BIOSENSING TOTAL BAC TERIAL LOAD IN LIQUI D MATRICES TO IMPROV E FOOD SUPPLY CHAIN SAFETY USING CARBOHYDRATE - F UNCTIONALIZED MAGNET IC NANOPARTICLES FOR CE LL CAPTURE AND GOLD NANOPARTICLES FOR SIGNALING By Leann Lerie Matta A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Biosystems Engineering Doctor of Philosophy 2018 A B STRACT BIOSENSING TOTAL BAC TERIAL LOAD IN LIQUI D MAT RICES TO IMPROVE FOOD SUPPLY CHAIN SAFETY USING CARBOHYDRATE - FUNCTIO NALIZED MAGNETIC NANOPARTICLES FOR CE LL CAPTURE AND GOLD NANOPARTICLES FOR SIGNALING By Leann Lerie Matta Economical rapid nano - biosensing methods with expedited electrochemical signaling, were developed using carbohydrate - functionalized magnetic nanoparticle s (MNP) and gold - nanoparticle s (AuNP) to detect pathogenic bacteria in liquid or homogenized food samples. MNP functionalized with glycan - and amino/glycan lig ands were able to rapid ly extract bacteria , while c oncentrated dextrin - coated AuNP labeling improved detection sensitivity. Carbohydrate ligands are more stable than antibodies, permitting long shelf life of MNP at room temperature and minimized AuNP aggre gation during simple refrigeration. Transmission electron microscopy (TEM) imag ed the electrostatic binding between MNP and Salmonella Enteritidis, E. coli O157:H7, Bacillus cereus , Listeria monocytogenes and E. coli C300 0, which mimics electrostatic bind ing by antibod ies , although with lower specificity. Capture index (CI) is defined as the parts - per - thousand (ppt) of bacteria extracted per initial bacterial presence . TEM mages showed that attached milk matrix components did not interfere with microb ia l . capture . Salmonella, E. coli, and Bacillus (3 to 5 log CFU/mL) capture in three milks was 2 ppt to 1 20 ppt C I. Capture in beef juice and apple cider was 0.002 ppt to 0.011 ppt for E. coli and Listeria , respectively , at 10 log cfu/mL due to accelerated microbial growth immediately following the spike . V iscous homogenized eggs, though, impeded MNP - Salmonella migration to the magnet during separation. Th is phenomenon was a motivating factor in creating - (MNP - strip). Rapid nano - biosensing of MNP - cell complexes in under 30 min from either suspended or strip capture was possible using electrochemical technology of spectrometry or a simple handheld potentiostat . C apture concentrates bacteria as MNP - cell fr om large volumes allowing strong cyclic voltammetric ( CV ) signalin g. N ormalized peak current responses (NPCR) for microbial detection from simple matrices ( PBS and beef juices ) showed sample (S) NPCR lower than negative controls (N) (S/N < 1.0). W hereas in complex matrices ( milk , apple cider , and homogenized eggs ) , S/N were significantly greater than 1.0. NPCR for negative controls were found to be linearly related to matrix components fats, proteins, and sodium (R 2 = 0.92). Except for E. coli in beef juices, a ll S/N were significant (p < 0.05) for contamination levels ranging between 6.2 to 12.3 log CFU/mL. E nhanced signaling of low pathogen presence in food was achieved using electrically active AuNP labeling. Electrochemical detection of MNP - cell - AuN P complexes with spectrophotometry or differential pulse voltammetry (DPV) w as significantly more sensitive, detecting 3 log CFU/mL and 5 log CFU/mL E. coli contamination in milk (p < 0.20), respectively. F ood component attachment to the complexes altered, but did not interfere , with distinguishing samples from negative controls . MNP carbohydrate ligands exposed to refrigerated milk matrix components ( fats, lipids, sugars, protein and sodium ) for up to 9 - days still extracted bacteria. This makes possible f uture biocompatible tag - on nano - biosensors inside individual food packaging. Pathogen presence could be monitored over the lifetime of the product, reducing consumption of contaminated foods. Reliable f requent testing along the food supply chain would faci litate reduce d human disease, while reducing industry financial losses due to foodborne outbreaks. F lexible carbohydrate - based MNP - cell /( CV ) and MNP - cell - AuNP /( DPV or spectrometry ) nano - biosensing with electrochemical detection c an provid e a truly rapid, economical test. iv Dedicated to my beloved family, supportive husband, Samir, of 32 blessed years of marriage, and children, Laran Alexi & Andreas Per, Carlos & Alexa Kay, Anastasia and Ibtisam. v ACKNOWLEDGMENTS This research was funded by The Axia Institute at Michigan State University (formerly MRIVCC) and supported by Dr. Evangelyn C. - Biosensors Lab . My sincere gratitude to Dan Jones, Ph.D., for his continued professional guida nce, friendship, and encouragement over many years, especially throughout my doctoral studies. vi TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ......................... i x LIST OF FIGURES ................................ ................................ ................................ ........................ x LIST OF SCHEMES ................................ ................................ ................................ .................. xiv ii KEY TO ABBREVIATIONS ................................ ................................ ................................ ....... xv Chapter 1. Introduction ................................ ................................ ................................ ................... 1 1.1 Introduction ................................ ................................ ................................ ........................... 1 1.2 Hypotheses ................................ ................................ ................................ ............................ 3 1.3 Objectives ................................ ................................ ................................ .............................. 4 1.4 Innovations ................................ ................................ ................................ ............................ 5 Chapter 2. Food Supply Chain Safety Against Pathogenic Microbial Contamination ................... 8 2.1 Introduction ................................ ................................ ................................ ........................... 8 2.2 Food supply regulation ................................ ................................ ................................ .......... 9 2.3 Pathogenic microbial contamination along the food supply chain ................................ ..... 10 2.4 Conventional methods in pathogen detection ................................ ................................ ..... 12 2.5 Conclusions ................................ ................................ ................................ ......................... 14 Chapter 3. Emerging Nano - biosensing with Suspended MNP Microbial Extraction and Electricall y - Active Nanoparticle Labeling ................................ ................................ ................... 16 3.1 Introduction ................................ ................................ ................................ ......................... 16 3.2 MNP - assi sted Biosensing Properties ................................ ................................ .................. 17 3.3 Suspended MNP - assisted Biosensing ................................ ................................ ................. 21 - ................................ ................................ ................................ ....... 24 3.5 Enhanced Biosensing with Electrically - Active Nanoparticles ................................ ............ 28 3.6 Carbohydrate Ligands Against Microbial Targets ................................ .............................. 32 3.7 Microbial Surface Epitopes for Carbohy drate - functionalized MNP - cell - AuNP Nano - biosensing ................................ ................................ ................................ ................................ .. 34 3.8 Conclusions ................................ ................................ ................................ ......................... 39 Chapter 4. Literature Review Technology ................................ ................................ ................. 41 4.1 Introduction ................................ ................................ ................................ ......................... 41 4.2 Electrochemical Signaling Using Cyclic Voltammetry and Differential Pulse Voltammetry ................................ ................................ ................................ ................................ ................... 42 4.2.1 MNP - cell - AuNP Detection ................................ ................................ ........................... 42 4.2.2 Voltammetric Methods ................................ ................................ ................................ . 43 4.2.3 Micro - bio - sensing ................................ ................................ ................................ ......... 45 4.2.4 Voltammetry Conventions ................................ ................................ ............................ 46 4.2.5 Improved Bio - sensing with Nanoparticles ................................ ................................ ... 49 4.2.6 Signal Interpretation ................................ ................................ ................................ ..... 50 vii 4.3 Spectrophotometric Verification of NP Presence and Quantity ................................ .......... 52 4.3.1 Spectrophotometric Conventions ................................ ................................ ................. 52 4.3.2 Spectrometry Bio - sensing ................................ ................................ ............................. 54 4.3.3 Spectral Integration ................................ ................................ ................................ ....... 55 4.4 Conclusion ................................ ................................ ................................ ........................... 57 Chapter 5. Analysis of Carbohydrate - Functionalized Magnetic Nanoparticles and Dextrin - Coated Gold Nanoparticles Binding Properties to Salmonella, E. coli, Bacillus, and Listeria Using Transmission Electron Microscopy ................................ ................................ .............................. 60 5.1 Introduction ................................ ................................ ................................ ......................... 60 5.2 Materials and Methods ................................ ................................ ................................ ........ 61 5.2.1 Materials ................................ ................................ ................................ ....................... 61 5.2.2 Microbial Culture ................................ ................................ ................................ ......... 62 5.2.3 MNP Microbial Capture and Deactivation ................................ ................................ ... 62 5.2.4 Concentrated AuNP Preparation ................................ ................................ .................. 63 5.2.5 AuNP Labeling of MNP - cell Complexes ................................ ................................ ..... 64 5.2.6 TEM Imaging ................................ ................................ ................................ ............... 64 5.3 Result s and Discussion ................................ ................................ ................................ ........ 65 5.3.1 Binding Forces Between Nanoparticles and Microbes ................................ ................. 65 5.3.2 MNP - Salmonella Enteritidis Binding ................................ ................................ ........... 67 5.3.3 MNP - E. coli O157:H7 Binding ................................ ................................ .................... 68 5.3.4 MNP - Bacillus cereus Binding ................................ ................................ ...................... 70 5.3.5 MNP - Listeria monocytogenes Binding ................................ ................................ ........ 71 5.3.6 AuNP Binding to MNP - cell Complexes ................................ ................................ ...... 72 5.4 Conclusions ................................ ................................ ................................ ......................... 75 Chapter 6. Rapid Extraction of Pathogenic Bacteria from Liquid Foods using Carbohydrate - Functionalized Magnetic Nanoparticles Including Cyclic Voltammetric Detection .................... 77 6.1 Introduction ................................ ................................ ................................ ......................... 77 6.2 Materials and Methods ................................ ................................ ................................ ........ 80 6.2.1 Materials ................................ ................................ ................................ ....................... 80 6.2.2 Bacterial Culture ................................ ................................ ................................ ........... 81 6.2.3 Suspended MNP Bacterial Capture in Milk ................................ ................................ . 82 6.2.4 Suspended MNP Multiple Bacterial Capture ................................ ............................... 83 6.2.5 Matrix Exposure Effects on Suspended MNP Capture ................................ ................ 84 6.2.6 Capture Index Calculation for Suspended MNP Capture ................................ ............. 84 6.2.7 MNP - strip Production ................................ ................................ ................................ ... 85 6.2.8 MNP - strip Bacterial Capture ................................ ................................ ........................ 86 6.2.9 Cyclic Voltammetry Detection of Cell Presence on MNP - strips ................................ . 87 6.2.10 Statistical Analysis ................................ ................................ ................................ ..... 88 6.3 Results and Discussion ................................ ................................ ................................ ........ 89 6.3.1 Aspects of Bacterial Capture. ................................ ................................ ....................... 89 6.3.2 Definition of a Complex Matrix ................................ ................................ ................... 90 6.3.3 Suspended MNP Capture in Milk Matrices ................................ ................................ .. 91 6.3.4 Suspended MNP Capture of Multiple Bacteria ................................ ............................ 94 6.3.5 Milk Matrix Exposure Effects on Suspended MNP Capture. ................................ ...... 98 viii 6.3.6 Maximum Capture Index due to Equilibrium ................................ ............................. 100 6.3.7 Suspended MNP Capture in Beef Juices, Apple Cider and Homogenized Egg ......... 102 6.3.8 Rapid Detect ion of Bacteria in Complex Liquids Using Cyclic Voltammetry .......... 103 6.3.9 Confirmation of MNP - strip Bacterial Capture in Simple Buffer ............................... 104 6.3.10 Rapid Detection of Bacteria in Complex Liquids Using MNP - strips with Cyclic Voltammetry ................................ ................................ ................................ ........................ 107 6.4 Conclusions ................................ ................................ ................................ ....................... 111 Chapter 7. Gold Nanoparticle Enhancement in Differential Pulse Voltammetric and Spectrophotometric Detection of MNP Extracted Bacteria ................................ ........................ 114 7.1 Introduction ................................ ................................ ................................ ....................... 114 7.2 Materials and Methods ................................ ................................ ................................ ...... 116 7. 2.1 Method of Development for Carbohydrate - based MNP - cell - AuNP Nano - Biosensing ................................ ................................ ................................ ................................ ............. 116 7.2.2 Materials ................................ ................................ ................................ ..................... 116 7.2.3 Bacterial Culture ................................ ................................ ................................ ......... 117 7.2.4 Antibody - functionalized AuNP ................................ ................................ .................. 118 7.2.5 Specificity of Capture of Salmonella Monoclonal Antibody and Cross - reactivity .... 119 7.2.6 Concentrated Dextrin - coated d - AuNP ................................ ................................ ........ 119 7.2.7 Temporal Attachment of Concentrated d - AuNP to MNP - cell Complexes ................ 120 7.2.8 Carbohydrate - based MNP - cell - AuNP Nano - biosensing ................................ ............ 120 7.2.9 Electrochemical Spectrophotometric or Differential Pulse Voltammetric Detection 121 7.3 Results and Discussion ................................ ................................ ................................ ...... 123 7.3.1 Electrochemical Signature Responses for MNP - cell - AuNP Biosensing ................... 123 7.3.2 Specificity of Salmonella Detection Using Antibody - functionalized AuNP ............. 127 7.3. 3 Dextrin - coated AuNP Biosensing with DPV and Spectrometry ................................ 132 7.3.4 Electrochemical Quantification of MNP - cell - AuNP ................................ .................. 136 7.4 Conclusions ................................ ................................ ................................ ....................... 140 Chapte r 8. Conclusions ................................ ................................ ................................ ............... 141 A PPENDIX ................................ ................................ ................................ ................................ . 151 BIBLIOGRAPHY ................................ ................................ ................................ ....................... 154 ix LIST OF TABLES Table 1. 1 Innovative advances in nano - biosensing using carbohydrate - based MNP - cell - AuNP capture and label with electrochemical technology. ................................ ................................ ....... 7 Table 3.1. Magnetization properties of various magnetic nanoparticles (MNP) with carbohydrate functional groups. ................................ ................................ ................................ ......................... 23 Table 3.2 Biosensing methods using suspended functionalized magnetic nanoparticles (MNP). 25 Table 3. 3 Nano - biosensing methods with functionalized MNP and NP labeling. ....................... 30 Table 3.4 Biosensing methods using carbohydrate functional groups for target capture. ............ 36 Table 6 .1 Composition and properties of liquid matrices. ................................ ............................ 92 Table 6. 2 Capture index (CI) of Salmonella, E. coli and Bacillus using suspended MNP - F#1 for large - increased cell concentration factor in each milk type following the standard five - fold serial dilution. + (n=3) . ................................ ................................ ................................ ............................. 95 Table 6. 3 Capture indexes (CI) of Salmonella, E. coli and Bacillus using suspended MNP - F#2 for large - increased cell concentration factor in each milk type fol lowing the standard five - fold serial dilution. + (n=3) ................................ ................................ ................................ .............................. 96 Table 6. 4 Microbial capture indexes using suspended MNP - F#1 for rapid extraction of three competitive bacteria, Salmonella, E. coli and Bacillus , either simultaneously (at 10 - 6 dilutions of the stock culture) or separately (at 10 - 5 dilutions) from PBS and vitamin D milk in small - volume capture (SVC) * . (n=3) ................................ ................................ ................................ ................... 97 Table 6. 5 Capture indexes for Salmonella in large - volume capture following lon g - term suspended MNP - exposure + . (n=3) ................................ ................................ ................................ ................ 100 Table 6. 6 Normalized peak current responses (NPCR) for MNP - strip negative controls in simple and complex matrices and the matrix composition used in linear regression statistical analysis. (n=3) ................................ ................................ ................................ ................................ ............ 109 Table 7. 1 Specificity of Salmonella detection using antibody - functionalized AuNP. ............... 128 Table S7. 2 Specificity of Salmonella detection using antibody - functionalized AuNP 153 x LIST OF FIGURES Figure 3.1 Schematic of the MNP - cell - AuNP biosensor for total bacterial load. Carbohydrate - functionalized MNP extract E. coli from artificially contaminated milk, while dextrin - coated EANP attach to captured cells at differen t cell receptors. Negative controls consist of sterile milk where MNP - EANP complexes form due to carbohydrate attraction between the functional groups. Cell quantification is possible using FDA bacteriological analytical methods (BAM) agar plating (top), w here neither MNP nor EANP inhibit bacterial growth and negative controls will have no cell growth present. Rapid biosensing is possible with spectrophotometric or electrochemical signaling, where negative controls will produce larger signals due to higher EANP presence than samples, due to saturation binding of EANP to MNP. ................................ ................................ . 18 Figure 3. 2 Examples of hysteresis magnetization for m agnetic nanoparticles (MNP) functionalized with carbohydrate groups, (A) from Alocilja Nano - Biosensors lab and (B) used in metal recovery [99] (emu = electromagnetic units). ................................ ................................ ..... 21 Figure 4.1 Electrochemical systems exemplified with (A) common voltaic cell, (B) miniaturized screen - printed carbon electrode (SPCE) sensor, and (C) a mock - up of voltammetric bio - sensing with an SPCE chip wi th capture bacteria labeled with AuNP. ................................ ..................... 44 Figure 4.2 Examples of graphical results of nano - biosensing responses using cyclic voltammetry for microbial detection with MNP - strips on (A) soft plastic or (B) hard plastic. ......................... 44 Figure 4. 3 Examples of graphical results of nano - biosensing responses using differential pulse voltammetry for microbial detection using MNP - cell - AuNP with (A) positive current convention and (B) negative current convention. ................................ ................................ ............................ 45 Figure 4.4 Equations for general theory of voltammetry including (A) the Nernst Equation which relates the applied potential force, E, across the electrode to the concentration of the reductive and oxidative species, c R and c O , respectively, where we R is the molar gas constant (8.3144 J mol 1K 1), T is the absolute temperature (K), n is the number of electrons transferred, F = Faraday constant (96,485 C/equivalents), and E 0 is the standard reduction potential for the redox couple; (B) the Butler - Volmer Equation further relates current, i, potential, E, and concentration, c R and c O , where = nF(E E0)/RT, k 0 is the heterogeneous rate constant, a is kno wn as the transfer , which is directly proportional to the analyte concentration gradient, c O / x, where D O is the diffusion coefficient of oxidative species a nd x is the distance from the electrode surface [205]. ................................ ................................ ................................ ................................ ....................... 47 Figure 5. 1 Examples of Salmonella Enteritidis (A) attachment to MNP - F#1 (B) and MNP - F#2 (C). ................................ ................................ ................................ ................................ ....................... 68 xi Figure 5. 2 Examples of E. coli O157:H7 (A) binding to MNP - F#1 (B) including cell lysing over extended exposure (C) and to MNP - F#2 (D) as well as cell lysing that occurs over extended exposure (E). ................................ ................................ ................................ ................................ . 69 Figure 5. 3 Examples of Bacillus cereus (A) binding to MNP - F#1 (B) and MNP - F#2 with short exposure (C) and extended exposure to the MNP (D). ................................ ................................ . 70 Figure 5.4 Examples of Listeria monocytogenes (A) binding to MNP - F#1 (B) and MNP - F#2 (C). ................................ ................................ ................................ ................................ ....................... 71 Figure 5. 5 Examples of MNP - F#2 - E. coli C3000 complex labeling with AuNP250 with excess AuNP250 present (A), showing aggregated AuNP binding to the cell wall (B) and at higher magnification (C) all in TSB, and finally persistence of AuNP binding to the cell structure even during lysing caused by extended MNP exposure in PBS. ................................ ......................... 73 3 Figure 5.6 Examples of unconcentrated d - AuNP (A), concentrated AuNP250 coating MNP - F#2 in PBS without prior milk exposure (B), with 2 - with 1 - ................................ .................. 