PHAGE-BASED MAGNETOELASTIC SENSOR FOR THE DETECTION OF SALMONELLA TYPHIMURIUM Except where reference is made to the work of others, the work described in this dissertation is my own or was done in collaboration with my advisory committee. This dissertation does not include proprietary, restricted or classified information. ______________________________ Ramji S. Lakshmanan Certificate of Approval: _________________________ _________________________ Zhong-Yang Cheng Bryan A. Chin, Chair Associate Professor Professor Materials Engineering Materials Engineering _________________________ _________________________ Valery A. Petrenko Dong-Joo Kim Professor Associate Professor Department of Pathobiology Materials Engineering _________________________ George T. Flowers Dean Graduate School PHAGE-BASED MAGNETOELASTIC SENSOR FOR THE DETECTION OF SALMONELLA TYPHIMURIUM Ramji S. Lakshmanan A Dissertation submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy Auburn University, Alabama December 19, 2008 PHAGE-BASED MAGNETOELASTIC SENSOR FOR THE DETECTION OF SALMONELLA TYPHIMURIUM Ramji S. Lakshmanan Permission is granted to Auburn University to make copies of this dissertation at its discretion, upon the request of individuals or institutions and at their expense. The author reserves all publication rights. _________________________ Signature of Author _________________________ Date of Graduation iii iv VITA Ramji S Lakshmanan, son of Anandhi Sitaraman and G. Sitaraman, was born on March 8 th , 1979 in Margao, Goa, India. He graduated from St Johns Senior Secondary School in March 1996. He joined Indian Institute of Technology- Bombay, Mumbai, India in August 1997 and graduated with a Bachelor of Technology in Metallurgical Engineering and Materials Science in May 2002 and a Master of Techonology in Ceramics and Composites in May 2002. He entered doctoral program in Materials Engineering at Auburn University in August 2002. He married Subhadravalli V. Chitti, daughter of Dr. Chitti Bhavanarayana Murthy and Chitti Rohini Annapurna, in May 2006. v DISSERTATION ABSTRACT PHAGE-BASED MAGNETOELASTIC SENSOR FOR THE DETECTION OF SALMONELLA TYPHIMURIUM Ramji S. Lakshmanan Doctor of Philosophy, December 19, 2008 (M. Tech, Indian Institute of Technology, Bombay, 2002) (B. Tech, Indian Institute of Technology, Bombay, 2002) 171 Typed Pages Directed by Bryan A. Chin In recent years, food-borne illness have garnered the attention of mainstream America with calls now coming from the media for more inspections to ensure the safety of our food supply. Food borne illness from the ingestion of S. typhimurium has been of great concern due to its common occurrence in food products of daily consumption. Annually approximately 80 million cases of food poisoning are reported in the United States alone. The ever growing need for rapid detection of pathogenic microorganisms present in food, environmental and clinical samples has invoked an increased interest in research efforts towards the development of novel diagnostic methodologies. Currently, the detection of bacteria in contaminated food relies on conventional microbiological methods that are time consuming and manpower intensive. This study presents the results of the characterization of a phage-based magnetoelastic biosensor for the detection of Salmonella typhimurium. This affinity- vi based biosensensor is comprised of a magnetoelastic material as the transducer and filamentous phage as the bio-recognition element. Magnetoelastic materials are ferromagnetic amorphous alloys that change dimensions in the presence of a magnetic field. This effect in combination with the reverse effect (inverse magnetostriction) is utilized in a typical sensor application. A time varying magnetic field causes these sensors to oscillate at a characteristic resonance frequency. The characteristic resonance frequency is dependent on the initial dimensions and physical properties of the material. These materials are of particular interest owing to their unique capability to perform remote (without direct wire contacts to the sensor) sensing, making in-vivo detection and detection in closed containers possible. The phage-immobilized magnetoelastic biosensor was characterized for specificity; dose response in water, spiked apple juice and in spiked milk; selectivity; and longevity. The sensor?s sensitivity is known to be higher for sensors with smaller dimensions. Hence sensors of different dimensions were studied to obtain better detection limits. The sensors? sensitivity increased from 98 Hz/decade to 1150 Hz/decade (decade of S. typhimurium concentrations) for a decrease in length from 5 mm to 500 ?m. The responses of 2 mm (length) sensors were studied in spiked fat free milk and in mixtures with other bacteria. Binding assays of tests conducted in water showed K d values of 149?76 cfu/mL with a binding valency of 2.42?0.02, whereas in fat free milk tests showed a K d value of 136?42 with a binding valency of 2.50?0.03. The similar responses obtained in two dissimilar liquids demonstrates the consistent performance of the sensor even in complex matrices. vii The effect of phage aggregation using varying counterion concentrations on the sensor performance was also studied. It was established that the formation of phage aggregates at higher counterion concentrations (>420 mM Na + ) was realizable. Its effect on the binding numbers was, however, contrary to the expectations. A sharp decline in the binding numbers was observed for higher counterion concentrations (>420 mM Na + ) owing to localized accumulation of these aggregates upon immobilization. Visual verification of bacterial binding to the phage-immobilized sensor was achieved through Scanning Electron Microscopy (SEM) studies of the sensor surfaces. High magnification SEM also provided an insight into the distribution characteristics of immobilized phage. In summary, specific and selective biosensor with a magnetoelastic transducer and filamentous phage was investigated and demonstrated to be suitable for the detection of S. typhimurium in liquid foods. viii ACKNOWLEDGEMENTS I would like to express my sincere gratitude to Dr. Bryan A. Chin for his expert guidance, support and persistent encouragement throughout my study period. It is a privilege to work under Dr. Chin with his extensive knowledge and genuine concern for his students. I would like to emphasize that his influence on me was not only in acquiring scientific knowledge but also as a person. I received utmost support from my committee members Dr. Z-Y. Cheng, Dr. Valery A. Petrenko, and Dr. Dong-Joo Kim and thank them whole heartedly. I would also like to express my thanks and appreciation to Dr. Stuart Wentworth and Dr Vitaly J. Vodyanoy for their thoughtful discussions and suggestions for my defense. Special thanks to Dr. Ben Fiebor, Dr. Hu Jing, Leslie C. Mathisson, Dr. Hong Yang, Shichu Huang, Michael Johnson Wen Shen and Dr. Jiehui Wan. I would like to thank my friends Dr. Jyoti Kumar Ajitsaria, Prakriti Choudhary and Dr. Rajesh Guntupalli, for their constant motivation and valuable discussions. I would like to thank my parents and brothers for their constant emotional support and encouragement during the course of my studies. I would like to dedicate this dissertation to my wife Subhadravalli Lakshmanan for supporting me and being patient through the ups and downs of the last 5 years. Without her love, prayers, support and help this work would not have been possible. Style manual or journal used Sensors and Actuators B Computer software used Microsoft Office XP and Microcal Origin 6.0 ix x TABLE OF CONTENTS TABLE OF CONTENTS.................................................................................................... x LIST OF FIGURES ......................................................................................................... xiv LIST OF TABLES........................................................................................................... xxi 1. INTRODUCTION ...................................................................................................... 1 1.1. Background.....................................................................................................................1 1.1.1 Bacterial food-borne pathogens...............................................................................................2 1.1.2 Salmonella Related Outbreaks.................................................................................................4 1.2. Diagnostic Biosensors: Definition and Components......................................................6 1.2.1 Bio-recognition Element..........................................................................................................7 1.2.2 Transducer Platform ..............................................................................................................10 1.3. Research Objectives .....................................................................................................12 1.3.1 Fabrication and characterization of the sensor platform........................................................13 1.3.2 Characterization of phage-immobilized magnetoelastic biosensor .......................................14 1.3.3 Characterization of phage immobilization on sensor surface ................................................14 REFERENCES ................................................................................................................. 16 2. LITERATURE REVIEW ......................................................................................... 22 2.1. Conventional micro-biological methods ......................................................................22 2.1.1 History...................................................................................................................................22 xi 2.1.2 Polymerase Chain Reaction (PCR)........................................................................................25 2.1.3 Enzyme-Linked Immunosorbent Assay (ELISA)..................................................................27 2.2. Diagnostic Biosensors ..................................................................................................30 2.3. Bio-recognition element ...............................................................................................31 2.3.1 Enzymes ................................................................................................................................31 2.3.2 Antibody................................................................................................................................31 2.3.3 Bacteriophage........................................................................................................................32 2.4. Transduction Methods..................................................................................................36 2.4.1 Electro-chemical....................................................................................................................36 2.4.2 Optical ...................................................................................................................................41 2.4.3 Mass based ............................................................................................................................47 REFERENCES ................................................................................................................. 51 3. MATERIALS AND METHODS.............................................................................. 65 3.1. Sensor Fabrication........................................................................................................65 3.2. Phage immobilization on the sensor surface ................................................................67 3.3. Resonance frequency measurement..............................................................................68 3.4. Bacterial suspensions....................................................................................................68 3.5. Scanning Electron Microscopy (SEM).........................................................................69 3.6. Estimation of bound bacteria based on sensor responses .............................................70 3.7. Estimation of bound bacteria based on SEM analysis..................................................71 3.8. Testing procedures........................................................................................................72 3.8.1 Static tests..............................................................................................................................72 3.8.2 Specificity..............................................................................................................................73 xii 3.8.3 Longevity...............................................................................................................................73 3.8.4 Dose response........................................................................................................................74 3.8.5 Selectivity and detection in food matrices.............................................................................76 3.9. Hill plot construction....................................................................................................77 REFERENCES..........................................................................................................................80 4. Theory and Measurement Circuit ............................................................................. 81 4.1. Theory ..........................................................................................................................81 4.2. Measurement Setup ......................................................................................................85 4.3. Equivalent circuit for the measurement setup ..............................................................86 4.3.1 Determination of Inductance, Capacitance and Resistance of the solenoid coil....................86 4.3.2 Equivalent circuit with sensor present in the coil ..................................................................93 4.3.3 Enhancement of sensor responses .........................................................................................98 4.4. Discussion ....................................................................................................................99 REFERENCES........................................................................................................................101 5. RESULTS AND DISCUSSION............................................................................. 