environment of the bacteria, combined with a knowledgeable interpretation of the results using the described model, have a definite impact on our application, as well as other (gene reporter) applications using the luxAB  system. For example, the coupled-enzyme luxAB system has been studied in vitro by another research group [54] that concluded that the kinetics should vary in relation to the ratio of the two enzymes. The luminescence was observed to increase to a plateau level when the reductase/luciferase ratio was low; with a high ratio, the luminescence was observed to peak and decay. It was also stated, in the same publication, that the commercial luciferase had a reductase/luciferase ratio leading only to continuously decreasing luminescence. We have studied in  vitro  luminescence of the commercially available coupled-enzyme system and were able to obtain plateau-like or peak-like response behaviours without changing either the ratio of the enzyme, nor the other substrate concentrations, but only their order of combination with each other. Our results showed that the general conclusions of this group are not entirely correct. This underlines the importance of an adequate model for interpretation of the response, and adequate knowledge of the influencing factors. 55 CHAPTER 4. TOXICIT Y ASSAY PRELIMINARY RESULT S As mentioned in the introduction, there are several existing methods for measuring toxicity. In this section, we will compare the approach based on C.  crescentus with some of the other existing methods and instruments. A short description of the available methods and their relative advantages and disadvantages will be presented. The presentation of a batch assay based on recombinant C.  crescentus  will follow, supported by a set of experiments demonstrating the potential of a toxicity sensing instrument based on a Caulobacter biosensor. 4.1. Measurin g toxicit y The conventional approach used to assess the relative toxicity of a chemical agent involves exposing (or administering) subgroups of a known population of living organisms to a predetermined set of concentrations (or doses) of the toxic agent, and recording the number of deaths that occur in a fixed amount of exposure time at each dose. In general, with experimental support in some cases, the relative toxicity assessed in relation to one species can be correlated in a meaningful way to other species. This motivates the development of alternative methods of toxicity testing (i.e. non-animal) that would address the concerns raised by animal's right activists, and would decrease the cost of toxicity testing as bacterial organisms are much cheaper to maintain than higher organisms. Conventional types of assays (described in the literature review) have debatable advantages and disadvantages. Considered advantageous is the fact that they have been widely used for many years, so it is very easy to compare the toxicity of different compounds based on these "conventional" organisms (e.g. fish, rats, mice). Data on a vast array of compounds of interest are available in the literature and/or material safety data sheets. However, the cost of these tests can be a major burden, especially for industrial users that are required by law to perform such tests on a routine basis. One factor that contributes significantly to this cost is the time required by personnel to perform these 56 tests (e.g. a standard fish assay can take several days). In addition, the high cost of special facilities, and the overall running cost of such an assay can be prohibitive for industries (e.g. pulp and paper) required to monitor their effluent. Also, increased public disapproval has arisen with the recent concern for laboratory animal's rights. There is now a growing demand for alternative means of measuring toxicity. In the late 1970's, a group of researchers proposed to use bacteria as an alternative organism instead of fish, mice, rats or others. Their research resulted in the development of a toxicity monitoring system which they later marketed under the trademark of Microtox® [11, 12, 13]. The principle of operation of this system is quite similar to the fish or mice assays and is described in Chapter 1. The Microtox® system has many advantages when compared with the fish and mice assays. The most important factor, as far as industry is concerned, is its much lower overall cost. The assay can be performed on a few millilitres of sample, using small quantities of the Microtox® reagents and the toxic compound of effluent sample under test. The assay of a particular compound might require up to 10 assay cuvettes, and can be performed in under an hour. A well-trained technician can test on the order of 10 compounds in a day. For these reasons, the Microtox® system has gained a lot of popularity with industry, and has therefore become a very successful product. Despite the above advantages, however, the Microtox® system has some serious limitations. The reproducibility of the data, from assay to assay, is very poor. In some instances, there are differences of two orders of magnitude in the data reported from different laboratories [44]. This is mainly due to the high sensitivity of the assay to small variations in the protocol used. It is apparent, therefore, that the microbial "transducer" (i.e. the P.  phosphoreiim Microtox® reagent) used in this test has not been adequately characterized to refine the protocol, and thus permit a consistent measurement and interpretation of the bioluminescence signals. 57 The fact that the Microtox® system is still widely used, despite its limitations, is due to its relative convenience and low cost compared to tests using fish or other higher organisms. The acceptance of this test, despite the low reproducibility of its results, is a strong indication of the need (and market demand) for practical alternatives to vertebrate animal testing. There is presently no competitive microbial system on the market, let alone one that gives reproducible results. An engineered Caulobacter-based biosensor system has the potential to offer better and more reproducible results than the Microtox® system, while preserving the same convenience and cost advantages. Furthermore, the potential exists to genetically modify the sensitivity of a Caulobacter assay, and even to engineer Caulobacter strains that are expressly sensitive to specific agents. Further development of a microbial toxicity assay, based recombinant strains of Caulobacter, are thus highly motivated. The first step toward the development of a toxicity monitoring system, based on Caulobacter, is to investigate its sensitivity and precision (i.e.  repeatability). A series of experiments were done to realize this objective using the knowledge acquired during the characterization of the construct reported in Chapter 3. A protocol was developed to produce data, relating to specific substances, that can be compared to similar data obtained using the Microtox® system. Signal versus  concentration (i.e.  dose-response) curves were obtained for three different toxic compounds. The precision of this dose-response data demonstrates the feasibility of a microbial toxicity measuring system, based on Caulobacters, that could be optimized in order to realize significant advantages over the Microtox® system. 4.2. Material s and method s APMT-based luminometer Model 1250 from LKB Wallac® (now BioOrbit®) was used to measure the bioluminescence intensity as described in Chapter 3. Standard 4 ml luminometer cuvettes were used to present the samples to the PMT. A fresh culture of 58 C. crescentus expressing the plasmid pRTB7 was prepared for each test as described in the previous chapter. The assay samples were prepared in the following way prior to the introduction of the toxic agent to be tested: 975 JX\ of medium (PYE kept at room temperature) was first mixed with 25 yi\ bacteria broth (culture is in exponential phase with OD600 of 0.6-0.7 and kept on ice) and left to incubate for 5 minutes (at room temperature). The cuvette containing the diluted bacteria was then placed in the luminometer and injected with 20 JX\ of decanal reagent dissolved in methanol and distilled water, leading to a working concentration of 130 /xM of decanal and 1% methanol in the cuvette. A small tube was then quickly inserted into the cuvette for sparging. The bioluminescence intensity was translated into a potential by the luminometer, and a chart recorder was used to record the signal. The prototypical temporal response, using a similar protocol, was previously characterized as having two phases. Once the luminescence intensity of the second phase had decreased to within 5% of the baseline, the sample was removed from the luminometer to allow a 20 p\  injection of a predetermined dilution (in water) of the toxic agent to be tested. The injection was followed by a 5-minute incubation period (at room temperature). The sample was then replaced in the luminometer and a second dose of decanal was injected to produce another bioluminescence profile. The pair of profiles thus obtained formed a single data point. The measured decrease in luminescence intensity of the secondary phase peak in relation to the first one corresponds to the toxic effect of the injected concentration of the toxic agent. Pairs of profiles were obtained following the above protocol for a series of different toxicant concentrations. The data were displayed by plotting the ratio of the maximum intensities of the second phase of the bioluminescence profiles (i.e.  after poison injection/before poison injection) versus  the concentration of the compound being assayed for toxicity. The data points were not obtained by injecting the toxic compound in a sequential (increasing or decreasing) order of the concentration, but in a pseudo-random 59 order. The three compounds tested were phenol (C6H5OH, stock solution was 90% in water), mercuric chloride (HgCl2, stock was a pure powder) and sodium hypochlorite (NaCIO, stock solution was a 6% commercial bleach). To determine the sensitivity of this protocol to sample incubation time, a series of experiments were also performed where one toxin was tested with varying periods of incubation of 5, 10, 15 and 20 minutes. 4.3. Result s and discussio n The toxicity assay for the three compounds lead to the expected sigmoidal shaped curves of standard dose-response studies and were well-behaved as shown in the presented graphs. The phenol dose-response data exhibited the best behaviour of all three. The 5-minute EC50 of phenol was found to be 6.5 mM as shown in Figure 4.1. Figure 4.2 shows the sodium hypochlorite dose-response characteristic. For this compound, the 5-minute EC50 was found to be 0.36 mM. The third compound assayed was mercuric chloride and Figure 4.3 shows the 5-minute EC50 to be 0.011 mM. 60 1.0H ,—_ & 0.8 -'c ba (0 .E 0.6 -M— o • • * = fi 0-4 -"§ 0.2 -Q_ 0.0-r " W / •  •  •  •  • i — i • ... • ™ w / i — 1 1  I  1  T " '•-... •"'-,. • -I <=>- E m C D 1 1 — i — i — i -i i  i  i 0.5 • ''•••• '••• • i i i  i  i ^ -• --_ --I r - ^ 0.00 M I I I I I I / / -0.0 1 10 100 1000 Phenol Concentration mM ppm Figure 4.1. Phenol dose-response characteristic. The Profile Ratio expresses the intensity of the secondary phase peak in the presence of phenol, relative to the intensity obtained in the absence of phenol, for the same sample of bacteria. The EC50 is the concentration of phenol at which the Profile Ratio equals 0.5. Recombinant C.  crescentus exhibits a 5-minute EC50 of 6.5 mM for phenol, corresponding to 570 ppm in the test sample. 61 1.4, ^ 1-2 -CC 'S=> c 1.0 -b i 2 0.8 -*+— • * * — • * ' O 0.6 -• * 3 <o „ o : 0.4 . !g 0.2 -o l _ ° - 0.0 --0.2-i i — i — i i  1 1 1 1 1 — i — i — i i  1 1 1 1 1 — r -1 -  -  . • \ M a 1 3 « CN O 1 1  '  1  1  1  1  1  |  1  1  1  1  1  1  1  1  |  1  1— ' 1 1  1  1  l l | 1  1 -. _ 0.66 " _ \ \ • -' • • • • • • H l l  l  l  l  l l l  l 0.1 0.01 0.1 1 '  " I 10 1 1  100 Sodium Hypochlorite Concentratio n mM ppm Figure 4.2. Sodium hypochlorite dose-response characteristic. The Profile Ratio and EC50 are defined as in Figure 4.1. Recombinant C.  cresceiitus exhibits a 5-minute EC50 of 0.36 mM for sodium hypochlorite, corresponding to 24.6 ppm in the test sample. 62 CO g CO c o Q_ 1.6 1.4 1.2 1.0 0.8-I 0.6 0.4-0.2-0.0 - 0 . 2 - ^ 0.00001 0.7 61 Q. , _ O T -O O co d 0.0001 ~ I — 1 I  1  1  I  J  1  I  1— I M T T ] 1 I 0.001 0.01 "T 1 — 1 ) 1 1 mM ppm 0.01 0.1 1 Mercuric Chlorid e Concentratio n Figure 4.3. Mercuric chloride dose-response characteristic. The Profile Ratio and EC50 are defined as in Figure 4.1. Recombinant C.  crescentiis exhibits a 5-minute EC50 of 0.011 mM for mercuric chloride, corresponding to 3 ppm in the test sample. An additional set of experiments were performed in order to determine the effect of the incubation time on the dose-response characteristic. We chose sodium hypochlorite to test whether or not the response would decrease significantly with longer incubation times. Data were collected for several different incubation times performed at three different sodium hypochlorite concentrations as shown in Figure 4.4. The variations in Profile Ratio obtained at each of the concentrations tested were within the range of variation observed at zero concentration for all incubation times. The results indicate that the assay is essentially independent of the exposure/incubation time up to 20 minutes (the longest incubation time tested) in the case of sodium hypochlorite. 