74 Figure 6.1 Photos of extracted bacterial growth on solid agar for Salmonella on BGA (A), E. coli on LBA (B), Bacillus on LBA (C), and TEM images of milk contamination bound to MNP (D). ................................ ................................ ................................ ................................ ....................... 94 Figure 6. 2 Evidence of MNP - strip Salmonella capture in PBS through TSB culture (A) and detection using cyclic voltammetry for soft plastic MN P - strips (B) and hard plastic MNP - strips (C), which showed an inverse linear relation (R 2 = 0.7132) to cells available for capture (D). Inset plots (B & C) show cathodic peaks in greater detail and the order of concentration listed in the legend (actual units log CFU/mL) follow the order of peak response. ................................ ....... 106 Figure 6.3 Effect liquid matrix components upon CV detection for MNP - strip cell capture of Salmonella in PBS (6.19 log CFU/mL available cells, S/N = 0.85, p = 0.0 4) (A) and E. coli in beef juices (7.7 log CFU/mL available cells, S/N = 0.92, p = 0.08) (B). All peak current responses here were normalized to the mediator reference run at the time of testing to compare matrix effects upon current responses. (error bars = standard deviation) ................................ ................................ ... 108 Figure 6.4 Effect of homogenized egg matrix component upon CV peak current response increased propor tionally as the purity of the matrix increased (A) such that MNP - strips exposed homogenized egg diluted with PBS showed higher peak current responses for higher PBS dilutions (B). All peak current responses here were normalized to the mediator reference ru n at the time of testing to compare matrix effects upon current responses. (error bars = standard deviation) .... 108 Figure 6. 5 Effect of complex liquid matrix components upon CV detection for Salmonella in homogenized egg (12.3 log CFU/mL available cells, S/N = 1.25, p = 0.04) (A), Listeria in apple cider (7.9 log CFU/mL cells available, S/N = 1.13, p = 0.06) (B), and E. coli in vitamin D milk (8.5 log CFU/mL cells available, S/N = 1.08, p = 0.04) (C) showed lower peak current responses for negative controls, while cell presence expended less of the chemical mediator, resulting in xii higher peak current responses. All peak current responses here were normalized to the mediator reference run at the time of testing to compare matrix effects upon current responses. (error bars = standard deviation) ................................ ................................ ................................ ...................... 110 Figure 7.1 Spectral scans of AuNP250 controls combined with MNP (A) and following attachment through the cell capture and labeling method (B). ................................ ................................ ...... 124 Figure 7. 2 Signature responses for MNP - cell - AuNP biosensing controls with differential pulse voltammetry. (Se= Salmonella Enteritidis) ................................ ................................ ................ 125 Figure 7. 3 Effect of sample complexity upon AuNP current response in DPV biosensing. ...... 127 Figure 7.4 Specificity testing using anti - Salmonella - antibody - functionalized AuNP to label MNP - Salmonella (Target Se1 at 5.69 log CFU/mL and Se2 at 4.69 log CFU/mL) and MNP - E. coli (Non - Target Ec at 5.94 log CFU/mL) complexes, as well as negative controls, for DPV detection of the AuNP label. ................................ ................................ ................................ ................................ . 130 Figure 7. 5 DPV current responses for concentrated AuNP responses (A) showing a linear relation between current response at 0.4 V and AuNP volume where (B) AuNP responses are indicated with an arrow. ................................ ................................ ................................ ............................. 133 Figure 7. 6 Spectral absorbance of concentrated AuNP (A) showing an extinction peak at approximately 520 nm; (B) AuNP concentration is linearly related to both the peak height and the step - i ntegrated spectral absorbance between 220 nm to 900 nm (n = 3). ................................ ... 134 Figure 7.7 Spectral signaling for carbohydrate - based MNP - cell - AuNP complexes showing (A) shifted AuNP peak from 520 nm slightly masked with the MNP peak at approximately 620 nm where (B) both the AuNP peak height and step - integrated spectral area 220 900 nm are linearly related to the ti me allotted for AuNP attachment. ................................ ................................ ...... 135 Figure 7.8 Electrochemical nano - biosensing for simultaneous carbohydrate - based MNP - E. coli - AuNP rapid extraction from milk and electrically - active labeling followed by (A) DPV or (B) log 5 CFU/mL E. coli E. coli . ................................ ...................... 137 Figure 7. 9 Graphs for electrochemical nano - biosensing for simultaneous carbohydrate - based MNP - E. coli - AuNP rapid extraction and ele ctrically - active labeling followed by (A) DPV or (B) spectral analysis of samples and negative controls for microbial detection in milk (n=3). Arrows = log 3 CFU/mL E. coli.. ................................ ................................ .................. 138 Figure 8.1 Cost - benefit analysis of microbial contamination detection in the food supply chain including (nano - ) biosensing methods. (see text for acronym definitions) ................................ 148 Figure S 8 . 2 Example of adjusted DPV current response for MNP - F#2 - cell - Ab - AuNP nano - biosensing, showing corrected baseline with respect to average current response between 0.9 V to xiii 0.6 V; identifying maximum (average) AuNP response potential (arrow, 0.38 here); and portion of a sample signal for clarity during visual data analysis 152 xiv LIST OF SCHEMES Schematic 6.1. Rapid detection of bacterial contamination in contaminated liquids using MNP - strips and CV detection. MNP - F#2 affixed to plastic strips (MNP - strip) (A) for user - friendly capture in only 10 min (B) by either dipping the strip into microbial contamination within undiluted liquids or applying 250 uL liquid onto the strip (C), then rapid detection by electrochemical testing of reacted chemical mediator on a screen - printed carbon electrode (SPCE) (D) usi ng cyclic voltammetry (E). Results between samples and negative controls are shown in (F) 88 xv KEY TO SYMBOLS AND ABBREVIATIONS Units o C = degrees Celsius A = microamps µL = microliter M = micromolar CFU/mL = colony forming units per millilite r cm = centimeter cm 2 = centimeters squared emu/gm = electromagnetic units per gram fg/mL = femtogram per milliliter gm = gram hr = hour in Hg = inches mercury log CFU/mL = log base 10 of colony forming units per milliliter M = molar mg = milligram mg/mL = milligram per milliliter min = minutes mL = milliliter ms = milli seconds mS/cm = milli - Siemens per centimeter xvi mV = millivolt mV/s = millivolts per second nm = nanometer nm = nanometers nM = nanomolar pM = picomolar ppm = parts per million ppt = parts per thousand rpm = rotations per minute tblsp = tablespoon V = volts v/v = volume of MNP to volume total ratio Chemical Nomenclature Ca 2+ = calcium ion CoPt 3 = cobalt platinum Fe 3 O 4 = iron oxide FePt = iron platinum H 2 O 2 = hydrogen peroxide HCl = hydrochloric acid Mn 2+ = manganese ion O = oxygen S = Sulphur Research - Related Acronyms xvii 1 - step biosensing = microbial detection using magnetic nanoparticle capture and gold nanoparticl e labeling simultaneously, followed by differential pulse voltammetric detection 2 - step biosensing = microbial detection using magnetic nanoparticle capture first, then gold nanoparticle labeling separately, followed by differential pulse voltammetric det ection Ab - AuNP = antibody - functionalized gold nanoparticles ANOVA = analysis of variance AuNP = gold nanoparticles BAM = bacteriological analytical methods BGA = brilliant green agar CI = capture index CV = cy c lic voltammetry d - AuNP = dextrix - coated gold nanoparticles DPV = differential pulse voltammetry FORMVAR = trademarked film coating for microscopy grids LBB = Luria Bertani broth LVC = large volume capture MANOVA = multi - variate analysis of variance MNP = magnetic nanoparticles MNP /EANP complexes consisting of magnetic nanoparticle and electrically - active nanoparticles MNP - AuNP = magnetic nanoparticles labeled with gold nanoparticles MNP - cell = complexes consisting of magnetic nanoparticles attached to cells MNP - cell/CV = cyclic v oltammetric detection of complexes consisting of magnetic nanoparticles attached to cells xviii MNP - cell - AuNP complexes consisting of magnetic nanoparticles attached to cells and labeled with (dextrin - coated) gold nanoparticles MNP - cell - AuNP/DPV = differential pulse voltammetric detection of complexes consisting of magnetic nanoparticles attached to cells and labeled with (dextrin - coated) gold nanoparticles MNP - F#1 = glycan - functionalized magnetic nanoparticles MNP - F#2 = amino/g lycan - functionalized nanoparticles MNP - strip = ( amino/glycan - functionalized ) magnetic nanoparticles adhered to plastic strips MNP - strip - cell = cells attached to magnetic nanoparticles adhered to plastic strips MNP - strip - cell/CV = cyclic voltammetric detect ion of cells attached to (amino/glycan - functionalized) magnetic nanoparticles adhered to plastic strips MOX = modified Oxford media N = negative control n = number of replicates NPCR = normalized peak current response PBS = phosphate buffered solution PCA = principle component analysis pH = negative logarithm base ten of the hydrogen concentration p - value = indicator of the strength of the null hypothesis R 2 = coefficient of determination representing the variability in the identified dependent variable with respect to a linear combination of independent variable regressors S = sample S/N = ratio of the electrochemical current response s of the sample to corresponding negative control SVC = small volume capture TEM = transmission electron magnification TSA = tryptic soy agar xix TSB = tryptic soy broth Literature Review and Background - Related Acronyms HAACP = Hazard Analysis and Critical Control Point HPLC - DAD = high - pressure liquid chromatography diode array detection HRP = horseradish peroxidase HUS = hemolytic uremic syndrome i = current response Ipa = anodic peak current Ipc = cathodic peak current LPS = lipopolysaccharide M s = magnetization saturation OMP = outer membrane protein PCR = polymerase chain reaction pKa = acid dissociation constant QD = quantum dots RF = radio frequency RNA = ribonucleic acid SPEL = signaling pathogens electrically in liquids SPR = surface plasmon resonance SWCNT = single - walled carbon nanotubes TN = titanium oxide nanoparticles USDA = United States Department of Agriculture UV - Vis = ultraviolet - visible light xx WHO = World Health Organization 1 Chapter 1. Introduction 1.1 Introduction Worldwide population growth is estimated to reach 9.8 billion people by 2050 [49]. This growth will further strain our food and water supplies. Currently over 28 billion meals are consumed each day around the globe [50] , b ut foodborne disease sickened 600 million people globally in 2010, causing over 420,000 deaths from foodborne hazards, including those that were bacterial in origin [1] . Unfortunately, children under the age of 5 years represented 40% of these illnesses [51]. Higher mortality rates persistently hit developin g countries in African sub - regions, South - East Asia and the Eastern Mediterranean [51], [52], which have limited economical resources to fight these diseases. Environmental microbial pathogens are therefore a major concern, especially creating a double - ed ged sword for developing countries. Food and water scarcity create conditions in which contamination is ignored by starving people who can also ill afford to contract diseases. Even i n the US , foodborne outbreaks sicken thousands yearly, with the majority caused by three pathogenic bacteria: Salmonella , Shiga Toxin - producing E. coli , and Listeria monocytogenes [53], [54]. Man y biosensing assays are too expensive or technically complicated to be economically feasible in many regions of the world. Designing a biosensor to meet more stringent regulations established by the US through the FDA and USDA will ensure rigorous assay per formance worldwide. Economical methods will ensure affordability in even developing countries and should require minimal storage and technology, with long shelf life, such as those using carbohydrate - functionalized magnetic nanoparticle ( MNP ) and dextrin - c oated gold nanoparticles (AuNP) [2] [6] . 2 In ensuring a secure m icrobial pathogen - free food supply, rapi d response detection of microbial contamination is of utmost importance. Many biosensor designs consist of self - contained test strips developed from the base up with compl ex nanoparticle chemistry and intricate ligand immobilization. Some methods are based upon ELISA 96 - well plate technology with fluorescent detection or may use expensive antibody receptors or DNA ligands. M any of these proposed designs are impractica l first address the broad goals of any biosensor: Field op erability, Inexpensive, with Real - time detection that is both Sensitive and Specific to target, while being as Trouble - free as possible (FIRST) [7] . Carbohydrate ligands are inexpensive, robust structures that are increasingly synthesized for higher selectivity. Used in conjunction with optical or electrical detection of gold nanoparticles (AuNP), carbohydrate - functionalized MNP - cell - AuNP nano - biosensing advances the goal of being the FIRST biosensor of choice in detecting microbial pathogens throughout our food s upply chain. Quantifying the level of microbial contamination within a food product using the FDA preferred bacteriological analytical methods ( BAM ) requires tedious, time - consuming and supply - intensive microbiological methods for decades in the US [8] . Before that, though, humans rarely knew the potential food hazards that may have sickened them , as discussed in food microbiology courses , much less went so far a s to investigate their cause [9] . With technological advances made in miniaturizing sensor components in the past twenty years, nano - sized sensors have reduced most aspects of microbial detection. Nano - biosensors are now prevalent within medical, environmental and even personal - use areas [10] [13] . Within industrialized countries, cost is of little concern to the user; and within under/developing countries, cost is the least of their concerns, since many of the nano - biosensors designed to date also require expensive test equipment, environmentally sensitive components, refrigeration, and advanced knowledge to use [14] . 3 Given the amount of sampling required throughout the food supply chain, reliable rapid methods are of necessity to expedite risk assessment for food products on the market. To improve the process of reliably tra cking pathogenic microbial contamination within our food supply, this research evaluated the ability of inexpensive, highly reactive magnetic and gold nanoparticles to develop rapid tests that quantify bacterial levels in liquid food products. O veral l, the goal of this research is to apply the capture and biosensing technology of carbohydrate - functionalized MNP and AuNP for rapid detection and quantification of microbial loads in complex liquid food matrices to improve the safety of our highly - integrated fo od supply chain and thereby reduce incidences of foodborne outbreaks. 1.2 Hypotheses This research had three hypotheses with regards to carbohydrate - based MNP/AuNP nano - biosensing. First, magnetic nanoparticles (MNP) coated with carbohydrates , without the presence of antibodies, will still rapidly bind , extract , and concentrate microbial pathogens from complex liquid food matrices to reduce the rate of false negatives through increased signal strength without initial sample preparation whether as suspended MNP or immobilized onto plastic strips for broad - based rapid pathogen detection in the liquid food industry . Second, electrochemical instrumentation will simplify testing methods for accelerate d path ogen detection of MNP - extracted microbial pathogens from complex food matrices despite the presence of food components. Third , concentrated, dextrin - coated gold nanoparticles, without the presence of antibodies , will rapidly label MNP - extracted microbial pathogens to reduce false positives through increased signaling by better separation from negative controls even in the presence of complex 4 liquid food matrices through enhance d electrochemical instrumental signaling for improved food security through freq uent economical testing . 1.3 Objectives Objective one of this research was to d etermine which MNP carbohydrate functional group extracts target bacteria at the highest amount, reported as a capture index (CI) . In fulfilling this objective, transmission electron microscopy (TEM) imaging was evaluated as a means to s ubstantiat e the cap ture dynamics of various pathogenic bacteria by the numerous carbohydrate - functionalized magnetic nanoparticles (MNP) . Then an e valuat ion was made of matrix component effects upon microbial extraction from complex matrices with suspended glycan - and amino/ glycan - functionalized MNP (MNP - F#1 and MNP - F#2 , respectively ) . To understand if i mprove d microbial extraction from complex matrices would result, MNP - F#2 were affix ed to plastic strips (MNP - strips) and then accelerate d detection using cyclic voltammetry (C V) signaling was evaluated. Finally, an extensive literature search was done to u nderstand the microbial surface epitope chemistry with respect to binding properties to MNP carbohydrate ligands. Objective two wa s to investigate whether AuNP labeling with e lectrochemical instrumental signaling would enhance pathogen detection. S pecificity of Salmonella capture by MNP - F#2 using anti - Salmonella antibody - functionalized AuNP (Ab - AuNP) was investigated using differential pulse voltammetry (DPV) . This antibody - bas ed method was compared to economical carbohydrate ligands dextrin - coated gold nanoparticle s (d - AuNP) as a means to enhance DPV signaling (biosensing) . Finally, accelerate d pathogenic bacteria detection through DPV or spectrophotometric signaling of carbohydrate - based MNP - cell - AuNP complexes was evaluated . 5 Objective three was to optimize carbohydrate - based MNP - cell - AuNP nano - biosensing parameters, including MNP and AuNP quantity , respective capture and l abeling times , and analyze matrix complexity effects upon signaling interpretation . Additionally, electrochemical instrumentation was evaluated to determine if it would reliably detect microbial pathogens extracted directly from complex food matrices of m ilk, beef juices, apple cider and homogenized eggs using spectrometry, cyclic voltammetry (CV) and DPV. 1.4 Innovations As will be shown later in sections 3.3 Suspended MNP - assisted Biosensing and 3.5 Enhanced Biosensing with Electrically - Active Nanoparti cles, many current nano - biosensors use expensive antibodies or DNA ligands, reducing their ability to detect pathogens across a broad spectrum, or use time - consuming and labor - intensive methods with complicated detection schemes. Many reported biosensors a re also limited in their ability to reliably detect bacterial loads below 1000 cfu/mL. Th e goal of this research was to develop a n e conomical microbial assay that minimizes the user - required technology , utilizes novel carbohydrate - functionalized nanoparticles with a proven long shelf - life, and reduces sample handling required to achieve an accurate result within minutes . Selective carbohydrates are an economical means to functionalize nanoparticles used in biosensing ap plications [15], [16] . Carbohydrate - functionalized MNP overcome matrix interference, easily extracting pathogens from undiluted liquid matrices. FDA prescribed BAM c ulture methods require substantial sample preparation and dilution and time - consuming centrifugation methods to concentrate cells. These steps are all eliminated with rapid MNP extraction, and these MNP - cell complexes can still be quantified with common ag ar plating methods. For thicker matrices, such as liquid homogenized eggs, these MNP can be affixed to 6 plastic strips (MNP - strip) and retain their capture ability. Either MNP extraction method concentrates microbial contamination to enhance microbial sign aling, especially for samples with low cell presence . These MNP - cell complexes can be labeled with electrically active d - AuNP. These concentrated MNP - cell - AuNP complexes enhance electrochemical signaling above background noise for either electrochemical vo ltammetric or spectrophotometric signaling, thereby reducing labor costs and providing real - time results. Table 1.1 highlights these innovations, along with the current state of the art and the particular research question each innovation addresses. 7 Table 1.1 Innovative advances in nano - biosensing using carbohydrate - based MNP - cell - AuNP capture and label with electrochemical technology . (Research Question) Shortcomings of Current Methods My Contribution to Advance Technology Literature References & Publication Question 1 (Hypothesis 1): What matrix effects will complex liquid foods have on the carbohydrate - functionalized MNP capture of pathogenic bacteria for different exposure times? Matrix Effects Require extensive sample preparation for matrix removal before exposure to biosensor Selectively bind microbial contamination amidst complex components in matrices 1 References [17] [20] 2 [3] Question 2 (Hypothesis 1): How does MNP affixed to plastic strips affect their ability to extract bacteria from complex liquids? What methods cans be used to detect microbial presence? MNP - strip Capture Complex manufacturing methods with specialized materials Develop cost - effective MNP - strip and verify ability for micro bial capture and detection References [21] [23] - [4] Question 3 (Hypothesis 1): How can literature concerning microbial surface chemistry and MNP - carbohydrate chemistry, along with TEM imaging, allow us to discern the dynamics of carbohydrate - functionalized MNP capture of pathogenic bacteria in simple and complex liquid food matrices? Microbia l Epitopes Selective methods depend upon expensive antibodies or DNA/RNA ligands Present carbohydrate selectivity from literature and outlin e biosensor selectivity vs cost trade - off References [24] [28] [29] Question 4 (Hypothesis 2): What factors affect selectivity of Salmonella detection using MNP capture, antibody - functionalized gold nanoparticles (Ab - AuNP) labeling and differential pulse voltammetry (DPV)? Detection Specificity Require complex nanoparticle labeling mechanisms for improved detection sensitivity Determine limitations within MNP - cell - Ab - AuNP/DPV biosensing method for future improvements 3 References [5], [6], [30] [32] Question 5 (Hypothesis 3 ): What factors must be considered when using dextrin - c oated gold nanoparticle (d - AuNP) attachment to MNP - cell complexes in liquid foods for (DPV) biosensing or spectral absorbance detection? d - AuNP Signaling Complex nanoparticle antibody or DNA ligand conjugation and extensive sample handling Developed carbohydra te - based MNP - cell - d - AuNP bio - sensing useful with DPV or spectrophotometry References [33] [35] g [29] Question 6 (Hypothesis 3): How can microbial surface attachment and AuNP binding be exploited to detect bacteria within liquids using electrochemical instrumentation? Electro - chemical Biosensing Lack of rapid detection methods Established electrochemical detection parameters using spectrometry and voltammetry References [36], [37] [29] & 4 [2] 1 Capture of Salmonella, E. coli, Bacillus and Listeria from milk, beef juices, apple cider and homogenized eggs. 2 Accepted manuscript by Journal of Food Protection , JFP - 18 - nanoparticles for centrifuge - E.C., 2018, awaiti ng publication fee funds. 3 Limitations to MNP - cell - Ab - AuNP/DPV biosensing is reviewed in chapter 7. 4 Research concurrent with wireless radio - frequency (RF) signaling of sensing pathogens electrically in liquids SPEL - cell - AuNP samples, published in Biose nsors & Bioelectronics . 8 Chapter 2. Food Supply Chain Safety Against Pathogenic Microbial Contamination 2.1 Introduction W orld - wide, governments in developed countries regulate the production and sale of food within their countries to ensure consumers are purchasing food free from microbial health risks [38], [39] , whereas in developing countries, food supply regulations are inadequate or poorly enforced [40] . In the US, the FDA regulates larger industry producers, w hereas small local sellers are not regulated. Although bacteriological methods recommended by the FDA are highly accurate in detecting microbial contamination, they are time - consuming and labor - intensive [8] . Using an inexpensive rapid microbial test would then substantially reduce both lost product and consumer health risks. For example, nutritionally rich drin ks such as milk and fruit juices allow rapid microbial growth . I t is estimated that over 210 pounds of dairy products are consumed per US citizen each year, with over 60% in milk form. In the liquid food supply chain, between the farm and fork, there are m ultiple avenues for pathogens to enter the food product, therefore frequent testing would ensure secure food production method s and reduce consumer risks of illness . But the complex components present in most liquid food products , as well as any homogenize d food sample, may interfere with pathogen detection. Many reported nano - biosensing methods require pre - test preparation to reduce this interference so that false negative results are not incurred. Nano - biosensors that bind more selectively to bacteria tha n surrounding liquid matrix components without any necessary dilution necessary would re d uc e both sample handling and false negative results. 