102 5.1. Outline ........................................................................................................................102 5.2. Preliminary studies .....................................................................................................103 5.2.1 Frequency response and mass sensitivity ............................................................................103 5.2.2 Sensor response to S. typhimurium in static conditions ......................................................107 5.2.3 Saturated response time.......................................................................................................110 5.3. Biosensor Characterization.........................................................................................112 5.3.1 Sensor specificity.................................................................................................................112 5.3.2 Biosensor dose response......................................................................................................114 xiii 5.3.3 Biosensor response in real food...........................................................................................119 5.3.4 Selectivity in the presence of high concentrations of masking bacteria ..............................124 5.3.5 Longevity of magnetoelastic biosensors..............................................................................128 5.4. Size dependent Sensitivity..........................................................................................133 5.4.1 Dose response of biosensors with different lengths.............................................................133 5.5. Phage immobilization.................................................................................................140 REFERENCES........................................................................................................................146 6. CONCLUSIONS..................................................................................................... 149 xiv LIST OF FIGURES Figure 1-1: Number of reported sources of Salmonella serotypes from human sources during the period of 2000 to 2005............................................................................... 3 Figure 1-2: Schematic showing the basic components of a diagnostic biosensor. ............ 7 Figure 1-3: Schematic showing basic components of fd filamentous phage. The tube in the center contains the genetic material. Minor coat proteins cap the ends of the tube. The inset shows a TEM picture (Magnification 45000X) of phage filaments at a concentration of 5?10 11 vir/mL. ................................................................................. 9 Figure 1-4: Chart depicting experiments to meet the research objectives....................... 13 Figure 2-1: Schematic showing basic steps involved in PCR.......................................... 26 Figure 2-2: Schematic showing basic steps involved in a typical Indirect ELISA, Sandwich ELISA assay and Competitive ELISA..................................................... 28 Figure 2-3: Schematic depicting selection procedure for phage...................................... 34 Figure 2-4: Basic reactions involved in amperometric detection of E. coli [35]. In this technique the intrinsic enzyme from the bacteria is released by phage-driven lysis of the cells. .................................................................................................................... 38 Figure 2-5: Schematic depicting a Light Addressable Potentiometric Sensor (LAPS)... 40 Figure 2-6: "Schematic of SPREETA (TM) system used" Adapted from [60]................... 46 Figure 3-1: Steps in the fabrication of magnetoelastic biosensor.................................... 67 xv Figure 3-2: SEM image of a magnetoelastic sensor surface after immunoreaction with 5 ?10 6 cfu/mL concentration of S. typhimurium. An overlay grid is shown to illustrate the bacterial counting technique on the sensor surface............................................. 72 Figure 3-3: Schematic of experimental setup for dose response studies. ........................ 75 Figure 4-1: Thin rectangular plate discussed in the derivation with a length L, width w and thickness t (<< L, w)........................................................................................... 81 Figure 4-2: Equivalent circuit for the solenoid coil.......................................................... 87 Figure 4-3: Typical frequency response of a solenoid coil wound around a glass tube (O.D.=200 ?m), measured using an Agilent 4395A network analyzer with an S- parameter test set (87511A)...................................................................................... 89 Figure 4-4: Impedance-Frequency spectrums used for calculating the inductance and self capacitance of solenoid coil. Curves also compare the responses obtained using the PSPICE model. ......................................................................................................... 90 Figure 4-5: Comparison of impedance data from PSPICE model and measured values. Similar trends in changes (frequency and magnitudes) showing the accuracy of the model......................................................................................................................... 93 Figure 4-6: Equivalent circuit for the sensor oscillations. The AC voltage source with an amplitude of 1 V was arbitrarily chosen to obtain a frequency spectrum. ............... 95 Figure 4-7: Comparison of actual signal and that obtained from PSPICE model. .......... 97 Figure 4-8: Comparison of frequency responses of a 200 ?m sensor in water with a tuned and an untuned circuit............................................................................................... 99 Figure 5-1: Frequency dependence on length (200 ?m-10 mm) of magnetoelastic sensors..................................................................................................................... 103 xvi Figure 5-2: Typical frequency response of a magnetoelastic sensor before and after addition of thin film of gold.................................................................................... 104 Figure 5-3: Frequency shifts calculated as function of mass of gold added to the sensor surface. The three different lengths shown are L=1.3, 1.0 and 0.5 mm. ................ 105 Figure 5-4: Dependence of length on the mass sensitivity (?f/?m) of magnetoelastic sensors with air as surrounding media.................................................................... 106 Figure 5-5: Typical response of a 1?0.2?0.015 mm sensor to 5?10 8 cfu/mL of S. typhimurium at different time intervals (t=0, 10, 20, 30, 40 and 50 minutes). A total frequency shift (?f) of 1290 Hz was observed........................................................ 108 Figure 5-6: SEM pictures (Magnification: 1000X) of two different regions on the sensor (2?0.4?0.015 mm) with a frequency shift of 1290 Hz. .......................................... 109 Figure 5-7: Responses of three different sensors (L=500 ?m) to S. typhimurium. An average saturation response could be seen at the end of 30 minutes...................... 111 Figure 5-8: Specificity of phage-immobilized sensors exposed to different pathogens (5?10 8 cfu/mL). The normalized area coverage density was calculated from SEM photomicrographs of the sensor surface (an average of 5 sensors each). ?f measured and ?f SEM are shown on the right side........................................................................... 113 Figure 5-9: Typical SEM images of phage-immobilized sensors exposed to (A) S. typhimurium; (B) S. enteritidis; (C) E. coli and (D) L. monocytogenes. ............. 114 Figure 5-10: Typical dynamic response curve for a sensor with dimensions 2?0.4?0.015 mm. 1 mL of each concentration of bacterial suspension was allowed to flow over the sensor at a flow rate of 50 ?L/min. Control sensor response shown is of a sensor devoid of phage....................................................................................................... 115 xvii Figure 5-11: Magnetoelastic biosensor?s responses, when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions on test sensors (2?0.4?0.015 mm) (?- Test sensor: sigmoidal fit ? 2 =0.048, R 2 =0.99) and control (2?0.4?0.015 mm) (? - Control sensor). Each data point is the average value obtained from five individual experiments (different sensors) carried out under identical conditions. ................................................................................................ 117 Figure 5-12: Typical SEM images depicting S. typhimurium attachment to phage- immobilized magnetoelastic sensor surface. Sensors exposed to S. typhimurium at concentrations of (a) 5?10 8 cfu/mL, (b) 5?10 6 cfu/mL (c) 5?10 3 cfu/mL and (d) control (biosensor devoid of phage and treated with 5?10 1 cfu/mL through 5?10 8 cfu/mL of bacterial sample).................................................................................... 118 Figure 5-13: Comparison of dose responses of magnetoelastic biosensor (2?0.4?0.015 mm), when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions in water ((?) ? 2 =0.442, R 2 =0.99), apple juice ((?) ? 2 =0.237, R 2 =0.99) and fat free milk ((?) ? 2 =0.194, R 2 =0.99). Control (?) represents the uncoated (devoid of phage) sensor?s response. The curves represent the sigmoid fit of signals obtained.............................................................................................. 121 Figure 5-14: Hill plots of binding isotherms showing the ratio of occupied and free phage sites as a function of bacterial concentrations spiked in different food samples. The straight line is the linear least squares fit to the data (water (?): slope=0.40?0.03, R=0.97; fat-free milk (?): slope=0.41?0.04, R=0.98; apple juice (?) slope=0.36, R=0.96). .................................................................................................................. 122 xviii Figure 5-15: Typical SEM images of S. typhimurium bound to a magnetoelastic biosensor surface (2?0.4?0.015 mm) in (a) fat-free milk, (b) water (d) apple juice and (c) control (biosensor devoid of phage and treated with 5?10 8 cfu/mL of bacterial sample). .................................................................................................... 123 Figure 5-16: Dose response curve of magnetoelastic sensors (2?0.4?0.015 mm) in response to S. typhimurium in mixture with other masking bacteria. Sensors (each data point is an average of the response from five sensors) exposed to only S. typhimurium (?-? 2 =0.44, R 2 =0.99), S. typhimurium in mixture with E. coli (?- ? 2 =0.18, R 2 =0.99), and S. typhimurium in mixture with E. coli + L. monocytogenes (?-? 2 =0.24, R 2 =0.99). The control (?-? 2 =0.048, R 2 =0.99), is the response of an uncoated (devoid of phage) sensor. The curves represent the sigmoidal fit of signals obtained................................................................................................................... 125 Figure 5-17: Hill plot constructed from the dose response curves, showing the ratio of occupied (Y) and free phage sites (1-Y) as a function of bacterial concentrations in different mixtures. The straight line is the linear least squares fit to the data (S. typhimurium (?): slope=0.40?0.03, R 2 =0.97; S. typhimurium + E. coli (?): slope=0.33?0.02, R 2 =0.98; and S. typhimurium + E. coli + L. monocytogenes (?) slope=0.34?0.02, R 2 =0.97)..................................................................................... 126 Figure 5-18: Typical SEM images of S. typhimurium bound to the phage-immobilized biosensor surface with increasing time (1, 3, 5, 15 34 and 62 days) at 65 ?C. ....... 129 Figure 5-19: Typical SEM images of S. typhimurium bound to the phage-immobilized biosensor surface with increasing time (1, 3, 5, 15, 34 and 62 days) at 45 ?C. ...... 130 xix Figure 5-20: Typical SEM images of S. typhimurium bound to the phage-immobilized biosensor surface with increasing time (1, 3, 5, 15, 34 and 62 days) at 25 ?C. ...... 131 Figure 5-21: Surface coverage densities (average number of cells/?m 2 ) calculated from SEM micrographs of stored magnetoelastic biosensors (25 ?C, 45 ?C, and 65 ?C) after exposure to S. typhimurium (5?10 8 cfu/mL). ................................................. 133 Figure 5-22: Comparison of magnetoelastic biosensor?s dose responses, when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions on two different sizes of sensors 2?0.4?0.015 mm (?- ? 2 =0.048, R 2 =0.99) and 5?1?0.015 mm (?- ? 2 =0.32, R 2 =0.99). The curves represent the sigmoidal fit of signals obtained....................................................................................................... 135 Figure 5-23: Comparison of magnetoelastic biosensor?s dose responses, when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions on two different sizes of sensors (1?0.2?0.015 mm (?- ? 2 =0.048, R 2 =0.99) and 2?0.4?0.015 mm (?- ? 2 =0.32, R 2 =0.99)). The curves represent the sigmoidal fit of signals obtained. Each data point is the average value obtained from five individual experiments (different sensors) carried out under identical conditions.................. 136 Figure 5-24: Comparison of magnetoelastic biosensor?s dose responses, when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions on two different sizes of sensors (0.5?0.1?0.015 mm (?- ? 2 =0.048, R 2 =0.99) and 1?0.2?0.015 mm (?- ? 2 =0.7231, R 2 =0.91) The curves represent the sigmoidal fit of signals obtained....................................................................................................... 137 xx Figure 5-25: Typical SEM images of the entire surfaces of assayed 500?m sensors at three different concentrations (5?10 2 cfu/mL, 5?10 4 cfu/mL and 5?10 8 cfu/mL). Bound S. typhimurium can be seen as black spots on the pictures. ........................ 138 Figure 5-26: High magnification (15000X) SEM images of sensor surface (a) before and (b) after immobilization of phage. .......................................................................... 141 Figure 5-27: SEM images showing the nature of phage distribution in presence of different Na + ion concentrations (a) 280 mM; (b) 420 mM; (c) 560 mM; (d) 840 mM.......................................................................................................................... 142 Figure 5-28: SEM images showing binding distribution on sensors immobilized with phage with varying counterion (Na + ) concentrations (a, b) 240 mM; (c, d) 420 mM; (e, f) 560mM; and (g, h) 840mM............................................................................ 144 xxi LIST OF TABLES Table 1-1: Salmonella outbreaks in various food products. .............................................. 5 Table 2-1: Timeline showing some of the important contributions in bacteriology......... 24 Table 2-2: ICTV classification of phages based on structural morphology and genetic material enclosed [32]............................................................................................... 35 Table 2-3: Summary of literature reports on the use of phage as a bio-recognition element in various assays.......................................................................................... 49 Table 3-1: Physical and magnetic properties of METGLAS? 2826MB........................ 65 Table 4-1: Calculated values for components of the equivalent circuit........................... 96 Table 5-1: Comparison of number of bacterial cells attached estimated from frequency shifts obtained and the extrapolated number calculated from SEM images........... 110 Table 5-2: The sensitivity, dissociation constant and binding valency of magnetoelastic sensors in different bacterial mixtures. ................................................................... 127 Table 5-3: Table summarizing the sensitivity and detection limits achieved for sensors with different dimensions. ...................................................................................... 