63 *"~v CO •4=1 c ^ CO c M— "*—' u -i= (0 (1) <£Z O i _ CL 1.6-1.4-• 1 ? -1 0 -0 8 -0.6-0.4-0.2-0.0-0 5.2 8 4 2 Sodium Hypochlorite Concentratio n Figure 4.4. Effect of incubation time on the sodium hypochlorite dose-response characteristic. The dose-response curves obtained are all qualitatively similar. The specific shape observed is an indication of an adequate assaying technique as the sigmoidal shape is typical of dose-response behaviour [58, 73], This also implies that the parameter chosen for measuring the effect of toxicity (i.e.  the maximum intensity of the secondary phase) is an appropriate parameter. Validation of the parameter also provides some support to the interpretation of the bioluminescence mechanism that we developed and discussed in Chapter 3. Each response curve can be divided into three zones that all have distinct meanings in regard to toxicity. The initial "flat" zone at low concentrations is known in the literature as the NOEL (No Observed Effect Level), and corresponds, as the name suggests, to the safe zone where the concentration of the compound causes no toxic effect. The second zone is called the responsive zone. It corresponds to the range where 64 the toxic effect varies in proportion to the concentration of the compound under test. In the responsive zone, there is always an unaffected component of the population. The third zone is located at the extreme upper level of the concentration scale. It corresponds to the zone in which the toxicity can not cause any additional effect, and hence is called the saturated zone. In this range of concentration, all of the population is affected. In addition to exhibiting excellent qualitative features, the quantitative qualities also appear promising when compared with other toxicity testing methods such as the Microtox® assay. For example, an important feature of these results is that they yield relative toxicity rank order for these compounds that agrees with the literature. The C. crescentits-based transducer shows the same order of relative toxicities as does the Microtox® system. The evaluation of the relative toxicities of different compounds is one of the major applications for this kind of system. The Microtox® system has been compared to a fish assay (fathead minnows [17]) and even though the relative toxicities of a broad range of chemicals did not correlate very well with the fish assay (with a R2 = 0.65), the correlation for semihomologous chemicals was much better (R2 = 0.96 for a series of alcohols). Since the Microtox® system is accepted and widely used in industry (despite poor correlation, in some cases, with other toxicity tests), then a C. crescentus-bastd system that agrees with Microtox® in terms of relative toxicity, would very likely be accepted too. The sensitivity of our C.  crescentiis  transducer for phenol (5-minute EC50 = 570 ppm) and mercury (5-minute EC50 = 3 ppm) is somewhat inferior to the sensitivity of the fish (rainbow trout, 96-h LC50 = 7.5-14 ppm), daphnid (small crustacean, 48-h LC50 = 24-43 ppm) and Microtox® (5-minute EC50 = 18-42 ppm) assays according to one study comparing the three methods [74], but can be compared to the S. volutcins  toxicity assay (5-minute MEC90 = 400 ppm, MEC90 is based on the motility inhibition of 90% of the bacteria [74]). The sensitivity of our C.  crescentiis  transducer for sodium hypochlorite (5-minute EC50 = 25 ppm) is also inferior to the Microtox® system (5-minute EC50 = 0.1 65 ppm), but we found only one reference treating this compound [11], and therefore the confidence on this value is unknown, especially given the variability that exists in the other Microtox® results reported in the literature. The apparently inferior sensitivity of our assay has only a small impact on the evaluation of a C.  crescentus  transducer for measuring toxicity. The first reason supporting this affirmation is that the parameters are not absolutely comparable. For example, comparison of 5-minute EC50 (for C. crescentus) to 96-hour LC50 (for fish) should not be taken too literally, since it is known that the sensitivity of many organisms to a toxic compound is related to the time of exposure to the compound. This is precisely why there are 24, 48 and 96-hour LC50 fish assay protocols. It should also be born in mind that the C.  crescentus  transducer can, in principle, be further optimized genetically to be more sensitive. To conclude this discussion on sensitivity, we should say that its importance also depends on the application. For quantitative detection of toxic chemicals, the sensitivity is of major importance and great care must be taken in the choice of the parameter and protocol, but for relative evaluation of the toxicity of several compounds, the sensitivity per se  is not critical. The precision, and hence reproducibility, of the results is a more important factor than is the sensitivity. Even though we have not carried out an extensive study at this stage to evaluate the reproducibility of our toxicity testing assay, the collected results obtained throughout the work forming this thesis demonstrated that, despite the non-optimized conditions under which the assays were performed, we can obtain results that are more reproducible than those obtained using the Microtox® assay. We are confident that, with better control of parameters such as temperature (our results were obtained at room temperature, which can vary slightly from day to day), and optimization of the assay conditions (as reagent concentrations and incubation time), techniques used (as injection and mixing) and the genetic optimization of the bacteria itself, we would obtain even 66 better precision then demonstrated here and possibly better than other highly characterized toxicity tests. We could have chosen to further investigate the quality of our assay by generating extensive data on several additional compounds for comparison with the existing toxicity tests, but this would be superfluous at this point in time since further development of the C. crescentus assay is planned (as discussed in the following chapter). Our initial goal here was to demonstrate that the principle of toxicity testing using a recombinant C. crescentus had significant merit and could potentially be used in a further optimized toxicity testing system. The results obtained satisfied these objectives and opened the door for the next phase of the project. Although many features of the results obtained using the present protocol are comparable with other commercial systems, the main drawback of the batch protocol used here is the time required to assay the toxicity of a given toxicant. Even though much faster than the conventional fish assay (24 to 96 hours), the protocol used here required up to 30 minutes per data point, for a total of 6 hours for a 12-point dose-response characteristic. This places the protocol used here behind that of the Microtox® assay. However, again, further work could be done to optimize the time required to generate meaningful toxicity data. An injection of a lower concentration of decanal, combined with a better controlled sparging/mixing system could easily cut in half (and probably more) the data generation time by directly diminishing the duration of each luminescence profile. Such an improvement would place the existing C. crescentus system in a better position to compete with the Microtox® system, even without the additional (genetic) optimization and fine environmental control mentioned previously. Even though it may be possible to develop a system with somewhat better performance than the existing Microtox® system, it would offer fundamentally the same features, if used in its present form. The protocol used in this chapter did not take advantage of the biofilm-forming qualities of the biological transducer. The possibility of 67 self-immobilization of the bacteria makes possible their use in a flow-through system, which can offer different and additional features from those of the "batch" protocol used here. Furthermore the use of a flow-through system to immobilize the bacterial transducer would facilitate the fine control of the environment surrounding the bacterial transducer. It would also allow the study of an inducible (or chemically specific) transducer, where the exposure to the target chemical must be transient and tightly controlled. It is the preliminary work toward the development of such a flow-through system that is treated in the next chapter. 4.4. Conclusion s In this chapter we developed a toxicity assay protocol based on our understanding of the bioluminescence mechanism gained in Chapter 3. The batch protocol was based on the temporal profile of the bioluminescence that changes in intensity in relation to the continuous decrease in decanal concentration due to sparging. The maximum intensity of the second phase of the profile was found to be related to the viability of the cells in the sample (e.g.  cell number). Dose-response curves were compiled from the observed correlation between the second phase maximum and the toxicity of the compounds tested, allowing extraction of relative toxicity data. The assay was used to generate preliminary toxicity data for the sake of assessing the feasibility of the transducer as a toxicity sensor, and for evaluating the potential for developing a more elaborate toxicity assay based on a flow-through system. The results obtained satisfactorily met these objectives and exhibited some encouraging features. The dose-responses characteristics were well-defined and exhibited the expected shape. The relative toxicities of the different compounds tested was in accordance with other results generated by existing toxicity assay methods. The batch protocol clearly demonstrated the feasibility of toxicity measurements based on the microbial transducer. However, our experience and knowledge acquired throughout the studies presented here suggested that 68 a flow-through protocol would be more appropriate and would have several advantages over the batch-mode assay used here. Further discussion of a system involving such a protocol will be treated in the next chapter. 69 CHAPTER 5. FLOW-THROUG H SYSTEM DESIG N The concept of applying the C. crescentus  transducer in a flow-through system had been considered during the very early stages of this project, but some indications of its feasibility and potential had to be demonstrated before focusing on the development of such a system. The characterization of the transducer performed in Chapter 3 allowed us to predict that the transducer could produce a meaningful steady intensity signal if it was used in a flow-through configuration. The toxicity data obtained in Chapter 4 demonstrated the value of the bacterial transducer when used as a toxicity sensor. These two important features were essential to prove that the bacterial transducer could be implemented in a complex flow-through toxicity assessment system. In order to reach this goal, a simple prototype flow-through system, comprising the most basic design requirements, had to be developed first. The design and development of such a prototype instrument, which takes advantage of the self-immobilizing characteristics of C. crescentus as well as the inherent advantages of flow-injection analysis technology, is the subject of this chapter. 5.1. Motivation s for the development o f a  flow-through syste m Compatibility o f the transducer with a  flow-through system. The assumption that the luminescence signal is stable when a constant decanal concentration is maintained was previously verified. Recall that we performed an experiment (based on In Vivo  Assay #1 in Chapter 3) with a very slow decanal decay {i.e.  we used the luminometer without sparging the sample) and the intensity of the second phase of the bioluminescence profile was virtually steady over several minutes. The overall shape of the temporal profile of the bioluminescence intensity was conserved, but the time required to go through the process was on the order of 20 times longer than with the sparging. Although this experiment is not likely to yield to quantitative results, due to the concerns raised previously regarding oxygenation and proper mixing, it did show that the bacteria were capable of sustaining 70 the bioluminescence signal over a long period of time (over at least one hour, possibly more). As we showed in Chapter 3, the decanal concentration dictates the intensity of this secondary phase luminescence, and hence by exposing the bacteria to a constant level of decanal, one will obtain a constant level of bioluminescence. The actual hardware (luminometer system) did not allow us to maintain a perfectly steady concentration of decanal during the bioluminescence assay. However, a flow-through system would allow us to maintain a constant decanal concentration around the bacteria by preparing a fixed concentration of decanal in the flow stream, and would thereby cause the transducer to produce a steady signal. Advantages o f a  flow-through  system. A flow-through system can take advantage of the already well-developed technology and methods of flow injection analysis (FIA) [8, 78, 79]. FIA technology also allows precise control of very small samples and reagents. This feature, combined with the self-immobilizing capability of the biofilm-forming C.  crescentus, allow almost total control of the environment surrounding the bacterial transducer (and much better control than a batch protocol can offer). For example, the exposure to toxic compounds can be limited, and can be introduced with different dose rate characteristics (e.g.  gradually or by steps). This type of control can ease the task of maintaining a "healthy" transducer that can perform for a longer period of time, and therefore allow the system to produce data of better quality in terms of both accuracy and precision. In addition, FIA technology allows a certain level of automation, which further improves the efficiency of FIA methods. The assessment of toxicity can be a lengthy and complex process that involves multiple steps in the preparation of the sample, and can significantly benefit from automation. A flow-through system, using FIA technology, would be much easier to automate and would show great potential for good reproducibility due to the high degree of accuracy that is available in the control of the reagents. 