9 2.2 Food s upply r egulation Worldwide , our food supply is monitored by a variety of agencies and regulations [38], [40], [41] . Unfortunately , the World Health Organization (WHO) reporte d that in the year 2010 an estimated 600 (95% uncertainty interval 420 900) million illnesses occurred. The most frequent foodborne disease agents were norovirus, Campylobacter , and non - typhoidal Salmonella enterica , causing diarrheal diseases [1] . The European Commis s ion (EC) instituted a Food Safety policy to ensure coherent farm - to - table monitoring of their food supply, including imports. The European Food Safety Authority (EFSA) provides science - based risk assessment to aid the EC in its regulations [38] . Similarly, the Food Standar ds Agency (FSA) of the United Kingdom monitors food and drink threats [42] . Finally, in the US, several agencies cover our local and imported food supply including the United States Department of Agriculture (USDA), Food and Drug Administration (FDA) and Centers for Disease Control and Prevention (CDC). The USDA provides inspection of domestic products, imports, and exports, and conduct egg supply against foodborne diseases [43] until it is determined that no microbiological contamination exist. Specifically, the presence of six strains of E. coli , due to their low infective dose , which has reduce d the number of ov erall recalls required [44], [45] . A foodborne outbreak response network was formed by the FDA in 2011 . This network surve ils outbreaks and provides post - response activity related to multiple - illness events involving food [46] . To reduce outbreak events, food producers Hazard Analysis and Critical Control Point ( HACCP ) method to identify points where contamination can 10 occur along their food supply chain [47] . T o determine any microbial contamination, t he FDA also established standardized bacteriological analytical methods (BAM). BAM methods are considered the gold standard that all reported nano - biosensing methods are commonly tested against in the drive to design more expedient means to ensure a safe food supply for consumers. Regulations instituted by different agencies may vary in different markets. For example, different microbial contamination levels are allowed in milk products in the European Union (EU) versus the USA. In the EU, pasteurized milk may contain no more than 50,000 CFU/mL total bacterial count and less t han 5000 coliforms, compared to 20,000 CFU/mL total bacterial load and below 10 coliforms in the US [48] . Suppliers of raw milk sold in the US for pasteurization are required to keep bacterial levels below 200,000 CFU/mL . US groups that have lobb ied the FDA to allow interstate raw milk sales , which are currently illegal, even concede the necessity to warn consumers about the presence of potential pathogens [49] . It would be expected that regulations across the globe would vary for most food products. The agencies assigned to protect consumers in each region from foodborne diseases continually monitor threats, occasionally setting new standards for allowed microbial levels, among other contaminants, as new evidence from event investigations and scientific research are reported. As these are reactive actions, o ne consistent factor, though, is the need to rapidly determine pathogen presence with hi gh reliability 2.3 Pathogenic microbial contamination along the food supply chain Sustainable food supply chains have a goal of providing a sufficient supply of safe, healthy food while still ensuring economical options for farmers and suppliers alike , along with addressing any environmental concerns during food production [50] . One important aspect of a safe food 11 supply is the continued monitoring of any pathogenic microbial con tamination in our food along the supply chain from farm to fork. This involves the diligent attention of farmers who grow and raise our food sources, packers and food - contact packaging providers, food product manufacturers, commercial sellers and the shipp ing industry [51] , since microbial contamination may be introduced at any step. T he Grocery Manufacturers Association encourage programs that focus on prevention and anticipation of potential problems, to ensure a safe finished product. This fits well contamination entry in th e food product. Common pathogenic bacteria found in our food supply that are tracked by the CDC include those that cause acute intoxication, toxico - infection, or invasive infections with a range of symptoms and possible organ failure. Clostridium botulinum spores and Bacillus cereus are both uncommon diseases in the US, but C. botulinum found in canned food may cause muscle paralysis if the food is not heated appropriately and B. cereus can curdle milk products. Clostridium perfingens and Staphylococcus aur eus cause quick onset of diarrhea when consumed in foods such as meatballs , causing up to 10% of foodborne disease in the US. Listeria monocytogenes is commonly found in ready - to - eat foods and fresh, unpasteurized apple cider. L. monocytogenes causes thous ands of invasive infections yearly , similar to Campylobacter jejuni found in raw milk and poultry. Together L. monocytogenes and C. jejuni result in 25% of yearly deaths due to microbial infection. Strains of E. coli cause some of the most serious diseases in the US , particularly enterohemorrhagic E. coli (EHEC) strain O157:H7. EHEC is one of the six E. coli strains the USDA lists as zero tolerance when found in ground beef since they can cause hemolytic uremic syndro me (HUS). Although most EHEC infections result in bloody diarrhea lasting up to eight 12 days, HUS results in kidney failure and death for 15% of infected children. Salmonella , a common peanut butter contaminant, can also result in HUS, causing over a million deaths a year, equating to 10% of all microbial illnesses and 30% of all deaths. Common contaminants of milk, an example of a rich, complex food matrix, include several bacteria. Staphylococcus aureus is commonly found on human skin and wounds and can eas ily contaminate milk products post - pasteurization and cause cramps, vomiting, and diarrhea from an infectious dose of 6 log CFU/mL. Yersinia is a common pork contaminant that is easily transferred from live hogs to raw milk on farms, causing vomiting and d iarrhea for 100,000 US citizens per year, even septicemia for 3% of suffers. Finally, Salmonella was the cause of a landmark outbreak case in 1985 in which cross - contamination of milk from Hillman Farms identified important possible sources of foodborne di sease. In fact, the FDA cautions against drinking raw milk and unpasteurized products due to the severity of diseases from over 90 microbial pathogens that may be present [52] . 2.4 Conventional methods in pathogen detection Bacteriological analytical methods (BAM) published by the FDA since 1965 are their preferred laboratory procedures to determine microbiological levels present in food [53] . These gold standard methods have been developed and refined over the years to detect pathogens in our food sources, even though they are time - consuming and require substantial supplies. Additionally, BAM methods are more reactive than proactive in response and used in industrial risk assessments of possible contamination presence. A concerted drive to develop sensitive, miniaturized biosensing methods or stand - alone nano - biosensors has succeeded in addressing numerous aspe cts trap 13 Biosensors can technically detect any quantifiable microbial component or biological material, but reliably relating the quantity of bacteria that are present may be problematic. Mandal et. al. reviewed biosensor designs reported throughout the first decade of the 21 st century [54] . Sensor limits above 100 CFU/mL limit their application within the food supply industry, especially for zero - tolerance pathogens , referring to the fact that if the pathogen is detected at levels set by the FDA, USDA or FSI S, for example, they may not be sold . M any of these methods may also not distinguish between live and dead cells nor be able to detect cells within complex food matrices. Numerous current and proposed biosensing methods like those reported later in section 3, below, suffer from matrix interference in achieving strong signaling responses . Particularly, molecular methods that detect biochemical markers and nucleic acid based methods that detect cellular DNA/RNA, causing higher detection limits and limited tes t applications [12], [20], [55] [58] . Many biosensors require extensive sample preparation before the final detection step, such as a sandwich - based sensor again st A eromonas or nucleic acid molecular beacons to discriminate between Bacillus strains [20], [59] . Flow cytometry BactoScan (R) methods approved by the FDA can detect down to hundreds of cells, but samples mu st be diluted to prevent system clogging and use more expensive fluorescent - tagged antibody labels if cell differentiation is desired [18], [19], [53], [60], [61] . In addition, flow cytometers are known for the intensive work necessary to obtain reliable results [18] . Multifaceted manufacturing methods using specialized materials make many biosensors effectively uneconomical for broad production and application [21] [23] . Specialized ligands improve sensitivity, but also increase production and handling costs. Investigation into car bohydrate based ligands that attach to microbial targets is promising more economical options [24], [26], [28], [62] . 14 Electrochemical methods show real promise in improving microbial sensitivity while reducing detection time. Gold nanoparticles are electrically acti ve labels that assist in improving signal strength using electrical signaling, as well as providing strong colorimetric responses for optical sensors [5], [6], [31], [63] . Biosensors incorporating gold nanoparticles into their design may also reduce sensor costs applying carbohydrate ligands for pathogen surface labeling [35], [64] . 2.5 Conclusio n s Understanding the seriousness of pathogenic bacterial contamination in our food supply to human health makes apparent the importance of reliable microbial detection. In extension, identifying which pathogens are most infective , their prevalence in ca using foodborne disease, and which food s they most contaminate will help to reduce human illness. Eliminating how this microbial contamination enters the food supply chain will then reduce supplier costs and improve consumer confidence in the food products . Food regulatory agencies around the world, such as the FDA and USDA, address bacterial limits allowed in foods, as well as steps that are necessary to ensure industry food products meet these levels before sale. Numerous pathogens exist in nature, making their way into our foods, but six strains of E. coli have been identified as so detrimental to foods such as milk may contain Bacillus or Listeria, causing illnesses from mere diarrheic to invasive infections. Emerging nano - biosensors show promise in providing rapid detection of food pathogens, as opposed to the gold standard BAM methods, but many are still uneconomically designed or lack reliability in sens itive detection. Development of economical biosensors that are user - friendly and easily adapted to new pathogen and food combinations may increase food testing 15 along the food supply chain and further reduce foodborne outbreaks. As more reliable nano - biosen sors are designed, food supply safety will be more reliable . 16 Chapter 3 . Emerging N ano - biosensing with S uspended MNP M icrobial E xtraction and E lectrically - A ctive N anoparticle L abeling 3. 1 Introduction Many biosensing assays are too expensive or technically complicated to be economically feasible in many regions of the world. Sensor component miniaturization has led to nano - sized microbial biosensors [17], [ 65] [69] . Nano - biosensors are now prevalent within medical, environmental and even personal - use areas [ 70] [72] . Their cost is of little concern to the industrial user; but in under - developed countries, cost is the least concern. Many nano - biosensors designed to date also require expensive test equipment, environmentally sensitive components, refriger ation, and advanced training [73], [74] . Economical methods should require minimal storage and technology, with long shelf life, such as those developed in our Nano - Biosensors lab using carbohydrate - functionalized MNP and dextrin - coated gold nanoparticles (AuNP). A truly successful biosensor design incorporates high sensitivity with fast responses that are easily deciphered and require little manual labor and supplies. Current nano - biosensing methods show specificity that ranges from non - selective bacterial attachment to surfaces, to selective capture using monoclonal antibodies or DNA - aptamer binding. ectrical and electrochemical signaling. Magnetic nanoparticles (MNP) are beneficial in all of these applications due to their large surface area to volume ratio and superparamagnetic properties. Immobilized surface ligands capture bacteria, and the MNP - cel l complex can then be separated rapidly from even complex solutions [5] . Carbohydrates are becoming an economical ligand option. Cell surfaces and carbohydrate functional groups form non - covalent, electrostatic forces, similar to the antigen - anti body bonds, which then persist throughout the extraction process. Recent development have found carbohydrates with increased specificity to target bacteria [69], [75] . Electrically - active 17 nanoparticles (EANP) additionally provide more sensitive electrochemical or spectrophotometric detection methods [2], [76], [77] . Reviewed here are state - of - the - art nano - biosensing methods designed for rapid bacterial extraction using suspended MNP. Selective versus non - selective functional groups are compared with respect to their capture ability. Those MNP methods proven in more co mplex matrices, which closely replicate food sampling, are also evaluated. Some MNP applications additionally use EANP to intensify optical detection or electrochemical methods. Carbohydrate ligands from a broad spectrum of research areas are discussed wit h regards to selectivity. Finally, a novel rapid (30 min) MNP - cell - AuNP nano - biosensing method is introduced that exclusively uses biocompatible carbohydrate ligands immobilized onto MNP and AuNP to detect the presence of E. coli in milk using spectral and electrochemical detection, as shown in Figure 3. 1. 3. 2 MNP - assisted Biosensing Properties Functionalized magnetic nanoparticles (MNP) allow rapid capture and extraction of pathogenic bacteria from liquid matrices in real - time, as well as target concen tration with minimal liquid handling. MNP have large surface area to volume ratios, providing for high surface functionalization. Many applications conjugate biological compounds to the MNP surface, such as antibodies or nucleic acid aptamers, to improve t heir specificity [31], [78] [85] Park et al. MNP - aptamer biosensing was able to visibly detect up to 5 log CFU/mL Salmonella [83] . But random DNA orientation or antibody cross - reactivity can reduce assay sensitivity [86] [88] , as also does matrix interference [89] . 18 Figure 3.1 Schematic of the MNP - cell - AuNP biosensor for total bacterial load. Carbohydrate - functionalized MNP extract E. coli from artificially contaminated milk, while dextrin - coated EANP attach to captured cells at different cell receptors. Negative controls consist of sterile milk where MNP - EANP complexes form due to carbohydrate attraction between the functional groups. Cell quant ification is possible using FDA bacteriological analytical methods (BAM) agar plating (top), where neither MNP nor EANP inhibit bacterial growth and negative controls will have no cell growth present. Rapid biosensing is possible with spectrophotometric or electrochemical signaling, where negative controls will produce larger signals due to higher EANP presence than samples, due to saturation binding of EANP to MNP. MNP are synthesized through a variety of methods, including microemulsion, co - precipitation and thermal decomposition, with a variety of final compositions, including iron oxides such as Fe 3 O 4 or alloys such as FePt and CoPt 3 . An excellent review of these synthesis methods was written by Lu et al. [90] . The aforementioned MNP functional groups further protect the nanoparticle s from degradation. Various carbohydrate and amino acid ligands have been used as antibody substitutes, making the MNP more biocompatible for biological sample applications. Examples of carbohydrates, and other carbon - based, ligands include lectins, mannose, oligonucleotides, galactose, glucose, caprylic acid, cysteine and chitosan [86], [91] [96] . As a group, carbohydrates 19 retain their struc tural integrity and chemical reactivity when immobilized onto the surface of MNP. The chemical reactivity of each surfactant then affects MNP suspension and capture dynamics. Chitosan is a cationic biopolymer derived from chitin with a high percentage of a mino groups that aid in MNP suspension. Chitosan may be used directly in MNP preparation using three different methods, micro - emulsion, cross - linking and covalent binding [97] . Amino groups improve nanoparticle suspension in aqueous solution [98] , and improve proximity between biological target and MNP. Galhoum et al. functionalized MNP with chitosan crosslinked to cysteine [99] , further increasing amino group presence. Reactive chitosan amino gr oups can form amide bonds with carboxyl groups from carboxylated sugars, such as oligosaccharides [100] . Oleic acid, citrate, and ethylene glycol are other biocompatible coatings used on nanoparticles [101] [103] . Each carbohydrate displays a different specificity towards biological target epitopes based upon their chemical moieties. Cysteine amino acid has three reactive moieties: carboxyl, amine, and thiol groups. It was proposed that the carboxyl O and thiol S groups bind to the iron oxide nanoparticle core [104], [105] , while the positively charged amino group electrostatically binds the highly negative cell surface [95] . Clues to carbohydra te specificity are found studying human cell epitope binding, which microbial cells also bind during infectious disease initiation. Mannose, glucose and galactose sugars show varying specificities to different cell surface markers [100], [106], [107] , binding the Fim - H surface molecule on Escherichia coli [108] . Although chitosan is a natural antimicrobial agent [109] [111] , its bonding strength to bacterial cells makes it a desirable MNP surfactant for extraction purposes. Its polycationic structure binds easily to the anionic surface markers of microorganisms [109] . The expediency of MNP extraction minimizes any deleterious effects. Also at high pH 10, chitosan is nearly neutral , 20 but at more neutral conditions such as in many liquid foods (pH 5 to 7) chitosan is more positively charged and more effective in cell attachment [109], [112] . The combination of the selected synthesis method, core composition and surfactant determi ne the final MNP properties, particularly free space permeability, particle dispersion or aggregation, biocompatibility, and their mechanical and electrochemical stability. MNP magnetization saturation (M s ) properties can be identified using hysteresis mag netization plots (Figs. 3.2 A & B), which show that application of an external magnet quickly saturates the nanoparticle magnetic field, aligning the magnetic moments, causing MNP to migrate to the magnetic force [100], [113], [114] . When the applied magnetic field is zero, superparamagnetic nanoparticles do not retain a net magnetic moment. Iron - oxide - MNP display strong superparamagnetism , especially below threshold diameters of 50 nm. These MNP do not aggregate without an external magnet present and therefore MNP are easily suspended evenly into liquid solutions, and following microbial capture, are easily separated from the supernatant u pon application of an external magnet. This eliminates the need for time - consuming centrifugation to separate bacteria from food matrices. Values for MNP M s vary, due to ferric oxide core diameter and carbohydrate functional group. As shown in Table 3. 1, s urfactants reduce the magnetic moment response. For example, MNP prepared by Mahdavi et al. without surfactant showed increasing M S as their diameter increased, but for the largest 25 nm diameter with oleic acid surfactants, M S dropped from 81 electromag ne tic units per gm ( emu/gm ) core also reduce saturation magnetization values. Therefore, a direct relationship between overall diameter and M S cannot be expected if other parameters vary. In comparin g the carbohydrate - 21 functionalized MNP produced in our Nano - Biosensors Lab against others, the M S values are relatively strong. Figure 3.2 Examples of hysteresis magnetization for magnetic nanoparticles (MNP) functionalized with carbohydrate groups, (A) from Alocilja Nano - Biosensors lab and (B) used in metal recovery [99] (emu = electromagnetic units). 3.3 Suspended MNP - assisted Biosensing magnetization prope rties and surface functionalization. Table 3. 2 provides detailed information concerning biosensing applications using suspended MNP, functionalized with either selective or non - selective surfactants. Antibody surfactants have higher specificity for their t argets, for example against Listeria monocytogenes showing a linear detection response between 10 3 to 10 6 colony forming units per mL (CFU/mL [77] . Jansaento et al. designed a deoxy ribonuclease acid (DNA)/polymerase chain reaction ( PCR) method that used aptamer - based MNP that achieved a detection limit of 4 log CFU/mL [115] . Although antibodies, genetic aptamers, and fluorescent tags are selective to their targets, improving assay sensitivity, they are expensive, requiring special preparation and storage [116], [93], [115], [74], [89], [117] . 22 Carb ohydrate ligands are economical options as MNP surfactants against microbes. Currently many reported carbohydrate ligands are non - selective, but still successfully detect various biological targets. Lin et al. detected 10 1 to 10 8 CFU/mL E. coli using MNP - chitosan/glutaraldehyde with wireless magneto - elastic resonance [118] . Vancomycin - polyethylene - glycol - MNP was used to extract Listeria , among other Gram - positive bacteria using UV trans - illumination de tection [119] . Wang et al. used cetyltrimethylamm onium bromide coated MNP to recover up to 1000 ng/L estrogen from pork samples [96] , whereas chitosan - MNP recovered over of spiked morphine from synthetic urine and processed serum [92] . Work in the Nano - Bi osensor Lab found glycan - coated - MNP and cysteine/glycan - coated - MNP captured 2 parts per thousand (ppt) to 120 ppt of microbial contamination directly from undiluted milk, and over 0.002 ppt to 0.011 ppt from beef juices, orange juice and apple cider [29] , providing sufficient bacterial presence for sensitive detection. A major advantage of MNP is their ability to separate biological targets from their matrices, even concentrating captured ba cteria, using only an external magnet. Most of these reported methods were tested in simple buffer or food rinsate solutions such as the E. coli detection by Liu et al. , but food matrix interference was not studied for subsequent detection using refractive index [80] , fluorescence spectrometry [127] , or cycli c voltammetry [1 28] . Concentrated MNP - target, though, can improve optical detection [87] . MNP may even be incorporated into subsequent detection steps, such as interdigitated microfluidic impedance [89] or chemiluminescence [82], [85] without signal interference. MNP even provided optical signaling in Salmonella quantification [86] . Two pitfalls in suspended MNP biosensing are cross - reactivity against non - target species and matrix interference with detection signaling. E ven using selective antibody ligands, Listeria 23 Table 3.1 Magnetization properties of various magnetic nanoparticles (MNP) with carbohydrate functional groups. Physical Diameter (nm) Surfactant (functionalization) Magnetization Saturation (emu/gm) Synthesis Method Reference 21 Oleate 17 Thermal Decomposition w/o O 2 [120] 21 Oleate 74 Thermal Decomposition w/ O 2 7.83 None 58 Hydrothermal [102] 8.29 70 9.41 78 < 25 None 81 Oleic Acid 58 120 Glutaraldehyde w/ antibody ~ 60 Solvothermal [121] 10 Chitosan - coated 45.09 Coprecipitation [122] 8 Chitosa n - coated w/ cellulase enzyme 37.58 (<2000) Cysteine - glutaraldehyde /Chitosan 1.15 [123] 7 25, with 150 - 250 aggregation Chitosan w/ epichlorohydrin crosslink to cysteine 21.51 Coprecipitation [99] 50 55 Alginate 59.3 Coprecipitation [124] 80 120 Chitosan 56.8 18 - 20 None 62 Coprecipitation [125] 103 23% Chit osan 39 58 15% Chitosan 25 14 Chitosan - FITC fluorescent marker 53 Coprecipitation [112] 14 None 55 8.5 None 58 Coprecipitation [126] Low mw chitosan 29 High mw chitosan 18 8 None 61 Thermal Decomposition Low mw chitosan 30 High mw chitosan 25 5 15 (none) 83.87 Coprecipitation [29] Amino Acid coating 47.06 Amino/Glycan 28.31 detection showed 50% cross reactivity [88] and a Staphylococcus assay showed 40% cross reactivity [87] , both against non - target pathogens. Another MNP - antibody method to detect Listeria cross - reacted with Staphylococcus [86] . Non - selective, carbohydrate surfactants show 24 broad spectrum capture, but vancomycin - PEG - MNP suffered only 20% cross reaction with Gram - negative bacteria when combined with PCR using L. monocytogenes specific primers [119] . When food matrices such as beef wash solution or pork extracts were introduced into the sampling method, detection sensitivity was reduced with capture at 70% [89] . Carbohydr ate - functionalized - Biosensors Lab showed similar matrix interference, but matrix effects were factored in the final signal interpretation using controls. 