139 1 1. INTRODUCTION 1.1. Background Food-borne illnesses or ?food poisoning,? as it is more commonly known is not of recent occurrence. Each year approximately 76 million illnesses, 325,000 hospitalizations and 5000 deaths are reported in the United States alone [1]. Food-borne illnesses are primarily caused by viruses, bacteria and parasites. These illnesses cause mild diarrhea to severe life-threatening neurological ailments. The reason for these large numbers of cases can be attributed to changing human demographics, human behavior, mass transportation of foods and microbial adaptation [2, 3]. Improved medical care has increased the median age of human beings [2, 3], however people of older age are at a higher risk due to weaker immune systems. The changes in human lifestyle and food consumption behavior are also to be blamed. Interest in international cuisine and frozen food products directly lead to longer transportation and storage times. The longer shelf life of foods allow for greater chances of spoilage and the production of foods in foreign countries where health standards vary significantly has lead to a plethora of additional contamination sources. Contamination of food may occur at any stage during processing, production, packaging, transportation or storage. The risks of food contamination can be decreased by the use of existing technologies. The Food and Drug Administration (FDA) and Centers for Disease Control (CDCs) have emphasized the problem of food-borne 2 illnesses to be one of the most serious, yet avoidable problems. With better awareness and proper hygienic practices, the losses due to food-borne illnesses can be reduced drastically. In spite of the significant impact of food-borne illnesses on society, it has never been a major issue of concern to the general public. Food supply bio-terrorism has invoked greater interest among researchers and authorities as a result of the looming terrorism threats following the September 11 th , 2001 attacks and the Anthrax mailings there after. The food supply system is most vulnerable to a bio-terrorism attack. The first deliberate and largest act of bio-terrorism in modern history of the United States was in 1984. Two members of the Rajneesh cult targeted the voters in Oregon to influence the outcome of the elections in their favor. These members contaminated the local restaurant salad bars with Salmonella typhimurium [4] sickening about 800 people. 1.1.1 Bacterial food-borne pathogens There are more than 250 food-borne pathogens that exist in the environment, including viruses, fungi and bacteria. About 90% of food-borne illnesses caused can be attributed to bacterial contamination of food. Bacterial pathogens such as Salmonella typhimurium, Salmonella enteritidis, Escherichia coli, Listeria monocytogenes, Staphylococcus aureus, Campylobacter jejuni, and Bacillus cereus are sources of bacterial contamination in the food products. Salmonella species are the most frequent and commonly occurring bacterial food-borne pathogens worldwide [5]. There are over 2000 distinct types of serovars of salmonella species. Figure 1-1 shows the 5 most reported serotypes isolated from human sources in the United States as indicated by the CDC annual reports 2000-2005 [7]. Salmonella enterica serovar Enteritidis followed by serovar typhimurium are the most frequently occurring serotypes (shown in Figure 1-1) isolated from human sources in the USA [6]. In the United States the occurrence of Salmonella typhimurium is more predominant. 0 1000 2000 3000 4000 5000 6000 7000 8000 2000 2001 2002 2003 2004 2005 Year Num b e r of re port e d sour ce s of Salmonel la Serotype Typhimurium Enteritidis Newport Heidelberg Javiana Figure 1-1: Number of reported sources of Salmonella serotypes from human sources during the period of 2000 to 2005. 3 4 1.1.2 Salmonella Related Outbreaks A number of infamous Salmonella related outbreaks of food poisoning around the world have affected a large number of people resulting in discomfort, grief and sometimes even death. The number of cases reported due to Salmonella related illnesses is between 6.5 million to 33 million annually, and account for up to 9000 deaths. The Economic Research Service (ERS) estimates the annual economic cost of salmonellosis as $142,552,427. The average cost per case (in the year 2006) was approximately $10,000 including medical costs and productivity loss. Some of the recent Salmonella typhimurium related outbreaks are caused by bacterial contamination of tomatoes, eggs, milk, orange juice, fresh vegetables, ground beef and peanut butter. A summary of the recent Salmonella outbreaks is given in Table 1-1. A quick look at the table indicates that most of the items on the list are foods of daily consumption, rendering the common man highly vulnerable to any such outbreaks. Salmonella typhimurium has been acknowledged to be one of the great threats to human beings, whether it is due to unintentional ingestion of contaminated food or due to a repugnant act by terrorists. In this dissertation, work on the development of a rapid, sensitive and accurate diagnostic biosensor for the detection of Salmonella typhimurium is reported. 5 Table 1-1: Salmonella outbreaks in various food products. Source Cases Cause News Article Headline Milk/ Ice- cream 10,000 S. typhimurium Schwain ice cream tainted with Salmonella [8] Milk 185,000 S. typhimurium Salmonella Typhimurium causes scare in Chicago milk [8, 9] Milk 80 S. typhimurium Chinese Milk Sickens 80 Children [8, 9] Several 70,000 S. typhimurium 700,000 killed by bad food and water in Asia each year: UN [10] Potato 3,400 S. typhimurium Salmonella in potato salad renders several ill [11] Oat Cereal 100 S. typhimurium Salmonella Outbreak Sickens 100 in 7 States [2, 12] Tomatoes 459 Salmonella Tomatoes Blamed in Salmonella Outbreak in Pittsburgh, PA [13] Beef 6,000 E. coli 5.7 Million Pounds Of Ground Beef Recalled Due To E.Coli. [14] Spinach 5,000 E. coli E. coli spinach scare increases to 21 states [15] Peanut butter 329 S. typhimurium Tests find salmonella in peanut butter. [16] 6 1.2. Diagnostic Biosensors: Definition and Components The ever growing need for rapid detection of pathogenic micro-organisms has resulted in an increased interest in the research and development of biosensor systems. Figure 1-2 illustrates the main components of a biosensor. A biosensor is defined as a device that incorporates a biological recognition element and a transducer for the detection of an analyte/target of interest. The transducer needs to be capable of transferring information about the specific interaction between the analyte and the recognition element into measurable signals with high sensitivity. The bio-recognition element is an integral part of any biosensor. The bio-recognition element is responsible for the selectivity, specificity and thermal stability of a biosensor. Bio-recognition Layer Signal Bio-recognition Layer TRANSDUCER ? Electro-chemical (Potentiometric, Amperometric) ? Optical (Waveguides, SPR) ? Mass Based (QCM, Magnetoelastic) No Signal TRANSDUCER ? Electro-chemical (Potentiometric, Amperometric) ? Optical (Waveguides, SPR) ? Mass Based (QCM, Magnetoelastic) Suitable electronics Enzyme After specific attachment of target Before specific attachment of target Suitable electronics Bacteriophage Antibody Enzyme Bacteriophage Antibody Figure 1-2: Schematic showing the basic components of a diagnostic biosensor. 1.2.1 Bio-recognition Element The bio-recognition element is vital to the design of a biosensor. There are several types of bio-recognition elements (enzymes (section 2.3.1), antibodies (section 2.3.2), and phage (section 1.2.1.1, section 2.3.3)) that are used for biosensor applications [17]. In this dissertation, bacteriophage was used as the bio-recognition element for the detection of Salmonella typhimurium. The application of different types of phages used for diagnostic biosensor applications has been discussed and reviewed in Chapter 2. 7 8 1.2.1.1 Filamentous Bacteriophage Bacteriophage derives its name from a combination of the word ?bacteria? and Greek word ?phagos? that in English mean ?bacteria eater.? The first report of the existence of phage was reported by a British scientist Dr. Ernest Hankin in 1896. He observed that the water from the river Ganges (largest river in India) had strong anti- microbial action (against Vibrio cholerae). Phage is a type of virus that infects bacteria at very specific sites. There are several types of phage that exist, classified based on their structure and the type of genes they carry. Filamentous bacteriophage, as its name suggests, is a type of phage that has a thread-like appearance and encloses a circular ss-DNA as its genetic material. Typically filamentous bacteriophage has its genetic components enclosed within a tube with the major coat proteins pVIII covering a large percentage of the entire surface along the tube length. The ends are then capped with minor coat proteins pVI and pIII on one end and pVII and pIX on the other end (Figure 1-3). ss- DNA Minor coat protein pVI Minor coat protein pVII Minor coat protein pIII Minor coat protein pIX Major coat protein pVIII TEM image (Magnification: 45000X) 20nm * Not drawn to scale Figure 1-3: Schematic showing basic components of fd filamentous phage. The tube in the center contains the genetic material. Minor coat proteins cap the ends of the tube. The inset shows a TEM picture (Magnification 45000X) of phage filaments at a concentration of 5?10 11 vir/mL. 9 10 Each phage filament is approximately 800-900 nm in length and has a diameter of about 6 nm. A landscape phage is a recombinant phage displaying 4000 copies of random peptides on their surface. Phage display technique is used to modify a small portion of the DNA. This modification leads to a display of desired peptides or ?organic landscapes? fused into the major coat proteins [18-23]. Several steps (usually 5-10) of affinity selection (biopanning) procedures are then performed to yield a phage clone displaying peptides that are highly specific to the target analyte. The development of filamentous phage using phage display techniques have been studied extensively with potential applications in the area of biosensors research [18-20]. However, the study of affinity selected filamentous bacteriophage as a bio-recognition probe for biosensor applications is limited [28-35]. 1.2.2 Transducer Platform In this dissertation, the use of a magnetoelastic material as a transducer platform for the rapid, selective and specific detection of S. typhimurium is reported. Recently, the use of magnetoelastic materials as transducer platforms for the remote monitoring of food-borne pathogens [34-39] and other applications in chemical detection and environmental monitoring [40-45] have attracted considerable interest. Magnetoelastic transducers have the unique advantage of detection in the absence of physical wire contacts to the sensor, enabling in-situ wireless measurements in sealed containers. Magnetoelastic materials are amorphous ferromagnetic alloys that usually include a combination of iron, nickel, molybdenum and boron. These materials work on the principle of magnetostriction, wherein, the material experiences changes in its dimensions in presence of a magnetic field. Upon application of a magnetic field, the randomly oriented magnetic domains in the material tend to align in the direction of the applied field. The alignment of the magnetic domains in the magnetoelastic material results in a change in the dimensions. By applying a time varying magnetic field, the magnetoelastic materials can efficiently convert the applied magnetic energy into mechanical oscillations. The characteristic fundamental resonance frequency of these oscillations is dependent on the physical properties (elastic modulus, density and Poisson?s ratio) and the dimensions of the material. Both the actuation of the sensor and the detection of the response of the sensor can be measured using changes in the impedance of a non-contact solenoid pick-up coil. Magnetoelastic sensors are actuated by the application of an AC magnetic field that causes the sensors to oscillate mechanically. When the frequency of the applied field is in resonance with the natural frequency of the sensor, the conversion of electrical energy to elastic energy is the largest. For a thin, planar, ribbon shaped sensor of length L, vibrating in its basal plane, the fundamental resonant frequency of longitudinal oscillations is given by [45, 46] L E f 2 1 )1( 2 ?? ? = (1-1) where, E is Young?s modulus of elasticity, ? is the density of the sensor material, ? is the Poisson?s ratio, and is the length of the sensor. L Any non-magnetoelastic mass added to the sensor surface reduces the mechanical oscillations causing the resonance frequencies to shift to a lower value. A scan through the range of applied AC frequencies yields a spectrum. The resonance frequency of the sensor can be read from this spectrum and can be tracked for any changes. The details of 11 the measurement setup have been described in detail in section 3.3 and section 4.2. The changes in the resonance frequency can then be related to the magnitude of non- magnetoelastic mass attached to the sensor surface. Addition of a small mass of ?m (much smaller than the original mass of the sensor (?m<50,000 Saturation Magnetostriction (ppm) 120 A 50 mm long strip was cut out of the roll of ribbon and polished using fine grit polishing paper (1000 micron and 2000 micron). The polishing was used to reduce the thickness to 15 ?m and to provide a smooth surface on both sides. The reduced thickness of the sensor (smaller mass) increases the mass sensitivity of the device. The polished strips were then diced to the desired sizes for experimentation using a computer controlled automatic micro-dicing saw. Any debris or grease remaining from the dicing process was removed by cleaning the diced sensors ultrasonically in acetone for 20 minutes. The cleaned sensor platforms were then subjected to a thermal anneal in a vacuum oven at 200 o C for 2 hours and oven cooled under vacuum (?10 -3 torr). Upon cooling, the sensors were transferred to a Denton? (Moorestown, NJ) high-vacuum, RF sputtering system. Chromium and Gold (in that order) were sputtered onto both sides of the sensor platforms. Chromium was sputtered (DC) to a thickness of approximately 50 nm and gold was sputtered (RF) to a thickness of about 100 nm. The layer of chromium was sputtered to improve the adhesion of the gold film to the substrate. In addition, chromium was also expected to electro-chemically protect the iron rich substrate from corrosion when exposed to saline environments for extended periods of time. The thin film of chromium was then covered with a thin layer of gold. The gold layer provides a suitable surface for phage immobilization. The fabricated sensor platforms were stored in a dessicator until their use. The sensor platforms were washed with hexane and air dried prior to immobilization of the bio-recognition element. The basic steps in the fabrication of the magnetoelastic biosensor are shown in Figure 3-1. 67 Figure 3-1: Steps in the fabrication of magnetoelastic biosensor. 3.2. Phage immobilization on the sensor surface The filamentous phage E2, used in this work, was provided by Dr. James M. Barbaree?s lab in the Department of Biological Sciences at Auburn University. The magnetoelastic sensor platforms were placed in vials containing 100 ?L of phage E2 (5?10 11 virions/mL). The immobilization of phage on the sensor surface was done by placing these vials on a rotor (running at 8 rpm) for 1 hour. The use of a rotor for the immobilization step resulted in a uniform distribution of phage on the sensor surface. These sensors were then washed three times with Tris-Buffered Saline (TBS) solution Incubation with phage for 1 hour Polished to a thickness of 15 ?m L 5 L As received METGLAS ? 