71 One of the main weaknesses of the batch protocol used in Chapter 4 to generate the preliminary toxicity data was its slower speed compared to the Microtox® system. The protocol was lengthy due to the fact that we had to wait for the decanal concentration to decrease between each step of the assay. Hence, we could have a much faster protocol if we did not require the decay of the decanal. This is possible by taking advantage of the possibility of generating a steady intensity signal when appropriately used in a flow through system (as discussed previously). This would allow much easier characterization, and faster data acquisition, since the detection and monitoring of the effects of toxic compounds on a steady signal is easier than the way it was done using the protocol of Chapter 4. In addition, the general approach of submitting the test population to a specific sample dose in a stagnant (e.g.  batch) system is acceptable for most compounds, but does not necessarily produce meaningful results for chemicals that are not stable in an aqueous environment [76]. For example, halogens such as chlorine and bromide chlorine rapidly decrease in concentration due to reaction with ammonia, normally found in aquatic environments. The products thus formed can further decay by several complex reactions, and hence the concentration of the target compound (in the original form) and its toxicological effects will remain in question. All of the above factors support the case for development of a flow-through toxicity system. 5.2. Desig n and development o f a  flow-through prototyp e instrumen t The physical development of a complete flow-through system involves a number of sub-systems: 1) the development of a flow-cell unit that will contain the biofilm, 2) the development of a fluid handling system that will regulate the content of the flow-cell, and 3) the development of adequate photodetection and signal conditioning instrumentation. A block diagram of such a system is shown in Figure 5.1. A realistic objective, at this stage, is to build a simple and inexpensive prototype of a flow-through system that will 72 allow the desired control of the decanal concentration and be able to detect with acceptable precision and accuracy the light output generated by the biofilm. These two features will yield capabilities lacking in the previously used PMT-based luminometer, and give us an essential tool to perform further tests and studies for the development of a toxicity biosensor system based on C. crescentus. Photodetection Instrumentation Figure 5.1. Block diagram of the flow-through system. Illustrated are the three main sub-systems that will be developed and treated separately. 5.2.1. Flow-cel l design The prototype described in this chapter was designed using glass microscope slides as the primary structural elements for the flow-cell and as carrier/substrates for the Caulobacter biofilms. We required a flow-cell which facilitated the easy exchange of biofilm slides for experimental flexibility and ease of characterization. Each biofilm was grown on a separate slide. The flow-cell was designed to accept the slide (covered with a biofilm), and enclose it in a transparent, non-permanently sealed cell for operation. Producing a reliable, non-permanently sealed enclosure was a major challenge and several early designs were rejected before all of the problems were solved. The final working prototype is depicted in Figure 5.2. This design uses silicone rubber tubing glued with silicone adhesive to large paper clamp as a convenient, removable closure device. The strength of the paper clamp was sufficient to press the tubing against the opening of the flow-cell and prevent leakage due to the pressure of the liquid flow stream. This pressure was the main reason for failure of the previous closure designs. The main body of the 73 flow-cell was also assembled using silicone adhesive (Silastic®) and two sizes of glass microscope slide (2.5 x 7.5 cm and 5.0 x 7.5 cm sizes). A future prototype of the flow-cell may not require the option of being easily and frequently opened. For such use, we investigated the feasibility of fabricating a flow-cell using silicon micromachining techniques. A small cavity can be etched in a wafer and covered by an anodically bonded Pyrex wafer or glass microscope slide (or cover slip). The biofilm could be grown on the interior wall of the flow-cell. A single wafer could potentially accommodate several small flow-cells that could be subsequently separated for individual use. The advantages of such a design are further miniaturization and the potential for large scale production of inexpensive disposable flow-cells. The details of this method of fabrication of a flow-cell are presented in Appendix A. A micromachined prototype flow-cell was fabricated and the flow characteristics were found to be suitable, but this approach was not pursued further due to the impracticalities of working with this type of flow-cell during the prototype development and characterization stage. 74 Flow cell sealing device Id.»J.F_ mk •.--B-jailA.i.liill-' - '  1U1L.Jy.-L.iJhi u — i— . 1 A l ! 1, *!l -Flow cell open end 1 -  J Tubing for reagent flow Removable biofilm support Silicon rubber seals Tubing for reagent flow \J^ Flow cell sealing device Open Seale d Side view Outside covers Removable biofilm support Flow cell lumen B Guide for the biofilm support Transverse view (a t the cut AB) Figure 5.2. Schematic diagram of the flow-cell fixture for supporting the biofilm and exposing it to the flow stream. The dimensions of the microscope slides used as outside covers are 5.0 x 7.5 cm. The biofilm support is a regular microscope slide of dimension 2.5 x 7.5 cm. The two guides for the biofilm support consist of three regular microscopes slides "sandwiched" together with a 1-2 mm offset on the middle slide. The reagent flow is introduced via silicone tubing (ID = 4.0 mm, OD = 6.0 mm). 75 5.2.2. Flui d handling system design The fluid handling system is perhaps the most simple sub-system of the overall flow-through instrumentation system previously described. The essential parts required to assemble such a system are: 1) tubing that will carry the flow stream, 2) a pump, 3) a controlling valve system allowing the choice of different solutions for the carrier flow stream, and 4) an injection valve for introducing compounds to be tested, into the carrier stream. Tubing an d flow contro l components . The tubing must be as inert as possible due to the wide variety of possible compounds and solvents that can be present in the flow stream. For this reason, all tubing was specified as Teflon® (PTFE), silicone rubber, or PVC. The valves also must resist a wide variety of chemicals, hence all valves were purchased with PTFE wet parts (Hamilton® low pressure valves). A 4-way distribution valve ensures adequate flexibility in switching between multiple carrier stream. A 6-port injection valve was necessary to allow introduction of a well defined sample plug into the flow stream. Pump specification . The pump must maintain aseptic conditions in the flow stream. A peristaltic type of pump would ensure that no parts of the pump directly contact the flow stream. The pump should also be dual-channel in order to be able to handle two parallel independent flow systems that will be required in later protocols using one reference and one test channel. However, the most important parameter to consider in the choice of the pump is its required range of flow rates. We will develop an estimate for this parameter in the following. One of the main concerns in designing this flow system was to be able to ensure viability of the biofilm by flowing the essential substrates for the bacteria. The main stream will contain PYE medium containing required nutrients, but a dense biofilm contained in a small volume flow-cell (potentially smaller once optimized) can exhibit much higher uptake rates than regular broth cultures because of the artificially high cell 76 density due to the confinement factor. Most nutrients can be provided in correspondingly larger concentrations in the flow stream. However, oxygen, which is essential, has a low solubility that limits its influx. Therefore, higher uptake has to be compensated by faster flow rates. The oxygen uptake rate of C. crescentus  could not be determined experimentally with our instrumentation. A realistic, conservative estimate of the required flow rate can be derived by assuming an oxygen uptake rate equivalent to that of some other organisms known to have large oxygen uptake rates such as some yeasts [3]. The rate assumed here, in terms of grams of oxygen per gram of cell per hour, is: Oxygen Uptake Rate = 0.3 g 02/(g cell hour) (5.1) The cell mass per area of biofilm was calculated based on a biofilm of maximum cell density {i.e. no void space between cells): Bacteria Cell Standing Height = 10 /xm Biofilm Cell Volume per Unit Surface Area = 0.001 ml/cm2 (5.2) Cell Density = 1 g cell/ml (5.3) therefore: Cell Mass per Unit Surface Area = 0.001 g cell/cm2 (5.4) By multiplying together Equations 5.1 and 5.4, we obtain the oxygen uptake rate per unit of surface area of biofilm: Oxygen Uptake Rate per Unit Surface Area = 0.0003 g 02/cm2 hour (5.5) The practical concentration of oxygen that a typical medium can carry in aqueous solution is: Dissolved Oxygen = 0.000007 g 02/ml medium (5.6) The pumping rate requirement per unit of surface area of biofilm is obtained by dividing Equation 5.5 by Equation 5.6: Required Medium Influx Rate = 43 ml/cm2 hour = 0.7 ml/min cm2 (5.7) 77 The flow-cell can contain up to 34 cm2 of biofilm; the maximum required flow rate can then be calculated by multiplying by Equation 5.7 by this area: Maximum Flow Rate Required = 26 ml/min A Gilson® Minipuls 3 peristaltic pump was chosen which can produce flow rates up to 34 ml/min. This model also offers digital (RS422A/485) or analogic external control, dual channel pump head, digital readout of the RPM value, and fine adjustment of the flow rate. The external control option facilitates future automation of the system, although the present system is controlled by manual settings. 5.2.3. Optica l and electronic system design The first step in choosing a light detection system was to evaluate the optical flux intensity that the detectors would have to measure. From our preliminary tests, we knew that photomultiplier tube (PMT) devices had adequate sensitivity to be considered as one of the possible choices. PMT's are capable of photon counting, however it would be more desirable to use semiconductor devices such as photodiodes, since these devices require simpler, low-voltage circuitry and are physically more robust than PMT's. These factors are important when considering devices to make an instrument capable of field operation. In addition, they can be more easily coupled to the flow-cell due to more flexibility in the size and geometry of available devices. Photodiodes are relatively inexpensive and yet offer adequate sensitivity for quantitative measurements of fairly low light levels. Charge coupled devices offer many of the same advantages as the simpler photodiodes, but the cost of each unit, along with the instrumentation required to operate at low light levels, sufficed to orient our choice toward photodiodes. To establish the feasibility of this approach (photodiodes based instrumentation), we needed to find the expected emission flux intensity of a biofilm to know if the commercially available photodiodes were sufficiently sensitive. We will first derive the luminometer instrument transfer function, relating the photon flux to the output voltage. This calculation will then be used to derive the photon flux per bacterium in a suspension culture, which can then be used to estimate 78 the flux per bacterium in a biofilm. This parameter dictates the required sensitivity of the semiconductor detection devices. Evaluation of the transfer function o f the Iuminometer. In all of our previous experiments, we used a Iuminometer made by LKB Wallac® (now BioOrbit®). The Iuminometer light detection device is a photomultiplier tube (PMT). It has a built-in standard for calibration and is setup for detection of luminescence in liquid samples held in a 4 ml cuvette. The relationship between the photon flux (<£x) seen by the PMT and the output voltage of the Iuminometer instrument can be expressed as ^ ( p ^ ( o U , p u , ( V ) . G F ) ( 1 1 ) ( 5 g ) s e c G v j / G i ^ QE^ hc/A Where: Gv0i0 = Current-to-Voltage Gain of the Iuminometer amplifier (V/A). GI0I; = Current Gain of the PMT (A/A). GF = Geometrical Factor taking into account the non-detected portion of the photon flux emitted by the sample due to the geometry of the apparatus. QEX = Quantum Efficiency of the PMT at a specified wavelength X (AAV). hc/X = Energy of a photon of wavelength X (J/photon). The instrument has to be calibrated so that the built-in ,4C standard causes a 10 mV output signal. From the specifications of the instrument, we can obtain the QEX and estimate the hc/X value for the photons emitted by the standard. The calibration protocol can be used to deduce the value of <&420 corresponding to 10 mV using the 14C standard. The combined value of the three remaining factors can then be deduced by substituting the values of QEX, hc/X, $420, and the calibrated output signal into Equation 5.8. 79 We will first obtain the value of $420 that results in a 10 mV output signal. The specifications of the apparatus state that the 14C standard has an activity of 0.26 /^Curies, for which the scintillation process produces photons at a rate of 250 photons/decay. The emission spectrum of the 14C standard is centered at 420 nm. For our purposes, we will neglect the shape of the spectrum and assume that a single wavelength is emitted at 420 nm. An approximation for the photon flux ($420) of the standard can then be made as follows: By definition: 1 Curie = 3.7xl010 - ^ Z sec 1 decay = 250 photons 14C Standard Emission: 0.26x10"6 Curie 14C Photon Flux = (0.26xl0"6 Curie)(3.7xl010 d e ° a y )(250 p h ° t 0 " ) sec • Curie decay 14C Photon Flux = $420 = 2.4xl06 p h o t 0 " (5.9) 42U sec The specifications of the PMT reveal a quantum efficiency of: QE420 = 40xl0-3 AAV (5.10) The energy of each photon emitted by the standard can be evaluated as follows: By definition: h = 6.62xl0'34 J-s c = 2.998xl08 m/s ,„™ x (6.62xl0"34 J-s)(2.998xl08 m/s) 4 „ tn.19 r / £ n . 