3.4 - Even though, superparamagnetic magnetic nanoparticles have been exploited in numerous biotechnology applications, their usefulness may be limited by thick liquid or homogenized food matrices where magnetic separation may be diminished. Affixing or embedding the functio nalized MNP onto a support has shown retention of their desirable properties such as thermal magnetic dissipation and easy functionalization while allowing expedited separation of chemicals and bacteria from contaminated liquids. MNP have increased surface area, and functionalized MNP retain their capture ability, with low mass transfer resistance. Methods that exploit immobilize d MNP include mixing them into carbon paste to improve optical and electrical properties of electrodes; or embedding them into pol ydimethylsiloxane to make use of MNP thermal properties by application of an AC magnetic field [132], [133] . MNP functionalized with tridecafluorooctyl triethoxysilane were deposited on various solid and fabric surfaces to create a hydrophobic surface, adhering through a simple air - drying [134] . Once the MNP are affixed to a solid surface, t hey are available to use for biological extractions within complex matrices, exploiting their numerous properties, but allow easy removal from the matrix following target binding. 25 Table 3.2 Biosensing methods using suspended functionalized magnetic nanop articles (MNP). Biosensor Design & Signal Generation Ligand and Target & Sample Matrix Method Aspects Linear Range Reference MNP (30 nm) capture & load polyurethane - coated Metglas w/ resonance frequency shift detection of magneto - elastic sensor loading N on - selective chitosan/ glutaraldehyde E. coli in phosphate buffer (8 min incubation) 8 min wireless frequency shift signaling vs untested against food matrix effects 10 1 10 8 CFU/mL [118] MNP (30 nm) capture w/ impedance detection on microfluidic interdigitated gold - microelectrode Selective antibody Listeria in phosphate buffer and food rinse water (2 hr incubation) User - frien dly & specific w/ rapid detection vs costly ligand, poor reliability & spiked beef wash solution matrix interference Threshold detection of 10 3 10 7 CFU/mL (poor reproducibility across range, w/ >70% capture efficiency) [89] MNP (10 - 20 nm) capture and ethanol separation w/ HPLC - DAD detection (1) Non - selective CTAB - coated MNP@caprylic acid MNP estrogens in pork (1 min incubation) Quick target recovery (standards) vs matrix effects upon signaling (food) 5 1000 ng/L for standards (w/ > 90% recovery) [96] MNP capture & PFBT dis - placement w/ fluorescence spectrometry detection of unbound PFBT (2) Non - selective q - MNP - 1 , 2, or 3/PFBT 8 bacteria in phosphate buffer (20 min incubation) Microbial discrimination w/o expensive ligands vs complicated chemistry & linear discriminant analysis 10 7 CFU/mL (poor LDA separation) vs OD600=0.2 (good LDA separation) [129] Aptamer/MNP (20 nm) dis - placement & TMB oxidation w/ absorbance spectrometry detection of supernatant (3) Selective MNP/ oligo - nucleotide Salmonella in water (15 min incubation) Easily modified for new bacteria & simple detection vs complicated 1 - step chemistry Tested at 7 .5 x 10 5 CFU/mL [83] MNP (100 nm) capture & PCTE filtration w/ HRP chemiluminescence detection (4) Selective HRP - MNP - antibody Salmonell a in PBS (30 min incubation) Minimal sample size vs high negative background signal & expensive PMT luminometer 10 1 10 4 CFU/mL [82] MNP capture & 1.2 µm pore nitro - cellulose/PDMS vacuum - filtration w/ color intensity (optical density) detection (5) Selective antibody Salmonella in PBS (30 min incubation) Minimal sample size vs vacuum - filtration & specialized ChemiDocTM MP detection 10 1 10 4 CFU/mL (varying capture efficiencies of 90 8%) [84] MNP capture & 1.2 µm po re nitro - cellulose/PDMS vacuum - filtration w/ color intensity (optical density) detection (5) Selective antibody Listeria in PBS (30 min incubation) Minimal sample size vs vacuum - filtration & cross - reaction with Staphylococcus and Listeria spp. 10 1 10 3 CFU/mL (varying capture efficiencies of 89 48%) [86] MNP capture & Au - chip loading w/ SPR detection of change in refractive index (6) Selective MNP - Ab1 - & Au - Chip - Ab2 Salmonella in PBS (40 min incubation) Signal amplification w/ MNP sample concentration 10 1 10 9 CFU/mL (3 - order improvement from 10 4 CF U/mL w/o MNP) [80] 26 Biosensor Design & Signal Generation Ligand and Target & Sample Matrix Method Aspects Linear Range Reference MNP capture w/ magnetic resonance relaxometry against a negative control Selective antibody Listeria in 2% milk stabilizer (40 min incubation) Specific Ab - silica - MNP capture vs almost 50% cross reactivity against 4 non - target pathogens 10 0 10 3 CFU/mL [88] MNP extraction, subsequent cell heat release, DNA extraction, PCR & gel electrophoresis w/ UV t ransillumination of ethidium bromide staining (7) Non - selective vancomycin - PEG Listeria & Gram(+) in PBS & lettuce homogenate (45 min incubation) MNP signal concentration vs vancomycin binds broad - spectrum Gram(+) w/ > 20% Gram( - ) false positives & overa ll 4 hr processing time Extracted bacteria between 10 1 10 6 CFU/mL at > 70% capture efficiency for PCR [119] MNP capture w/ enzymatic chemiluminescenc e detection of AMPPD (8) Selective MNP - Ab1 & ALK - Ab2 E. coli (1 hr processing) Good specificity w/ little non - target cross reaction against 5 pathogens Tested at 10 3 CFU/mL [85] MNP extraction w/ magneto - phoresis separation Selective antibody E. coli in PBS High capture for possible bulk process detection Tested at 10 3 CFU/mL w/ > 97% capture [79] PANi - MNP capture & CNN membrane separation w/ resistance detection (9) Selective MNP - Ab1 & CNN - Ab2 - E. coli in Tris buffer Lateral flow design vs minimal stable detection window after 6 min 10 1 10 4 CFU/mL with threshold detection > 10 5 CFU/mL [81] ASMNP separation, PCR & gel electrophoresis w/ UV transillumination of ethidium bromide staining (10) Non - selective ASMNP & selective DNA PCR in buffer Broad spectrum DNA extraction (post cell lysis) vs complex sample preparation from contaminated raw milk Extracted DNA lysed from 10 0 10 4 CFU/mL Salmonella & Listeria [78] Au/MNP concentration & LFA inverse color line detection (11) Selective half - fragment Salmonella in milk (resuspended in Tween) Minimal matrix interference vs inverse threshold indicator Threshold detection above 10 3 CFU/mL [31] Au - MNP concentration & 0.8 µm CA filtration w/ optical density detection (12) Selective antibody & Staphylococcus in PBS and milk (30 min incubation) User - friendly, optical detection vs high LOD & 20 40% cross reactivity against 4 non - target bacteria Threshold detection 10 3 CFU/mL in PBS and 10 5 CFU/mL in milk (> 70% capture efficiency) [87] CMNP - CPE attachment & filtration w/ ferrocyanide mediated DPV detection (13) Non - selective chitosan morphine in PBS or synthetic urine & serum Improved sensitivity vs required centrifugation/ filtration to minimize matrix interference & 10 mL voltammetric cell volume 0.01 2 µM and 2 720 µM w/ > 90% recovery in synthetic urine & serum [16] MNP extraction & PCR amplification w/ optical HRP enzyme fluorescence detection (14) Selective DNA Campylob acter (3 hr method) after chicken skin culture in broth (24 hr) & cell lysis Fast PCR amplification w/ MNP - DNA amplification vs 24 - hr enrichment media incubation of chicken skin pg DNA or 10 4 CFU equivalents/ mL [115] 27 Biosensor Design & Signal Generation Ligand and Target & Sample Matrix Method Aspects Linear Range Reference CMN extraction & CN electrospun LF biosensor separation w/ resistance detection (15) Selective CMN - mAb & biosensor - pAb E. coli O157:H7 in peptone water Fast lateral flow, real - tim e detection vs expensive ligand & limited 8 - min detection window 10 1 10 4 CFU/mL [81] EAM extraction & electrochemical CV transducer detection (16) Selective EAM - Ph - Pro - DNA & SPCE - PRO - bio - DNA amplified Bacillus anthracis DNA in water High target sensitivity vs extended preparation time (w/ PCR amplification & purification) Sensitive 0.01 [128] AMN extraction & TEM analysis (17) Non - selective glycan carbohydrate E. coli O157:H7 Cell concentration w/o centrifugation proof - of - concept (not applicable) [130] MNP extraction & electrochemical redox ferrocyanide - mediated CV detection (18) Broad - based chitosan or chitosan/cysteine Salmonella, E. coli, Baci llus & Listeria in PBS, milk, beef juices & apple cider (10 min incubation) Total bacterial quantification from undiluted complex liquids, minimal capture loss due to liquid components & optional dip - stick method vs necessity to calibrate biosensor per liq uid type 10 4 10 8 CFU/mL w/ capture in undiluted liquids of 2 ppt to 120 ppt in mi lk and 0.002 ppt to 0.11 ppt in beef & apple cider (500 uL 25 mL volumes) [131] (Table 3. 2 biosensing definitions: (1) HPLC - DAD = high performance liquid chromatography diode array detection, CTAB = cetyltrimethyl ammonium bromide; (2) PFBT = in - lab produced anionic polymer functionalized with side - chain carboxylic acid group, q - MNP - 1, 2 or 3 = quartenized MNP functionalized 3 types of organic bromides, bacteria = S. oneidensis, V. fischeri, M. luteus, E. tarda, E. coli. V.alginolyticus, P.aeru ginosa, and P.pastoris; LDA = linear discriminant analysis; (3) TMB = 3,3,5,5 - tetramethylbenzidine (oxidized via H 2 O x upon aptamer displacement from MNP); (4) PCTE = 0.6 - - etched filter, HRP = horse - radish peroxidase, PMT = photo - multiplier tube; (5) PDMS polydimethylsiloxane film possessing six holes with 3 mm diameters, PBS = phosphate buffered saline; (6) SPR = surface plasmon resonance, Ab1 = monoclonal and Ab2 = polyclonal anti - Salmonella antibodies, Au - chip = gold chip; (7) PEG = polyethylene glycol, Gram(+) = Gram - positive bacteria, Gram( - ) = Gram - negative bacteria, PCR = polymerase chain reaction; (8) AMPPD = 3 - (2 - spiroadamantane) - 4 - methoxy - 4 - (3 - phosphoryloxy) phenyl - 1,2 - dioxetane substrate of ALK enzyme, Ab1 = polyclonal a nti - E. coli antibody, ALK = alkaline phosphatase enzyme, Ab2 = monoclonal anti - E. coli antibody; (9) PANi - MNP = (electrically conductive) polyaniline MNP, CNN = cellulose nitrate nanofibers, Ab1 = monoclonal anti - E. coli antibody, Ab2 = polyclonal anti - E. coli antibody; (10) ASMNP = amino - modified silica - coated magnetic nanoparticles, PCR = polymerase chain reaction, DNA = deoxyribonucleic acid; (11) Au/MNP = gold - coated magnetic nanoparticles, LFA = lateral flow immunoassay; (12) Au - MNP = gold nanoparticle - coated MNP, CA = cellulose acetate, LOD = limit of detection; (13) CMNP - CPE = chitosan magnetic nanoparticle (functionalized) carbon paste electrode, DPV = differential pulse voltammetry; (14) PCR = polymerase chain reaction, HRP = horse - radish peroxidase enzyme; (15) CMN = conductive magnetic nanoparticle s w/ electrically - active PANi shell, CN = cellulose nitrate, LF = lateral flow, mAb = monoclonal antibody, pAb = polyclonal antibody; (16) EAM = electrically - active magnetic nanoparticles w/ electrically - active PANi shell, CV = cyclic voltammetry, Ph - Pro - DNA = phosphorylated DNA probes, SPCE - PRO - bio - DNA = screen printed carbon electrode biotinylated DNA probe; (17) AMN = Alocilja magnetic nanoparticles; (18) CV = cyclic vo ltammetry , ppt = parts per thousan d .) 28 3.5 Enhanced Biosensing with Electrically - Active Nanoparticles Electrically - active nanoparticles (EANP) have strong surface plasmon resonance (SPR) properties that make them ideal for nano - biosensing methods. EANP generate local electromagnetic fields through conductance of electrons, detectable by optical, spectral and electrochemical instrumentation. Microbial surfaces labeled with EANP show amplified signaling for higher assay sensitivities and detection limits down to 10 1 CFU/mL. Nanoparticles used in MNP - cell - EANP nano - biosensing (Table 3. 3) include those produced from titanium, gold and other transition metals, yttrium and other rare metals, quantum dots from lead sulfide and silica dots [77 ], [121], [135] [137] . AuNP surface plasmon resonance (SPR) occurs within the UV - visible region between 380 to 780 nm, making them highly suitable as colorimetric biosensors [34], [35], [63], [138], [139] or microbial concentration can also be determined spectrophotometrically [140] . Particles with diameters below 30 nm show absorbance maximums at approximately 520 nm. Other SPR absorbance maximums are 230 nm for titanium oxide nanoparticles (TN) or 630 nm for quantum dots (QD) photoluminescence . Rare earth metal nanoparticles can be designed to fluoresce at varying wavelengths, allowing simultaneous detection of multiple targeted bacteria [76] . As with MNP, EANP surface ligands determine their binding properties to targeted biological compounds. Citrate is a conventional surfactant for AuNP [138], [141], [142] , but other capping agents have been investigated for microbial bin ding specificity. In evidence of surfactant specificity, cysteine - capped AuNP were able to bind E. coli more effectively than citrate - capped AuNP, purportedly from positively charged amino group attraction to negatively charged microbial surfaces [95] . 29 Few reported biosensing methods combine suspended MNP capture with EANP labeling. Notably, many of the reported methods summarized in Table 3. 3 target Salmonella in their development, using antibody or aptamer ligands on both MNP and EANP with optical or spectral signaling for detection [76], [136], [137] . In direct signaling, the MNP - cell - EANP complex is directly quantified. For example, Duan et al. (20 12) used rare earth nanoparticles functionalized with DNA aptamers to detect Salmonella and Staphylococcus (10 1 10 5 CFU/mL) from filtered environmental water samples [76] . Inverse methods quantify microbial presence by the level of unbound EANP in the supernatant. Levels of unattached EANP identified Salmonella presence (10 2 10 5 CFU/mL) in diluted milk with AuNP labeling [135] , or residual TN quantified Salmonella (10 2 10 8 CFU/mL) in milk [121] . AuNP functionalized with both antibody ligand and urease enzyme was used in a biosensing assay that monitored impedance changes to detect Listeria [77] . Dextrin is a novel capping agent used to reduce gold chloride during AuNP production [143], [144] . The prevalent negatively charged hydroxyl groups of dextrin bind to the positivel y charged surface gold atoms, protecting the nanoparticles against aggregatio n in liquid suspension. This oligosaccharide coating also improves AuNP/dextrin suspension in aqueous - based solutions by its low molecular weight, pKa of 12.3, and external hydroxyl chemical groups. Dextrin is also biocompatible, for example reducing selen ium nanoparticle toxicity from 95% against human cells to 15% [145] . Dextrin - c oated AuNP developed in our Nano - Biosensors Lab successfully detected Salmonella and E. coli from undiluted milk in a proof - of - concept design, described below. Comparison of MNP - cell (Table 3. 2) versus MNP - cell - EANP (Table 3. 3) nano - biosensing methods show s EANP SPR properties improves optical and spectral detection. This advance s development of rapid, simple and economical microbial assays, improving subjective visual tests 30 Table 3.3 Nano - biosensing methods with functionalized MNP and NP labeling . Biosensor Design & Signal Generation Ligand and Target & Sample Matrix Method Aspects Linear Range Reference MNP capture & AuNP label w/ optical & absorption spectrometry detection (520 nm) of unbound AuNP (1) Selective MNP - & AuNP - aptamer1,2 Salmonella in strained & 1:20 diluted milk (45 min incubation) User - friendly 1 - step specific method w/ simple optical detection vs >25% cross reaction w/ E. coli & Listeria at 10 5 CFU/mL 10 2 10 5 CFU/mL (inversely propor tional to signaling) [135] MNP capture, TN label w/ absorption spectrometry detection (230 nm) of unbound TN (2) Selective MNP - mono - & TN - poly - clonal antibody Salmonella in milk (20 min incubation each) User - friendly & specific w/ simple detection vs costly ligand, 2 - step method & non - target signaling 10 2 10 8 CFU/mL (inversely proportional to signaling) [121] MNP capture & QD labeling w/ photoluminescence spectrometry detection (630 nm) (3) Selective MNP - Ab1 and QD - Ab2 Salmonella in phosphate buffer (30 min incubation) 1 - step capture - n - label & simple detection vs expensive Ab ligands 10 3 10 8 CFU/mL [136] MNP capture & YYbErTmNP labeling w/ fluorescence spectrometry detection ( Salmonella at 452 nm; Staphylococcus at 660 nm; for 980 nm excitation) (4) Selective aptamers DNA1 Salmonella and DNA2 Staphylococcus in buffer (40 min incubation) High specificity w/ easily synthesized aptamer & 1 - step batch method vs complicated NP chemistry & high cross - reactivity to E. coli & Enterobacter with DNA1 10 1 10 5 CFU/mL (>70% capture efficiency from filtered environmental water samples: lake, stream & puddle) [76] MNP capture & Blue - SiNP labeling w/ visual optical color density confirmation (5) Selective MNP - Ab1 & Blue - SiNP - Ab1 Salmonella in saline (pH 2 to pH 13) (30 min MNP & 15 min Blue - SiNP incubations) Strong visual calib ration & no non - target pathogen cross - reaction vs 2 - step capture & label w/ multiple liquid changes 10 1 10 9 CFU/mL (1 mL mixed target bacteria) 10 4 10 9 CFU/mL (20 uL mixed target bacteria) [137] MNP capture, AuNP labeling, & urea reaction w/ supernatant im pedance change detection of inter - digitated microelectrode loading Selective MNP - mono - & urease - AuNP - poly - clonal antibodies Listeria in PBS or lettuce wash (45 min MNP & 30 min AuNP incubations, 30 min urease catalysis) Small sample and nanoparticle volu mes vs multi - step process w/ multiple liquid handling steps 10 3 10 6 CFU/mL (> 90% recovery from spiked lettuce samples at same concentrations; no E. coli cross - reactivity) [77] MNP capture & AuNP labeling w/ differential pulse v oltammetry (DPV) (6) Selective MNP - mAb & AuNP - pAb E. coli in PBS (45 min MNP, AuNP & DPV) Small samples and expedited detection vs complicated Ab conjugation chemistry and random MNP - AuNP binding 10 1 10 6 CFU/mL (log - linear between contamination concent ration vs DPV - AuNP current response) [6] 31 Biosensor Design & Signal Generation Ligand and Target & Sample Matrix Method Aspects Linear Range Reference MNP capture, AuNP labeling w/ electrochemical HCl - activated DPV (0.4 V) or spectrometry detection (520 nm , or novel step - integrated spectral absorbance ) Broad - based MNP - amino/glycan & AuNP - dextrin carbohydrates Salmonella & E. coli in PBS & milk (30 min incubation ; total 45 min detection ) 1 - step capture & label of total bacterial load from undiluted complex liquid vs proof - of - concept stage Tested at 10 3 10 5 CFU/mL [29] (Table 3. 3 biosensing method definitions: (1) MNP = magnetic nanoparticles, AuNP = gold nanoparticles, aptamer 1 & 2 = Salmonella DNA sequences; (2) MNP= magnetic nanoparticles, TN= t itanium oxide nanoparticles, CFU = colony forming units; (3) QD = quantum dot, Ab1 and Ab2 = two different monoclonal anti - Salmonella antibodies; (4) YYbErTmNP = (rare earth metals) yttrium ytterbium erbium thulium nanoparticle s which are Lanthanide - doped near - infrared (NIR) - to - visible upconversion - nanoparticles (UCNPs) , DNA1 = aptamer for Salmonella, DNA2 = aptamer for Staphylococcus; (5) Blue - SiNP = silica nanoparticles functionalized with Reactive Blue 14 (Zhejiang Shunlong Chemical Co. Ltd, China).) 32 and replacing complicated methods requiring expensive instrumentation. Spectral detection of MNP can mask bacterial presence due to their wide absorbance band [2] , therefore secondary reactions, such as TMB ( 3,3,5,5 - tetramethylbenzidine ) oxidation [83] or HRP (horseradish peroxidase) enzymatic chemiluminescence [82] (Table 3. 2) were required. All selecti ve and non - selective methods suffer false positive results from cross - reactivity, which minimizes assay efficacy. The trade - off between expensive selective ligands, complicated technology and time - consuming procedures versus economical carbohydrate ligands and user - friendly methods with easily - adapted signaling should be a driving force in improving carbohydrate - based MNP - cell - EANP nano - biosensing. 3.6 Carbohydrate Ligands Against Microbial Targets Carbohydrate ligands with selective binding against targe ted pathogens can be identified from many sources. Microbial virulence is initiated through bacterial surface protein lectin attraction to tissue carbohydrate ligands. These protein - carbohydrate interactions can be exploited to target pathogens in our food sources. Selectivity can even be improved by addition of binding moieties to the ligand [146] , and synthesis methods are producing a broad range of possible - glycosides of branched o ligosaccharides inhibited type 1 fimbriated E. coli attachment to yeast or tissue cells [147] . Higher monosaccharide numbers in the carbohydrate structure, particularly mannose and fucose, improve multivalent interaction to incr ease binding strength [148] . Table 3. 4 lists several of these carbohydrate ligands identified through these glycobiology or synthesis chemistry investigations . Several seemingly unrelated review articles have identified selective carbohydrate ligand options. Type 1 fimbriated uropathogenic E. coli (EPEC) pathogen was inhibited from binding 33 - D - mannos ide virulence lectin residue [149] . Tissue cell haemaglutinin lectins associated with bacterial virulence were iden tified by replicating the protein - carbohydrate binding using glycoconjugate receptors [150] . Researchers found that C - type lectin immu lectin - 2 from the tobacco hornworn, Manduca sexta binds E. coli or Staphylococcus - 1,3 glucan [75 ] . Even a hospital epidemic identified how Clostridium difficile strains acquired the ability to bind trehalose, a food substitute [151] . Carbohydrate ligands offer economical options to larger antibodies and DNA ligands. Fratila et al. reviewed magnetic glyconanoparticles with increased carbohydrate functionalization for rapid cell extraction [100] , which is a technique utilized in our Nano - Biosensors lab. Various carbohydrate ligands have shown promising binding properties, as summarized in Table 3. 4. Several of these biosensing methods use mannose or fucose residues [69], [152] [155] , or plant lectins [156], [157] to target pathogenic bacteria. In addition, the terminal sugar residues of C - type lectins were identified as the primary targets for binding interaction [75], [158] . Protein - carbohydrate bonds identified in Table 3. 4 all use electrostatic forces, similar to those used by antibodies. The E. coli Fim - H surface molecule uses hydrophobic binding to attach to mannose - ri ch oligosaccharides [69] . Hydrogen binding force s are present between sugar hydroxyls and amino acid residues in microbial surface C - type lectins [75], [158] . Type 1 piliated E. coli HB101 attached organic polymer through multivalent acidic and mannose hydrogen bonds [153] . Some bind ing schemes also require cation cofactors Ca 2+ or Mn 2+ for binding between surface lipopolysaccharides and lectins. 34 3.7 Microbial Surface Epitopes for Carbohydrate - functionalized MNP - cell - AuNP Nano - biosensing Microbial pathogens have two main goals: survi survival necessitates its ability to attach, either to surfaces or host organisms; involves secretion of toxins against competitors and regulatory cells; and requires uptake of nutrients for growth, regeneration and repro duction. Microbial cell surfaces are enveloped with molecular structures that fulfill their cellular functions. All of these same microbial surface markers, or epitopes, may a lso be used to target those microbes that contaminate our environmental and food supply sources. The human immune system itself produces antibodies and complement that target pathogen epitopes. Following is a brief review of these microbial epitopes which can be targeted by carbohydrate ligands in nano - biosensing assays. Microbial surface epitopes may be attached directly to the membrane surface, transverse surface through non - covalent forces. These molecules are complex, fr equently constructed from several sub - structures excreted to the cell surface. Epitopes are temporal structures, changing over time with growth and environmental conditions. Due to the complex nature of surface epitopes, many regions along the structure wh ich are not necessarily used in cellular processes may provide novel areas for selective MNP targeted capture. Nanoparticles can be functionalized with surfactants that target microbial surface epitopes for extraction and quantification from liquid foods . In immune responses, antibody binding is one of the most specific and strongest non - covalent binding mechanisms in cellular processes [159] . But antibody ligands are quite expensive and time - consuming to produce and require special handling. Curr ent carbohydrate ligands, such as those summarized in Table 3. 