2826MB Dicing Ultrasonic cleaning Vacuum anneal @200 C Cr/Au Sputtering 68 and two times with sterile distilled water in order to remove any unbound or loosely bound phage. The resonance frequency of the sensors was measured at this point to obtain a baseline signal. 3.3. Resonance frequency measurement A solenoid coil wound around a glass tube was used to measure responses of the sensor platform. The sensor platform was placed in the center of the solenoid coil. A time-varying current applied to the solenoid generates a time-varying magnetic field. This time-varying magnetic field causes the sensors to oscillate [1-4]. Along with the AC magnetic field, an external DC magnetic field was used to bias the magnetoelastic sensor. At the mechanical resonance frequency of the sensor, there is a maximum transfer of mechanical energy to magnetic energy. The resonance frequency of the sensor?s oscillations was determined using a HP8751A network analyzer. The measurements using the network analyzer have been described in detail in section 4.2. The solenoid coil is connected to the network analyzer and is calibrated before the sensor is placed in the coil. This calibration step cancels any ambient electrical noise. 801 points were recorded over the frequency range of interest with a 11.31 sec sweep time. The data measured by the network analyzer was transferred to a personal computer with the help of the HP-VEE program. 3.4. Bacterial suspensions The S. typhimurium (ATCC13311), S. enteritidis, E. coli and L. monocytogenes used in this work were provided by Dr. James M. Barbaree?s lab in the Department of Biological Sciences at Auburn University, Auburn, AL. One colony of bacteria from a 69 master plate was inoculated into NZY (Casein Hydrolysate (10 g), Yeast extract (5 g), NaCl (5 g), MgCl 2 (2 g)) broth and incubated at 37 ?C for 12 hours. Following incubation, the culture was gently mixed for homogeneity and transferred into sterile tubes. The sterile tubes containing the bacterial cells were then centrifuged. The supernatant solution obtained from the centrifugation was decanted. After the first round of centrifugation and decantation, the bacterial cells were resolubilized in distilled water. The S. typhimurium cultures obtained from Dr. Barbaree?s lab were obtained in the form of a suspension at a concentration of 5?10 8 cfu/mL. The concentration of the cultures was confirmed using a spectrophotometer. The suspensions were serially diluted in water or desired food matrices to prepare bacterial suspensions ranging from 5?10 1 to 5?10 8 cfu/mL. Prior to each dilution, the solution was mixed using a vortex mixer to ensure homogeneity of concentration in the solution. All test solutions were prepared on the same day as the biosensor testing. The test solutions were stored at 4?C (during transfer and storage) and equilibrated to room temperature in a water bath prior to the experiments. 3.5. Scanning Electron Microscopy (SEM) The surface of the biosensors was viewed at different stages of experimentation using scanning electron microscopy. SEM provided a visual verification of Salmonella cells bound to the sensor surface. Assayed sensors were washed one time with sterilized distilled water. The sensors were then placed on cylindrical aluminum stubs with the help of an adhesive double sided carbon conductive tape (Ted Pella Inc.). These sensors were then allowed to dry in air for 15 minutes. The aluminum stubs with the sensor 70 platforms were placed in a Petri dish containing a small volume (60 ?L) of Osmium tetraoxide (OsO 4 ) for 10 minutes. Osmium tetraoxide is commonly used as a stain for biological samples for electron microscopy. The diffusion of Os (heavy metal) into the cell membrane provides a better contrast to the images in an electron microscope. For some of the initial set of experiments, a gold layer of 50nm was sputtered (PELCO sputter coater, SC-7) onto the sensor surface. A JEOL-7000F Field Emission Scanning Electron Microscope was used for imaging. The images were taken at an accelerating voltage of 10 or 15 kV, a working distance of 10 mm, an aperture of 3, and a probe current of 54 ?A. SEM images of the sensor surface at desired magnifications were recorded in electronic format (jpeg files) using JEOL-Imaging software. The number of bacterial cells attached to the sensor was estimated by counting the number of bacterial cells bound to several different regions on the sensor surface. These results were then extrapolated to obtain the total number of cells bound to the sensor surface. The sample preparation technique limited the imaging to only one side of the sensor. The number of cells bound to either side of the sensor platform was assumed to be the same. Hence, the total number of cells attached to the sensor surface was calculated as two times that obtained for one side of the sensor. 3.6. Estimation of bound bacteria based on sensor responses The sensor?s response to a small added mass (?m), much smaller than the initial mass (M) of the sensor, will result in a change in frequency (?f) given as (rewritten from equation 1-2): )13( 2 ? ? ?=? f fM m For a sensor with initial dimensions of 2?0.4?0.015 mm, the frequency (f) can be calculated using equation 1-1 by substituting parameters from Table 3-1. The initial mass was calculated using the density (?) and initial volume of the sensor (L?w?t). Upon substitution of these values, the equation 3-1 can be rewritten as )23(5863 ???=? fm where ?f is in Hz and ?m is in picograms. For all calculations that were used to estimate the number of bacteria attached to the sensors, the mass of each bacterium was assumed to be 2 pg. Hence the number of bacteria attached (n calculated ) was determined using the following equation )33( 2 ? ? = m n calculated 3.7. Estimation of bound bacteria based on SEM analysis SEM micrographs were taken at 10 different regions on the sensor surface. The number of bacterial cells attached on the sensor surface was counted manually for each of the pictures taken. The average number of bound cells per unit area was calculated. The resulting number was multiplied by the entire surface area of the sensor to obtain the total number of bound cells. Each picture was divided into nine equal sections as shown in Figure 3-2. The number of bacteria in each of these sections (A n , B n , and C n (n=1, 2, 3)) was counted. The total number was calculated as shown in equation 3-4. )43()( 3 1 ?++= ? = nn n nST CBAn 71 The viewing area (A P ) for each of the SEM image was used calculate the surface area coverage density for the bacterial cells. The average area coverage density obtained from the 10 pictures was then extrapolated to obtain the total number of cells bound to the sensor surface. 72 A 1 A 2 A 3 B 1 B 2 B 3 C 1 C 2 C 3 Figure 3-2: SEM image of a magnetoelastic sensor surface after immunoreaction with 5 ?10 6 cfu/mL concentration of S. typhimurium. An overlay grid is shown to illustrate the bacterial counting technique on the sensor surface. 3.8. Testing procedures 3.8.1 Static tests In static tests, the sensor resonance frequency was measured before and after exposure to solutions containing the pathogens of interest. The prepared biosensor was 73 allowed to dry after final immobilization with phage had been completed. At this point, the resonance frequency of the biosensor was measured. The sensor was then incubated in a vial containing a desired concentration of pathogens (suspended in water). After 45 minutes of incubation, the sensor was again dried. The resonance frequency of the sensor was then compared with that measured prior to incubation. The difference in the resonance frequency of the biosensor before (f before ) and after (f after ) incubation was calulated (?f measured =f before - f after ). The assayed biosensors were then analyzed using SEM and compared with the obtained frequency shifts. Static tests were used in the study of specificity and longevity of the biosensors. 3.8.2 Specificity The specificity of the biosensor was studied by exposing the biosensor to pathogens other than S. typhimurium. The affinity of the biosensor to other pathogens was compared with the biosensor?s affinity to S. typhimurium. Three gram-negative pathogens (S. typhimurium, S. Enteritidis and E. coli) and one gram-positive pathogen (L. monocytogenes) were used in this test. The biosensors were exposed to each of the pathogens individually. After exposure the ?f measured was obtained using the static test procedure (section 3.8.1). The number of pathogens attached to the biosensor was calculated using the methods described in section 3.7. The ?f measured for the assayed biosensors was then compared with ?f SEM . 3.8.3 Longevity The thermal stability of the biosensor was studied at three different temperatures (25 ?C, 45 ?C and 65 ?C). A large number of biosensors were prepared and distributed 74 into three groups. The three groups of sensors were incubated in humidified ovens at 25 ?C, 45 ?C and 65 ?C, respectively. At specified intervals of time (in days) 5 biosensors, from each of the three groups, were taken out of the ovens and equilibrated to room temperature. These 15 biosensors were then exposed to a high concentration (5?10 8 cfu/mL) of S. typhimurium for 45 minutes. The biosensors were then dried and prepared for SEM. SEM micrographs were taken of all the tested biosensors. The average area coverage densities for the three temperatures of incubation were calculated and plotted as a function of time. 3.8.4 Dose response Dose response of the biosensor was studied by monitoring the resonance frequency of the biosensor, upon exposure to different concentrations of S. typhimurium in a single flow-through mode. The different concentrations (5?10 1 cfu/mL through 5?10 8 cfu/mL) were prepared by successive dilutions of the as-received S. typhimurium (5?10 8 cfu/mL). Each successive dilution reduced the concentration of S. typhimurium by a factor of 10. The biosensor was placed in the center of a solenoid coil wound around a glass tube (Figure 3-3). Peristaltic Pump Analyte Reservoir Network Analyzer DC Biasing Magnetic Phage-immobilized Solenoid Coil Glass tube Waste Reservoir Direction of flow Figure 3-3: Schematic of experimental setup for dose response studies. A peristaltic pump was used to flow the analyte through the glass tube. A low flow rate (50 ?L/min) was maintained to ensure a laminar flow (Reynold?s Number=2.37) over the sensor. The biosensor was exposed to flowing water for 20 minutes after obtaining a steady-state resonance frequency. The resonance frequency of the biosensor 75 76 at t=20 minutes was used as the baseline resonance frequency. At time t=20 minutes, the biosensor was exposed to a single pass, flowing suspension containing S. typhimurium (5?10 1 cfu/mL). One ml of suspension was passed across the biosensor at a flow rate of 50 ?L/min every 20 minutes. The change in the resonance frequency due this concentration was calculated as the difference between the baseline resonance frequency and the resonance frequency of the biosensor at t=40 minutes. The resonance frequency of the biosensor was monitored using the network analyzer. The biosensor was then successively exposed to the different concentrations of S. typhimurium (5?10 2 cfu/mL through 5?10 8 cfu/mL) in an increasing order. The resonance frequency at t=60, 80, 100, 120, 160 and 180 minutes were used to calculate the resonance frequency changes due to the different concentrations. The resonance frequency was measured at two minute interval throughout the entire experiment. A computer recorded the resonance frequencies and a plot of frequency as a function of exposure time (exposure concentration) was produced. The assayed biosensors were then analyzed using SEM (section 3.5). 3.8.5 Selectivity and detection in food matrices Milk and apple juice samples were spiked with known concentrations of Salmonella typhimurium (5?10 1 cfu/mL through 5?10 8 cfu/mL). The spiked samples were prepared using fat free milk (Parmalat ? brand) and clear apple juice (Kroger ? brand) purchased from a local grocery store. The frequency response of the biosensor platforms was then studied as described in section 3.8.4. Selectivity tests were performed to determine whether Salmonella typhimurium could be detected in a mixed microbial population. The selectivity tests were conducted using two different masking mixtures: 1) a mixture of S. typhimurium with one masking bacteria (E. coli) and 2) a mixture of S. typhimurium with two masking bacteria (E. coli and L. monocytogenes). The dilutions of S. typhimurium (5?10 1 cfu/mL through 5?10 8 cfu/mL) were prepared in such a way that all the suspensions contained a high concentration (5?10 7 cfu/mL) of the masking bacteria. A high concentration of the masking bacteria was used to investigate the maximum masking effect that a non-specific bacteria may have on the performance of the biosensor. The biosensors were then exposed to the two different masking mixtures (containing 5?10 1 cfu/mL through 5?10 8 cfu/mL of S. typhimurium) using the procedures explained in the section 3.8.4. 3.9. Hill plot construction A Hill plot, constructed from the dose response curves, was used to study the kinetics of the Salmonella-phage binding. This plot was also used to determine the apparent dissociation constant and binding valency. The S. typhimurium binding to the phage immobilized on the sensor?s surface can be expressed as a reversible reaction: 5)-(3 .phageSTphagenST n ?+ where 1/n represents binding valency of the ST (S. typhimurium) attaching to the phage immobilized on the sensor surface. The association (k a ) and the dissociation constant (k d ) for the above reversible reaction is: 6)-(3 1 ].[ ][][ dn n a kphageST phageST k == 77 The reaction of association is primarily due to the movement of bacteria in solution and its interaction with the immobilized phage. The dissociation is governed by the strength with which bacteria binds to the immobilized phage on the sensor surface. The calculations to determine the values of dissociation constant (k d ) and binding valency (1/n) were done by constructing a hill plot. The value of k d can be determined as the reciprocal of the ordinate intercept. The binding valency is determined as the reciprocal of the slope of the hill plots. The binding valency is the number of phage binding sites that interact with Salmonella typhimurium. 7)-(3 apparent)( 1 n dd kk = The Hill plots were traditionally used to study kinetics of reactions, where the reactant and the product concentrations were expressed using molar concentrations. However, in this method the concentration of S. typhimurium has the units of cfu/mL. The use of k d (apparent) takes into account the scaling factor due to the use of different units for the concentrations of the analyte. The k d (apparent) is defined as the concentration at which half of the available binding sites are occupied. Hence, a stronger Salmonella-phage binding is indicated by lower values for k d (apparent). The Hill plot was constructed by plotting log X versus the log[ ]ST , where X is given as 9)-(3 & 8)-(3 )1( maxmax f f m m Y Y Y X ? ? = ? ? = ? = 78 Here ?f denotes the frequency shift obtained as a response to the binding of each concentration of S. typhimurium to the biosensor. ?f max denotes the maximum frequency that can be detected before the biosensor has reached saturation. ?f max was calculated 79 from the sigmoidal curve fitting of the dose response curve obtained for the biosensor. The values of the apparent dissociation constant and binding valency were used to compare the biosensor?s performance in different matrices. 