1 Photon (420 nm) = - ^ - r =4.73x10 iy J (5.11) 420x10"9m By combining Equations 5.9, 5.10 and 5.11 into Equation 5.8, we obtain the following: Gvoio'Gioii = r 1 v 1 output (V) GF Whc/XK $x } Gv0i0 -GI0I, = 1 1 output (V) GF 40xlO"3A/W 4.73xl0"19 J/photon 2.4xl06 photon/sec Gv0i -GI0I, = 2.2x10" V/A (5.12) GF 80 With this value, we can now deduce the efficiency transfer function of the luminometer at the wavelength emitted by our bacteria which is 490 nm. From the specifications of the instrument we obtain: QE490 = 22xl0-3 AAV (5.13) and we can calculate the energy per photon at 490 nm as , A (6.62xl0-34 J-s)(2.998xl08 m/s) , „ r i „ . l 9 l , L hc/ \= - ^ - '-  =4.05x10 19 J /photon (5.14) 490x10"9m By substituting Equations 5.12, 5.13 and 5.14 into Equation 5.8, we obtain the following expression for the photon flux at 490 nm: *« ( £^ > " (7^ ) (r7I ) (7T^><°utput (V)) sec QE490 hc/X G V J / G I J , ^ ( ^ ' S x l O M ^ X O u t p o K V ) ) (5.15 ) sec s e c • v Equation 5.15 yields the approximate relationship between the output signal of the luminometer and the actual photon flux (at a wavelength of 490 nm) seen by the PMT. It is important to note that this is only an approximation for the photon flux captured and detected by the PMT, and not the total photon flux emitted by the sample. The accuracy of this transfer function is affected by the error present in the values used for the quantum efficiency of the PMT and the 14C standard emission rate. Unfortunately, it is not possible with the means available to us, to improve on the accuracy of these values, and they are not available from the manufacturer. However, the accuracy is sufficient to obtain a rough approximation that enables us to determine the feasibility of employing a photodiode detection system for the flow-cell. The bioluminescenc e flux  densit y fro m a  biofilm . An estimate of the photon flux density (photon/(sec-cm2)) emitted by a biofilm is required in order to determine if the sensitivity of the photodiode devices is sufficient. The photon flux emitted per bacterium can be estimated by measuring the cell density of a culture of C. crescentus  and using the 81 luminometer transfer function (Equation 5.15) to calculate the photon flux emitted by the measured number of cells. Our evaluation of the cell density of a typical broth by microscopic visual count using a microcytometer is: Cell Density = l.OxlO9 cell/(ml-OD6 0 0) (5.16) A conservative value, obtained in our experiments, of the output signal of the luminometer per O D 6 0 0 units for a volume of 1 ml of broth is: Signal Density = 25 Volt /(ml-OD6 0 0) (5.17) Therefore, the normalized output signal per bacteria cell is: Signal Density = 25 V o l t / ( m l - Q D 6 0 0 ) Cell Density lx lO 9 c e l l / ( m l - O D 6 0 0 ) S i g n a'D e n S i t y = 25xl0-9 ™ (5.18) Cell Density cell By substituting Equation 5.16 into Equation 5.81, we obtain a value corresponding to the photon flux per bacterium: *'490 = ( 5 x 1 0 * * ^ X 2 . 5 x 1 0 - * ^ ) sec • V cell * ' « » = 1 3 ^ (5.19) *yu sec-cell Multiplying this factor by Equation 5.14 will give us the equivalent power per cell: Power , , , / photon NW he , J — — = ( * ' 4 9 o ( - — n ^ T ^ - ^ - ^ cell sec-cell A photon Power ,8 W =5xl048 (5.20) cell cell A realistic (but non-optimized) value for the biofilm density is: cell Biofilm Cell Density = 107 — - (5.21) cm Therefore, an estimate of the power density emitted by a biofilm is: 82 Power Density = (5xl0"18 )(10' —=-) cell cm n W Power Density = 5x1 CT11—- (5.22) cm2 This value of power density is a conservative, rough estimate of the potential power density that a biofilm can emit. This calculation contains parameters that are not optimized, such as the signal density (Equation 5.17) and the biofilm density (Equation 5.21). These parameters could lead to a considerably larger value for the optical power density once optimized. Choice o f bioluminescenc e photodetectio n instrument . The conservative estimate of the power density from a biofilm is within the range of detection of photodiodes. Commercial photodiodes are available with a detection limit of 1012W. The designed flow-cell accommodates a microscope slide that has about 17 cm2 of area per side, and both sides can be covered by a biofilm for a total available area of approximately 34 cm2. Photodiodes are commonly available with active areas of up to 6 cm2. Therefore, with a few cm2 of active area exposed to a larger biofilm, we could measure the bioluminescence with acceptable sensitivity. We chose to use a photodetection instrument based on photodiodes used in a photovoltaic mode. This mode allows low noise and high sensitivity. Two instruments were chosen: a single-channel Tramp® multi-gain amplifier, and a dual-channel multi-gain optometer Model 380 both instruments were manufactured by UDT Instruments® (now Graseby Optronics®). The single-channel multi-gain amplifier has the advantages of being smaller and battery operated, therefore it can be easily placed in a light tight Faraday cage (which will contain the flow-cell and photodiodes for light detection) in order to be less affected by external noise for high sensitivity light detection. Unfortunately, it only offers a 0 to 5V analog output, nevertheless it will be used in the early trials of the flow-through system due to its slightly higher sensitivity. The dual-channel optometer offers many features such as built-83 in memory containing photodiode calibration data, IEEE 488 communications port, 0 to 5 V analog output, digital readout of the value of the signal in various scales and units, baseline signal subtraction and others. This instrument will be used once the preliminary characterization of the protocol produces optimum luminescent signals. It is an instrument compatible with more complex protocol (e.g.  eventual toxicity assay protocol) which could be automated to some degree. The diodes chosen were PIN photodiodes Model pin-25D manufactured by UDT Sensors®, which are designed to be operated in a photovoltaic mode and are compatible with the described amplifier and optometer instruments. They have an active area of 613 mm2 and have a detection limit of lO12 W as required. Instrument modification s fo r improve d signal-to-nois e ratio . The first tests done with the newly acquired instrumentation revealed more noise than expected. We proceeded to design a filter to remove noise from the signal at the output of the instrument. Before the description of the filter, we will introduce a simple method to further improve the signal-to-noise ratio (S/N) using several photodiodes in parallel. This type of set-up allows ensemble averaging of the summed signals, which improves the S/N ratio. The effectiveness of ensemble averaging can be explained as follows: Assume that one photodiode produces a signal SP. By connecting n photodiodes in parallel their signals are summed to n-SP. Each signal SP contains a noise component Np. It can be shown that, provided the noise components are uncorrected, then the n noise components will add together as the square root of n (i.e.  Vn-Np) [88]. Therefore, by combining n photodetectors together in parallel, the signal-to-noise ratio is improved by a factor of Vn as shown below: N Vn-Np Np 84 We calculated the signal-to-noise ratio improvement factor based on the maximum noise amplitude of an assumed constant intensity dark current signal. The result of the measurement of S/N obtained when diodes are connected in parallel, divided by the measured Sp/Np obtained with only one diode, leads to a value of 1.4 (i.e.  7/5), which is identical to the theoretical value of v 2 . We will now describe the design of the noise filtering instrument that was designed in order to remove excessive remaining noise at the output of the photodiode amplifier. Since our bioluminescent signals are expected to be steady or very slowly changing with time in the flow-cell, a low-pass filter would address the noise problem with a minimum impact on our signal. The necessity to incorporate a filter also brought the opportunity to incorporate into the design a few additional features to enhance the flexibility of the instrument. These additional electronics were designed, fabricated, and tested with the assistance of S. Jubenvill and D. Robinson as part of their fourth year project laboratory (see Appendix B). We required the filter to have a cut-off frequency of 0.5 Hz to obtain adequate noise reduction in the output signal. To allow the option of using the filter for signals changing faster than 0.5 Hz, we added 3 more selectable cut-off frequencies (1.0, 10 and 100 Hz). The UDT Instrument® Model 380 optometer has two channels, hence we decided to incorporate into the design a two-channel summation option. The only other addition to the design was the implementation of a variable gain option for each channel (0-100 V/V). These modifications would give greater flexibility for future use, and provide compatibility with the luminometer. The final instrument was a dual-channel filter/amplifier with summation option [42]. It is capable of accepting 0-10 V input, and produces 0-5 V output. The output of our filter/amplifier also has a limiter circuit for surge protection to provide future compatibility with an A/D board that will be used for data acquisition. A block diagram of the instrumentation is shown in Figure 5.3. 85 Instrumentation Amplifier Instrumentation Amplifier Gain Amplifier Filter Clipping Detector Clipping Detector Gain Amplifier Filter Limiter Summer Limiter 7 Clipping Detector Limiter Figure 5.3. Diagram of the filter/amplifier. The instrument features various selectable cut-off frequencies and infinitely adjustable amplifier gain (0-100 V) options on two channels that can be summed. The output signal is limited to a maximum of 5 Volts. The instrument can be described as a set of parallel sub-systems connected together at the input of the summer. Each channel input passes through an instrumentation amplifier of unity gain that provides high input impedance and high common-mode rejection ratio. The next amplifier stage provides infinitely variable gain between 0 and 100 V/V. The amplified signal then passes through the filtering stage, which consists of a two-pole, active, low-pass Butterworth filter. Four different cut-off frequencies are available as mentioned above. The filtered signal passes through a voltage limiter that ensures the 0 to 5 V output required for compatibility with the A/D board. The summation option is a standard "weighted summer", with both inputs weighted equally. Additional clipping detection was implemented with a LED to provide warning of limited signal. The performance of the instrument in terms of signal-to-noise improvement was tested following the same basic protocol used for evaluating the effect of parallel diodes. The evaluation of the S/N obtained when filtering the signal, divided by the Sp/NP, obtained when filtering was not used, gave the corresponding improvement due to the filter. The signal-to-noise improvement was calculated in two different situations using 86 two cut-off frequencies for comparison. In the first case, the ratio was obtained from the dark current signal of the photodetectors placed in the light tight box, while the Tramp® amplifier was outside of the box (also used as a Faraday cage). The ratio of S/N divided by Sp/Np was found to be 4.3 for the 0.5 Hz cut-off filter and 2.5 for the 1.0 Hz filter. In the second case, the ratio was obtained from actual bioluminescence signals from a biofilm, while pumping reagents through the system. The pump motor introduced a large noise component into the signal. The ratio of the S/N (while pumping) divided by Sp/Np (without pumping) was 0.2, which corresponds to a five fold decrease in the signal-to-noise ratio. The use of the filter was almost essential in this case. A signal-to-noise improvement ratio of 3.3 was measured when using the 0.5 Hz cut-off filter, and 1.7 when using the 1.0 Hz cut-off filter. The combination of the custom filter/amplifier and the UDT® Model 380 optometer represents the complete photodetection apparatus used in the flow-through system to measure the intensity of bioluminescence signals emitted from the biofilm. One pair of photodiode detector heads, connected in parallel, is installed on each side of the flow-cell to allow maximal detection of the bioluminescence (illustrated in Figure 5.4). 87 Flow cell sealing device Photodiodes. Inlet tubing Outlet tubing Parallel connections To the photodection instrument Figure 5.4. Photodetector coupling to the flow-cell. The parallel connection of the photodiodes is illustrated with each diode facing the biofilm. The distance separating the diode active area from the biofilm is approximately 4 mm. Two similar set-ups could be implemented in parallel and connected to the two channels of the optometer. 