4, are less selective 35 in their binding targets but have proven their ability to attach to a broad spectrum of surface epitopes at fractions of the cost of antibodies and DNA. Plus, more selectiv e carbohydrate ligands are being discovered and synthesized. Lipopolysaccharide (LPS) and peptidoglycan moieties make up a substantial amount of prokaryotic cell wall structure. Gram - negative cell walls contain higher amounts of LPS, whereas Gram - positive cell walls are composed of a thick layer of peptidoglycan. LPS is toxic to the human body [160] and cause antibody immune responses. LPS is composed of a hydrophobic lipid A region, responsible for toxicity, which anchors the structure within the cell membrane, followed by a core polysaccharid e region, extending into the O - antigen domain which consists of repeated polysaccharides [161] . Peptidoglycan is composed of sugars and amino acids, with residues of N - acetyglucosamine or N - acetylmuramic acid. Peptidoglycan moieties can be furth er conjugated to secondary cell wall polymers (SCWP), covalently link to the exposed N - terminus [162] . Spread across the cell walls of all microbial types are peptidases and outer membrane proteins that assist in nutrient flow, pathogenesis, motility and cell wall turnover. Peptidases are specialized proteins secreted to the surface that interact with surface carbohydrates and lipids and act as enzymes with catalytic sites involved in peptidoglycan modification, cell wall turnove r and cellular locomotion [164], [165] . Adenosine triphosphate (ATP) - binding cassettes (ABC) are one major group of nutrient transporter proteins. Due to the ABC structure specificity among microbial species they are strong antigens, or targets, for immunoge nic responses, resulting in highly selective antibodies [166] . Outer membrane proteins (OMP) are porins that transverse the cell wall providing entry across the cell wall for a broad spectrum of nutrient exchange [161], [167] ), using active - transp ort of nutrients at low - concentration, and more passive diffusion of higher - concentration nutrients. Due to porin discriminatory transport of hydrophilic compounds by size 36 Table 3.4 Biosensing methods using carbohydrate functional groups for targ et capture. Carbohydrate Ligand & Modifications Microbial Target & Surface Epitope Identified Binding Moiety Binding Efficiency Reference Mannose - containing oligosaccharides with PTAM, PCAM, DAMP, glucosamine & chitosan; (immobilized onto gold electrodes and SPR surfaces) (1) FimH of E. coli Hydrophobic phenyl residues & aliphatic side groups Relative selectivity for E. coli at 86% vs 2.6% & 8.6% for Citrobacter freundii & Staphylococcus epidermis (detection for MED across 2 to 6 log CFU/mL, o r for SPR across 3 to 11 log CFU/mL) [69] Organic polymer glycopolythiophene with mannose Type 1 piliated E. coli HB101 (additional lectin & Influenza virus targets also described) Multivalent polymer acidic group electrostatic bonding and mannose H - binding along microbial receptors Visual red - shi ft with ligand binding to target [153] - D - mannose - pyranoside altered w/ attached fluorescent DBD dye (2) FimH of E. coli D (not addressed) Signal to noise optimized binding at ligand concentration of 10^ - 12 mol/L with 10 - min binding window [152] Plant - based lectins ConA, LCA, WGA, MAL, & UEA; (immobilized onto quartz crystal electrode) (3) Campylobacter jejuni & Helicobacter pylori LPS of lysed cells (ConA & LCA to mannose - N - acetylglucosamin e, UEA to fucose) Binding requires Ca 2+ & Mn 2+ ions 4 log CFU/mL microbial effect on crystal oscillation frequency against lysed cell samples, e.g. for C. jejuni strains ConA - HS:3 @ 60 Hz; evidence of selective lectin binding [156] C - type lectin SSL from Atlantic Salmo salar (4) Aeromonas salmonicida surface molecule LPS (which is also mannose - binding) and glycans w/ terminal GlcNAc; also binds Pseudomona s fluorescens, A. hydrophilia, Pichia pastoris, & Saccharomyces cerevisiae 3 to 4 hydroxyl groups from the sugar hydrogen bind with acidic and amino acid side chains of LPS in coordination with Ca 2+ Western blots showed SSL binding to bacteria & fluoresce nce - labeled SSL bound 10 glycans w/ GlcNAc terminal labels [158] Plant - based lectins from: WGA, Con A, UAE, PNA & MAL (5) E. coli O157:H7 surface mono - and oligosaccharide polysaccharide structures SEM imaging of WGA - coated biosensor surface with captured E. coli cells Linear response between 5 to 8 log CFU/mL for WGA ligand w/ higher signaling than Con A, UEA, PNA & MAL ligands [1 57] Fucose - bearing oligosaccharides Le b & blood group H type 1 FMNP (6) H. pylori J99 BabA cell surface fucose binding adhesin lectin Monomeric sugar binding to protein Confocal microscopy evidence of FMNP - cell binding & TEM evidence of FMHP - AAL aggre gation [155] 37 Table 3.4 ( Carbohydrate Ligand & Modifications Microbial Target & Surface Epitope Identified Binding Moiety Binding Efficiency Reference Purified immulectin - 2 C - type lectin (w/ extended terminal carbohydrate loop) from hemocytes cell surface containing 2 carbohydrate - recognition domains (7) LPS from E. coli K12 & E. coli 011:B4; peptidoglycan from E. coli K12, Staphylococcus aureus & Bacillus subtil is; zymosan & mannan from Saccaraomyces - 1,3 - glucan) Glutamate - proline - asparagine motif binds sugar motifs ELISA confirmation of ligand - target binding in 96 well plate @ 10 nM and 40 nM concentrations [75] Mannose monosaccharide equipped with ethanolamine linker to immobilize on microarray E. coli ORN178 (not addressed) Levels above 10 6 E. coli ORN178 stained with nucleic acid dye provided signaling above background, whereas E. coli ORN209 showed 4 - fold lower binding [154] 20 D - mannose derivative s (sulfonamides & cinnamides) Pseudomonas aeruginosa LecB (PA - IIL) surface lectin virulence factor Via three hydroxy groups from trisaccharide Lewis a , w/ ability to target adjacent protein cleft for improved affinity Inhibition ligand - target testing showed derivatives increased binding 4 to 24 - fold, which minimized pathogen tissue binding [163] Tridecafullerenes w/ 120 - count mannose residues attached to azide - substituted glycof ullerene scaffold with alkyl chain linkers (8) DC - SIGN lectin of EBOV infection DC - SIGN recognizes multivalent sugar residues on oligosaccharides 100% inhibition of 2.5 x 10 5 DC - SIGN @ >10 4 pM, with IC 50 of 667 nM [148] (Table 3. 4 biosensing definitions: (1) PTAM = p - thiolphenylaminomannose, PCAM = p - carboxyphenylaminomannose, DAMP = 1 - deoxy - 1 - aminomannopyranoside, SPR = surface plasmon resonance, FimH = mannose - specific fimbrial lectin adhesin molecule, and MED = metal - enhanced electrochemical detection; (2) DBD = [1,3]dioxolo[4,5 - f][1,3]benzodioxole; (3) ConA = Concanavalin A, LCA = Lens culinaris agglutinin, WGA = Wheat Germ agglutinin , MAL = Maackia amurensis lectin, UEA = Ulex europaeus agglutinin, & LPS = lipopolysaccharides; (4) SSL = Salmo salar lectin, LPS = lipopolysaccharides, & GlcNAc= N - acetylglucosamine; (5) WGA = Triticum vulgaris wheat germ agglutinin, Con A = Canavailia en siformis jack bean concanavlin A lectin, UAE = flowering plant Ulex europaeus aggluntinin,, PNA = Arachis hypogaea peanut agglutinin, MAL = tree species Maa ckia amurensis leukagglutinin, & SEM = scanning electron microscopy; (6) Le b = Lewis b, FMNP = fluor escent magnetic nanoparticles, BabA = blood group antigen binding adhesin A & AAL = fucose binding Aleuria aurantia lectin; (7) LPS = lipopolysaccharide; (8) DC - SIGN = dendritic cell - specific intercellular adhesion molecule - 3 - grabbing non - integrin receptor , EBOV = Ebola virus, & IC 50 = 50% inhibition concentration.) 38 and charge, they are targets of specific antibody production, especially for OmpD porins [168] [170] . Microbial cells detect their environment through their cellular wall molecular components. Chemotaxis proteins sense signals secreted from other microbes, OmpC porins sense changing osmolarity, and MSCRAMM (microbial surface com ponents recognizing adhesive matrix molecules) recognize connective tissue of mammalian tissue [171] [173] . Flagella are locomotive organelles attached to the surface of the microbe, composed of monomeric polypeptides with high variability in their protein composition with high specificity per microbial species as H antigens in immuno - reactivity serotyping [164], [1 74], [175] . Curli and fimbriae are hair - like projections involved in adhesion to surfaces and aggregation with other cells, containing hydrophobic amino acids lysine, proline, and serine, as well as a mannose - binding region [94], [176] [180] . Pili also acts in surface binding through hydrophobic amino acids that act [28], [181] [185] . As major nutrients, amino acid and sugars uptake is facilitated by transporter structures. Cysteine transfer across the cell membrane is performed in cystine form for E. coli [24] , Bacillus subtilis [186] , and Listeria monocytogenes [187] . Glycosylated milk compounds inhibited lectin proteins of Pseudomonas aeruginosa from galactose, fucose and mannose uptake [62] . Chronobacterium viola ceum lectin binding protein binds both fucose and mannose, but at different protein orientations [188] . Chitosan is a polycationic polysaccharide derived from acetylated chitin that easily binds the negative charged surfaces of microbial species, such as seen for E. coli, Pseudomonas, and Salmonella [189], [190] . The positively charged amino groups of chitosan also bind Bacillus vegetative cells to block nutrient uptake [191], [192] . 39 3.8 Conclusions Carbohydrate functionalized MNP are an economical means to expedite microbial extraction in mere minutes from simple and complex matrices such as milk , without the need for time - consuming centrifugation from large sample volumes. Upon extraction, MNP - cell complexes can be concentrated to less than 1 mL, resulting in stronger detection signaling. Although the initial carbohydrate ligands reported here bi nd and extract the total bacterial load present in samples, higher selectivity is possible through study of the cell surface molecular structures against a range of carbohydrates. Biocompatible ligands are desired as antibody - replacements due to the cost o f antibodies, particularly given the persistence of cross - reactions with non - target microbes. Regardless of the ligand used, though, it was shown that several methods were able to significantly quantify microbial presence with the MNP still attached to the ir target. Beyond the potentially intensified microbial biosensing via MNP concentration, EANP labeling, such as with quantum dots, TN or AuNP, of the MNP - cell complexes further improves test sensitivity with less complicated spectral and electrochemical detection methods. Microbial contamination along the food supply chain could be detected quickly using this economical method in frequent product checks, assuring suppliers the safety of their product and reducing substantially consumer illness. Additional ly, functionalizing MNP and AuNP with inexpensive biocompatible carbohydrates, such as dextin, further decreases assay costs, while reducing hazardous waste production through green chemistry. Combining carbohydrate - functionalized MNP cell capture and con centration, along with dextin - coated AuNP labeling resulted in promising nano - biosensing development towards the goal of feasible, less complicated tests. Much work in nano - biosensing development has focused heavily on designing highly specific, one - time - u 40 technology to produce. These are not cost - effective for many worldwide food suppliers, though. Even many of the suspended MNP/EANP biosensing assays reviewed here are still out of reach for many users, requiring special storage or highly technical equipment. Carbohydrate ligands are easily incorporated onto nanoparticles, and require little special handling, showing a long shelf life at room temperature. As more selective carbohydrate ligands are identified and i ncorporated into MNP - cell - AuNP assays, consumers can be assured of safer food supplies at affordable costs. While the proof - of - concept test presented here requires optimization for detection at lower cell concentrations, it was successful in detecting E. coli presence directly from milk, whereas many reported tests were studied only in simple buffer solutions. The future in nano - biosensing is not in developing more technically complex sensors, but applying technological know ledge to develop inexpensive, user - friendly microbial assays. 41 Chapter 4. Literature Review Technology 4.1 Introduction Designing reliable, user - friendly nano - biosensors to detect pathogenic bacteria does not require intricate combinations of the multitude components that define biosensors. The intricate design of the glucose meter, which uses a sensitive enzymatic reaction to quantify sugar levels in blood samples through miniaturized electronic components , is a bio - sensor style that has not been easily rep licated in other biological testing at an economical cost. Instead, separating components of the bio - sensing mechanism while utilizing inexpensive ligands, rapid extraction methods, and affordable signaling instrumentation will allow the design of a sensit ive, reliable and adaptable nano - biosensing system to aid in ensuring the safety of food supplies globally. Carbohydrate - based MNP - cell - AuNP bio - sensing complexes can be used to quantify bacterial presence utilizing either commonly available UV - Vis spectr ophotometer technology or an inexpensive handheld potentiostat , which uses electrically produced signaling . As spectrophotometers are more commonly available lab instrumentation, bacterial contamination levels can be rapidly quantified from MNP - cell - AuNP a bsorbance peaks in the UV - visible spectral region of 900 nm to 200 nm. Within developing countries, though, access to even minimally equipped labs is problematic, therefore, an in expensive handheld potentiostat for repeated on - site testing is a reliable op tion. electrochemical force, such as potential (voltage), current (amps), or due to non - equilibrium chemical concentration. These species are targets being quanti fied and may be organic or inorganic chemicals [193] [195] , with simple or complex molecular structures [196] [198] , even pharmaceuticals [15] , generated through chemical reactions or biological metabolism, an d may 42 even be individual cells of mammalian or microbial origin [199] . Cyclic voltammetry (CV) and differential potential voltammetry are two such electrochemical methods minimally applied in novel biosensing methods, as indicated in Tables 3.2 and 3.3 [6], [92], [128] . These methods provide sensitive detection of nano - levels of compounds with highly technical instrumentation. But, user interpretation of detection results can be designed by developers to provide simplified, straight - forward responses to their contaminat ion concerns. S pectrophotometry is another electrochemical method that offers a sensitive approach to design rapid microbial contamination detection methods. In the continuing drive to develop improved nano - biosensing methods of pathogenic bacteria, spect rometry is a more simplistic method that uses simple light beam technology [76], [80], [8 3], [121], [127], [135], [136] . Spectrophotometry measures the absorbance (or transmission) of an incident light beam upon a suspended compound due to its surface plasmon resonance (SPR) [80] . The ultimate success in the application of spectrometry to nano - biosensing likewise depends upon the development of a user - friendly method to interpret signaling responses. 4.2 Electrochemical Signaling Using Cyclic Voltammetry and Differential Pulse Voltammetry 4.2.1 MNP - cell - AuNP D etection MNP - cell - AuNP species described in th e previous chapter are electroactive and their properties can be studied through electrochemical means. In the case of amperometry (electrical current), applied potentials across an electrochemical test cell result in the chemical changes in the form of ch electron flow. This electron flow is either positive or negative current that can be monitored using a potentiostat instrument. The redox properties of the species result in characteristic graphs of the 43 current response versus the applied potential [21], [197], [200], [201] . Common uses of studying chemical redox reactions, along with measurement of reaction kinetics and corresponding chemi Finally, quantification of species is possible through the characteristic current response peak height for either oxidation or reduction, or both [65], [202], [203] . 4.2.2 Voltammetric M ethods Voltammetry applies the principles of electrochemistry in the study of the electroactive species, or target analyte. These experiments may be carried out within voltaic cells, which are commonly studied in general chemistry, using separated anodic and cath odic electrodes with a connecting salt bridge for macro - analysis (e.g. Fig. 4.1A). Over the past two decades, though, miniaturized cells have been developed with grafted, printed or etched electrically conductive chemicals forming the separated electrodes, which are then bridged by a micro - volume solution containing the target analyte (e.g. Fig. 4.1B and C). Currently sensors can detect analyte levels as low as nano - and pico - grams or micro - moles, and also analyze the properties of the analyte [132], [195], [1 97], [204] . Cyclic voltammetry (CV) consists of applying a linearly increasing then decreasing potential (voltage) across the electrode surface and measure the current as a function of applied potential and establishes whether it has reversible electroa ctive properties (Fig s . 4 .2 A & B ). Reversible chemical species are those that can be regenerated even after reaction. Differential pulse voltammetry (DPV) instead applies increasingly (or decreasingly) sized pulses 44 Figure 4.1 Electrochemical systems exe mplified with (A) common voltaic cell, (B) miniaturized screen - printed carbon electrode (SPCE) sensor, and (C) a mock - up of voltammetric bio - sensing with an SPCE chip with capture bacteria labeled with AuNP. of potential (voltage) across the electrode, to measure electrochemical currents from the various components of the sensor and target, since different component s will respond at c haracteristic potentials (Fig s . 4 .3A & B). DPV additionally provides improved sensitivity for measuring analyte concentrations compared to CV, showing similar current responses of 15 A to 20 A for morphine detection for concentrations at one order of magnitude lower, e.g. 0.5 M in DPV testin g versus 50 M in CV testing [132] . Figure 4.2 Examples of graphical results of nano - biosensing responses using cyclic voltammetry for microbial detection with MNP - strips on (A) soft plastic or (B) hard plastic. 45 4.2.3 Micro - bio - sensing Miniaturized sensors consist of a working electrode, reference electrode and counter electrode all contained within squared centimeter areas (Fig. 4.1B). Potential changes are applied across the working and reference electrode, with liquid sample sizes of only microliters bridging the electrode, and the current response monitored at the working electrode. A potentiostat instrument connected to the electrodes provides both the desired potential changes and measures the resulting curren t response. During a voltammetric experiment, applied potential creates an electrical force across the sensor surface. If the target analyte is in an electrically reduced state and a more positive potential is applied with respect to the working electrode , electrons will be stripped Figure 4.3 Examples of graphical results of nano - biosensing responses using differential pulse voltammetry for microbial detection using MNP - cell - AuNP with (A) positive current convention and (B) negative current convention. from the analyte, causing oxidation, and flow from the solution into the working electrode. These current responses will be registered as negative with respect to the working electrode , since the electrode is not providing electrons (Fig. 2A) . A chemically reversible analyte can then be returned to its reduced state by applying reversed potential differences across the electrodes, resulting in positive current responses as electrons flow out of the working electrode to the analyte . 46 A similar response is ac hieved in differential pulse voltammetry, except current convention can vary per instrument (Figs. 4.3A & B). Initially, the activated MNP - cell - AuNP suspension sample applied to the electrode is oxidized. Then, as more positive negative potential is applie d across the electrode, electrons will flow from the working electrode to reduce the analytes in the solution. This electron flow may be registered as either positive current (Fig. 4.3A) or negative current (Fig. 4.3B) depending on the direction of the cur rent . When a reversible analyte is monitored using cyclic voltammetry , the oxidative and reductive redox potentials can be determined (Fig. 4.2A). A reversible analyte is one that rapidly reacts to potential changes so that the system continuously approach es equilibrium for the new conditions. During oxidation of the analyte within the test solution, the analyte at the electrode surface will lose electrons and the speed at which this occurs depends upon the size of the chemical gradient within the solution and diffusive properties of the analyte through the bulk solution to the electrode. When reduced analyte finally reaches zero concentration at the electrode, the absolute peak oxidation current is reached at a characteristic potential. At this time, the chemical gradient is reversed, slowing additional analyte flow to the working electrode and oxidation eventually ceases, tailing off in current response. Upon reversing the poten tial difference, analyte reduction will begin in reverse order as oxidation. 4.2.4 Voltammetry C onventions These analyte responses can be described by several physical laws, such as the Nernst equation (Fig. 4. 4 A) which defines the newly established conc entrations of the oxidized and reduced forms of the analyte, c O and c R , respectively, due to the applied potential. The Butler - Volmer equation (Fig. 4. 4 B) further relates these same analyte concentrations to both the applied 47 potential, E, and the current r esponse, i 4 C) relates the analyte concentration Figure 4.4 Equations for general theory of voltammetry including (A) the Nernst Equation which relates the applied potential force, E, across the electrode to the concentration of the reductive and oxidative species, c R and c O , respectively, where we R is the molar gas constant (8.3144 J mol 1K 1), T is the absolute temperature (K), n is the number of electrons transferred, F = Faraday constant (96,485 C/equivalents ), and E 0 is the standard reduction potential for the redox couple; (B) the Butler - Volmer Equation further relates current, i, potential, E, and concentration, c R and c O , where = nF(E E0)/RT, k 0 is the heterogeneous rate constant, a is known as the transfer coefficient, and A is the area of the electrode; and (C) , which is directly proportional to the analyte concentration gradient, c O / x, where D O is the diffusion coefficient of oxidative species and x is the distance from the electrode surface [205] . gradient within the sample solution, c O / x, at the sensor to the flux of matter, [205] , and the analytes diffusion coefficient, D o . In studying a new analyte, these relations provide valuable information into the redox reaction occurring on the sensor, especially during cyclic voltammetry. In some systems, a well - defined electrochemically reversible mediator chemical, or redox probe, such as ferrocene or potassium ferricyanide , may be included into the test sample to assist electron transfer if the analyte shows irreversible chemistry [205] [207] . In these situations, analyte quantification is determined with respect to its reactive effect on the med iator presence. Reversible mediators allow the easy application of cyclic voltammetry to irreversible or more complex systems when desiring only to quantify analyte. 48 In voltammetry, polarity convention with respect to the working electrode must be determin ed before interpreting any resulting voltammograms. Different conventions result in oxidation and reduction peaks assigned to different quadrants of their respective plots [207] . Oxidation occurs when electrons are stripped from the analyte due to increasing potential d ifference across the electrodes, making the working electrode more positive than the analyte and more attractive to free electrons. This current flow can either be considered as positive into the working electrode, or negative away from the analyte. The op posite is true for reduction, when the working electrode is more negative than the oxidized analyte, the working electrode then supplies excess charge to the analyte, reducing it. Again this current can be defined as negative flowing away from the working electrode, or positive flow to the analyte. Inclusion of redox probes reversible mediators or redox - active receptors into the test sample or directly onto the sensor can improve sensitivity and reduce detection limits of the desired analyte. This is through improved electron transport or flux to the electrode as the strongly oxidizing or reducing mediator causes a large shift of the redox potential to a higher rate constant than the sample system [208], [209] . In addition to chemicals such as potassium ferricyanide mentioned earlier, biological enzymes may be incorporated into a sensing system. In fact, immobilizing catalytic electron - transfer proteins , such as cytochrome C and ferrodoxins [208] , directly to a sensor surface improves access to the buried catalytic region and improves voltammetry sensing without the need for electroactive intermediates [210] . Similar catalytic proteins may populate the surface of microbial targets, also. Other studies have exploited glucose oxidase and/or horseradish peroxidase to react upon secondary species such as hydrogen peroxide (H 2 O 2) , glucose or thionine to quantify analyte at concentrations down to 6 pM DNA, 10 µM H 2 O 2 , and 3.33 fg/mL alpha fetoprotein within human serum [198], [211], [212] . Even so, adding a redox 49 potential, shifting the redox peaks from the autocatalytic potentials to more a dvantageous regions of the voltammogram [211] . 