80 REFERENCES 1. Vincent, J.H., Further experiments on magnetostrictive oscillators at radio- frequencies. Proceedings of the Physical Society, 1931. 43(2): 157-165. 2. Muzzey, D.S., Some Measurements of the Longitudinal Elastic Frequencies of Cylinders Using a Magnetostriction Oscillator. Physical Review, 1930. 36(5): 935-947. 3. Pierce, G.W., Magnetostriction Oscillators. Proceedings of the IRE, 1929. 17(1): 42-88. 4. Vincent, J.H., Experiments on magnetostrictive oscillators at radio frequencies. Proceedings of the Physical Society, 1928. 41(1): 476-486. 5. "Agilent 4395A Network/Spectrum/Impedance Analyzer Operation Manual," Agilent Technologies. 6. Landau, L.D. and Lifshitz, E.M., Theory of Elasticity. 3 ed. Vol. 7. 1986: Butterworth-Heinemann. 187. 4. Theory and Measurement Circuit 4.1. Theory A theoretical derivation of the basic equations to determine the resonance frequency of magnetoelastic sensor platform is presented in this section. The sensor platforms used in this work are thin rectangular plates. A thin rectangular plate of length L, width w and a thickness t (<< L, w) is shown in Figure 4-1. ?x x Z XY x x a aa ? ? ? + ? ? a ?? Displacement: At x =u x x u ? ? ? At x+?x =u+ Area of cross-section=A Figure 4-1: Thin rectangular plate discussed in the derivation with a length L, width w and thickness t (<< L, w). For the thin plate shown, we consider an element of length, ?x,extending from X=x to X=x+?x. The center of the thin plate is defined at X=0. Let ? T be the total stress and u be the displacement of the element ?x at X=x. For mathematical simplicity, the effects of eddy currents and the viscoelastic damping have been neglected. The thickness of the 81 plate was assumed to be negligible in comparison to the width and the length of the plate (t<>50 ?). To check the validity of the circuit representing the solenoid, a capacitor was added in series with the solenoid coil. Simulations were run for two different values of the added series capacitance. In Figure 4-5 there are two resonance peaks: one for a series resonance (seen as a minima); and one for a parallel resonance (seen as a maxima). The series resonance is due to the added capacitance (C a ) in series with the inductance (L l ) of the solenoid coil. The parallel resonance is due to the self capacitance of the solenoid coil (C s ) in parallel with the inductance of the solenoid coil (L l ). It is also important to observe that the measured and modeled values in terms of the magnitude of the peak and the resonance frequency followed similar trends. The observed trends due to the addition of capacitors in parallel or in series with the coil (Figure 4-5) support the validity of the model. 10 100 1000 10000 100000 1000000 1000000 6000000 11000000 16000000 21000000 Frequency (Hz) I m pe danc e ( ? ) No C (Measured) 150pF in Series (Measured) 22pF in Series (Measured) Model 150pF in series Model 22pF in series Model (No C) With No C With 22pF With 150pF Figure 4-5: Comparison of impedance data from PSPICE model and measured values. Similar trends in changes (frequency and magnitudes) showing the accuracy of the model. 4.3.2 Equivalent circuit with sensor present in the coil In this section, a similar model to describe the electrical circuit of the free- standing thin rectangular sensor platforms is presented. Free-standing magnetoelastic sensor platforms are allowed to freely oscillate in the presence of a biasing field. The biasing field moves the sensor response into the linear region of strain and magnetic field. S. Butterworth [6] used equations for changes in magnetic flux in the solenoid coil and arrived at an equation of the form: 93 27)-(4 11 ? ? ? ? ? ? ? ? += mc ZZ VI Thus, the sensor in the solenoid coil consists of two impedances (Z c and Z m ) in parallel. In addition to these two impedances, Butterworth also defined another impedance Z l . The three different impedances described by Butterworth were called: 1. Leakage Impedance (Z l ): The impedance of the coil when the sensor platform is not present in the coil. 2. Core Impedance (Z c ): The contribution of the permeability of the magnetoelastic sensor platform to the impedance of the coil. 3. Motional Impedance (Z m ): The contribution of the oscillating magnetoelastic sensor platform in the circuit. The equivalent circuit described by Butterworth with the inclusion of the self capacitance of the coil is shown in Figure 4-6. This self capacitance was not considered by Butterworth in his work. It is important to note that the effects of eddy currents on the sensor platform?s responses were neglected. This assumption is reasonable because of the small cross-sectional area of the sensor that is parallel to the applied magnetic fields. The contributions due to the eddy currents could be incorporated into the circuit by modifying the expressions for Z c . Hence, the overall equivalent circuit representing the oscillations of the magnetoelastic sensor is shown in Figure 4-6. To establish the validity of the model, simulations were performed for the above circuit using PSPICE. 94 95 Figure 4-6: Equivalent circuit for the sensor oscillations. The AC voltage source with an amplitude of 1 V was arbitrarily chosen to obtain a frequency spectrum. When the sensor is placed in the solenoid coil, in the absence of a biasing magnetic field, the inductance of the coil is increased. This increase is due to the presence of a ferromagnetic material in the coil. The increased inductance results in a decrease in the self resonant frequency of the solenoid coil. The placement of the sensor in the coil will not alter the self capacitance (C s ) or the DC resistance (R l ) of the coil. The core inductance (L c ) can then be calculated using equation 4-17. The values for the motional inductance (L m ) and motional capacitance (C m ) were calculated from the measured frequency spectrum for the sensor platform. A sensor platform with dimensions of 500?100?8 ?m was used to calculate the values for L m and C m . The measured frequency spectrum shown in Figure 4-7 has a maxima and minima for impedance. The two frequency values corresponding to the maxima and minima of the frequency spectrum were used to calculate the values for L m and C m . The values for the leakage resistance (R l ), leakage inductance (L l ) and self capacitance (C s ) used in the models were determined by methods described in section 4.3.1. 96 The calculations of the core resistance (R c ) and motional resistance (R m ) are very complex. For mathematical simplicity, values were chosen by trial and error to match the measured signal damping. It is important to mention that the values of Z c and Z m used in this model were constant. However, according to the equations described by Butterworth, these values are frequency dependent. In the vicinity of the resonance frequency, this assumption is reasonable because the frequency dependence of Z c and Z m is small. Table 4-1 summarizes the results of the model for a coil with a self resonance frequency of 18.7 MHz. This table lists all the calculated values for components of the equivalent circuit. The calculated values (Table 4-1) for the components of the equivalent circuit were then input in the PSPICE program. The frequency response obtained from the PSPICE model was compared with the measured frequency response for a 500?m (Figure 4-7) long sensor. The frequency response obtained from the PSPICE model matched the experimental measurements. This result validates the equivalent circuit model (Figure 4-6) for the measurement of magnetoelastic sensor responses. Table 4-1: Calculated values for components of the equivalent circuit. Component Value Calculation method R l 19.5 ? DC multimeter. L l 10.5 ?H By adding parallel capacitor and calculating using method described in Section 4.1.1. C s 5.4 pF By adding parallel capacitor and calculating using method described in Section 4.1.1. R c 1 ? Chosen by trial and error to match signal damping. L c 0.5 ?H Calculated by placing the sensor in the coil in the absence of DC magnetic bias and from resonance frequency measurements describes in Section 4.1.1. R m 10 ? Chosen by trial and error to match signal damping. L m 100 ?H Calculated using measured resonance and anti-resonance frequency of sensor. C m 14.25 pF Calculated using measured resonance and anti-resonance frequency of sensor. Figure 4-7: Comparison of actual signal and that obtained from PSPICE model. 97 98 4.3.3 Enhancement of sensor responses In this section, a modification to the measurement setup is described that enhances the signal amplitude of the resonance frequency response (enhanced oscillation displacements) of magnetoelastic sensors. A solenoid coil measures changes in magnetic flux caused by changes in the magnetization of the oscillating magnetoelastic sensor platform. A sensor with smaller dimensions will produce smaller changes in the magnetic flux. This will affect the frequency response of the sensors. The smaller sized sensors had lower amplitudes of the resonance peaks. During the development of the equivalent circuit (section 4.3.2), it was observed that the addition of a capacitor in series with the solenoid coil resulted in enhanced signal amplitudes. A variable capacitor was added in series to the solenoid coil. By tuning the variable capacitor, the series resonance frequency of the circuit changes. When this series resonance frequency of the circuit is equal to the sensor?s mechanical resonance, an increase in the amplitude of the resonance peak was observed. The addition of the tuned capacitor in the circuit resulted in an increase in the signal amplitudes. The signal amplitudes increased by 6 times for a 200 ?m length sensor and by 20 times for a 500 ?m length sensor. Figure 4-8 shows a comparison of the S 11 frequency responses of a tuned and an untuned circuit for a 200 ?m length sensor. 99 In Water 10600000 10800000 11000000 11200000 11400000 0.970 0.975 0.980 0.985 0.990 0.995 1.000 Norma l iz ed R e lative power loss Frequency (Hz) Untuned Circuit Tuned Circuit Figure 4-8: Comparison of frequency responses of a 200 ?m sensor in water with a tuned and an untuned circuit. 4.4. Discussion In the late 1920?s and early 1930?s several authors published their work describing the theoretical aspects of magnetostrictive oscillators G. W. Pierce [2] was the first to provide a detailed theoretical understanding of magnetostrictive oscillators. Further work by J. H. Vincent [3], D. S. Muzzey Jr. [4], F. D. Smith [5], S. Butterworth [6] and J. M. Ide [7] studied different geometries of the magnetostrictive oscillator, using the theory described by G. W. Pierce. G.W. Pierce used a variable capacitor in series with the solenoid coil to determine the resonance frequency of the sensor platform. By tuning the capacitance to a critical 100 value, he observed a sharp change in the ?plate current.? He used the values of the coil inductance and the critical value of the capacitance to evaluate the resonance frequency of the sensor platform. An enhanced sensor response due to the addition of a capacitor in series with the solenoid coil was discussed in section 4.3.3. Theoretically an inductor and capacitor in series will have a low impedance value at resonance. The low value of impedance would result in a higher current in the coil. The higher current in the coil produces larger oscillations on the sensor. The increased oscillations (elastic strains) of the sensors is hypothesized to be the reason for the observed enhanced responses (section 4.3.3). In summary, an equivalent circuit model has been developed to describe the behavior of the measurement system and magnetostrictive platform. Methods to calculate the circuit parameters have been demonstrated. The use of a tuned circuit was shown to enhance the sensor?s resonance peak amplitude. Additional work needs to be performed to verify that the increased resonance peak amplitudes are the result of larger elastic oscillations of the sensor. Measurement of these larger elastic oscillations maybe confirmed using a laser vibrometer. 101 REFERENCES 1. "Agilent 4395A Network/Spectrum/Impedance Analyzer Operation Manual," Agilent Technologies. 2. Pierce, G.W., Magnetostriction Oscillators. Proceedings of the IRE, 1929. 17(1): 42-88. 3. Vincent, J.H., Further experiments on magnetostrictive oscillators at radio- frequencies. Proceedings of the Physical Society, 1931. 43(2): 157-165. 4. Muzzey, D.S., Some Measurements of the Longitudinal Elastic Frequencies of Cylinders Using a Magnetostriction Oscillator. Physical Review, 1930. 36(5): 935-947. 5. Smith, F.D., The magnetostriction constant for alternating magnetic fields. Proceedings of the Physical Society, 1930. 42: 181-191. 6. Butterworth, S. and Smith, F.D., The equivalent circuit of the magnetostriction oscillator. Proceedings of the Physical Society, 1931. 43: 166-185. 7. Ide, J.M., Magnetostrictive alloys with low temperature coefficients of frequency. 1934. 22(2): 177-190. 8. Dapino, M.J., Smith, R.C., Calkins, F.T., and Flatau, A.B., A Coupled Magnetomechanical Model for Magnetostrictive Transducers and its Application to Villari-Effect Sensors. Journal of Intelligent Material Systems and Structures, 2002. 13(11): 737-747. 102 5. RESULTS AND DISCUSSION 5.1. Outline This chapter begins with the results of experiments designed to validate the theoretical behavior of the magnetostrictive platforms. Experiments were conducted to measure the effects of size, mass, and mass loading on the resonance frequency of the sensor platforms (Section 5.2.1). In section 5.2.2 the relationship between the biosensor?s resonance frequency and exposure to a specific concentration of Salmonella typhimurium bacteria in a static system is established. Section 5.2.3 describes the results of experiments to determine the saturation response times of the biosensor when exposed to two different concentrations of Salmonella typhimurium using a flowing system (section 5.2.3). In section 5.3, the biosensor?s specificity (section 5.3.1), longevity (section 5.3.5) and dose response in different media (water (section 5.3.2), spiked milk, spiked apple juice (section 5.3.3) and in bacterial mixtures (section 5.3.4)) are discussed. In section 5.4, the dose response of biosensors of 4 different lengths (L=5 mm, 2 mm, 1 mm and 500 ?m) in response to increasing concentrations of S. typhimurium is presented and discussed. In section 5.5, electron microscopy results are described that detail the structure and distribution of immobilized phage on the sensor surface. Different immobilization conditions were also investigated in these experiments. 5.2. Preliminary studies 5.2.1 Frequency response and mass sensitivity The resonance frequencies of sensor platforms with different lengths were recorded. Sensor platforms, ranging in lengths from 500 ?m to 30 mm, were hand-cut from the as-received METGLAS TM 2826MB. The measured values of the resonance frequency were compared with the theoretical values (equation 4-14). Figure 5-1 shows a comparison between the theoretical values and the experimentally measured values. The measured values of the resonance frequency for the sensor platforms matched fairly well with the theoretically predicted values. 