5.3. Preliminar y tests of the flow-through system A complete prototype of the flow-through system has been assembled in order to determine experimentally whether or not such a system is adequate for detection of the biofilm bioluminescence. The prototype includes a flow regulation system consisting of two reagent bottles (one neutral buffer or PYE media and the other containing the decanal solution) linked to the flow-cell via PTFE, PVC and silicone rubber tubing. The Gilson® peristaltic pump is used to control the flow of one of the reagents chosen by a multiport 88 distribution valve. The waste is collected in a container directly linked to the outlet of the flow-cell. The photodiodes are fixed on both sides of the flow-cell, which is placed in a light-tight box made of aluminium. The ports allowing the tubing and wiring to enter and exit the box were carefully inspected to be light-tight. The photodiodes are linked to the photodetection apparatus consisting of the optometer and our custom made filter/amplifier instrument. The output signal of the photodetection apparatus is recorded using a chart recorder. A schematic of the system is illustrated in Figure 5.5. Light-Tight Enclosure Waste Distribution Valve Photodetection Apparatus Chart Recorder Figure 5.5 Diagram of the overall flow-through system. The reagent bottles contain a neutral solution for rinsing/flushing the system and a solution containing a decanal solution for the bioluminescence process. The flow is driven by a peristaltic pump. The injection valve is used to insert defined amounts of a compound to be analyzed, into the carrier stream for exposure to the biofilm. The outflow is collected in a waste bottle. The photodetection amplifier is shown and can have one or two channels depending on the application. This system satisfies all the instrumentation requirements necessary to allow the detection of bioluminescence from a biofilm (c.f.  bacterial suspension), and to allow 89 control of decanal and other reagent concentrations. It is a simple configuration that allows for easy modification to facilitate future characterization of different types of biofilms and could eventually be automated by using computer controlled pumps, valves and data acquisition. 5.3.1. Material s and methods Biofilm growth . The bacteria were grown at 30°C in PYE media with streptomycin sulfate as described in Chapter 3. Microscope slides were placed in the media and immobilized in a customized slide holder to allow bacterial growth on both sides. The bacteria were allowed to grow until an OD600 of the suspension culture reached 0.6 to 0.8. The culture containing the biofilm was then put on ice while awaiting to be used. Measurement o f bioluminescenc e fro m biofilms . The described flow-through system was used with two different carrier stream reagents. One reagent consisted of 1% methanol in PYE media (air saturated) containing 28 /xM decanal, and the other was a blank solution consisting only of 1% methanol in PYE media (air saturated). The flow-through system was flushed with the blank reagent prior to insertion of the biofilm (attached to a microscope slide) in the flow-cell. Once the biofilm was in place in the flow-cell, the light tight enclosure was sealed and monitoring of the bioluminescence was started and continuously recorded on the chart recorder. The reagent containing the decanal was then pumped at a fixed flow rate of 26 ml/min. The bioluminescence was detected with two circular photodiode photodetectors of 6.13 cm2 active area located on either side of the flow-cell. The signal from the photodetectors was added by connecting the photodiodes diodes in parallel and amplified with the Graseby Optronics® a low-noise Tramp® amplifier (adjustable transimpedance between 103 to 1010 V/A). 5.3.2. Result s and discussion The results obtained in this simple experiment provided important experimental validation of the flow-through system approach, and of the use of semiconductor 90 photodiode detectors for biofilm measurements. The experiment was designed to demonstrate the capacity to produce and detect a steady light output signal from a real biofilm. Decanal presen t i n the carrier stream. When first inserted in the flow-cell, the biofilm is exposed to the blank reagent (1% methanol in PYE). At this point, the decanal concentration in the flow-cell should be very close to zero, but trace amounts remaining from an incomplete flushing of the flow-cell (i.e.  dead volume) between independent biofilm luminescence measurements leads to a non-zero initial bioluminescence intensity. Once the flow is switched to decanal-containing reagent, the bioluminescence intensity quickly increases to a maximum and shortly after, stabilizes to a lower level, as illustrated in Figure 5.6. 91 0.25-—^' o £ . 0.20 -Q) O c _  „ 0 0.15 -O CO 0) ~ 0.10 -5 0.05 ^ f) 00 c 1 1 Decanal Inflow T 7 i i 1 X--5. 1 11 1 i ) 5 • i  »  i  •  i No Decanal Inflow -_ - i ^S fe^ ^ " ^ f e -*^^W ^ Dar k Signal Level 1 i  >  i  •  i 10 1 5 2 0 Time (min) Figure 5.6. Biofilm bioluminescence monitoring. "Decanal inflow" indicates that the bioluminescence was recorded while pumping (26 ml/min) 28 piM of decanal in 1% methanol in PYE in the flow-cell. "No decanal inflow" indicates that the bioluminescence was recorded while pumping PYE only. The dark signal level indicates the limit of detection of the photodetection system. The error bars illustrate the maximum noise amplitude of the signal. The dynamic shape of the signal is once again related to variations in the decanal concentration seen by the bacteria. As the decanal containing reagent is pumped through the flow-cell, the concentration of decanal in the flow-cell increases gradually until reaching the reagent concentration of 28 piM. Once the concentration reaches steady state, the bioluminescence signal exhibits the same behaviour as discussed earlier. The reason that the steady signal intensity is less than the maximum intensity is that the previously determined optimum decanal concentration for in  vivo  bioluminescence was slightly over estimated. Therefore, the decanal concentration increased to a constant, but 92 inhibitory, level. By supplying a lower (constant) decanal concentration in the flow stream, the steady signal intensity increases toward the maximum value corresponding to the true optimum. This level (data not shown) was found to lie between 10 and 20 fiM. Another experiment was performed that submitted a biofilm to an almost optimum concentration of decanal for an extended period of time. The flow rate was decreased to a value of 5 ml/min which was found to be sufficient to maintain a steady bioluminescence signal. The experiment lasted 2.25 hours, during which the bioluminescence intensity was maintained practically steady with slight variations of less than 5% (due to flow rate perturbation) of the total signal amplitude. This experiment clearly showed that the biofilm was capable of producing a steady intensity bioluminescence signal while exposed to a constant decanal concentration for a prolonged period of time. Decanal absen t fro m th e carrie r stream . The use of a decanal concentration larger than the optimum decanal concentration allowed us to measure the maximum intensity of the biofilm bioluminescence signal, which allowed a meaningful comparison with the predicted maximum intensity estimated in Section 5.2.3 (vide  infra).  The reproducibility of the signal was verified by rinsing the biofilm with the decanal-free (blank) reagent and repeating the procedure. We recorded the signal while rinsing it (Figure 5.6), and observed the expected release of inhibition by decanal as evidenced by an increase in signal intensity immediately following the beginning of the rinsing operation. Once again, the decrease of the decanal concentration from an inhibiting value to a limiting value causes the bioluminescence signal to pass through a maximum. The lower intensity of this maximum, compared to the maximum intensity obtained when the decanal concentration passes from non-inhibiting levels to inhibiting levels is attributed to a difference between the kinetics associated with the onset of inhibition and the release of inhibition. If the changing decanal concentration due to the rinsing process decreases at a faster rate than the release of inhibition, then the optimum decanal concentration could be reached before all of the inhibition effect is reversed. This would lead to a lower 93 maximum intensity going from high to low concentrations than would arise going from low to high concentrations, even though the decanal concentration passes through the optimum value in both cases. This explanation is only hypothetical at this stage, but now that a working FIA system is available, these processes can (and will) be investigated in future studies involving the biofilms. Quantitative analysi s o f the signal . Repeated bioluminescence signals obtained from the same slide were very reproducible and stable. Averaged triplicates yielded a maximum intensity signal (decanal inflow) of 0.256 V, with a standard deviation of 0.017 and a coefficient of variation of 6.6%, which is excellent for biological signals. Similarly repeatable signals were obtained from different biofilms, and the variations in intensities from film to film were relatively small; the maxima of the signals ranged between 0.22 and 0.32 Volts. The difference in intensities are mostly due to the differences in cell density of the biofilms, which varied between 6.7 and 8.2xl06 cell/cm2. Methods of addressing these variations in the cell density will be a subject of further studies planned for the optimization of the growth protocol. A photomicrograph of a typical biofilm prepared according to the procedure described in Section 5.3.1 is shown in Figure 5.7. The measured intensities allow us to derive a better estimation for the emitting power of the biofilm: Biofilm Density = 7xl06 cell/cm2 (5.23) Measured Signal = 2xl0"12 A/cm2 (5.24) Photodetector Efficiency = 0.25 AAV (5.25) * ™ * = !<>-» W- (5.26) cell cell Power Density = 7x1012 - ^ - (5.27) cm" Our preliminary estimation of the power density was 5xl0-nW/cm2 for an estimated biofilm density of 107 cell/cm2 which leads to a cellular emitted power of 10"17 W/cell. 94 The experiment revealed a power density of 7xl0"12 W/cm2 for a biofilm density of 7xl06 which leads to a cellular emitted power of lO18 W/cell. The difference between our previous estimation and the measured signal is mostly due to the rough estimation of the PMT-based luminometer transfer function, although some component of the discrepancy is also due to uncertainties in the cell density, the biofilm density and the effective active area of the photodiodes. The luminometer instrument can not be accurately calibrated without an external standard, which was not available, however we can now use both instruments to obtain a relative calibration between the PMT and the photodiode devices. The quality of measured biofilm signals is very good in terms of signal-to-noise ratio. The use of the filter/amplifier instrument with a cut-off frequency of 0.5 Hz yielded satisfactory results. Further optimization of the protocol is expected to improve the signal-to-noise ratio even more since, for example, the density of the biofilm could realistically reach as high as 108 cell/cm2. This would increase the signal significantly compared to the noise level. 95 4 * ' t • '  r ' v  \  -  ' < J  i  ^  / ' , v •» i  *  . J ^  *  "l** * * * * * *  . - . * • * - V ^ .. _ , » _ _ _ _ *  j * * L Figure 5.1. Photomicrograph of a section of a typical biofilm used in the flow-through system. The width of the frame is 81 um. The image was acquired with a CCD camera mounted on a microscope. 96 5.4. Conclusion s A flow-through bioluminescence measurement system for use with Caulobacter biofilm has been designed, fabricated and tested. The system performed according to the design parameters. The flow-through system allowed us to generate a steady intensity bioluminescence signal, and permitted much greater control over the microenvironment of the bacteria than did the cuvette-based luminometer system. The results presented here, combined with the toxicity results presented in the previous chapter, confirm the feasibility of a toxicity assessment system based on a flow-through mode of operation. The choice of device used for photodetection was also shown to be adequate from the point of view of the signal-to-noise ratio of the observed (steady) bioluminescence signal. The project is now ready to be launched into the next phase focusing on the characterization of inducible genetic constructs ,and on the development of a refined protocol for a toxicity sensing instrument based on the flow-through system. Specific directions for this future work will be discussed in the next chapter. 97 CHAPTER 6. CONCLUSION S The work accomplished during this project has brought many insights concerning the development of a bacteria-based toxicity biosensor using engineered bioluminescence in C.  crescentus. Our results helped to proceed one step closer to the development of a practical flow-through toxicity assessment system. These results can also be applied to the characterization and understanding of useful scientific tools based on similar genetically engineered in  vivo bioluminescence constructs. In this section, we will present a summary of the main conclusions coming from our work, and will discuss some of the future work to be addresses in subsequent phases. 