4.2.5 Improved B io - sensing with N anoparticles MNP with high surface area to volume ratio may also improve voltammetric sensing through when present in the sample applied to the flat electrode surface. MNP have a high surface area to volume ratio for increased analyte a d gh surface reactivity promotes electron transfer [132] . In testing morphine levels within urine and serum, the presence of 10% MNP functionalized with chitosan improved DPV current response at 0.4 V for - fold, a thr ee - fold higher current response over CV signaling for the same sample , as well as allowing reliable detection of 0.5 morphine . The combination of MNP and DPV resulted in a detection limit of - molar amou nts [132] . Another sensor including MNP, this time functionalized with an N - acyl - homoserine - lactone (AHL) homologue, was able to detect a limit of 0.1 nM AHL using DPV [213] . Felisilda et. al. w ere also able to lower the limit of detection for a lysozyme enzyme to 0.010 [196] Incorporating AuNP into the sensor design can also improve electrochemical signaling responses for similar reasons as mentioned for MNP, along with autocatalytic capabilities. Fournier et. al. immobilized cytochrome B enzyme to the surface of an AuNP - modified sensor to study the electron transfer properties of its heme factors, which showed current pea ks for reduction at - 0.12 V and oxidation at - 0.06 V with CV [193] . Improved signal transfer through AuNP is particularly beneficial when using sensors to detect microbial targets. Kang et. al. detected Bacillus 50 bacteria down to 1 log CFU /mL using horseradish peroxidase reduction of H 2 O 2 and thionine integrated into an AuNP - antibody - sandwich sensor; this same signaling method was used by Fei et. al. to detect Salmonella at a concentration of 4 log CFU /mL using a n AuNP - DNA - sandwich sensor; and Cheng et. al. could detect 2 log CFU /mL Listeria in milk also using an AuNP antibody - sandwich sensor [21], [200], [214] . 4.2.6 Signal I nterpretation In analyzing the current responses from voltammetric interrogation of an electroactive system, DPV measurements provide direct comparison of peak heights between varying an alyte concentration and sensor composition. CV measurements of a test sample not only indicate redox potentials but can be used to quantify analyte concentrations through specific anodic and cathodic peak heights. For reversible systems, or an irreversible system that applies a reversible mediator, the cathodic peak response, Ipc, for reduction is determined as the distance between the maximum absolute current response and a line extrapolated to the reduction potential of the initial reduction slope. The an odic peak response, Ipa, for oxidation is simply the maximum absolute peak current value. A truly reversible system has a ratio of Ipa/Ipc equal to 1, whereas irreversible systems have a ratio of their anodic and cathodic peak currents greater than 1. Addi tionally, the sum of Ipa and Ipc provide a method to compare analyte concentrations between test samples. Even though pathogenic microbes have a net negative surface charge from the high levels of peptidoglycans or transmembrane lipopolysaccharides (LPS) present, it might be expect ed that they could be detected directly as analytes within an electrochemical system. Voltammetric signaling of microbial samples , though, requires redox - active moieties to be present in the microbial surface structures, and ava ilable for oxidation in a high concentration . This is exemplified by cytochrome c proteins which utilize tyrosine and tryptophan amino acids for 51 electron transport [215] . But, direct detection of pathogenic microb ial electro - activity shows minimal promise with low current responses for biofilm detection of totaled 10 to 20 mA over 10 hr o f pathogen detection monitored over 10 to 24 hr in milk enriched with target substrates [216], [217] . Voltammetric methods have been developed, though, that exploit microbial surface enzymes or transport systems connected to metabolic processes to provide electron flux for current response, but these methods suffer in poor rapid ext raction of the target [70], [218] [220] . Recent examples, though, include cytochrome c membrane electron transport of Shewanella w as connected to a photoanode to supply electrons in which to reduce environmental carbon. The flux of electrons through this proof - of - concept microbial fuel cell circuit was monitored through CV and found to have a reductive peak at - 329 mV and oxidative p eak at - 25 mV with peak sizes relative to the combined microbial electron flux of the system [218] . Antibody - captured E. coli were quantified via a surface membrane redox enzyme, D - glucose dehydrogenase, reaction with glucose substrate and ferrocenemethanol to assist in electron transfer which p roduced an oxidation peak at 0.3 V in CV measurements, able to detect as low as 1 log CFU/mL [70] . Pseudomonas aeruginosa was quantified using cellular acridine orange (AO) uptake. Pure AO shows two oxidation peaks within CV at 0.72 and 0.88 V, while DPV interrogation of cells exposed to AO showed a 100 nA peak at 0.8 V for 6 log cfu/mL [219] . Additionally, secreted electroactive m olecules detected using CV distinguished between E. coli and Saccharomyces yeast cells when stimulated with glucose nutrient [220] . 52 4.3 Spectrophotometric Verification of NP Presence and Quantity 4.3.1 Spectrophotometric C onventions Metallic nanoparticles contain a large surface area to volume ratio, resulting in a high quantity of surface electrons that can be exploited for spectral signaling, compared to bulk materials. Electrical and spectral detection schemes excite these surface electrons, creating a surface plasmon resonance (SPR) phenomenon. For example, the reduced gold nanoparticle surfaces are coated with dextrin which containing an abundance of electrically charged hydroxyl groups and oxygen atoms. When a spectral incident light beam at a resonant wavelength strikes the nanoparticle surface , the electrons flow across the surface, absorbing a fraction of that energy related to their quantity, in dilute systems. SPR is dependent upon the dielectric properties of the sample com pound and its environment, among other factors . C hanges in the dielectric constant will result in a characteristic reportable change in spectral absorbance. But, plasmon excitations are only possible if the incident light has a longer wavelength , lower ene rgy, than the nanoparticle diameter. Therefore, nanoparticles with diameters smaller than 100 nm show properties different from bulk materials. These nanoparticles can be excited, and thereby measured, using electromagnetic radiation in the visible light s pectrum with wavelengths between 400 nm and 700 nm. Electromagnetic radiation in the form of light waves carries an electric field that oscillate s at specific frequencies, causing movement of electrons on the nanoparticle surfaces, creating concentrated ne gative regions and positive (ion) regions, or surface plasmons. Light photons with the appropriate energy are absorbed by the nanoparticle s causing the electron motion across the rising from electric charge and current distributions and these equations can be solved for spherical nanoparticles using 53 10 nm will absorb most visible lig ht, whereas those with diameters of 100 nm will scatter most visible light, which results in signature peak absorbances useful in nano - biosensing. The commonly reported spectral absorbance spectrum is more accurately described as an e xtinction spectrum whi ch measures the amount of light absorbed and scattered at each wavelength . For a given sample, the resulting spectral results represent the maximum , or overall, absorbance value at each wavelength. Spectrometry of dilute systems is linearly related to concentration and can be used to quantify the amount of compound present in the sample. Increased nanoparticle concentration results in increased surface plasmon, and therefore an increase in absorbance of the incident light. This property can b e exploited, for example, in using increases in AuNP labeling to report the the scattered light recorded at the spectrometer sensor, I sample , with respect to a reference sample, I 0 , as A = log (I 0 / I sample ). In addition, the nanoparticle concentration, C, can be related to the absorbance through the Beer - l particle and l is the sample cell lengt h of the cuvette. For nanoparticles with a narrow range of diameters in dilute suspensions , their total concentration can be directly related to the maximum peak absorbance, and to their integrated peak area. For gold nanoparticles with a diameter below 30 nm, the peak extinction maximum is approximately 520 nm, whereas magnetic nanoparticles diameters of 200 nm have a peak extinction maximum at approximately 620 nm. Additionally, any absorbance at higher wavelengths can be monitored with spectrometry. 54 4.3.2 Spectrometry B io - sensing Exploiting nanoparticle surface area properties allows unique nano - biosensing applications. Quantifiable shifts in the spectral absor bance will result in relation to changes within the sampling suspension due to varying surface properties. As discussed earlier, the high surface area to volume ratio provides substantial room to functionalize nanoparticles with surfactants that will be a ttractive to either a variety or specific biological targets. Therefore, spectrometry provides a means to rapidly evaluate detection parameters when investigating the binding properties of nanoparticle surfactants to microbes. As reported by Bilankohi, as the thickness of the functionalized nanoparticle shell decreases relative to the core metal, surface plasmon increases, that both the core material and its diamet er established the location of the peak absorbance maximum [221] . A shift in the peak maximum will also occur as biological targets bind to the surfactant, both in total absorbance and with respect to the wavelength of the incident light, thus providing two measures to quantify target analyte. Although numerous examples exist in the application of magnetic and/or gold nanoparticles to improve extraction and detection of pathogenic bacteria within environmental samples as shown earlier [6], [21], [31], [74], [89], [93], [121], [127], [201], [222] [225] , a few even apply visual methods to simplify detection [66], [93], [95], [226] , but most examples that combine spect ral detection with na noparticle capture are used to target and detect chemicals [34], [72], [138] [140] , with none appear ing to use spectroph otometry to directly detect MNP - captured bacteria combined with AuNP enhanced signaling. Biological components show a strong absorbance within the UV - visible spectral range between 200 to 400 nm. Their characteristic absorbance pattern, minus nanoparticle presence, may even allow identifying different species 55 such as that shown by Ghosh et. al. in identifying 5 species of Bacillus using the first derivative of their individual spectr a [227] . Although microbes can be easily detected and quantified using spectrophotometry methods, cellular extraction from many environmental samples is tedious, requiring centrifugation or serial dilu tions in the initial sample preparation steps. Centrifugation is used to separate the targeted bacteria from the surrounding matrix, concentrating the microbial load and removing chemicals that may interfere with subsequent detection. Serial dilution is ne cessary when for suspected high microbial concentrations to ensure linear responses during detection steps. Functionalized MNP provide an efficient method to extract bacteria from even complex liquids, but MNP can then mask the spectral properties of captu red cells. AuNP attachment to captured cells, though, can provide a strong spectral signal around their peak extinction maximum (520 nm) for the MNP - cell - AuNP complexes. 4.3.3 Spectral I ntegration Since the combination of MNP, cells, and AuNP in a spectral sample have overlapping absorbances, s pectrum obtained from MNP - cell - AuNP samples show convoluted peaks throughout the spectral range. Additionally, as the nanoparticles aggregate, as will be seen in Chapte r 5 through transmission electron microscopy (TEM), the peaks will broaden about their respective maximum extinction peaks, increasing peak convolution , or overlapping, and the difficulty in identifying the true nanoparticle response. The additive effect o f the absorbance for the individual MNP, cells and AuNP peaks may also be further compromised by any matrix residue, such as milk components. When cells are included in the MNP - cell - AuNP complexes, the additive effect for all three components causes even m ore masking of the ir individual peaks. But step - integra tion of spectral absorbance for the complex sample across the tested visible region between 220 nm to 900 nm may reliably represent the AuNP peak height signaling. Since 56 absorbance spectrum incorporate the responses of each component of the MNP - cell - AuNP sample, including any matrix components discussed later, integrating the entire region in a step - integration manner will provide a metric that compares samples based on their complete composition. In c ontrast, Joo et. al. used MNP to capture Salmonella bacteria from a milk matrix, magnetically separated them from the matrix, then further labeled the MNP - cell complexes with titanium nanoparticles (TNP) in fresh buffer solution, but only measured spectral absorbance of the unattached TNP remaining in the supernatant, not the MNP - cell - TNP complexes. TNP peak extinction values at 230 nm were inversely related to the concentration of captured cells. When tested, though, in a complex matrix such as milk, a ttac hed milk residue caused noticeable broadening of TNP peaks [121] , masking the pure TNP response . With the necessity to identify more reliable means to interpret convoluted, multi - component spectrum, analysis through absorbance integration was considered. Streit et. al. found that integration across the spectral scans for single walled carbon nanotubes (SWCNT) were more reliable in predicting their concen tration in solution in place of peak heights [228] . Even serially diluted fluorescent probes showed a linear relationship between their final concentrations and the integration across their spectral scans [229] . Demissie et. al. agreed with Streit et. al. that using integration across the spectral range was more reliable and found a strong linear relationship to the concentration of caffeine found in green coffee beans. In addition, caffeine spectral peaks were - law, to normalize the quantity across different samples [230] . 57 4.4 Conclusion Al though the US as a developed country has seemingly inexhaustible resources , microbial foodborne diseases occur regularly each year, even given the heavily regulated food industry. Resources and regulations in under/developing countries are both insufficient to protect t heir populace which live on dollars per day [231], [232] . Feasible methods to determine the safety of a and minimal technological instrumentation and knowledge. Inexpensive, reliable, u ser - friendly instrumentation would open the door for food suppliers to incorporate more frequent food testing into their supply chain procedures. Rapid identification of pathogen contamination sources and presence in food is possible using voltammetric and spectrophotometric methods with the MNP - cell - AuNP complexes described in the previous chapter. This nano - biosensing method contrasted with the numerous methods also reported earlier that require extensive sample preparation, expensive supplies, and compl e x instrumentation application. Electrochemical signaling can provide expedient, reliable and sensitive detection of microbial contamination interrogating MNP - cell - AuNP complexes. This electrochemical signaling may consist of spectrophotometric absorbance o f the entire complex, or current responses to activated samples where cyclic voltammetry measures total complex while differential pulse voltammetry reports current responses for individual components. As mentioned earlier, MNP may also be affixed to plast ic strips to minimize contaminated liquid handling and expedite extraction of bacteria from more complex liquids, and also tested with CV. DPV, CV , and spectrophotometric detection methods may use either direct quantification of microbial presence based on sample responses alone, or threshold indicators based upon negative control responses. Ideally, MNP - cell - AuNP nano - biosensing methods would be 58 developed per microbial target and food matrix, optimizing parameters to obtain linear responses across a range of bacterial contamination. This optimization would be personalized per available instrumentation and required minimal detection levels. For example, E. coli O157:H7 and Listeria monocytogenes are both zero - tolerance pathogens, requiring detection of even 1 microbe per sample size, or ensuring undetectable levels of pathogen presence of less than 1 microbe in 25 gm samples. Specifically, these different electrochemical detection methods require different analyses to interpret their instrumental responses i n determining target concentration. Calibration is necessary in all applications between the microbial contamination and the particular instrument response to the specific MNP - cell( - AuNP)/matrix samples being tested. In CV, the combined, absolute anodic an d cathodic current responses are evaluated ; for DPV, the absolute AuNP current response at 0.4 V ; and with spectrophotometry, the total step - integrated spectral absorbance (220 nm to 900 nm) is used. Each detection method will also entail its own set of detection limits and range , based not only on the chosen electrochemical method, but also optimizing any chemical mediator or initiator parameters, and mitigating matrix interference through minimizing nanoparticle quantities . Within new liquid food mat rices, rapid detection at threshold levels may be quickly established without extensive development by applying a sample - to - negative ratio detection method. Since negative controls replicate the sample system, but do not contain contamination, a sample res ponse that differs from the negative control indicates microbial presence. Given the pertinent parameters of the microbial - food system, particularly level of microbial tolerance and matrix component effects, the necessary nanoparticle quantities and chemic al mediator/initiator can be quickly identified. In this design, fewer parameters would require optimization since only 59 MNP, and possibly AuNP for DPV detection, would need to be optimized to identify quantities that would allow significant (p < 0.05) sepa ration between a contaminated sample and a negative control. Application of this nano - biosensing method could then find broad application across the food supply chain. 60 Chapter 5. Analysis of C arbohydrate - F unctionalized M agnetic N anoparticles and D extrin - C oated G old N anoparticles B inding P roperties to Salmonella, E. coli, Bacillus, and Listeria U sing Transmission Electron Microscopy 5.1 Introduction Transmission electron microscopy (TEM) provides a means to assess capture of cells by magnetic nanoparticl es (MNP) and label ing by gold nanoparticles (AuNP). As mentioned earlier in chapter 3 , the carbohydrate functional groups coating MNP and AuNP surfaces bind to the various surface molecular structures of the cells via ionic attraction , van der Waals forces, dipole - dipole or ion - dipole interactions, or hydrophobic attraction. Although carbohydrate functional groups show less specificity in binding than antibodies, binding between MNP/AuNP and pathogenic cells persist s following extracti on and labeling . This allow binding patterns between the nanoparticle functional group and targeted microbe to be studied for information regarding quantity and location of nanoparticle attach me nt . MNP - F#1 are functionalized with larger glycan carbon struc tures, containing amino and hydroxyl groups, whereas MNP - F#2 have additional amino acid functionalization, increasing the number of carboxyl and amino groups available for binding. Dextrin coating on AuNP (d - AuNP) contain s substantial numbers of negatively charged hydroxyl groups , particularly in basic solutions . Proper steps are necessary in preparing nanoparticle - cell complexes for viewing using TEM . Due to the rigid ity of MNP and AuNP structures at the nano - level, common TEM embedding methods used to st udy tissue structure at the sub - molecular level are not applicable, since nanoparticle presence cause s tearing during section ing . Therefore, negative staining techniques were utilized to image nanoparticle - cell attachment, whereby nanoparticle - cell complex es are deposited to a film - covered copper grid and coated with a heavy uranyl acetate metal stain for contrast in TEM. 61 B acteria naturally attach to most surfaces to form biofilms [233] [237] . Although cells will attach to sample containers, evidence of cell binding to the MNP from TEM images provide initial evidence that bacterial colony growth following MNP extraction are directly related to the true bacterial load within the sample. TEM images of the binding between MNP - F#1, MNP - F#2 and d - AuNP used in this research identified some characteristics specific to the nanoparticle s and particular pathogen cell surface. Microbial cells studied were both gram negative, Salmonella Enteritidis and E. coli , and gram positive, Bacillus cereus and Listeria monocytogenes. The binding properties of the nanoparticles displayed differences in both cell binding and nanoparticle clumping due to the molecular surface structures and the carbohydrate functional groups. 5.2 Materials and Methods 5.2.1 Materials Cell strains of Salmonella Enteritidis, E. coli O157:H7, E. coli C3000, Bacillus cereus, and Listeria monocytogenes - Biosensors Lab at Michigan State University (MSU). In - house, proprietary glycan - and amino/ glycan - functionalized magnetic nanoparticles (200 nm in diameter, MNP - F#1 and MNP - F#2, respectiv ely) and d extrin - c oated gold nanoparticles ( approximately 2 0 nm in diameter, d - AuNP ) were used as received from the Nano - Biosensors Lab, MSU , and have been described previously [6], [143] . Phosphate buffer solution, pH 7.4 (Neogen, Lansing, MI), was prepared as directed. Brilliant green agar (BGA), tryptic soy broth (TSB), tryptic soy agar (TSA), Luria Bertani broth (LBB), and Luria Bertani agar (LBA) were purchased from Sigma Aldrich (St. Louis, MO), while CHROMagar TM E. coli was purchased from DRG International (Springfield, NJ), and prepared as directed. A Sphero FlexiMag Separator was purchased from Spherotech Inc (Lake Forest, IL). Sodium c acodylate buffer (0.2 62 M, pH 7.4), p araformaldehyde 16% aqueous solution and glutaraldehyde 25% aqueous solution both in 0.1 M c acodylate buffer , copper grids (200 line s/inch), and FORMVAR 1% solution were purchased from VWR (Radnor, PA). 5.2.2 Microbial Culture Salmonella Enteritidis, E. coli strains , and Listeria monocytogenes colonies were rejuvenated from frozen ( - 70 o C) storage frequently on BGA, CHROMagar TM E. c oli and LBA , respectively . A transfer loop - full of Bacillus cereus was heat - shocked at 75 o C for 15 min in 10 mL LBB in 50 mL flasks , grown with foam stoppers overnight at 37 o C and 100 rpm , then streaked onto LBA. All master plates were incubated at 37 o C overnight and stored at 20 o C for a maximum of 3 months. Microbial cultures were generated fresh for each experiment. A single colony was isolated from the master plate , inoculated into broth and grown overnight at 37 o C. Salmonella and E. coli were grown in TSB, and Bacillus and Listeria in LBA, with Bacillus heat - shocked at 75 o C for 15 min in 10 mL LBB in 50 mL flasks , grown with foam stoppers overnight at 37 o C and 100 rpm . One milliliter of the liquid culture was transferred to a new tube of appropriate broth (9 mL) and incubated at 37 o C for 4 h r , with shaking for Bacillus, before each experiment . 5.2.3 MNP Microbial Capture and Deactivation Each sample consisted of one type of MNP and one type of bacteria . MNP (5 mg/mL in PBS) were sonicated for 20 min then dispersed with shaking between sampling. Based upon ongoing TEM imaging results, between 100 to 1000 µL of MNP were added to 1000 µL of undiluted mic robial culture within a 2 mL flat bottom centrifuge tube. MNP capture was allowed for 10 to 30 min, then MNP - cell complexes were separated with a magnetic separator for 10 min. 63 The supernatant was removed, and the complexes were resuspended into 500 µL of 2.5% (v/v) glutaraldehyde and paraformaldehyde in 0.1 M c acodylate buffer. Cells were deactivated for 30 min, the complexes re - separated, and suspended into 1000 µL 0.1 M c acodylate. MNP quantity was based upon prior thickness interfering with electron tra nsfer, and deactivated samples may even be subsequently diluted as 10 µL MNP - cell solution into 1000 µL fresh buffer. Additionally, when investigating complex liquid food matrix effects upon capture, only non - pathogenic E. coli C3000 bacteria were used, ne gating the necessity of a deactivation step to eliminate any rinsing effects upon the electrostatic MNP - cell bonds . 