0.1 1 10 100 0.1 1 10 Length (mm) F r e que nc y ( M H z ) Theoretical Prediction f = 2.05 (1 / L (in mm)) Mhz Figure 5-1: Frequency dependence on length (200 ?m-10 mm) of magnetoelastic sensors. 103 Another experiment was done to study the change in the resonance frequency caused by the addition of mass on the sensor surface. Mass was added to the sensor surface by sputtering gold onto the sensor platform. The resonance frequency of the sensor platform was measured before and after the gold deposition. Figure 5-2 shows how the resonance frequency changes for a typical magnetoelastic sensor after the sputtering of gold onto the sensor platform. The resonance frequency of the sensor platform was found to decrease in response to the added mass. 0 0.005 0.01 0.015 0.02 0.025 0.03 72500 73000 73500 74000 74500 75000 75500 Frequency (Hz) Signal (V) Before Gold CoatingAfter Gold Coating Figure 5-2: Typical frequency response of a magnetoelastic sensor before and after addition of thin film of gold. Theoretical equation 4-15c predicts that the mass sensitivity of magnetoelastic sensor platforms is higher for sensors with smaller initial mass (smaller dimensions). To experimentally establish this relationship, resonance frequency measurements were made 104 on sensor platforms of different lengths ranging from 200 ?m to 30 mm. Sputtered thin films of gold were used to simulate mass loading on the sensor surface. Gold films were chosen because they can be uniformly deposited and are chemically inert and oxidation resistant. Five successive layers of gold (3 nm each in thickness) were sputtered on the sensor platforms. Resonance frequencies were measured before and after the deposition of each layer. The mass added (?m gold ) was calculated as a product of the density and the volume of gold deposited on the surface. Figure 5-3 shows the frequency shifts (?f) obtained for successive layers of gold added to three different sensors (L=1.3 mm, 1.0 mm and 0.5 mm). The mass sensitivity (?f/?m) was calculated as the slope of ?f vs ?m gold curve. ? f 1.3 = 0.0159?m Gold ? f 1.0 = 0.0444?m Gold ? f 0.5 = 0.0913?m Gold 0.00000 1.00000 2.00000 3.00000 4.00000 5.00000 6.00000 7.00000 8.00000 0 50 100 150 200 250 300 Mass of gold added (?m Gold ) F req uency Sh i f t | ? f | (k H z ) 1.3 mm 1 mm 0.5 mm Linear (1.3 mm) Linear (1 mm) Linear (0.5 mm) Figure 5-3: Frequency shifts calculated as function of mass of gold added to the sensor surface. The three different lengths shown are L=1.3, 1.0 and 0.5 mm. 105 The results of the mass sensitivity measurements are shown in Figure 5-4 with ?f/?m (Hz/pg) as the ordinate and length (mm) as the abscissa. The theoretical equation shown in Figure 5-4 was calculated by substituting for f and M in equation 4-15c. The thickness (28 ?m) and the length to width ratio (5:1) of all the sensor platforms used in this experiment, were held constant. The plot clearly shows that sensor platforms with smaller physical dimensions have a higher mass sensitivity. It is important to note that for the sensitivity study described in this section, the frequency response measurements were performed in air. 0.000001 0.00001 0.0001 0.001 0.01 0.1 1 10 0.1 1 10 100 Length (mm) | ? f/ ? m (H z / p g )| Theoretical df/dm ( Hz/pg) = 0.0434[Length (mm)] -3 Figure 5-4: Dependence of length on the mass sensitivity (?f/?m) of magnetoelastic sensors with air as surrounding media. 106 107 5.2.2 Sensor response to S. typhimurium in static conditions In this section, the change in resonance frequency of the biosensor upon exposure to S. typhimurium was studied. This experiment was performed using static test procedures described in section 3.8.1. The biosensor was exposed to S. typhimurium (5?10 8 cfu/mL) for 30 minutes. The resonance frequency of the biosensor was measured before and after the exposure. The total number of bacteria attached to the sensor surface was counted from the SEM micrographs of different regions on the biosensor surface. The measured frequency responses and the SEM micrographs for a 2mm length biosensor are shown in Figure 5-5 and Figure 5-6, respectively. The attachment of bacterial mass to the biosensor resulted in a measured frequency shift of 1290 Hz. The total number of S. typhimurium attached (counted from SEM micrographs) to the biosensor was 1.232?10 5 cells. This bacterial mass attachment corresponds to a theoretical frequency shift of 1346 Hz. The resonance frequency shifts obtained for 5 different biosensors using this procedure are summarized in Table 5-1. 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1090000 1095000 1100000 1105000 1110000 1115000 1120000 1125000 Frequency (Hz) Refl ected P o wer l o s s ( a .u.) t = 0 minutes t = 30 minutes ? f = 1290Hz Figure 5-5: Typical response of a 1?0.2?0.015 mm sensor to 5?10 8 cfu/mL of S. typhimurium at different time intervals (t=0, 10, 20, 30, 40 and 50 minutes). A total frequency shift (?f) of 1290 Hz was observed. 108 109 Figure 5-6: SEM pictures (Magnification: 1000X) of two different regions on the sensor (2?0.4?0.015 mm) with a frequency shift of 1290 Hz. 110 Table 5-1: Comparison of number of bacterial cells attached estimated from frequency shifts obtained and the extrapolated number calculated from SEM images. Measured Frequency Shift (Hz) Number of cells estimated from frequency shifts Total number of cells from SEM Estimated Salmonella typhimurium mass (pg) Sensor 1 1291 118107 123200 1.987 Sensor 2 1163 106397 152000 1.451 Sensor 3 1048 95876 142000 1.399 Sensor 4 973 89014 118600 1.556 Sensor 5 947 86636 115200 1.559 From the results shown in Table 5-1 it was confirmed that the measured frequency shifts were due to the attachment of S. typhimurium to the biosensor. The mass of S. typhimurium was estimated from the observed experimental frequency shifts. Equation 4-15a was used to calculate the mass of one S. typhimurium cell. The dimensions of the sensors used in this calculation were 2?0.4?0.015 mm. The average mass of Salmonella typhimurium obtained from the calculation was 1.59?0.23 pg. 5.2.3 Saturated response time This experiment was performed to study the time required to achieve a saturated frequency response of the biosensor upon exposure to S. typhimurium. The biosensors were placed in the center of a solenoid coil. Water was then allowed to flow over the sensor at a flow-rate of 50 ?L/min. Fifteen minutes after obtaining a steady-state resonance frequency in flowing water, the biosensor was exposed to 1mL of S. typhimurium suspension (5?10 8 cfu/mL). Figure 5-7 shows the changes in the resonance frequency of the biosensor as a function of time. The introduction of the S. typhimurium (5?10 5 cfu/mL) to the biosensor resulted in a gradual decrease in the resonance frequency during the first 20 minutes of exposure. After 20 minutes, the resonance frequency of the biosensor reached saturation. Similarly, when the biosensor was exposed to S. typhimurium of 5?10 8 cfu/mL concentration, the saturated resonance frequency response was achieved in 30 minutes. Hence, for all the static tests performed, the sensors were exposed to S. typhimurium for more than 30 minutes to ensure saturation. 0.997 0.9975 0.998 0.9985 0.999 0.9995 1 1.0005 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 Time (minutes) No rma l i z ed Fr eq ue nc y ( f / f 0 ) ? f = 6k Hz ? f = 12k H z 5?10 5 cfu/ml 5?10 8 cfu/ml Control Sensor Figure 5-7: Responses of three different sensors (L=500 ?m) to S. typhimurium. An average saturation response could be seen at the end of 30 minutes. 111 112 5.3. Biosensor Characterization 5.3.1 Sensor specificity In evaluating the performance of a biosensor, it is essential to establish specificity (cross-reactivity with other pathogenic species) of the immobilized bio-recognition element. The response of the biosensor upon exposure to S. typhimurium, E. coli, S. Enteritidis and L. monocytogenes was studied. A significantly lower affinity of the immobilized phage to pathogens other than S. typhimurium will establish the biosensor?s specificity. The biosensors were exposed to each pathogen using the static test procedure. In the section 5.2.3, it was established that the biosensor resonance frequency achieved saturation in 30 minutes. In this experiment the biosensor was exposed to the pathogenic solution for 45 minutes. Figure 5-8 shows the measured values of normalized area coverage density, ?f measured and ?f SEM obtained for the biosensors. These values were compared for biosensors exposed to the different pathogens. The normalized area coverage density shown in Figure 5-8 was calculated as the ratio of the area coverage density of a certain pathogen (N P ) to the area coverage density of S. typhimurium (N ST ). The ?f measured and ?f SEM were calculated using procedures described in section 3.7. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Salmonella typhimurium Salmonella enteridis Eschericia coli Listeria monocytogenes Nor mali z ed a r ea co ver age de ns it y ( N / N ST ) 0 120 240 360 480 600 720 840 960 1080 1200 F r e que nc y Shi ft ( H z ) Bacterial Counts ?f measured ?f SEM Figure 5-8: Specificity of phage-immobilized sensors exposed to different pathogens (5?10 8 cfu/mL). The normalized area coverage density was calculated from SEM photomicrographs of the sensor surface (an average of 5 sensors each). ?f measured and ?f SEM are shown on the right side. The normalized area coverage density of the biosensors exposed to S. typhimurium, S. enteritidis, E. coli and L. monocytogenes were 1.00, 0.17, 0.06 and 0.03, respectively. Large differences in the distribution density of different pathogens bound to the biosensor?s surface can be observed in the SEM micrographs (Figure 5-9). The values of resonance frequency shifts obtained upon exposure to different pathogens followed the same trend as the normalized distribution density. The values of ?f measured were slightly higher than the values of ?f SEM . The difference in the values of ?f measured 113 and ?f SEM can be attributed to extrapolations used in the calculation of ?f SEM . Figure 5-8 and Figure 5-9 establish that the biosensor had a significantly higher affinity towards S. typhimurium. (D) (C) (A) (B) Figure 5-9: Typical SEM images of phage-immobilized sensors exposed to (A) S. typhimurium; (B) S. enteritidis; (C) E. coli and (D) L. monocytogenes. 5.3.2 Biosensor dose response The dose response of magnetoelastic biosensors (Dimensions: 2?0.4?0.015 mm) to various concentrations of S. typhimurium suspensions was studied. The dose response 114 tests were carried out using procedures described in section 3.8.4. Figure 5-10 shows a typical resonance frequency response as a function of time for a biosensor (L=2 mm) exposed to different concentrations of bacterial suspensions. The biosensor?s resonance frequency was recorded every 2 minutes. Exactly 1 mL of each concentration was allowed to flow over the sensor at a flow rate of 50 ?L/min. At this flow rate, it takes 20 minutes for 1mL of the analyte to flow over the biosensor. As described in section 3.8.4, the resonance frequency at the end of every 20 minutes was used to calculate the resultant frequency shift for that particular concentration. 1.0685 1.069 1.0695 1.07 1.0705 1.071 1.0715 1.072 0 20 40 60 80 100 120 140 160 180 Time (minutes) Fr eq ue nc y (M Hz ) 1.047 1.0475 1.048 1.0485 1.049 1.0495 1.05 1.0505 Fr eq ue nc y (M Hz ) 5?10 2 cfu/ml 5?10 4 cfu/ml 5?10 3 cfu/ml 5?10 6 cfu/ml 5?10 1 cfu/ml 5?10 7 cfu/ml 5?10 5 cfu/ml WATER 5x10 8 cfu/ml Control Sensor Test Sensor Figure 5-10: Typical dynamic response curve for a sensor with dimensions 2?0.4?0.015 mm. 1 mL of each concentration of bacterial suspension was allowed to flow over the sensor at a flow rate of 50 ?L/min. Control sensor response shown is of a sensor devoid of phage. 115 116 There was no change in the resonance frequency upon exposure to the lowest concentrations (5?10 1 cfu/mL and 5?10 2 cfu/mL) of S. typhimurium. The first measurable decrease in the resonance frequency occured when the biosensor was exposed to a concentration of 5?10 3 cfu/mL of S. typhimurium. The resonance frequency decreased with the introduction of each successive concentration (5?10 4 cfu/mL through 5?10 8 cfu/mL) of S. typhimurium. The control sensor that was devoid of any phage on the sensor surface had a negligible change in its frequency even upon exposure to very high concentrations (5?10 8 cfu/mL) of bacteria. In order to confirm the results obtained from frequency shifts (Figure 5-11), pictures were taken of different regions on the assayed biosensors using SEM. Intentionally interrupted tests were performed to visually verify (using SEM) that a lower number of S. typhimurium were bound to biosensors after exposure to lower concentration solutions. Interrupted tests were performed by intentionally stopping the dose response tests once the biosensors were exposed to all concentrations up to 5?10 6 cfu/mL in one test and up to 5?10 3 cfu/mL in another test. The SEM micrograph shown in Figure 5-12(a) is of an assayed biosensor that had gone through an uninterrupted dose response test (exposure to 5?10 1 cfu/mL through 5?10 8 cfu/mL). SEM micrographs for the biosensors subjected to the two interrupted dose response tests are shown in Figure 5-12(b) and Figure 5-12(c). Figure 5-12(d) is an SEM micrograph of the control sensor (devoid of bio-recognition element) after a completed dose response test. A lower area coverage density (Figure 5-12 (a-c)) was observed for lower concentrations of S. typhimurium. The negligible change in resonance frequency of the control sensor was further confirmed by the nominal binding observed in the SEM micrograph (Figure 5-12 (d)). 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 10 1 0 200 400 600 800 1000 L = 2mm Sensor Control Sensor F requ ency Shi f t | ? f | (H z ) Concentration (cfu/ml) Figure 5-11: Magnetoelastic biosensor?s responses, when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions on test sensors (2?0.4?0.015 mm) (?- Test sensor: sigmoidal fit ? 2 =0.048, R 2 =0.99) and control (2?0.4?0.015 mm) (? - Control sensor). Each data point is the average value obtained from five individual experiments (different sensors) carried out under identical conditions. 117 (d) (c) (b) (a) Figure 5-12: Typical SEM images depicting S. typhimurium attachment to phage- immobilized magnetoelastic sensor surface. Sensors exposed to S. typhimurium at concentrations of (a) 5?10 8 cfu/mL, (b) 5?10 6 cfu/mL (c) 5?10 3 cfu/mL and (d) control (biosensor devoid of phage and treated with 5?10 1 cfu/mL through 5?10 8 cfu/mL of bacterial sample). 118 119 5.3.3 Biosensor response in real food The detection of S. typhimurium suspended in different media (water, milk and apple juice) was essential to establish the field applicability of magnetoelastic biosensors. The sensors (2?0.4?0.015 mm) were exposed to milk and apple juice spiked with increasing concentrations of S. typhimurium (5?10 1 through 5?10 8 cfu/mL). The biosensor?s dose responses to spiked samples were evaluated using procedures described in section 3.8.5. The first detectable shift in resonance frequency occured at a concentration of 5?10 3 cfu/mL. An average of five different sensor responses was used to construct the dose response curves depicted in Figure 5-13. The dose response of the sensors exposed to spiked milk, apple juice and water samples was similar except at higher concentrations. The resonance frequency shifts obtained for spiked milk samples were lower than that of spiked water and spiked apple juice samples. The dose response was linear over five aliquots of concentrations (5?10 3 through 5?10 7 cfu/mL) for the three different media. The sensitivity of the biosensor was calculated as the slope of the linear region of the dose response curve (Hz per decade of concentration change). The sensitivity of biosensors exposed to spiked water, apple juice, and milk was 161 Hz/decade, 155 Hz/decade and 118 Hz/decade, respectively. The control sensor (Figure 5-13) had a negligible change in resonance frequency in response to even high concentrations of S. typhimurium. The control sensor showed a maximum resonance frequency shift of 50 Hz, while a maximum resonance frequency shift of 980 Hz was observed for the biosensor. This significant difference in the measured frequency shifts (control versus measurement sensor) indicates negligible, non-specific binding of bacteria to the bare gold surface. 