6.1. Summar y o f main conclusions an d scientific contribution s The experimental work presented in this thesis was divided into three main parts: 1) the characterization of the luminescence signals obtained from the C.  crescentus transducer, 2) the acquisition of some preliminary toxicity data using the C.  crescentus transducer in a suspension culture, and 3) the design of a flow-through system prototype that will enable us to efficiently measure the bioluminescence of C. crescentus  immobilized in biofilms. Characterization o f the C. crescentus transducer . The results obtained in this phase of the project helped us to furtherer our understanding of the biochemical mechanisms of in  vivo  bioluminescence. The characterization studies involving In  Vivo Assays #1 and #2 showed that the engineered in  vivo  bioluminescence caused by the luxAB genes exhibits a complex two-phase temporal behaviour that can be adequately explained in terms of a coupled enzyme model involving six net biochemical reactions leading to the emission of light in C. crescentus. The intensity profile during the so-called first phase is directly due to the depletion of the luciferase-FMNHOOH complex. This complex is formed in the bacteria (provided FMNH2 is available) prior to its contact with exogenously added decanal, which is one of the principal substrates of the luciferase 98 reaction. Following the initial high-intensity peak, the decay seen throughout the remainder of the first phase is due to the effect of decanal inhibition on the uncomplexed luciferase enzyme. The inhibition slows the formation of new luciferase-FMNHOOH complex, resulting in lower light emission, even though the same amount of luciferase enzyme is present; the excess decanal thus acts like an "overdose" to the enzyme that is no longer resistant due to complexation with the other substrate FMN. The intensity profile of the second phase reveals the recovery of the luciferase from decanal inhibition with an accompanying increase in the intensity of the bioluminescence as the decanal concentration decreases. The maximum intensity of the second phase corresponds to the optimum concentration of decanal, after which the luminescence intensity decreases again due to decanal limitation as the decanal concentration continues to decrease; the opposite situation to the overdose effect that occurs at decanal concentration higher than the optimum. The eventual depletion of decanal turns off the bioluminescence reaction cycle and the light emission drops to very nearly zero. The relative positions of the maximum intensities of the two phases depends on the initial amount of decanal present and its rate of change. We established that out-gassing was the principal cause of the decanal concentration decrease in our case. The rate of out-gassing can be influenced by sparging and/or mixing the sample. Therefore, the overall shape of the profile can be modified by appropriate control of the environment or protocol involved. The understanding of the bioluminescence signal that was acquired in this project is crucial to the efficient development of biosensors, or gene expression reporters, based on the luxAB  system. It allowed us to understand which parameters were meaningful for a particular application, and how to control the experiments in order to achieve the most precise or repeatable results. For our specific application as a toxicity biosensor, the second phase can be used as it is linked to the "real-time" availability of the two substrates, FMN and 0-NADH, that reflect the viability of the cell, which is influenced by the effects of toxic compounds. The first phase indicates luciferase activity 99 and the intensity of this phase thus correspond to the amount of luciferase available in the sample. This is a good indicator for molecular biologists as it gives some indication of quality and/or efficiency of the construct and its expression as it reflects the total amount of luciferase that has been produced; it can also act as a reporter of heterologous gene expression {i.e. if the htxAB genes are co-transcribed with some other recombinant gene(s) of interest). It was also found that the maintenance of a constant level of decanal could lead to a constant level of bioluminescence intensity and that such a constant level was sustainable over a prolonged period of time. Generation o f toxicit y data . The development of a simple poisoning protocol was facilitated by the results and conclusions derived form the previous section. The toxicity of three different compounds was assayed and their toxic dose-response characteristics were obtained. The dose-response data were well-behaved and the typical sigmoidal characteristic provided evidence that the results were meaningful in toxicological terms. This further helped to validate our choice of secondary phase peak intensity as the appropriate measured parameter to relate to toxicity. The protocol used here lead to results comparable to established assays such as the Microtox® system. The relative toxicity of the three compounds tested were in agreement with published results and the range of sensitivity was comparable to some of the other systems. The amount of reagent required was much smaller than the conventional fish assay and comparable to the Microtox® assay. The time involved in producing the data was less than the conventional methods, although slightly longer than the Microtox® system. The quality of the results obtained in this section confirmed the feasibility of a toxicity sensor based on engineered bioluminescence in C. crescentvs. The design of a flow-through system was then seen as the next logical step. The conclusions from both previous sections were indicators of the feasibility of a toxicity monitoring instrument based on a flow-through system. It was established that such an instrument could be based upon the monitoring of a constant light output from a 100 biofilm. The biofilm could be submitted to controlled sample test conditions using FIA techniques and any detected changes in the luminescence would indicate a toxic effect. Design of a prototype flow-through system. We developed a first prototype of a flow-through system for use with biofilms. The flow-cell offered a large area available for the detectors in order to obtain the largest signal possible. Our evaluation of the bioluminescent power density emitted by the biofilm lead us to conclude that photodiodes operating in a photovoltaic mode could yield sufficient sensitivity for light measurement. The light detection system was supplemented by a custom made filter/amplifier instrument that significantly improved the signal-to-noise ratio. The flow control system is still at the manual stage, but after further characterization of the ensemble system, a high degree of automation should be possible and should yield an efficient and rapid toxicity assay with several advantages over the commercial Microtox® system. In summary, the system under development showed many promising qualities and our objective of designing a prototype flow-through system for subsequent work was fulfilled. Contribution o f thi s research . The bioluminescence mechanism is currently being applied toward the development of many kinds of novel scientific tools (such as gene expression reporters), and various biosensors (including sensors for toxicity, as well as specific toxic agents). The reasons are manifold, but one of the most important factors is the great sensitivity with which one can detect and measure the light produced. This sensitivity of detection allows a very direct and accurate way to monitor the mechanisms involved. However, the attempts so far directed toward the development of novel applications of the luxAB  system have yet to yield reproducible data. This lack of precision is undoubtedly a result of the lack of adequate characterization of the mechanisms underlying the bioluminescence, and hence of the signal itself. This thesis presents the only detailed characterization of a bioluminescent organism expressing the luxAB genes, and should therefore greatly help the development and/or improvement of many kinds of biosensors or other applications. In fact, the results presented already 101 permit a more accurate interpretation (see Chapter 3) of some of the current literature results concerning the analysis of the bacterial luciferase mechanism. In addition to this signal characterization, the dose-response characteristics presented here are the first known toxicity data obtained from a recombinant bacterial transducer (biosensor), thus demonstrating the feasibility of this approach. The demonstrated compatibility of the bacterial transducer with FIA technology makes possible many desirable features for the development of a toxicity assessment system. 6.2. Futur e wor k Considerable research and development remains to be done in order to realize an instrument for toxicity assessment that can potentially be commercialized. First, further optimization of the prototype flow-through system should be done in order to improve the efficiency of toxicity data acquisition. This should be followed by the complete characterization of the system in terms of the accuracy and reproducibility of the generated data. This will be performed by generating a toxicity data-base and comparing it to the other toxicity testing methods. In the following, we will briefly describe some of the specific tasks planned for the next phase of this work. The capacity to maintain a steady decanal concentration, and to detect the light output of practical biofilms, will enable us to characterize the behaviour of the biofilm under tightly controlled conditions. This will lead to a better understanding of the (immobilized) bacterial transducer and will facilitate the further optimization of the instrumentation system. The actual genetic construct used as a transducer is functional but has not been perfected. More molecular (genetic) work in this field could lead to higher performance constructs. Improvements in terms of signal-to-noise ratio, sensitivity of detection, and specificity for particular chemicals are some examples of what should be possible. A set of bacterial transducers having different attributes could be developed specifically for use in different applications, such as general toxicity sensing or specific 102 compound sensing. In parallel with the development of these different constructs, the bioluminescence mechanisms which could vary from one construct to another, have to be re-characterized to ensure the development of optimal protocols for each case. This re-characterization (of new constructs) should be much more efficiently accomplished, based on our present knowledge, and using the flow-through system for experiments. In addition to the development of better constructs, the development of protocols for producing the biofilms has to be studied in order to ensure optimum biofilm density, reproducibility of the results, and maximum signal output. The results obtained from the re-characterization of different constructs and their biofilms may bring about the design of a better flow-cell. Another factor influencing the flow-cell design is the efficiency of coupling the light output to the photodetection system. Total volume, flow capacity, and geometry are some examples of other flow-cell features that may have to be optimized. The use of a flow-cell is made further attractive by advances in the CCD imaging systems capable of high sensitivity and resolution. The combined use of better flow-cells and a high-sensitivity imaging system could make the study of single cells possible. In our case, the use of a high-sensitivity CCD camera, capable of low-light-level detection, and a "laser tweezer" instrument could be of great benefit with regard to biofilm characterization and optimization; this could help to elucidate some of the physical properties of the biofilm and the two-dimensional spatial variations in the bioluminescence emitted by the biofilm. For example, some of the fluid dynamic properties of the flow-cell (in terms of mixing, availability of nutrients, edge effects, etc.)  could be visualized with a CCD camera and would aid in the design of a better flow-cell. Strength of adhesion of the bacteria to the glass substrate could be directly measured with laser tweezers and could also indicate some appropriate design modifications for the flow-cell in terms of flow rate limit and turbulent stress limits. All these factors will have a direct effect on the development of an efficient, cost-effective and highly repeatable FIA toxicity instrument. 103 7. Reference s 1. Abu-Soud , EL, L.S. Mullins, T.O. Baldwin, and F.M. Raushel , 1992 Stopped-flow kinetic analysis of the bacterial luciferase reaction. Biochemistry 31:3807-3813. 2. Athwal , G.K. , E.A . El-Kohly , an d A.E.G . Cass , 1990 Amplified bioluminescence assays, p. 67-70. In  Bioluminescence and Chemiluminesence: Current Status, P.E. Stanley and L.J. Kricka (ed.), John Wiley & Sons, N.Y. 3. Bailey , J.E., and D.F. Ollis. 1986. Biochemical engineering fundamentals, second edition, McGraw-Hill Book Company, N.Y. 4 Baldwin , T.O. , T . Berends , T.A . Bunch , T.F . Holzman , S. K Rausch , L Shamansky, M.L . Treat , an d M . M . Ziegler . 1984. Cloning of the luciferase structural genes from Vibrio  harveyi  and expression of bioluminescence in Escherichia coli.  Biochemistry 23:3663-3667. 5. Baldwin , T.O. , L.H . Chen , L.H . Chlumsky , J.H . Devine , an d M.M . Ziegler . 1989. Site-directed mutagenesis of bacterial luciferase: Analysis of the essential thiol. J. Biolum. Chemilum. 4:40-48. 6. Barret , J.T. , C.S . Rhodes , D.M . Ferber , B . Jenkins , S.A . Kuhl , an d B . Ely . 1982. Construction of a genetic map for Cavlobacter  crescentvs.  J.  Bacteriol. 149:889-896. 7. Bass , R., P. Gunzel, D. Henschler, et. al. 1982. LD50 versus acute toxicity. Arch. Toxicol. 51:183-186. 8. Betterridge , D. 1978. Anal. Chem. 50:832A-846A 9. Blum , L.J. , S.M . Gautier , an d P.R . Coulet . 1989. Design of luminescence photobiosensor. J. Biolumin. Chemilumin. 4:543-550. 10. Brock , T.D. , an d M.T . Madigan . 1991. Biology of microorganisms, Sixth edition. Prentice Hall Inc., N.J. 104 11. Bulich , A.A. , K.-K . Tung , an d G . Scheibner . 1990. The luminescent bacteria toxixity test: its potential as an in  vitro  alternative, J. Biolumin. Chemilum. 5:71-77. 12. Bulich , A.A., an d D.L . Isenberg . 1981. Use of the luminescent bacterial system for the rapid assessment of aquatic toxicity. ISA Trans. 20(l):29-33. 13. Bulich , A.A. , M.W . Green , an d D.L . Isenberg . 1981. Reliability of bacterial luminescence assay for determination of the toxicity of pure compounds and complex effluents, p. 338-347. In  Aquatic Toxicology and Hazard Assessment: Fourth Conf, ASTM STP 737, D.R. Branson and K.L. Dickson (ed.), American Society for Testing and Materials. 