5.2.4 Concentrated AuNP Preparation Concentrated d - AuNP was prepared by removing excess dextrin and solution to improve signaling through increased labeling with minimal dextrin interference . Volumes ranging between 25 to 250 L of the original d - AuNP suspension were transferred to 2 mL pinched - tip centrifuge tubes, sonicated for 10 min, and centrifuged at 15,000 rpm at 4 o C for 20 min. Foll owing supernatant removal, the remaining 20 L d - AuNP pellet was sonicated for 15 min to evenly distribute the concentrated d - AuNP evenly, then used as the starting aliquot. Even distribution of the concentrated d - AuNP was assured using spectrophotometric scans (900 - 200 nm wavelength, Shimadzu UV - 3101PC, UVProbe software, 1000 L PBS reference) compared to non - concentrated d - AuNP standards of the same original volume, showing peak values at 520 nm for both and no evidence of nano particle aggregation , whic h was described earlier as purpose during AuNP production . This lack of aggregation therefore also ensures s ufficient dextrin coating for subsequent microbial attachment. Tubes were stored at 4 o C and used within 24 h r of 64 preparation. Concentrated d - AuNP are referred to using their initial volume, e.g. a 75 L concentrated d - as appropriate . 5.2. 5 AuNP Labeling of MNP - cell Complexes MNP - cell complexes, following magnetic separation and removal of the supernatant, were resuspended into the appropriate concentrated AuNP. An additional 100 uL of PBS was added to augment mixing and labeling. The complexes were labeled for 30 min, magnetically separated, and the supernatant re moved. Labeled complexes were resuspended into 100 to 1000 µL of PBS, depending upon ongoing TEM imaging results. Negative MNP controls consisted of 20 to 100 µL MNP (5 mg/mL in PBS) without cells present, labeled with the same procedure. 5.2. 6 TEM Imagin g A 5 µL drop of the MNP - cell or labeled MNP - cell - AuNP complex solution was deposited onto a copper grid coated with FORMVAR film with excess moisture wicked away using filter paper. Complexes were stained with 5 µL micro - filtered 1% uranyl acetate in wate r, again with excess liquid wicked away using filter paper. Grids were allowed to dry for at least 15 min before imaging. Grids were imaged using a JEOL 100CX TEM instrument (MSU) containing a lanthanum hexaboride, LaB 6 , electron gun encased within a vacuu m system at magnifications between 5000 x to 100,000x. Caution was taken to minimize imaging times of MNP clumps to reduce sample loss due to electron over - excitement. 65 5.3 Results and Discussion 5.3.1 Binding Forces Between Nanoparticles and Microbes As d escribed earlier in chapter 3 , carbohydrate ligands bind to microbial surfaces using the same electrostatic forces used in antigen - antibody binding, but with less specificity and therefore strength. As libraries of carbohydrates are tested for their select ivity against pathogenic bacteria, those ligands that more tightly bind microbial surface structures will be identified. These carbohydrate ligands will further advance detection methods since their biocompatible properties will allow development of possible individual in - package monitoring throughout the food supply chain. TEM imaging of the attachment between the carbohydrate coated nanoparticles and microbial cells used in this study provide information into the binding forces prese nt. Current glycan - coated MNP - F#1 and amino/glycan - coated MNP - F#2 that will be shown to expeditiously extract bacteria from simple and complex liquid matrices attached to differing molecular surface structures or locations among the pathogens. Images of de xtrin - coated d - AuNP labeled samples , as concentrated AuNP, also lent information, and verified to some extent, unforeseen electrochemical responses achieved by their presence, which is reported later. MNP clumping appearance is particularly specific to th e two types of MNP used. Glycan coating during production of the MNP - F#1 causes a uniform, spherical distribution of the iron oxide (Figs. 5.1A, 5.2B & C, 5.3B, and 5.4B) as it reduces the surface molecules during formation . The presence of an amino acid u sed in MNP - F#2 production along with the glycan instead causes a more nondescript shape to the nanoparticles (Figs. 5.1C, 5.2D & E, 5.3C & D, and 5.4C). All of these examples are with MNP - cell complexes prepared in PBS for imaging. The nanoparticles also a ggregate differently, with MNP - F#1 forming smaller groupings when attached to the gram - 66 negative microbes , Salmonella and E. coli (Figs. 5.1A and 5.2B & C), but when binding gram - positive Bacillus and Listeria microbes they aggregate in larger groupings, wi th microbes attached around the edges of the grouping (Figs. 5.3B & 5.4B). Meanwhile MNP - F#2 form larger amorphous groupings seemingly covering the bacteria, regardless of the microbial surface structures. Overall, extended MNP - F#2 exposure to the microb ial surfaces caused sufficiently more cell wall lysing than MNP - F#1, particularly for gram - negative cells. Gram - positive cell walls are composed of a single thick layer of peptidoglycan protecting the inner cell membrane , while g ram - negative cell walls are composed of an outer membrane that contain s higher amounts l ipopolysaccharide (LPS) encasing a thinner wall of peptidoglycan . LPS is composed of a hydrophobic lipid A region extending into the O - antigen domain which consists of repeated polysaccharides [238] . Peptidoglycan is composed of sugars and amino acids, with residues of N - acetyglucosamine or N - acetylmuramic acid. Since MNP - F#2 bind all of the microbial surfaces directly, the single layer of LPS for Salmonella and E. coli is not as resista nt to the electrostatic forces that both bind and pull the membrane apart. Most likely the glycan ligand hydrophobically bind the Lipid A regions, while the amino groups then pull at the membrane as they also try to bind the negatively charged surface. Nei ther microbe, though, showed evidence of lysing after 8 hr exposure to MNP - F#1. This lysing effect is especially seen with E. coli , even though gram - negative cells have a second, albeit thinner, peptidoglycan layer. In fact, E. coli cells exposed only to PBS, and likewise bound to MNP - F#1 in PBS, for 8 hr showed substantial cell membrane lysing and cell dissolution (images not shown). Salmonella exposed to PBS for 8 hr, though, did not lyse, but did show minimal lysing when exposed t o either MNP - F#1 or MNP - F#2 for that length of time. 67 Similar to how gram - positive microbes resist the forces of acetone or ethanol in removing the peptidoglycan layer, retaining their violet coloring during common staining methods , they also resist the ly sing forces of bound MNP - F#2. Therefore, Bacillus and Listeria are able to retain their cell structure for longer periods of time upon exposure to the MNP. Results did show eventual minimal cell lysing for Bacillus and Listeria following 8 hr exposure to M NP - F#2 (image not shown), though, when the complexes were suspended in PBS, which does not provide nutrients to repair wall tears. Lysing effects, though, are of minimal concern during any microbial detection as the extraction process is completed within m inutes versus the hours needed here . 5.3.2 MNP - Salmonella Enteritidis Binding Salmonella Enteritidis contain significant number of flagella (Fig. 5.1A) that participate in binding to MNP - F#1 (Fig. 5.1B), but MNP - F#2 bind the surfaces of the microbe (Fig. 5.1C). Flagella found on pathogens such as Salmonella (Fig. 5.1A) and Bacillus (Fig. 5.3A) are locomotive organelles attached to the surface of the microbe that assist in cellular motility. The se flagella are composed of monomeric polypeptides and are diph asic, producing two different types of flagella, but never both at the same time. Due to the high variability of their protein composition and micrometers - in - length extension from the surface, they provide high specificity per microbial species as H antige ns in immuno - reactivity serotyping [164], [174], [175] . The protein structures along the flagella contain alternating projecting carbonyl electron rich oxygen atoms and amine hydrogens, along with possible aliphatic side chains for binding. The glycan stru cture in MNP - F#1 seemingly bind in specific locations along the flagella, not coating the entire length. In addition, the flagella do not appear to wrap around the surface of the MNP. Therefore, it may be that regions with high 68 Figure 5.1 Examples of Salmonella Enteritidis (A) attachment to MNP - F#1 (B) and MNP - F#2 (C). content of aliphatic side groups hydrophobically bind to similar carbon chain lengths of the glycan coating. Although Salmonella are peritrichous, with flagella extending from the entire cell surface, no flagella are present when binding to MNP - F#2 (Fig. 5.1C). It is possible that the lysing properties of MNP - F#2 immediately attack the flagella , which are more available for initial binding, lysing flagella faster than seen in cell wall lysing . This lysing may occur during TEM sample preparation during the 30 - min deactivation process. Once the flagella have been removed, the MNP - F#2 bind the surface of the cell, with the positive amine groups present in the ligands interacti ng with the negatively charged cell surface. Aggregation of the MNP against the ionic PBS environment may be the reason that the entire cell surface is not covered, causing an equilibrium between microbial binding and ionic exclusion forces upon the MNP . 5.3.3 MNP - E. coli O157:H7 Binding As stated earlier, exposure of E. coli over extended times to PBS or either MNP can have deleterious effects upon the cell membrane causing lysis (Figs. 5.2C & E). These effects do not 69 occur immediately (Figs. 5.2B & D), w hich allows for pathogen detection during the timely extraction from liquid matrices. Less aggregation of MNP - F#1 occurs in E. coli attachment (Fig. 5.2B) than was present for Salmonella , and the nanoparticles bind directly to the cell wall, which was also reported by Lim et al. [130] and MNP - chitosan/glutaraldehyde used by Lin et al. [118] . This probably hastens the eventual lysing effects against the membrane. MNP - F#2 again do not bind the entire surface of the cell (Fig. 5.2C) bu t aggregate more at the ends than seen for Salmonella . Again, there appears to be an equilibrium established between the ionic P BS environment that suspend the complexes and their electrostatic binding of the surface. Additionally, each microbe exudes a va rying composition of chemica ls which also alter the liquid surrounding the MNP - cell complexes. Figure 5.2 Examples of E. coli O157:H7 (A) binding to MNP - F#1 (B) including cell lysing over extended exposure (C) and to MNP - F#2 (D) as well as cell lysing th at occurs over extended exposure (E). 70 5.3.4 MNP - Bacillus cereus Binding Bacillus cereus are peritrichous cells (Fig., 5.3A), but flagella are not present when bound to either MNP - F#1 nor MNP - F#2. Whereas MNP - F#1 bound the length of E. coli , they bind specific locations along the Bacillus cell wall (Fig. 5.3B), again aggregating in smaller groupings similar to that seen to the Salmonella flagella when not located next to the larger grouping. Little binding occurs at the cell ends as the path ogen lines the periphery of the aggregated nanoparticles attached along their lengths. MNP - F#2 similarly binds the cell length (Fig. 5.3C), aggregating in Figure 5.3 Examples of Bacillus cereus (A) binding to MNP - F#1 (B) and MNP - F#2 with short exposure (C) and extended exposure to the MNP (D). larger clumps that also overlay the surface. Some cell lysing is evident when the Bacillus cells are exposed to MNP - F#2 for extended periods of time (Fig. 5.3D). 71 5.3.5 MNP - Listeria monocytogenes Binding Rod - like L isteria (Fig. 5.4A) expand in diameter when attached to MNP - F#1 (Fig. 5.4B) and MNP - F#2 (Fig. 5.4C). Their binding to MNP - F#1 are similar to that of Bacillus , lining the periphery of the aggregated nanoparticles. MNP - F#2 more clearly show their round appea rance comparable to that of MNP - F#1 when bound to the Listeria cell surface. Meng et al. studied nanoparticle binding against Listeria monocytogenes using MNP - Vancomycin/polyethylene glycol bound in small clumps around the surface of the microbe [119] . The nanoparticles d id not aggregate as much , which may be an aspect of the Listeria surface molecular structure binding forces pulling the MNP more tightly to the surface as individual particles. Internalins are leucine - rich adhesive proteins that assist pathogen ic Listeria to insert themselves into epithelial cells and reproduce intracellularly [23 9], [240] , while Ctap is a cysteine binding surface transporter for Listeria monocytogenes which is expressed during low cysteine presence [187] . Leucine is rich in methyl groups, creating hydrop hobic pockets for targeted binding to the MNP moieties , while MNP - F#2, specifically, will bind Ctap surface epitopes . Figure 5.4 Examples of Listeria monocytogenes (A) binding to MNP - F#1 (B) and MNP - F#2 (C). 72 5.3.6 AuNP Binding to MNP - cell Complexes E. coli C3000 non - pathogenic microbial surrogate cells were used to exemplify binding eliminating the need to deactivate the cells. Due to the non - covalent electrostatic forces between carbohydrate ligands and cell surfaces, a fraction of either MNP - cell and AuNP - cell binding is lost in fresh supernatant. During capture, extraction and labeling, each subsequent suspension of these complexes in fresh liquid matrix results in a new equilibrium established between complexed and separated cells, even during cellular deactivation. This concept will be further elucidated later, but t o reduce lost binding during increased rinses, AuNP250 binding properties were studied using E. coli C3000 (Figs. 5.5A D) to eliminate these deactivation steps . AuNP250 bi nding to the cellular surface through the dextrin ligand was studied under several conditions. When excess AuNP250 is not rinsed from the E. coli - AuNP250 complex (Fig. 5.5A) the AuNP250 nanoparticles can be seen dispersed throughout the remaining dextrin m atrix. Magnification of rinsed MNP - F#2 - E. coli - AuNP250 complexes shows the aggregation of the AuNP into 50 100 nm clumps along the cell surface, but seemingly coating the MNP more evenly (Fig. 5.5B). This sample was suspended in nutrient rich TSB to mini mize cellular lysis when exposed to MNP - F#2 through cell membrane regeneration. As can be seen in Fig. 5.5C, the AuNP retain their dextrin coating when binding the cell surface structure. These attachments persist even when the cell is allowed to lyse when suspended in PBS (Fig. 5.5D). Cellular polypeptides contain positively charged amino groups which provide binding sites for the negatively - charged dextrin hydroxyl groups. Pockets of positively charged amino acids within the polypeptide chain will cause AuNP aggregation. MNP - F#2 have a more uniform appearance when bound to the E. coli surface within the presence of AuNP (Figs. 5.5A & B) due 73 to the combined binding properties of dextrin coating individual MNP. Dextrin will saturate the MNP - F#2 surfaces due to the presence of the amino acid ligand. Therefore, binding to the cell Figure 5.5 Examples of MNP - F#2 - E. coli C3000 complex labeling with AuNP250 with excess AuNP250 present (A), showing aggregated AuNP binding to the cell wall (B) and at higher mag nification (C) all in TSB, and finally persistence of AuNP binding to the cell structure even during lysing caused by extended MNP exposure in PBS. surface is more selective than to the MNP surface, which will later be shown to cause larger negative control MNP - AuNP electrochemical responses than sample MNP - cell - AuNP responses. Binding is the transduction event in biosensing that leads to signaling. Electrostatic forces dominate the binding dynamics between targeted microbial cells to the AuN P dextrin - coat ing , and to the carbohydrate ligand of MNP. Attachment mechanisms are of varying selectivity. The bonds formed during MNP - cell - AuNP nano - biosensing between the cells and the carbohydrate functional groups are non - covalent, electrostatic force s in form, similar to the antigen - antibody 74 potentially bind to positively charged amino acid pockets of the cellular surface molecules, leading to seemingly selectiv e attachment of AuNP on E. coli cell surfaces (Fig. 5.5B) . MNP bind readily to cells and appears to be selective as seen above for each pathogen tested. AuNP also easily bind the MNP functional groups at saturation levels (Figs. 5.6B & C) . AuNP saturated binding to the MNP, though, are potentially between the hydroxyl groups of the dextrin to the high number of positively charged amines in the MNP carbohydrates. Binding between MNP and cells, though, impedes d - AuNP binding to both the MNP and cell, reducin g AuNP presence in MNP - cell - AuNP Figure 5.6 Examples of unconcentrated d - AuNP (A), concentrated AuNP250 coating MNP - F#2 in PBS without prior milk exposure (B), with 2 - - ctly in the milk matrix (D). 75 complexes (Figs. 5.5A & B). These binding dynamics will impact signaling interpretation in this biosensing method. AuNP labeling on MNP - F#2 remains saturating whether carried out in PBS or a complex matrix such as milk. The A uNP used in these experiments were of sizes below 20 nm (Fig. 5.6A) therefore a substantial quantity of the concentrated AuNP may coat MNP - F#2 during labeling (Fig. 5.6B). In experiments when MNP - cell capture is carried out in a milk matrix before AuNP lab eling (2 - step), versus when MNP cell capture and AuNP labeling is concurrent in milk (1 - step), AuNP still saturate the surface of the MNP. In 2 - step biosensing, the PBS rinse step following MNP - cell capture removes the attached milk residue from the MNP - F# 2 surface , as visualized in Fig. 5.6C compared to the presence of milk residue in Fig. 5.6D . This exposes the MNP surface again to allow saturation amounts of AuNP binding. 1 - step biosensing, though, result still in the MNP - F#2 surface being completely covered with AuNP, albeit over the attached milk residue layer. The binding of milk components and AuNP to the samples will be shown later to have created contrasting electrochemical signaling results. 5.4 Conclusions Along with TEM imaging providing evidence of nanoparticle binding to cellular surfaces, evidence is also found to understand subseq uent nano - biosensing results. Of particular interest w ere the difference s in MNP - F#1 and MNP - F#2 binding properties to the different pathogenic bacteria studied, showing selectivity in the surface markers that they attach. This points to different aspects of electrostatic binding between MNP and cell, which are the first steps in distinguishing biocompatible carbohydrate ligands for specific microbial targets. Lysing effects caused by MNP - F#2 attachment to E. coli , and to a lesser extent Salmonella , over ex tended times indicate tests 76 against gram - negative microbial targets will require quick response times of at least less than the 40 min extraction and deactivation time for TEM sample preparation , which are possible through MNP - cell extractions. As importan t is the verification of saturation binding levels of AuNP to the MNP - F#2 surfaces versus the more selective binding to E. coli surfaces, which assisted in explaining later electrochemical responses for MNP - cell - AuNP responses below those of negative contr ols. 77 Chapter 6 . Rapid Extraction of Pathogenic Bacteria from Liquid Foods using Carbohydrate - Functionalized Magnetic Nanoparticles I ncluding Cyclic Voltammetric Detection 6 .1 Introduction Fluid milk products are consumed at over 100 pounds per year per person in the US [241] . Since raw milk may contain pathogenic bacteria from the pasture, feed or milking equipment [242], [243] ction contain no more than 100,000 CFU/mL [244] . But bacterial contamination may persist through pasteurization or enter through secondary contamination [52], [236], [245] . Whole milk is a complex matrix that is 89% water containing fats, proteins, carbohydrates and enzymes suspended in whey serum [246], [247] . This complex matrix encourages microbial growth, while also interfering with conventional microbial extraction methods and detection sensitivity. Both raw and pasteurized milk products are consumed in the U.S., even though raw milk was involved in 60% of dairy associated foodborne outbreaks in 2006 [248] . The Food and Drug Administration (FDA) cauti ons against drinking raw milk and unpasteurized products, due to the severity of diseases that may be contracted from over 90 microbial pathogens that may be present, such as E. coli , Salmonella , Streptococcus , Staphylococcus , and Listeria [52] . In 2005 2006, 80% of children (2 11 years) consumed milk [249] , but they are especially susceptible to infections. Pasteurized milk products a re safer for human consumption, given they are not cross - contaminated post - sterilization, as occurred in numerous foodborne outbreaks between 1976 to 2007 [250] . To protect this valuable commodity and its consumers, farmers and producers alike require rapid, frequent, an d reliable testing of milk quality and contamination. Nanoparticles provide a large surface - to - volume ratio onto which multiple ligand molecules may bind [35], [251] [254 ] . Although antibodies are ideal ligands to target pathogenic bacteria, their high cost, short shelf - life and special handling requirements increase detection costs. 78 Therefore, less expensive, stable antibody - mimics that can selectively attach to bacter ia are desired [16], [44] [46] . Carbohydrate ligands are an economical option, able to attach to bacterial surface markers through stable non - covalent bonds that mimic antibody binding [118], [163], [257] . These i nexpensive carbohy drate ligands retain their structural integrity when immobilized onto MNP. Their multivalent moieties can both target pathogenic bacteria and improve suspension within liquids by increasing proximity between MNP and their microbial target . M agnetic nanoparticles (MNP) functionalized with a variety of bio - recognition ligands have been used to target bacteria from various liquid and gaseous matrices [30] [33] . MNP from ferrous oxide are superp aramagnetic, becoming magnetized in the presence of an external magnetic force, allowing for quick extraction from solution. Admittedly, as mentioned, antibody ligands provide highly selective biosensors, but they require expensive storage with a limited s helf life. In comparison for the same assay performed in our lab, carbohydrate - functionalized MNP cost less than $0.10 per assay, whereas the cost of the antibody ligands alone used in a similar extraction method [11] increas ed costs to $0.40 per assay. Suspended MNP biosensing methods cover a broad range of ligands, targeted pathogens and detection methods, as extensively reviewed in Tables 3.2 and 3.3 above. Although many studies report on the extractive ability of MNP, mos t of their work was carried out in simple buffer solutions or following time - consuming pre - sampling steps of over an hour that first removed matrix components [2], [3], [5], [6], [8], [49], [50] . Many of these n ano - biosensing applications showed promising results using selective antibodies or oligonucleotides to target Salmonella, E. coli and Listeria [79], [82] [86], [88], [89], [259] to detect bacterial levels between 10 1 to 10 8 CFU/mL . I nitial studies into MNP capture in the complex milk matrix centered around designing a biocompatible MNP - based biosensor to detect microbial pathogens in i ndividual liquid food 79 packages over the life of the product. This requirement preclude d using antibody ligands. Other proposed biosensors have used carbohydrate functional groups on the MNP, such as chitosan/glutaraldehyde [118] , which showed a similar range of detection in buffer solution s . A major advantage of MNP is their ability to separate biological targets from their matrices, even concentrating captured bacteria, using only an external magnet. Concentrated MNP - target, though, can improve optical detection [87] . MNP may even be incorporated into subsequent detection steps, such as interdigitated microfluidic impedance [89] or chemiluminescence [82], [85] without signal interference. Other s ignaling methods fo r these biosensing systems included magneto - elastic, HPLC - DAD, optical density and chemiluminescence [84] [86], [96], [118] . Most of these reported methods require d matrix removal pre - detection, instead of identifying matrix effects and incorporating these effects into th e signaling . E xtensive sample preparation using substantial supplies and time were required to first separate the bacteria from the matrix [12], [20], [55] [58] . After t he bacteria have been removed from the matrix, the final detection step is then rapid. F unctionalized magnetic nanoparticles prepared in the Nano - Biosensors lab with an iron oxide magnetite core functionalized with glycan (MNP - F#1) or amino/glycan (MNP - F# 2) carbohydrates were used to extract pathogenic bacteria from complex matrices in suspended and affixed forms. MNP - F#2 were also affixed to plastic strips (MNP - strip) and used to rapidly separate bacteria directly from complex matrices with reduced matrix handling . Capture using either method did not require matrix dilution or complicated pre - separation steps. Electrochemical detection using cyclic voltammetry (CV) was used to expedite detection during MNP - strip biosensing. Suspended MNP extraction provided a rapid means to concentrate bacteria within minutes for improved detection using preferred bacteriologica l analysis methods (BAM) with agar 80 growth detection after 18 h incubation. This method is most economical for under/developing locales to determine contamination levels in their liquid food supply. Meanwhile, MNP - strips used in conjunction with a simple ha ndheld potentiostat minimized liquid handling, concentrated targeted bacteria for enhanced signaling, and allowed detection within 30 min. Currently both methods report total cell presence in the liquid food sample s . As an initial design, the MNP - strip e xtraction with CV detection biosensing method provides simple, rapid and affordable microbial detection in complex matrices for higher food security through more testing . 