120 A Hill plot constructed from the dose response data was used to study the kinetics of the Salmonella-phage binding reaction. This plot was constructed using methods described in section 3.9. A Hill plot, constructed using the three different dose response curves, is shown in Figure 5-14. A binding valency of 2.4, 2.5 and 2.3 were calculated for the biosensors in response to spiked water, milk and apple juice, respectively. These values indicate that the binding of S. typhimurium to immobilized phage was multivalent in nature. More than two phage binding sites participated in the capture of one S. typhimurium cell. The values of K d (apparent) obtained for biosensors exposed to spiked water, milk and apple juice were 1.82?10 5 , 2.51?10 5 and 2.16?10 5 cfu/mL, respectively. The higher value for K d (apparent) obtained for biosensors exposed to milk is also evident from the lower resonance frequency shifts observed. The lower frequency shifts and higher K d (apparent) values obtained for biosensors exposed to spiked milk samples is hypothesized to result from the milk proteins blocking of some of the available binding sites. 121 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 -100 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 Figure 5-13: Comparison of dose responses of magnetoelastic biosensor (2?0.4?0.015 mm), when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions in water ((?) ? 2 =0.442, R 2 =0.99), apple juice ((?) ? 2 =0.237, R 2 =0.99) and fat free milk ((?) ? 2 =0.194, R 2 =0.99). Control (?) represents the uncoated (devoid of phage) sensor?s response. The curves represent the sigmoid fit of signals obtained. Frequency Shift | ? f | ( H z ) Concentration (cfu/ml) Control Milk Water Apple Figure 5-14: Hill plots of binding isotherms showing the ratio of occupied and free phage sites as a function of bacterial concentrations spiked in different food samples. The straight line is the linear least squares fit to the data (water (?): slope=0.40?0.03, R=0.97; fat-free milk (?): slope=0.41?0.04, R=0.98; apple juice (?) slope=0.36, R=0.96). SEM micrographs of assayed biosensors were used to provide visual verification of measured experimental data shows SEM micrographs of the biosensor surfaces after dose response testing to S. typhimurium. For biosensors exposed to spiked milk samples, the binding appeared lower than that for spiked water or spiked apple juice samples. The smaller number of bound S. typhimurium cells explains the lower values of frequency shifts obtained for milk samples. The milk proteins can be seen as white spots in the SEM micrograph for the biosensor exposed to spiked milk (Figure 5-15(a)). In this test, 122 it was established that the biosensors can detect the concentrations of S. typhimurium present in both water and food samples. (d) Figure 5-15: Typical SEM images of S. typhimurium bound to a magnetoelastic biosensor surface (2?0.4?0.015 mm) in (a) fat-free milk, (b) water (d) apple juice and (c) control (biosensor devoid of phage and treated with 5?10 8 cfu/mL of bacterial sample). 123 124 5.3.4 Selectivity in the presence of high concentrations of masking bacteria The effect that masking bacteria has on the measurement of S. typhimurium on the response of the biosensor was studied. This was done by exposing the biosensors to three different sets of prepared suspensions: 1). S. typhimurium, 2). S. typhimurium + E. coli and 3). S. typhimurium + E. coli + L. monocytogenes). The test procedures described in section 3.8.5 were used to obtain the dose response curves. The obtained dose response curve is shown in Figure 5-16. The biosensor?s dose response had similar sigmoidal trends. The dose response curves indicated that the frequency shifts for biosensors exposed to the two mixtures (suspensions 2 and 3) were lower than the ones exposed to S. typhimurium alone. However, there was no significant difference in response between the mixtures with one or two masking bacteria. The dose responses for biosensors exposed to the three suspensions were linear over the range of 5?10 3 cfu/mL to 5?10 7 cfu/mL. The sensitivity of the biosensor exposed to S. typhimurium in the absence of masking bacteria was 161 Hz/decade. In presence of one masking bacteria (E. coli) and two masking bacteria (E. coli and L. monocytogenes) in the mixtures the sensitivity was 131 Hz/decade and 127 Hz/decade (R 2 =0.97), respectively. 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 10 10 0 200 400 600 800 1000 1200 1400 F r e que nc y shi f t | ? f | ( H z ) In S. typhimurium only With E. coli With E. coli+L. monocytogenes Control Concentration (cfu/ml) Figure 5-16: Dose response curve of magnetoelastic sensors (2?0.4?0.015 mm) in response to S. typhimurium in mixture with other masking bacteria. Sensors (each data point is an average of the response from five sensors) exposed to only S. typhimurium ( ?-? 2 =0.44, R 2 =0.99), S. typhimurium in mixture with E. coli (?-? 2 =0.18, R 2 =0.99), and S. typhimurium in mixture with E. coli + L. monocytogenes (?-? 2 =0.24, R 2 =0.99). The control (?-? 2 =0.048, R 2 =0.99), is the response of an uncoated (devoid of phage) sensor. The curves represent the sigmoidal fit of signals obtained. 125 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 S. typhimurium S. typhimurium+E. coli S. typhimurium+E. coli+L. monocytogenes lo g ( Y/ 1 - Y) Concentration (cfu/ml) Figure 5-17: Hill plot constructed from the dose response curves, showing the ratio of occupied (Y) and free phage sites (1-Y) as a function of bacterial concentrations in different mixtures. The straight line is the linear least squares fit to the data (S. typhimurium (?): slope=0.40?0.03, R 2 =0.97; S. typhimurium + E. coli (?): slope=0.33?0.02, R 2 =0.98; and S. typhimurium + E. coli + L. monocytogenes (?) slope=0.34?0.02, R 2 =0.97). 126 127 A hill plot (Figure 5-17) was constructed using the dose response curves [1-4] to determine the binding kinetics. The binding valency was similar for all the three prepared suspensions. The binding valencies obtained from the hill plots were 2.42, 2.79 and 2.91 for the suspension 1, suspension 2 and suspension 3, respectively. This reaffirms the multivalent nature of phage-Salmonella interaction on the biosensors. The biosensors exposed to S. typhimurium in the presence of masking bacteria had a higher value of the apparent dissociation constant. The lower sensitivity and the higher dissociation constants obtained for the mixtures can be attributed to the presence of the masking bacteria. The presence of masking bacteria in the mixtures reduces the probability of interaction between S. typhimurium and the binding sites on the biosensor. A summary of results from the calculations for sensitivity, binding valency and apparent dissociation constant for the assayed biosensors discussed in section 5.3.3 and section 5.3.4 are presented in Table 5-2. Table 5-2: The sensitivity, dissociation constant and binding valency of magnetoelastic sensors in different bacterial mixtures. Bacterial mixtures Sensitivity (Hz/decade) Binding valency (1/n) K d (cfu/mL) K d(apparent) =K d n (cfu/mL) S. typhimurium 161 2.42 149 1.82?10 5 S. typhimurium + E. coli 131 2.79 82 2.19?10 5 S. typhimurium + E. coli + L. monocytogenes 127 2.91 87 4.41?10 5 Spiked Apple Juice 155 2.77 89 2.51?10 5 Spiked Milk 118 2.5 136 2.16?10 5 128 The detection limit of the biosensor (2?0.4?0.015 mm) in all the above described tests was 5?10 3 cfu/mL. The biosensor was capable of detecting small amounts of S. typhimurium, even in the presence of high concentrations of masking bacteria. In the preceding sections, it was established that the biosensor could detect S. typhimurium with high specificity and selectivity. 5.3.5 Longevity of magnetoelastic biosensors In this section, the longevity/thermal stability of the magnetoelastic biosensors was studied at three different temperatures (25 ?C, 45 ?C and 65 ?C). The transducer and bio-recognition element of a typical diagnostic biosensor should to be robust to endure the variations in temperatures that they might experience in real applications. Conventionally used antibodies face a major disadvantage, due to their poor performance at higher temperatures. Antibodies are known to lose their activity at 25 ?C in less than 20 days [5-8]. However, filamentous phage has been shown to display better stability in adverse conditions of temperature and pH [9-12]. The procedures used for the longevity testing have been described in section 3.8.3. All of the assayed sensors were prepared for SEM (section 3.5) and the area distribution density as a function of time and temperature were evaluated. Figure 5-18, Figure 5-19 and Figure 5-20 show typical SEM images of the assayed biosensors incubated at 65 ?C, 45 ?C and 25 ?C, respectively. The area coverage density was observed to decrease with increasing time and temperature. The number of cells present on the sensor was then counted using the procedures described previously (section 3.7) 1 ?m 1 ?m 1 ?m 1 ?m 1 ?m1 ?m 1 ?m Day-1 Day-5 Day-15 Day-34 Day-62 Day-3 Figure 5-18: Typical SEM images of S. typhimurium bound to the phage-immobilized biosensor surface with increasing time (1, 3, 5, 15 34 and 62 days) at 65 ?C. 129 1 ?m Day-62 1 ?m 1 ?m 1 ?m 1 ?m1 ?m Day-1 Day-3 Day-5 Day-15 Day-34 Figure 5-19: Typical SEM images of S. typhimurium bound to the phage-immobilized biosensor surface with increasing time (1, 3, 5, 15, 34 and 62 days) at 45 ?C. 130 Day-62 Day-34 Day-15 Day-5 Day-1 Day-3 1 ?m 1 ?m 1 ?m 1 ?m 1 ?m 1 ?m Figure 5-20: Typical SEM images of S. typhimurium bound to the phage-immobilized biosensor surface with increasing time (1, 3, 5, 15, 34 and 62 days) at 25 ?C. 131 132 The changes in the surface distribution density of the bacteria bound to the biosensor surface as a function of time (in days) for the different incubation temperatures are shown in Figure 5-21. A gradual decrease in the binding affinity with time can be seen for the biosensors incubated at the three temperatures. However, the biosensors retained 59%, 45% and 33% of their binding affinity at 25 ?C, 45 ?C and 65 ?C, respectively after a period of 63 days. Figure 5-21 also shows the result of a similar experiment done with magnetoelastic sensors immobilized with a polyclonal antibody (data kindly provided by Guntupalli et.al. [8]). The initial (day 1) area coverage densities for phage-immobilized biosensors was approximately two times higher than antibody- immobilized biosensors. However, there was a rapid loss in the binding activity of the antibody-based sensors following day 1; losing all binding activity in less than 10 days even at 45 ?C. On the other hand, phage-immobilized biosensors had significant binding even at the end of 2 months. 133 10110 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 Av era g e A rea Co v era g e D e n s ity ( cells/ ? m 2 ) Time (Days) Phage at 25 ?C Phage at 45 ?C Phage at 65 ?C Antibody at 45 ?C Antibody at 65 ?C Figure 5-21: Surface coverage densities (average number of cells/?m 2 ) calculated from SEM micrographs of stored magnetoelastic biosensors (25 ?C, 45 ?C, and 65 ?C) after exposure to S. typhimurium (5?10 8 cfu/mL). 5.4. Size dependent Sensitivity 5.4.1 Dose response of biosensors with different lengths As discussed previously, an enhanced sensitivity of magetoelastic sensors could be achieved by reducing the physical dimensions of the sensor. Hence we studied the dose response of sensors with different dimensions upon exposure to different concentrations of S. typhimurium. Since the sensitivity of the sensor depends on the 134 length of the sensor, biosensors with dimensions of 5?1?0.015 mm, 2?0.4?0.015 mm, 1?0.2?0.015 mm and 0.5?0.1?0.015 mm were prepared. The dose response tests were conducted (section 3.8.4) for biosensors with the four different dimensions. Typical curves comparing dose responses (each curve represents an average of 5 different biosensors) obtained for 5 mm and 2 mm is shown in Figure 5-22. It can clearly be seen that higher frequency shifts were obtained for a smaller sensor. The detection limit for a 5 mm sensor was 5?10 4 cfu/mL and for a 2 mm sensor was 5?10 3 cfu/mL. The sensitivity of the biosensor in the linear region of the dose response curve increased from 98 Hz/decade to 161 Hz/decade. Similar comparisons of dose response for a reduction in length from 2 mm to 1 mm and from 1 mm to 500 ?m are shown in Figure 5-23 and Figure 5-24, respectively. An increase in sensitivity and reduction in detection limits were obtained for sensors with smaller dimensions. The detection limits went from 5?10 4 cfu/mL for a 5 mm sensor down to 5?10 2 cfu/mL for a 500 ?m sensor. The sensitivity of the response in the linear region was measured to be 770 Hz/decade for a 1mm sensor and 1152 Hz/decade for a 500 ?m sensor respectively. 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 0 200 400 600 800 1000 1200 2 ? 0.4 ? 0.015 mm 5 ? 1 ? 0.015 mm F r e que nc y Shift ? f (Hz) Concentration (cfu/ml) Figure 5-22: Comparison of magnetoelastic biosensor?s dose responses, when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions on two different sizes of sensors 2?0.4?0.015 mm (?- ? 2 =0.048, R 2 =0.99) and 5?1?0.015 mm (?- ? 2 =0.32, R 2 =0.99). The curves represent the sigmoidal fit of signals obtained. 135 10 0 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 0 1000 2000 3000 4000 5000 6000 F requency Shif t | ? f | (H z ) Concentration (cfu/ml) 1 ? 0.2 ? 0.015 mm 2 ? 0.4 ? 0.015 mm Figure 5-23: Comparison of magnetoelastic biosensor?s dose responses, when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions on two different sizes of sensors (1?0.2?0.015 mm (?- ? 2 =0.048, R 2 =0.99) and 2?0.4?0.015 mm (?- ? 2 =0.32, R 2 =0.99)). The curves represent the sigmoidal fit of signals obtained. Each data point is the average value obtained from five individual experiments (different sensors) carried out under identical conditions. 136 10 0 10 1 10 2 10 3 10 4 10 5 10 6 10 7 10 8 10 9 -2000 0 2000 4000 6000 8000 10000 12000 14000 1.0 ? 0.2 ? 0.015 mm 0.5 ? 0.1 ? 0.015 mm Frequency Shift | ? f | (Hz) Concentration (cfu/ml) Figure 5-24: Comparison of magnetoelastic biosensor?s dose responses, when exposed to increasing concentrations (5?10 1 to 5?10 8 cfu/mL) of S. typhimurium suspensions on two different sizes of sensors (0.5?0.1?0.015 mm (?- ? 2 =0.048, R 2 =0.99) and 1?0.2 ?0.015 mm (?- ? 2 =0.7231, R 2 =0.91) The curves represent the sigmoidal fit of signals obtained. Due to the small size of a 500?100?15 ?m biosensor, the entire surface of the biosensor could be imaged using SEM. Five pictures were taken and attached to view the entire surface. SEM micrographs of the assayed biosensors are shown in Figure 5-25. 