14. Burlage , R.S. , G. S Sayler , an d F  Larimer . 1990. Monitoring of naphthalene catabolism by bioluminescence with nah-lux  transcriptional fusion. J. Bacterid. 172:4749-4757. 15. Carmi , O.A., G.S.A.B. Stewart , S. Ulitzur, and J. Kuhn. 1987. Use of bacterial luciferase to establish a promoter probe vehicle capable of nondestructive real-time analysis of gene expression in Bacillus spp. J. Bacteriol. 169:2165-2170. 16. Chan , P.K.P. , an d A.W Hayes . 1989. Principles and methods for acute toxicity and eye irritancy, p. 169-220. In  Principles and methods of toxicology, Second edition, A.W. Hayes (ed.), Raven Press, Ltd., N.Y. 17. Curtis , C , A . Lima , S.J . Lozano , an d G.D . Veith . 1982. Evaluation of a bacterial bioluminescence bioassay as a method for predicting acute toxicity of organic chemical to fish, p. 170-178. In  Aquatic Toxicology and Hazard Assessment: Fifth Conference, ASTM STP 766, J.G. Pearson, R.B. Foster, and W.E. Bishop (ed.), American Society for Testing and Materials. 18. Dooley , C.A., and P. Kenis. 1987. Response of bioluminescent bacteria to alkytin compuonds. Oceans 87-Proc: The ocean, an internationnal workplace, pp. 1517-1524. 105 19. Dower , W.J. , J.F . Miller , an d C.W . Ragsdal . 1988. High efficiency transformation of Escherichia coli by high voltage electroporation. Nucleic Acids Res. 16:6127-6145. 20. Ely , B. , an d R.H . Croft . 1982. Transposon mutagenesis in Caulobacter crescentus. J. Bacteriol. 149:620-625. 21. Ely , B. , an d R.C . Johnson . 1977. Generalize transduction in Caulobacter crescentus. Genetics 87:391-399. 22. Ely , B. 1985. Vectors for transposon mutagenesis in non-enteric bacteria. Mol. Gen. Genet. 200:302-304. 23. Ely , B. , an d C.J . Gerardot . 1988. Use of pulsed-field-gradient gel electrophoresis to construct a physical map of the Caulobacter crescentus genome. Gene. 68:323:333. 24. Engebrecht , J. , K . Nealson , an d M . Silverman . 1983. Bacterial bioluminescence: isolation and genetic analysis of function from Vibrio  fisher i. Cell 32:773-781. 25. Engebrecht , J., M. Simon, and M. Silverman. 1985. Measuring gene expression with light. Science 227:1345-1347. 26. Farinha , M.A. , an d A.M . Kropinski . 1990. Construction of broad-host-range plasmid vectors for easy visible selection and analysis of promoters. J. Bacteriol. 172, No. 6:3496-3499. 27 Frazier , J.M. , S.C . Gad , A.M . Goldberg , an d J.P . McCulley 1987. A critical evaluation of alternatives to acute ocular irritation testing. In  Alternative Methods in Toxicology, Vol. 4, A.M Goldberg (ed.), Mary Ann Liebert Inc., N.Y. 28. Gad , S.C , an d C.S . Weil . 1989. Statistics for toxicologists, p. 435-483. In Principles and methods of toxicology, Second edition, A.W. Hayes (ed.), Raven Press, Ltd., N.Y. 106 29. Gilchrist , A. , an d J . Smit . 1991. Transformation of freshwater and marine Caulobacters by electroporation. J. Bacteriol. 173:921-925. 30. Guzzo , A., and M.S. Dubow. 1991. Construction of stable, single-copy luciferase gene fusions in Escherichia coli. Arch. Microbiol. 156:444-448. 31. Harvey , E.N. 1952. Bioluminescence. Academic Press Inc., N.Y. 32. Hastings , J.W. , an d C . Balny . 1975. The oxygenated bacterial luciferase-flavin intermediate. J. Biol. Chem. 250, No. 18:7288-7293. 33. Hastings , J.W. , A . Eberhard , T.O . Baldwin , M.Z . Nicoli , T.W . Cline , an d K.H. Nealson . 1978. Bacterial bioluminescence. Mechanistic implications of active center chemistry of luciferase. p.369-380. In  Chemiluminescence and bioluminescence, M.J. Cormier, D.M. hercules, and J. Lee (ed.), Plenum Press, N.Y. 34. Hastings , J.W., an d H.Q. Gibson . 1963. Intermediate in the bioluminescence oxidation of reduced flavin mononucleotide. J. Biol. Chem. 238:2537-2544. 35. Hastings , J.W., C.J. Potrikus , S.C. Gupta , M. Kurfurst, an d J.C. Makemson. 1985. Biochemistry and physiology of bioluminescent bacteria. Adv. Microbiol. Physiol. 26:235-291. 36 Hastings , J.W. , S.C . Tu , J.E . Becvar , an d R.P . Presswood . 1979 Bioluminescence from the reaction of FMN, H202 and long chain aldehyde with bacterial luciferase. Photochem. Photobiol. 29:383-387. 37. Hogan , M.D. , an d D.G . Hoel . 1989. Extrapolation to man, p. 879-891. In Principles and methods of toxicology, Second edition, A.W. Hayes (ed.), Raven Press, Ltd., N.Y. 38. Holzman , T.F. , an d T.O . Baldwin . 1983. Reversible inhibition of the bacterial luciferase catalyzed bioluminescence reaction by aldehyde substrate: Kinetic mechanism and ligand effects. Biochem. 22:2838-2846. 107 39. Ismailov , A.D. , A.Y. Sobolev , an d V.S . Danilov . 1990. Bioluminescence decay kinetics in the reaction of bacterial luciferase with different aldehydes. J. Biolum. Chemilum. 5:213-217. 40. Jablonski , E. , and M . DeLuca . 1978. Studies of the control of luminescence in Beneckea han>eyi:  properties of the NADH and NADH:FMN oxidoreductases. Biochemistry 17, No. 1:672-678. 41. Jassim , S.A.A. , A . Ellison , S.P . Denyer , an d G.S.A.B . Stewart . 1990. In vivo bioluminescence: a cellular reporter for research and industry. J. Biolum. Chemilum. 5:115-122. 42. Jubenvill , S. , an d D . Robinson . 1992. Instrumentation amplifier and filter, Unpublished technical report submitted to R.F.B. Turner. 43. Kaaret , T.W. , an d T.C . Bruice . 1990. Electrochemical luminescence with n(5)-ethyl-4a-hydroxy-3-methyl-4a, 5-dihydrolumiflavin. The mechanism of bacterial luciferase. Photochem. and Photobio. 51:629-633. 44. Kaiser , K.L.E. , an d V.S . Palabrica . 1991. Photobacterwm bacterium  toxicity data index. Water Poll. Res. J. Canada 26, No 3: :361-431. 45. Karube , I. 1987. Micro-organism based sensors, in  Biosensors-Fundamentals and applications. Turner, Karube and Wilson, ed., Oxford University Press. 46. Karube , I., et al 1988. Anal. Chim. Acta. 213:69-77. 47. Karp , M . 1989. Expression of bacterial luciferase genes from Vibrio  harveyi in Bacillus subtilis  and Escherichia coli.  Biochim. Biophys. Acta. 1007:84-90. 48 Keen , N.T. , S . Tamaki , D . Kobayashi , an d D . Trollinger . 1988 Improved broad-host range plasmids for DNA cloning in gram-negative bacteria. Gene, p. 191-197 49. Kennedy , G.L. , Jr. , R . Ferenz , an d B.A . Burgess . 1986. Estimation of oral toxicity in rats by determination of the approximate lethal dose rather than the LD50. J. Appl. Toxicol. 6:145-148. 108 50. Korpela , M. , an d M . Karp . 1988. Stable-light producing Escherichia coli. Biotech, lett. 10, No. 6:383-388. 51. Korpela , M. , P . Mantsala , E.M . Lilius , an d M . Karp . 1989. Stable - light -emitting Escherichia coli as a biosensor. J. Biolum. Chemilum. 4:551-554. 52. Kosower , E.M . 1980. A proposed mechanism for light emission by bacterial luciferase involving dissociative electron transfer. Biochemical and biophysical research communications 29, No. 2:356-364. 53. Kricka , L.J . 1988. Clinical and biochemical applications of luciferases and luciferins. Anal. Biochem. 175:14-21. 54. Lavi , J. , R. Raunio , Y. Malkov, and T . Lovgren. 1983 The effect of luciferase and NADH:FMN oxidoreductase concentrations on the light kinetics of bacterial bioluminescence. Biochemical and Biophysical Res. Comm. 111:266-273. 55. Lebsack , M.E. , A.D . Anderson , G.M . DeGraeve , an d H.L . Bergman . 1981. Comparison of bacterial luminescence and fish bioassay results for fossil-fuel process waters and phenolic constituents, p. 348-356. In  Aquatic Toxicology and Hazard Assessment: Fourth Conference, ASTM STP 737, D.R. Branson and K.L. Dickson (ed.), American Society for Testing Material. 56. Lee , J. 1985. The mechanism of bacterial luminescence, p. 401-437. In  J. Burr (ed.), Chemi- and Bioluminescence, Dekker, New York. 57. Lilius , E.M., M . Korpela , an d M . Karp . 1986. Luminescent Escherichia coli as an immunosensor. Biolum. and Chemilum.: New Perspectives, p. 431-434. 58. Lork e D . 1983 A new approach to practical toxicity testing. Arch. Toxicol. 54:275-287. 59. Macheroux , P. , J . Eckstein , an d S . Ghisl . 1987. Studies on the mechanism of bacterial bioluminescence. Evidence compatible with one electron transfer process and a CIEEL mechanism in the luciferase reaction, p.613-619. In  D. Edmonson, 109 and D. McMormick (ed.), Flavins and Flavoproteins, Walter de Gruyter & CO., New York. 60. MacRae , J.D., and J. Smit. 1991. Characterization of Caulobacters isolated from wastewater treatment systems. Appl. Environ. Microbiol. 57, vol. 3:751-758. 61. Mager , H.I.X., an d R. Addink. 1984. On the role of some flavin adducts as one-electron donors, p.37-40. In D. Edmonson, and D. McMormick (ed.), Flavins and flavoproteins, Walter de Gruyter & CO., New York. 62. Makemson , J.C., and J.W. Hastings. 1991. Bovine serum albumin interacts with bacterial luciferase. J. Biolum. Chemilum. 6:131-136. 63. Mathews , C.K. , an d K.E . va n Holde . 1990. Biochemistry. The Benjamin/Cummings Publishing Company, Inc. CA. 64. Meighen , E.A. 1988. Enzymes and genes from the lux  operons of bioluminescent bacteria. Ann. Rev. Microbiol. 42:151-176. 65. Meighen , E.A . 1990. Molecular biology of bioluminescence, p. 3-10. In  P.E. Stanley and L.J. Kricka (ed), Bioluminescence and Chemiluminesence: Current Status, John Wiley & Sons, N.Y. 66. Meighen , E.A. , an d LBartlet . Complementation of subunits from different bacterial luciferases. Evidence for the role of the B  subunit in the bioluminescent mechanism. J. Biol. Chem. 255:11181-11187. 67. Meighen , E.A. , an d R.E . Mackenzie . 1973. Flavine specificity of enzyme-substrate intermediates in the bacterial bioluminescence reaction. Biochemistry 12:1482-1491. 68. Miller , J.H . 1972. Experiments in molecular genetics. Cold Spring Harbour Laboratory, Cold Spring Harbour, N.Y. 69. Muller , H. , an d H.P . Kley . 1982. Retrospective study on the reliability of an "approximate LD50" determined with a small number of animal. Arch. Toxicol. 51:189-196. 110 70. Pointdexter , J.S. 1964. Biological properties and classification of the Caulobacter group. Bacteriol. Rev. 28:231-295. 71. Poindexter , J.S. 1981. The Caulobacters: Ubiquitous unusual bacteria. Microbiol. Rev. 28:231-295. 72. Poindexter , J.S . 1981. Oligotrophy. Feast and Famine exixtence, p.63-83. In Advances in Microbial Ecology, Vol. 5, Chapter 2, M. Alexander , ed., Plenum Publishing Corp. 73. Poole , A. , an d G.B . Leslie . 1989. A practical approach to toxicological investigations. Cambridge University Press, N.Y. 74. Qureshi , A.A , K.W . Flood , S.R . Thompson . S.M . Janhurst , C.S . Inniss , an d D.A. Rokosh . 1982. Comparison of a luminescent bacterial test with other bioassays for determining toxicity of pure compounds and complex effluents, p. 179-195. In  Aquatic Toxicology and Hazard Assessment: Fifth Conference, ASTM STP 766, J.G. Pearson, R.B. Foster, and W.E. Bishop (ed.), American Society for Testing and Materials. 75 Rattray , E.A.S , J.A . Prosser , K . Killham , an d L.A . Glove r 1990 Luminescence-based nonextractive technique for in  situ  detection of  Escherichia coli in soil. Appl. Eviron. Microbiol. 56:3368-3374. 76. Roberts , M.H . Jr . 1980. Flow-through toxicity testing system for molluscan larvae as applied to halogen toxicity in estuarine water, p. 131-139. In  Aquatic Invertebrate Bioassays, ASTM STP 715, A.L. Buikema, Jr., and J. Cairns, Jr. (ed.), American Society for Testing Materials. 77. Rogowski , P.M. et  al.  1987. Regulation of the vir genes of Agrobacterium tumefaciens plasmid pTiC58. J. Bacteriol. 169:5101-5112. 78. Ruzicka , J., and E.H Hansen. 1988. Flow injection analysis, second edition. John Wiley and Sons, New York. 79. Ruzicka,  J., and E.H. Hansen. 1975. Anal. Chim. Acta. 78:147 111 80. Sambrook , J. , E.F . Fritsch , an d T  Maniatis . 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 81. Schauer , A.T . 1988. Visualizing gene expression with luciferase fusions. Trends in Biotechnology. 6:23-27. 82. Schmetterer , G. , C.P . Wolk , an d J . Elhai . 1986. Expression of luciferase from Vibrio han>eyi  and Vibrio  fisheri  in filamentous cyanobacteria. J. Bacteriol. 167:411-414. 83. Scholz , P. , V . Haring , B . Wittmann-Liebold , K . Ashman , M . Bagdasaria n and E . Scherzinger . 1989. Complete nucleotide sequence and gene organization of the broad-host-range RSF1010. Gene 75:271-288. 84. Schutz , E. , and H . Fuchs . 1982 . A new approach to minimizing the number of animals used in acute toxicity testing and optimizing the information of tests results. Arch. Toxicol. 51:197-220. 85. Sevign y P., an d F. Gossard . 1990. Construction of cloning vectors using the Vibrio harveyi luminescence genes hixA and laxB as markers. Gene 93:143-146. 86. Shaw , J.J. , an d C.I . Kado . 1986. Development of a Vibrio  bioluminescence gene-set to monitor phytopathogenic bacteria during the ongoing disease process in non-disruptive manner. Bio/Technology. 4:560-564. 87. Sigma , technical note on the bacterial luciferase. 88. Skoog , D.A., an d J.J. Leary . 1992. Principles of instrumental analysis, Fourth edition. Saunders College Publishing, Harcourt Brace Janovich College Publisher, N.Y. 89. Smit , J., and N. Agabian. 1984. Cloning of the major protein of the Caulobacter crescentus periodic sruface layer: Detection and characterization of the cloned peptide by protein expression assays. L. Bacteriol. 160:1137-1145 112 90. Stewart , G.S.A.B . 1989. The detection of microorganisms using in-vivo bioluminescence. p. 175-182. In  Biotech '89. Blenheim Online Publications, London, UK. 91. Stewart , G.S.A.B . 1990. A review, hi  vivo  bioluminescence: new potentials for microbiology. Letters Appl. Microbiol. 10:1-8 . 92. Symons , B.D. , an d R.C . Sims . 1988. Assessing detoxification of a complex hazardous waste, using the Microtox assay. Arch. Environ. Contam. Toxico. 17:497-505. 93. Szittner , R. , an d E. Meighen . 1990. Nucleotide sequence, expression, and properties of luciferase coded by lux  genes from a terrestrial bacterium. J. Biol. Chem. 265:16581-16587. 