137 5?10 8 cfu/ml 5?10 4 cfu/ml5?10 2 cfu/ml Figure 5-25: Typical SEM images of the entire surfaces of assayed 500?m sensors at three different concentrations (5?10 2 cfu/mL, 5?10 4 cfu/mL and 5?10 8 cfu/mL). Bound S. typhimurium can be seen as black spots on the pictures. 138 139 A clear decrease in the number of bacteria bound to the sensor can be observed for the biosensors exposed to the three different concentrations (5?10 2 cfu/mL, 5?10 4 cfu/mL and 5?10 8 cfu/mL). The number of bacteria attached to the entire biosensor surface (one side) was counted manually. The total number of attached S. typhimurium cells were 14278, 3286 and 157 for the three different biosensors exposed to concentrations of 5?10 8 , 5?10 4 and 5?10 2 cfu/mL, respectively. An attachment of 14278 cells on the sensor surface theoretically corresponds to a frequency shift of 9.5 kHz. This number is very similar to the measured resonance frequency shift of 10 kHz for the biosensor. A summary of the observations for sensors with different dimensions is shown in Table 5-3. This table lists the sensitivity and detection limits achieved for sensors of each of these dimensions. An average increase in sensitivity by 400% was achieved by reducing the dimension of the biosensor by 50%. It can clearly be inferred that better sensitivity can be achieved by reducing the dimensions of the biosensor. Table 5-3: Table summarizing the sensitivity and detection limits achieved for sensors with different dimensions. Sensor Dimensions Sensitivity (Hz/decade) Order of detection Limit 5.0?1.0?0.015 mm 98 10 5 cfu/mL 2.0?0.4?0.015 mm 161 10 4 cfu/mL 1.0?0.2?0.015 mm 770 10 3 cfu/mL 0.5?0.1?0.015 mm 1150 10 2 cfu/mL 140 5.5. Phage immobilization An understanding of the distribution characteristics of the immobilized phage is essential in achieving better sensitivity and stability of biosensor responses. Olsen et. al. [13] had discussed two possible distributions of phage (assuming no bundle formation) on the sensor surface and calculated an optimum concentration of phage needed for immobilization. In this dissertation, the concentration of phage (5?10 11 vir/mL), suggested by Olsen et. al. [13], was used for immobilization. They also hypothesized that the phage aggregates would be more prevalent rather than individual phage filaments on sensor surfaces. The aggregation of like charged polyelectrolytes (such as phage) to form ?bundles? caused due to the presence of counterions, has been studied extensively [14-16]. However, the application of phage as a bio-recognition element is limited in the literature. There is very little understanding of whether a uniform distribution of phage filaments or a uniform distribution of phage ?bundles? on the surface would be better for a typical sensor application. Also there is limited understanding [17] of the binding interaction between the gold surface and phage filaments. Electron Microscopy was used to study the distribution of phage on the sensor surface. For SEM studies, phage immobilization was carried out using procedures described in section 3.2. The immobilization step was followed by an exposure of the biosensor to 5?10 8 cfu/mL of S. typhimurium for 45 minutes. The assayed sensors were then mounted on aluminum stubs with the help of a double-sided carbon tape and exposed to OsO 4 vapors for 15 minutes. The main challenge in imaging immobilized phage on the biosensor surface was to resolve the nanometer sized phage filaments on the sensor surface. In our earlier experiments, the assayed sensors were prepared for SEM by sputtering a very thin layer of gold (?30 nm) on the sensor surface. However, sputtering of gold films on the sensor surface covered the phage filaments completely. After trying several different thicknesses of the gold film, we discovered that assayed biosensors prepared without a gold film allowed imaging of immobilized phage on the sensor surfaces. The presence of Os diffused into the bacterial cells, the smaller size of the sensors and the use of a low probe current, minimized the charge accumulation on the samples in the SEM chamber. A lower probe current also resulted in a better resolution for the SEM micrographs. Figure 5-26 shows an SEM picture of a sensor surface before and after immobilization of phage. (b ) (a ) Figure 5-26: High magnification (15000X) SEM images of sensor surface (a) before and (b) after immobilization of phage. Upon immobilization of phage (5?10 11 vir/mL suspended in 1X TBS) using physical adsorption on the sensor surface, it was observed (Figure 5-26 (b)) that randomly oriented phage filaments saturated the entire surface. Phage was visible as 141 thread like features observed in the SEM micrograph shown in Figure 5-26. The as- received phage was suspended in 1XTBS solution containing 140 mM of NaCl. To study the effect of the concentration of counterions we varied the concentration of the NaCl (280 mM, 420 mM, 560 mM and 840 mM) in the TBS solution. The different Na + ion concentrations were used in order to facilitate formation of phage ?bundles.? Typical high magnification (15000X) SEM micrographs for each of the sensor surfaces at these concentrations are shown in Figure 5-27 (a) through (d). 142 Figure 5-27: SEM images showing the nature of phage distribution in presence of different Na + ion concentrations (a) 280 mM; (b) 420 mM; (c) 560 mM; (d) 840 mM. 5?10 11 (280mM NaCl) 5?10 11 (420mM NaCl) 5?10 11 (560mM NaCl) 5?10 11 (840mM NaCl) (a) (b) (c) (d) 143 At lower concentrations of the counterion (280 mM and 420 mM of NaCl) a uniform distribution of individual phage filaments was observed (Figure 5-27(a), (b)). However, an increase in the counterion concentration resulted in the formation of phage ?bundles.? The phage ?bundles? formed can be seen in Figure 5-27 (c) and Figure 5-27 (d). The typical distribution of the binding of S. typhimurium on the biosensor surface for the different concentrations of the counterion (Na + ) is shown in Figure 5-28. A uniform distribution of individual filaments on the entire surface enabled a uniform distribution of bound S. typhimurium (Figure 5-28 (a) through (d)) for low counterion concentrations (? 420 mM). SEM micrographs of the sensor surface at a lower magnification indicated that the ?bundles?, unlike the individual phage filaments, were not uniformly distributed over the entire surface. At higher counterion concentrations (> 420 mM), due to localized accumulation of bundles (seen as bright white spots in Figure 5-28 (e) through (h)) the binding of S. typhimurium was also localized. This resulted in an overall reduction of the number of S. typhimurium attached to the sensor surface (Figure 5-28). 144 Figure 5-28: SEM images showing binding distribution on sensors immobilized with phage with varying counterion (Na + ) concentrations (a, b) 240 mM; (c, d) 420 mM; (e, f) 560mM; and (g, h) 840mM. (e) (a) (b) (c) (d) (f) (g) (h) 145 From the studies done so far we can conclude that formation of phage ?bundles? can be achieved by altering the counterion concentration. The SEM observations could mislead to a conclusion that the presence and formation of phage ?bundles? would be detrimental for typical sensor applications. A phage ?bundle? consists of more than thousand individual phage filaments. This would result in large number of binding receptors to be concentrated on each phage ?bundle?. However the localized accumulation of these ?bundles? resulted in a decrease in the number of attached S. typhimurium cells. A hypothesis for the significant difference in the nature of distribution of phage is the difference in effective concentration of the bio-recognition probe. When individual phage filaments are present, each filament is a bio-recognition probe. Each phage ?bundle? would consist of several thousands of individual phage filaments hence the effective number of these ?bundles? available for binding would be lower than that for individual phage filaments. Additionally, these ?bundles? have a tendency to accumulate at certain regions on the sensor surface rather than a uniform distribution. Additional work is required to achieve a uniform distribution of the bundles and to ascertain the nature and conditions that would optimize the performance of phage as a bio-recognition probe for sensor applications. 146 REFERENCES 1. Guntupalli, R., Lakshmanan, R.S., Johnson, M.L., Hu, J., Huang, T.S., Barbaree, J.M., Vodyanoy, V.J., and Chin, B.A., Magnetoelastic biosensor for the detection of Salmonella typhimurium in food products. Sensing and Instrumentation for Food Quality and Safety, 2007. 1 (1): 3-10. 2. Nanduri, V., Sorokulova, Iryna B., Samoylov, Alexandre M., Simonian, Aleksandr L., Petrenko, Valery A., Vodyanoy, Vitaly., Phage as a molecular recognition element in biosensors immobilized by physical adsorption. Biosensors and Bioelectronics, 2007. 22(6): 986-992. 3. Nanduri, V., Kim., G., Morgan, M.T., Ess, D., Hahm, B.-K., Kothapalli, A., Valadez, A., Geng, T., and Bhunia, A.K., Antibody Immobilization on Waveguides Using a Flow?Through System Shows Improved Listeria monocytogenes Detection in an Automated Fiber Optic Biosensor: RAPTOR?. Sensors, 2006. 6: 808-822. 4. Petrenko, V.A. and Vodyanoy, V.J., Phage display for detection of biological threat agents. Journal of Microbiological Methods, 2003. 53(2): 253-262. 5. Van der Linden, R.H., L.G.F., de Geus, B., Harmsen, M.M., Ruuls, R.C., Stok, W., de Ron, L., Wilson, S., Davis, P., and Verrips, C.T., Comparison of physical chemical properties llamma VHH antibody fragments and mouse monoclonal antibodies. Biochimica Biophysica Acta, 1999. 1431: 37-46. 147 6. Usami, A., A.O., Takahama, S., and Fujii, T., The effect of pH, hydrogen peroxide and temparature on the stability of human monoclonal antibody. J. Pharm. Biomed. Anal, 1996. 14: 1133-1140. 7. Dooley, H., S.D.G., Harris, W.J., and Porter, A.J., Stabilization of antibody fragments in adverse environments. Biotechnol. Appl. Biochem, 1998. 28: 77-83. 8. Guntupalli, R., Lakshmanan, R., Wan, J., Kim, D.J., Huang, T., Vodyanoy, V., and Chin, B., Analytical performance and characterization of antibody immobilized magnetoelastic biosensors. Sensing and Instrumentation for Food Quality and Safety, 2007. DOI: 10.1007/s11694-007-9025-x: Accepted/ Online First. 9. Goldman, E.R., Pazirandeh, M.P., Mauro, J.M., King, K.D., Frey, J.C., and Anderson, G.P., Phage-displayed peptides as biosensor reagents. Journal of Molecular Recognition, 2000. 13(6): 382-387. 10. Brigati, J.R. and Petrenko, V.A., Thermostability of landscape phage probes. Analytical and Bioanalytical Chemistry, 2005. 382(6): 1346-1350. 11. Petrenko, V.A. and Sorokulova, I.B., Detection of biological threats. A challenge for directed molecular evolution. Journal of Microbiological Methods, 2004. 58(2): 147-168. 12. Sorokulova, I.B., Olsen, E.V., Chen, I.H., Fiebor, B., Barbaree, J.M., Vodyanoy, V.J., Chin, B.A., and Petrenko, V.A., Landscape phage probes for Salmonella typhimurium. Journal of Microbiological Methods, 2005. 63(1): 55-72. 148 13. Olsen, E.V., Sorokulova, I.B., Petrenko, V.A., Chen, I.H., Barbaree, J.M., and Vodyanoy, V.J., Affinity-selected filamentous bacteriophage as a probe for acoustic wave biodetectors of Salmonella typhimurium. 2006. 21(8): 1434. 14. Angelini, T.E., Liang, H., Wriggers, W., and Wong, G.C.L., Direct observation of counterion organization in F-actin polyelectrolyte bundles. The European Physical Journal E - Soft Matter, 2005. 16(4): 389. 15. Lyubartsev, A.P., Tang, J.X., Janmey, P.A., Nordenskiold, L., Electrostatically induced polyelectrolyte association of rodlike virus particles. Physical Review Letters, 1998. 81(24): 5465-5468. 16. Tang, J.X., Janmey, P.A., Lyubartsev, A., and Nordenskiold, L., Metal Ion- Induced Lateral Aggregation of Filamentous Viruses fd and M13. Biophysical Journal, 2002. 83(1): 566. 17. Souza, G.R., Christianson, D.R., Staquicini, F.I., Ozawa, M.G., Snyder, E.Y., Sidman, R.L., Miller, J.H., Arap, W., and Pasqualini, R., Networks of gold nanoparticles and bacteriophage as biological sensors and cell-targeting agents. Proceedings of the National Academy of Science USA, 2006. 103 (5): 1215-1220. 149 6. CONCLUSIONS In this dissertation, it was established that a phage-based magnetoelastic biosensor can be used for the specific and selective detection of S. typhimurium. The biosensor?s performance was characterized for longevity, specificity, selectivity, and size dependent sensitivity. The dose response curves obtained for the magnetoelastic biosensor show that different concentrations of S. typhimurium can be detected. The biosensor is capable of detecting S. typhimurium present in water and food matrices (milk and apple juice). The biosensor was also able to detect S. typhimurium in the presence of high concentrations of one and two masking bacteria. It was demonstrated that a minimum concentration of 5?10 3 cfu/mL could be detected by a biosensor with dimensions of 2?0.4?0.015 mm. The results obtained from the study of binding kinetics using Hill plots indicated a strong binding of S. typhimurium to the biosensor. It was also established that phage and S. typhimurium had a multivalent interaction. It was demonstrated that the sensitivity of the magnetoelastic biosensors could be increased by decreasing the dimensions of the sensor. A biosensor with dimensions of 5?1?0.015 mm showed a minimum detection limit of 5?10 4 cfu/mL and a sensitivity of 98 Hz/decade. However, a 0.5?0.1?0.015 mm showed a minimum detection limit of 5?10 2 cfu/mL and a sensitivity of 1150 Hz/decade. 150 For real-time field applications it was essential to establish the thermal stability of magnetoelastic biosensors. Biosensors stored at elevated temperatures (65 ?C) had significant binding activity even after 60 days. A high specificity of the biosensor to S. typhimurium was demonstrated. The normalized area coverage densities observed for the biosensor exposed to S. typhimurium, S. enteritidis, E. coli and L. monocytogenes were 1.00, 0.17, 0.06 and 0.03, respectively. These values indicated that the immobilized phage had negligible affinity for pathogens other than S. typhimurium. Scanning Electron Microscopy provided a visual verification of bacterial attachment to biosensor surfaces. High magnification (15000X) imaging of biosensor surfaces also provided an insight into the nature of phage binding characteristics (present in the form individual filaments or phage bundles). By increasing the concentration of counterions (Na + ) in the phage solutions, it was demonstrated that the formation of phage bundles could be achieved. It was established that phage stored in suspensions containing less than 420mM of the monovalent ion will have a uniform distribution of phage filaments on the sensor surface. It was observed that the formed ?bundles,? upon immobilization, had a tendency to accumulate at certain regions on the biosensor. However, further studies need to be performed to achieve a uniform distribution of the phage ?bundles.? In summary, a magnetoelastic transducer immobilized with filamentous phage as a sensitive, specific, selective and robust diagnostic biosensor platform for the detection of S. typhimurium was demonstrated. Such a biosensor system presents several advantages over the existing detection methodologies including low cost, wireless nature, environmental robustness, ease of operation and the possibility of miniaturization.