94. Ulitzur , S. , an d J . Kuhn . 1986. Introduction of lux  genes into bacteria, a new approach for specific determination of bacteria and their antibiotic susceptibility, p. 463-472. In  Bioluminescence and Chemiluminescence: New Perspectives, J. Scholmerick et al. (ed.), John Wiley ans Sons, N. Y. 95 Watson , J.D. , M . Gilman , J . Witkowski , an d M . Zolle r 1992 Recombinant DNA, second edition, Scientific American Books. W.H. Freeman and Company. N.Y. 96. Wolk , C.P. , Yupin g C , an d Panof f J.-M . 1991. Use of a transposing with luciferase as a reporter to identify environmentally responsive genes in cyanobacterium. Proc. Natl. Acad. Sci. USA 88:5355-5359. 97. Hong , X. , A . Dingwall , an d L . Shapiro . 1989. Negative transcriptional regulation in caulobacter flagellar hierarchiy. Proc. Natl. Acad. Sci. USA 86:6656-6660. 98. Yanisch-Perron , C , J . Vieira , an d J . Messing . 1985. Improved M13 phage cloning vectors and host strains: nucleotide sequences of the M13mpl8 and pUC19 vectors. Genes. 33:103-119. 113 Zbinden, G. , an d M . Flury-roversi . 1981. Significance of the LD50-Test for toxicological evaluation of chemical substances. Arch. Toxicol. 47:77-99. Ziegler, M.M. , an d T.O . Baldwin . 1981. Biochemistry of bacterial bioluminescence. Curr. Top. Bioenerg. 12:65-113. 114 8. Appendi x A A flow-cell prototype based on silicon micromachining technology was designed and fabricated. The initial motivation for doing this work was to investigate the feasibility of making flow-cells using a technology that could allow easy and inexpensive large-scale production. The precision offered by this technology for small-dimension machining is also attractive since it could allow miniaturization of some FIA components and could thereby lead to a smaller and more easily portable overall flow-through system. The flow-cell described here is a first generation prototype, and is therefore very simple. The design is configured to support the growth of biofilm only on its interior walls. Therefore, it can not be opened as the one used in Chapter 5, where the biofilm could be removed from the cell for further study (e.g.  microscopy). Figure 8.1. illustrates the photolithographic mask layout for the etching of a 3-inch silicon wafer, and Figure 8.2. is a scale drawing of the assembled flow-cell. The glass windows shown in Figure 8.1 and 8.2 were made of a glass microscope slide and a glass cover slip fixed to the wafer using a parafilm membrane that was heated to ensure adhesion and a good seal. The windows have the double function of allowing photodetection of the luminescence that would be emitted from the inside, and ensure the closure of the open silicon structure. The tubes used for flow outlet and inlet were also made of glass and were attached with silicone rubber adhesive (Silastic®). The flow-cell was successfully tested for leaks and pressure resistance. This type of design would also allow inspection of the biofilm with a microscope (possibly fitted with a CCD camera for light detection). 115 £ 1 x 2= 3. T£ . liiii J i i / / / / / / / 1 iiii 12 l_l i^ Figure 8.1. Schematic of the photolithographic layout for the flow-cell. The upper part represents a view from the top, and the lower part represents a side view of the cut chamber. The filled zones illustrate the glass added to form windows for photodetection and to close the cell. 116 A E o I Microscope slid e Fl ow c h a m b e r - ^ 22.440m m ^ Si wafe r inner tub e diamete r 2 . mm 8.100mm .638mm 510mm mm Figure 8.2. Scale drawing (side view) of the flow-cell. The filled zones are the glass windows used to seal the chamber. The scaling is different for the x-axis than for the y-axis. 117 Even though miniaturization is an important factor, it is not the only factor influencing our decision to investigate the potential of flow-cell fabrication using micromachining technology. This technology offers many options in terms of tools and devices that can be implemented on a silicon wafer. Flow control devices (such as valves and injection ports), temperature control devices (such as heaters and thermocouples) could be built-in, along with a network of micro-channels that could be configured to form a miniature flow-injection analysis system. Therefore, most of the components of a future generation of the flow-through system could be integrated into a single micromachined device. 118 9. Appendi x B 1. INTRODUCTION Following is a brief description of the design and fabrication of a filter/amplifier instrument that was used to improve the signal-to-noise ratio at the output of the photodetection system used in the flow-through system described in Chapter 5 of this thesis. The design parameters and specifications were given to S. Jubenvill and D. Robinson, who proceeded to build the filter/amplifier instrument under the supervision of the author of the thesis. The instrument was required to be dual-channel, capable of handling input signals ranging between 0 and 10 Volts, and ensure a limited output signal ranging between 0 and 5 Volts (for compatibility with a commercial data acquisition system used in the laboratory). The filtering capabilities of the instrument were required to feature a low-pass filter with four selectable cut-off frequencies (0.5, 1.0, 10, 100 Hz). The amplification capabilities of the instrument were required to offer an infinitely adjustable gain ranging between 0 and 100 V/V. In addition, a summation option was required in order to add together the output signals from the two channels of the instrument. The following is excerpted from the technical report submitted by S. Jubenvill and D. Robinson detailing the design and testing of the built instrument. 119 2. SYSTEM OVERVIEW A block diagram of the custom amplifier/filter system is shown in Figure 9.1. The system is a two input system, with optional summation producing a third output. The instrumentation amplifier provides high common-mode rejection ratio, and high input impedance. The instrumentation amplifier's gain is unity. The second amplification stage, labeled "Gain Amp", provides an infinitely variable gain of 0 to 100 V/V. A clipping detector then looks at the output of the gain amplifier, and activates an LED if the voltage is in excess of 5.0 V. A two-pole active low-pass filter stage is capable of filtering 0.5, 1.0, 10, or 100 Hz noise. A limiter then ensures output voltage is limited to 5.0 V and provides a low system output impedance. The summer is switch activated, and is provided with clipping detection and a limiter. INSTRUMEN-j TATION AMP. GAIN AMP. FILTER LIMITER CLIPPING DETECTOR SUMMER o— INSTRUMEN-TATION AMP GAIN AMP CLIPPING DETECTOR FILTER LIMITER LIMITER CLIPPING DETECTOR Figure 9.1 Block diagram of amplifier/filter. 120 3. DESCRIPTION OF FUNCTIONAL BLOCKS In this section, each of the blocks in the block diagram of Figure 9.1 are broken down and discussed. Instrumentation Amplifier The instrumentation amplifier chip is shown symbolically in Figure 9.2, This amplifier is a commercial i.e., and was chosen because it provides high input resistance and high common mode rejection for the input signal. The data sheet for this i.e. is in Appendix C. Figure 9.2 Instrumentation amplifier. Gain Amplifier Figure 9.3 depicts the gain amplifier. A precision 100k ohm potentiometer with rotary counting dials is responsible for gain adjustment. The linear accuracy of the 100k ohm pot is 0.25 %.  A 100 ohm pot on the input terminal of the amp in series with the 953 ohm resistance makes it possible to calibrate the amplifier to compensate for component tolerances and input offset voltage. The resistor from the "+" terminal of the op amp to ground reduces the effects of input bias currents. iooa 953n AA/V 100K.TI Figure 9.3 Gain amplifier. 121 Clipping Detector In Figure 9.4 the clipping detector is shown. A less expensive, quad op amp i.e. is employed. The 10k ohm in series with the 20k ohm pot enables the voltage at the input of the "-" op amp terminal to be exactly 5.0 V by voltage division. When the input to the clipping detector is less than 5.0 V, the LED has no voltage across it, and thus does not illuminate. If the input swings slightly greater than 5.0 V, the LED is forward biased and lights up, letting the user know the signal is greater than 5.0 volts. +15 V Figure 9.4 Clipping detector. Filter Figure 9.5 depicts the filter. The filter is an active two-pole low pass Butterworth filter. This filter's gain in the passband is unity. A pole order of two was deemed to be sufficient, and helped keep the cost and complexity from becoming unnecessarily high. 122 2M21 SI Figure 9.5 Filter Limiter The limiter circuit is shown in Figure 9.6, The limiter circuit is provided with a 500 ohm input potentiometer in series with the nominal 9760 ohm input resistance so the gain of the limiter circuit can be adjusted to unity, compensating for the 10k ohm resistance tolerance. As in the case of the gain amp, the resistance in the "+" terminal of the op amp is there to reduce the effects of input bias currents. It can be determined that its value should be equal to the parallel equivalent of the feedback resistance and that of the input resistance. The diode in the feedback path is a 5.1 V zener diode. If it was desirable that the 0 to 5 V limitation should be expanded to, for example, 0 to 5.6 V, the clipping detector circuitry could be adjusted by adjusting the 20k ohm, and the 5.1 V zener in the limiter circuit could be replaced with a 5.6 V zener 500n 9K76tt -yxfv-—Wv Figure 9.6 Limiter. 123 Summer The summer is shown in Figure 9.7. This summer is the standard "weighted summer", with both inputs weighted equally in this case. A single on/off switch activates the summation. ]OKSl -AAA/ IOKTI AAAr J< 10KU Figure 9.7 Summer. Power Supply Figure 9.8 shows the power supply circuit. The power supply was designed for simplicity and low cost. The input is fused to protect against excessive currents which might be caused by accidental shorting of the power supply. The outputs are +/- 15 volts, regulated by two regulator i.c.'s. The common is connected directly to earth ground. o-1/2A 128:28 VAC 1A & 1 7815 220quF -I 2200/yF 7915 -1 }fjF -o+]5V -X V? oGND -0-15V Figure 9.8 Power supply. 124 4. FABRICATIO N DETAIL S The system is contained in an aluminum box. The chassis is connected to earth ground. Holes were punched, drilled, and filed for component mounting. Channel inputs and outputs are of the banana plug variety. LED's are mounted with grommets. The power supply and main circuitry are on two different boards with a removable cable taking power from the power supply to the circuit board. Both boards are mounted to the chassis with nylon bolts. Gain adjustment is facilitated with precision lockable dials, and self-canceling selection switches are provided for filter selection. All user interface components are clearly labeled on the box. The front view, top view, and rear view of the box are shown in the figures 9.9,9.10and9.11below. Output A Ou-fyJ-  A-hB  Output  B O O  O  O  O  O Clipping Clipping  Clipping HIW Sefecf A Filkr 6ekc/- B I o O I I o ft+B W"*1 4. O Gain  B 0*/Off Figure 9.9 Top view of box. 125 i^pul A o o X^ipu-f B o o Q-\5 Figure 9.10 Front view of box. Fuse Figure 9.11 Rear view of box. 126 5. SYSTEM PERFORMANCE The amplifier/filter system was tested in many stages. The functional blocks which did not require any specialty components that were on order were tested on a breadboard. This consisted of testing of the limiter, summer, and the filter. Noise immunity was looked at as well as functionality, and in each case the design performed well. In the construction stage, each functional block was tested as it was built. The instrumentation amps and the gain amps perform noise-free and meet specifications. The gain amps were calibrated. The clipping detectors were seen to be accurate, and were calibrated to activate at 5.0 V (easily adjusted with 20k ohm pots). The filters were tested at each break frequency. InFigure 9.12below, the test results for a signal of input frequency just over 2 Hz are shown. This frequency was the lowest frequency attainable by the function generator available. Figure 9.12 shows the attenuation achieved with each filter selection. 1 2.Q0V f— O.OOs lOO?/ fi RU N 4—O.OOs l O O g / f l RU N i -1 -1• i • j -1 -1-1• i •I -1-V-1 . i - j -1 -1 - 1 . i - j - 1 . i . i . /-V|a-pCn=9.313 V \4 Figure 9.12 Filter test results (cont. on next page). 127 10 Hi 1 2.QOV O.OOs IOCS/ f l RUN, f l RUN vp-pcn=i.50o v 1 2.0QV a5frz fl RUN Vp-pCn=562.5mV Tests were performed on the limiter circuitry, and the results are shown in Figure 9.13. The top trace is the limiter transfer characteristic, and the bottom shows a sample input and output of the limiter. Note: the noise present in the 128 limiter output was eliminated in the construction stage by tightening the component layout. From the tests, the limiter is shown to perform very well. 1 3,2.oo v 2  "3,2.oo v iftwT'iv. a. «..'-r RUN •I -I- H I- |- •!• -I- -I- j- I- •!• I- •]• I- -H -|. .j. .|. + .|.,j. .,. .,. |. .|. .|. .j. .|. .y. .|. .j. ,H .,. .,. .j. .,. .,. .,. .j. .,. .,. .,. . . .,. .,. .,. .j. ., 1 ii.oo v 2 l.oov {-Q.QOs 2.00g / fZ RU N Figure 9.13 Limiter test results. The overall system was tested, and was found to perform as required. 129 6. CONCLUSIONS The implementation of the proposed design was successful, at a total cost of $ 292.54, itemized in Appendix A. The implemented analog network satisfies the functionality requirements for the project. The amplifier/filter system built provides infinitely adjustable gain with a range of 0 to 100 V/V, optional filters with selectable break frequencies of 0.5, 1.0, 10, and 100 Hz, 5.0 V voltage limiters, clipping detection circuitry for overvoltage indication, and is powered by a built in ± 15 V power supply that runs off the standard 120 VAC, 60 Hz line. 130 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items