6.2 Materials and Methods 6.2.1 Materials Cell strains of Salmonella Enteritidis, E. coli O157:H7, E. coli C3000, Bacillus cereus, and Listeria monocytogenes - Biosensors Lab at Michigan State University (MSU). In - house, proprietary glycan - and amino/ glycan - functionalized magnetic nanoparticles ( 200 nm in diameter, MNP - F#1 and MNP - F#2, respectively) and d extrin - c oated gold nanoparticles ( approximately 50 nm in diameter, d - AuNP ) were used as received from the Nano - Biosensors Lab, MSU [6 ], [143] . Phosphate buffer solution, pH 7.4 (Neogen, Lansing, MI), was prepared as directed. Brilliant G reen agar (BGA), tryptic soy broth (TSB), tryptic soy agar (TSA), Luria Bertani broth (LBB), and Luria Bertani agar (LBA) were purchased from Sigma A ldrich (St. Louis, MO), CHROMagar TM E. coli was purchased from DRG International (Springfield, NJ), and Oxford Listeria Agar base and Modified Oxford Listeria Supplement for modified O xford agar (MOX) was purchased from Neogen Corp. (Lansing, MI), and prepared as directed. 81 Square 1 oz (30 mL) clear glass bottles with phenolic caps were purchased from Thomas Scientific (Swedesboro, NJ). A Sphero FlexiMag Separator was purchased from Sph erotech Inc (Lake Forest, IL). Vitamin D, 2% reduced fat, fat free milk, apple cider, and homogenized eggs were purchased from a local commercial seller, while beef juices were procured from the MSU meat store , all stored at 4 o C, and used fresh. Before us e, all food containers were sterilized externally using 70% ethanol and the bottles thoroughly mixed by inversion. Salmonella Enteritidis, E. coli O157:H7 and E. coli C3000, Bacillus cereus , and Listeria monocytogenes were rejuvenated from frozen storage f requently on BGA, CHROMagar TM E. coli , LBA following heat shock treatment , and MOX, respectively. 6.2.2 Bacterial C ulture Salmonella Entertidis , E. coli O157:H7 and E. coli C3000 were cultured in 10 mL TSB, while Listeria monocytogenes was cultured in 10 mL LBB, in 15 mL Eppendorf tubes overnight at 37 o C. Bacillus cereus was heat - shocked at 75 o C for 15 min in 10 mL LBB in 50 mL flasks and grown with foam stoppers overnight at 37 o C and 100 rpm. A 10% spike was then grown with the same respective conditions for four hours to log phase immediately preceding all experiments. Tenfold serial dilutions of the stock culture, from 10 - 1 to 10 - 7 , were prepared using PBS before each experiment. Similar tenfold dilutions were made in the appropria te liquid food from 10 - 1 up to 10 - 12 to account for the immediate accelerated bacterial growth during dilutions and used as indicated below for cell capture. Viable cells for both cultures from PBS dilutions and liquid food were enumerated by microbial pla ting. Salmonella was plated on BGA while E. coli , Bacillus, and Listeria were plated on LBA and plates incubated at 37 o C for up to 16 hr. Target plate counts were between 25 and 300 colonies per plate. 82 6.2.3 Suspended MNP B acterial C apture in Milk Each experiment consisted of one type of bacteria and one type of milk. All samples and gm of food matrix, approximated as 25 mL, used without further dilution. To ensure sample and control volume co nsistency, all sample and control bottles were first filled with 23 mL of milk, then 750 µL of milk was added to sample and positive control bottles and 1000 µL of milk was added to the negative and positive control bottles. MNP (suspended in the correspon ding matrix at 5 mg/mL) were next added to sample and negative control bottles in 1 mL aliquots. Sample and positive bottles were then inoculated with - 3 in milk, resulting in a final 10 - 5 dilution for capture. Bacteri a were found to acclimate to the food matrices in under 10 min during the stock culture dilutions, with new microbial growth increasing cell concentrations even three - fold over those in PBS. During the experimental process, separate replicates for each typ e of MNP were run concurrently, with each replicate spiked from a fresh culture dilution in the appropriate food matrix to minimize differences in the cell concentrations at the time of capture. This increase in concentration was also quantified using agar plate counts from the 10 - 6 dilutions in their respective matrices as: Increased Cell Concentration Factor in Milk = - 4 in the appropriate L of matrix liquid, and 200 µL of MNP. 83 Prepared bottles or tubes were mixed fo r 10 min for LVC or 5 min for SVC and separated on a magnetic rack for 5 min to allow MNP - cell complexes to migrate to the magnetic side of the bottle. Supernatant was removed wh ile the bottles remained on the magnet. All bottles had their contents resuspended in 1 mL of fresh m atrix for LVC and 500 µL for SVC. This in essence concentrated the MNP - cell complexes of samples from 25 mL down to 1 mL. Then 100 µL was plated onto appro priate agar in duplicate. Resulting positive control colony counts represent those cells that initially affixed to the vessel wall but resuspended in fresh matrix and were reported as and controls were run in triplicate and plate counts were averaged. For TEM imaging of fungal milk contamination , MNP - cell complexes were inactivated , as described earlier, with a glutaraldehyde - paraformaldehyde solution and resuspended in cacodylate buf fer. Then 5 µL volumes were applied to Formvar ® coated copper grids and stained with 5 µL of 1% uranyl acetate immediately before imaging. 6.2.4 Suspended MNP Multiple B acterial C apture Using SVC with suspended MNP capture , milk and PBS matrices were ino culated simultaneously with competitive bacteria Salmonella , E. coli and Bacillus to test multiple bacterial capture. These tests were carried out at a final 10 - 6 dilution from culture to provide appropriate final plate counts. Resulting MNP - cell complexe s were plated on both BGA and LBA. Salmonella were counted from BGA plates, while E. coli and Bacillus were counted from LBA plates. Salmonella appeared as red colonies on BGA, distinguished from yellow E. coli colonies, while Bacillus did not grow on BGA. All three bacteria on LBA were distinguished by their colony appearance: E. coli colonies were larger than Salmonella , with a darker outer ring, while the 84 Bacillus colonies were composed of smaller, conglomerated clear globes. Total t arget plate counts were 25 to 300 colony forming units. Capture indexes (C I ) were calculated both individually per bacterial strain and for total captured bacteria. 6.2.5 Matrix E xposure E ffects on Suspended MNP C apture Cell capture after extended MNP exposure to different fresh milk types was tested. In SVC long - term exposure, MNP was exposed to milk stored at refrigerated (4 o C) conditions for 5 days before inoculation. On day 5, Salmonella, E. coli or Bacillus were sep arately spiked into the tubes to determine MNP capture ability . In LVC long - term MNP exposure, bottles of milk with MNP were prepared in vitamin D milk, 2% reduced fat milk, and fat free milk (9, 16 and 8 days from expiration, respectively) on day one and refrigerated for 9 days, then spiked with Salmonella for capture. 6.2.6 Capture Index C alculation for Suspended MNP Capture D ilution of the bacterial cultures within the liquid food matrices caused accelerated microbial growth, therefore suspended MNP ca pture was calculated with respect to stock culture diluted in the corresponding matrix . In addition, Increased Cell Concentration Factor in Milk was calculated by dividing the stock concentration diluted in matrix by that in PBS. Positive control counts, resulting from inoculated samples without MNP present, represent those cells that originally adhered to the capture vessel but resuspended into the fresh matrix , being therefore unavailable for MNP capture in the sample bottles . Th erefore, the resulting positive 85 control cell concentrations were subtracted from the stock culture concentration to correct for the actual cells available for MNP capture. Sample and positive control cell concentrations for LVC were determined as: where 250 µL of the 10 - 3 diluted stock culture was inoculated into LVC bottles and concentrated to a fi nal 1000 µL after magnetic separation and removal of the supernatant, followed by plating 100 µL of MNP - cell on appropriate agar media. Capture index (CI) was calculated based on the capture matrix serial dilution cellular concentration, as shown below, an d reported with its standard deviation, both as parts per thousand (ppt) . 6.2.7 MNP - strip P roduction Plastic strips (5 x 1 cm) were numbered at the top 1 cm, heated in a vacuum oven at 100 o C and 1.4 in Hg for 1 hr, allowed to cool overnight under vacuum, then weighed. MNP was prepared in sterile water (5 mg/mL) and 250 uL (1.25 mg M NP) was applied to the lower 4 cm length as 5 to 6 evenly - spaced dots. Strips were returned to the pre - heated vacuum oven at 100 o C and 1.4 in Hg for 1 hr, allowed to cool overnight under vacuum, and re - weighed. Resulting MNP weights were calculated and av eraged per production batch. Positive control strips were handled in the same manner, except without MNP application. 86 6.2.8 MNP - strip B acterial C apture Each experiment consisted of one type of bacteria and one type of liquid matrix ( Salmonella in homogeni zed egg, Listeria in apple cider, E. coli in milk, or E. coli in beef juices). MNP - strip capture followed the method of LVC described above replacing suspended MNP. S amples, negative controls and positive controls consisted of 25 mL of matrix used without further dilution. B ottles were filled with 24 mL of the liquid matrix, then 750 uL of matrix was added to sample and positive control bottles and 1000 uL of the matrix added to the negative control bottles. Sample and positive control bottles were inocula ted with 250 uL of log - phase 10 - 2 serially diluted bacterial culture in matrix for a final 10 - 4 dilution. Prepared MNP - strips were inserted into each bottle; positive control strips contained no MNP. Capture in beef juices was carried out in 10 mL volumes, due to limited resources, using the same final 10 - 4 final dilution for E. coli . Bottles were then mixed by inversion several times and incubated for 10 min to allow the MNP - strips to capture bacteria. Then the MNP - strips were removed, rinsed on both sides with 1 mL of PBS over prepared agar plates, and placed individually MNP - side up in a sterile petri dish for cyclic voltammetry (CV) testing. Remaining bacteria within the supernatant were determined by plating 100 uL on appropriate agar. Agar plates were incubated overnight at 37 o C. Target plate counts were between 25 and 300 CFU. Al l samples and controls were run in triplicate. Capture of Salmonella in PBS at varying concentrations was carried out using the same method. To verify cell capture, MNP - strips were inserted into 10 mL TSB, incubated at 37 o C for 4 hr, serially diluted twi ce, and 100 uL plated on TSA, incubated at 37 o C overnight, and counted. Target plate counts were between 25 and 300 CFU. All samples and controls were run in triplicate. As described above, m icrobial cells showed new growth in the liquid food matrices and increased concentration above that of their resting concentration in PBS, particularly in 87 homogenized egg. At the 10 - 4 dilution for capture of Salmonella in PBS, E. coli in beef juices, Salmonella in homogenized egg, Listeria in apple cider, and E. coli i n milk, cell concentrations of available cells were 5.19, 7.72, 12.30, 7.94 and 7.04 log CFU/mL, respectively. 6.2.9 Cyclic V oltammetry D etection of C ell P resence on MNP - strips Cyclic voltammetry (CV) testing was carried out using a handheld potentiostat (model PG581 Uniscan Instruments, Bio - Logic Scientific Instruments, Seyssinet - Pariset, France), driven by UiE Chem Software. For each test, a disposable screen - printed carbon electrode (SPCE, Gwent Electronic Materials Ltd, United Kingdom) with a surface a rea of 1 cm 2 , was loaded into the port of the potentiostat. Ferricyanide (150 µL), a chemical mediator, was applied to the MNP spots on the MNP - strips, reacted for 5 min, then 75 µL was carefully collected using a pipette and applied to the SPCE. The resid ual ferricyanide ion concentration in the collected liquid following the reaction was indirectly related to the reactive components on the MNP - strip and was then quantified. CV was run between +1.2 V to - 1.2 V, applied across the electrode, at a scan rate of 100 mV/s with 30 mV increments for 300 ms step time. A CV voltammogram was recorded for each sample. All sampling was done at room temperature. Total CV voltammogram was calculated by adding the absolute values of the anodic and cathodic peak currents . Averaged sample peak currents were normalized against averaged negative controls as S/N for biosensing results. Sample and negative control peak currents were normalized against the chemical mediator reference peak currents normalized peak current respo nses (NPCR) for matrix effects analysis and reported with corresponding standard deviation. This process is exemplified in Schematic 6.1. All samples and controls were run in triplicate. 88 Schematic 6. 1. Rapid detection of bacterial contamination in conta minated liquids using MNP - strips and CV detection. MNP - F#2 affixed to plastic strips (MNP - strip) (A) for user - friendly capture in only 10 min ( B ) by either dipping the strip into microbial contamination within undiluted liquids or applying 250 uL liquid on to the strip (C) , then rapid detection by electrochemical testing of reacted chemical mediator on a screen - printed carbon electrode (SPCE) ( D ) using cyclic voltammetry (E) . Results between samples and negative controls are shown in ( F). 6.2.10 Statistical A nalysis All suspended MNP capture experiments were run in triplicate, and the average C I was reported with corresponding standard deviation. Two - way ANOVA for LVC was run for milk and bacteria type with replications per MNP type, as well as f or either milk or bacteria type against MNP type, and significant differences in the average C I were determined using the q - test, both with significance p < 0.05. Unpaired t - tests were run comparing MNP type for LVC and for Salmonella capture in fresh vers us long - term MNP exposure milks for all MNP and milk types, all with significance of p < 0.05. All MNP - strip experiments were run in triplicate, with NPCR reported with corresponding standard deviation s . Two - way ANOVA for matrix type (PBS, beef juices, hom ogenized egg, apple cider, and milk) and treatment type (sample and negative control NPCR ) with three replicates was 89 run along with linear regressions to analyze matrix component effect upon negative control NPCR. Finally, unpaired t - tests were used to com pare sample to negative control NPCR in all matrix types. All tests were run with significance of p < 0.05. 6.3 Results and Discussion 6.3.1 Aspects of B acterial C apture. MNP capture here is defined as a stable bond between the MNP carbohydrate ligand an d the cell membrane, as an MNP - cell complex, allowing rapid extraction of the pathogenic bacteria from liquid food using a simple magnet, in place of time - consuming centrifugation. Carbohydrate - functionalized MNP used in these experiments have hydroxyl, am ino. and hydrophobic regions that interact with the bacterial membranes and their surface epitopes forming non - covalent electrostatic bonds similar to those found in antibody bonding [159] . Bacteria have varied transmembrane and surface molecular s tructures composed of lipids, sugars and proteins that determine their binding dynamics to ligand functional groups such as the F#1 and F#2 carbohydrates [177], [185], [260] [262] . For example, flagella are composed of polypeptides and extend micrometers from the cell surface, making them highly available for MNP binding [164], [174] . Microbial surface expression varies with respect to their surrounding environment, therefore MNP capture must be evaluated in matrices mor e representative of the food supply chain, such as milk, when cells are growing versus in buffer solutions, when cells enter a resting state. The positively charged amino groups and negatively charged hydroxyl groups present in the MNP functional groups al so aid in MNP suspension within aqueous - based simple and complex matrices such as PBS or milk , improving proximity for MNP and cell binding opportunities. 90 In the previous section, TEM images showed MNP - cell complexes that were analyzed for their binding p roperties. For example, Salmonella bind MNP - F#1 (Fig. 5. 1A) primarily through their flagella in selective areas, with some MNP also adhering to the cell wall, while MNP - F#2 bind along the cell wall (Fig. 5.1B ) . Also shown were MNP - F#1 clearly binding both E. coli (Fig. 5.2 B) and Bacillus (Fig. 5.3B ) cell membranes, while MNP - F#2 aggregate along their walls (Figs. 5.2D and 5.3C , respectively). As will be shown below, u ser - friendly MNP used in this study rapidly extracted pathogenic bacteria from undiluted mil k , beef juices and apple cider. This was in part due to their superparamagnetic property. Superparamagnetic particles show random electron spins in the absence of an external magnet, allowing them to disperse easily within liquids. As shown in earlier (Fig . 3.2A), both MNP types display strong magnetization in the presence of an external magnet, comparable to a similarly carbohydrate - functionalized MNP [99] , but when this external magnetic force is removed, or zero, MNP lose magnetization. Therefore, MNPs are easily suspended in most liquid foods and likewise rapidly separate d . Other methods to rapidly detect pathogenic bacteria have proposed various bioreceptors and support materials, including antibodies [10], [16], [63] [68] - consuming separation of the target bacteria from the food matrix to minimize signal interference, or culturing to first increase target quantity. 6.3.2 Definition of a Complex Matrix As mentioned earlier in the literature review , numerous biosensors work well against various species of pathogenic bacteria in simple buffer and water suspensions or following extensive cell - matrix separation steps. In this work , the freely suspended MNP and MNP - strips 91 were able to directly extract bacteria from liquid matrices that contained high amounts of carbohydrates in the form of fats, sugars and/or proteins. In addition, as will be shown later, this biosensing method was able to distinguish between samples and negative controls (p < 0.05), even though there was evidence of matrix binding to the MNP see in Fig. 5.6D . Most food - based liquids are rich in carbohydrates, as well as minerals. Even though milk, apple cider a nd homogenized egg are all over 80% water - based, they contain substantial amounts of carbohydrates and other nutrients (Table 6.1) . In a single serving (240 mL), milk has 28 gm of sugars, lipids and proteins in a whey serum, while, apple cider has 30 gm of sugars. In the same volume, homogenized eggs contain over 32 g of carbohydrates, most as protein. These matrices, therefore, are here defined as complex liquids since their carbohydrates will be electrostatically attracted and attach to the MNP car bohydrate ligand. The binding between the matrix and ligand carbohydrates are through electrostatic forces due to their similar chemistries. In contrast, PBS does not contain carbohydrates, and beef juices have minimal myoglobin proteins released from the muscle. These differences in the complexity of the matrix will be shown to have different results on nano - biosensing using MNP. 6 .3.3 Suspended MNP Capture in Milk Matrices Both MNP - F#1 and MNP - F#2 were able to extract bacteria directly from undiluted milk, which is a thick , nutritionally rich and chemically complex matrix. Capture indexes (C I ) and increased cell concentration factor s for large volume capture (LVC) in milk are listed in Table 6.2 for MNP - F#1 and Table 6.3 for MNP - F#2. Salmonella, and E. coli capture in all milks was between 2 ppt to 20 ppt CI . Meanwhile, Bacillus showed significantly higher capture of 40 ppt to 120 ppt across all milk types (p < 0.05) regardless of MNP type , except in fat free milk with MNP - 92 F#2 . These CI for capture in c omplex milk matrices were for samples spiked with 3 to 5 log CFU/mL bacteria and compar ed well with capture of Salmonella in PBS , a simple matrix , at 20 ppt ( 20 ppt ± 6 ppt at 3.78 log CFU/mL, n=3). Capture for Bacillus, a Gram - positive microbe, may be higher than Salmonella and E. coli , both Gram - negative microbe s , due to the higher Table 6.1 Composition and properties of liquid matrices. Composition Milk Beef Juice Apple Cider Homogenized Egg PBS Serving Size 1 cup (240 mL) 8 oz (240 mL) 8 oz (240 mL) # Calories (Kcal) 150 120 133 Fat (gm) 8 * 0 0 Cholesterol (gm) 25 0 0 Sodium (gm) 105 135 480 Potassium (gm) 0 60 373 Carbohydrate (gm) 12 (sugars) 30 (sugars) 5 Protein (gm) 8 + 0 26 Water (%) ~80 ~75 ~80 ~80 100 pH (25 o C) 6.54 5.27 3.21 8.58 7.44 Conductivity (mS/cm) 4.92 11.84 19.2 8.83 14.2 * 8 gm for Vitamin D, 5 gm for 2% reduced fat and 0 gm for fat free milks + minimal amount of myoglobin proteins released from muscle tissue # based upon reported 3 tablespoon ( tblsp ) (46 gm) serving size; 3 tblsp = 1.5 oz concentration of lipopolysaccharide in its cell wall which may bind the common glycan ligand present on each MNP. The highest accelerated bacterial growth rates in milk wer e 2.26 for Salmonella in 2% reduced fat milk; 2.86 for E. coli in both 2% reduced fat milk and fat free milk; and 1.38 for Bacillus in fat free milk. Naturally bacteria attach to most surfaces to form biofilms [233] [237] . But even though cells will attach to sample containers, evidence of cell binding to the MNP were shown in the 93 previous section, ensur ing that the colony growth following MNP extraction was directly related to the true bacterial load within the sample. Statistically, no r elation was found between MNP type and either milk or bacteria type for capture , except Bacillus as mentioned above , nor any significant difference in capture between MNP type for all milk and bacteria types ( = 0.05) . Within a particular MNP, though, the re were significant differences in capture between milk and bacteria types and these differences are noted in Tables 6.2 and 6.3 . Specifically, for MNP - F#1 there are true differences in the higher capture of Bacillus in vitamin D and fat free milk s . Extrac tion using MNP - F#2 also showed a true difference exis ted in capture of Bacillus in 2% fat free milk. All of these combinations have C I greater than 75 ppt and show that biocompatible carbohydrate ligands may be identified to increase targeted pathogen dete ction , particularly Gram - positive Bacillus using the amino/glycan - functional groups for capture in milk . Conventional bacteriological methods are time - consuming [54], [268] , whereas MNP extraction e fficiently replaces cumbersome dilution and centrifugation steps in liquid food sampling, while rapidly concentrating bacteria from large volumes, improving test sensitivity. Bacterial elution from MNP was even unnecessary since its presence does not affec t cell growth on agar plates following capture; and MNP was even safely included in polymerase chain reaction (PCR) without altering DNA amplification [70] . As mentioned in the methods section , Salmonell a colonies were enumerated on BGA (Fig. 6.1A ), while E. coli colonies (Fig. 6.1B ) and Bacillus colonies (Fig. 6.1C ) were enumerated on LBA. Capture was carried out at a five - fold serial dilution of the stock culture, resulting in 3 to 4 log CFU/mL final co ncentrations in fresh milk reported in 94 Figure 6.1 Photos of extracted bacterial growth on solid agar for Salmonella on BGA (A), E. coli on LBA (B), Bacillus on LBA (C), and TEM images of milk contamination bound to MNP (D). Tables 6.2 and 6.3 , which is in the range of maximum contamination acceptable for raw milk (100,000 CFU/mL) and allowed for pasteurized milk (20,000 CFU/mL) [244] . After capture, MNP - cell complexes were concentrated to a final 1 mL volume. Bacterial concentration also reduces sample handling since microbial pre - enrichment is not necessary. Occasionally fungal colonies would appear on incubated media plates following milk sampling (Fig. 6.1D ). Numerous spoilage and pathog enic bacteria have been reported in raw and pasteurized milk and milk products [52], [269], [270] . The fungal colonies consisted of aggregated clear globes but did not appear in every experiment. These colonies were cultured in TSB and i maged with TEM, showing rod - like cells of 10 to 20 m in length (Fig. 6.1D ). No appreciable effect was seen on the capture indexes (C I ) for the target pathogens when this milk contamination was present. Additionally, data for any milk product that tested p ositive for the presence of other pre - existing non - fungal contamination was rejected and the experiment repeated. 6 .3.4 Suspended MNP Capture of Multiple Bacteria Products contaminated with multiple types of competitive bacterial species occur infrequently in the food industry but results here show that each different species presen t coul d 95 Table 6. 2 Capture index (CI) of Salmonella, E. coli and Bacillus using susp ended MNP - F#1 for large - volume capture type following the standard five - fold serial dilution.+ (n=3) Milk Bacteria Concentration at Capture (log CFU/mL) Capture Index Average ( ppt ) Standard Deviation ( ppt ) Positive Control Retained Cell Concentration (log CFU/mL) Increased Cell Concentration Factor in Milk Vitamin D Bacillus cereus m 2.93 118.81 b 6.75 - 0.18 1.03 E. coli O157:H7 4.46 5.92 a 2.03 1.98 2.09 Salmonella Enteritidis 4.33 7.50 a 1.13 2.07 3.42 2% Reduced Fat Bacillus cereus 3.96 111.66 b 20.54 1.84 0.94 E. coli O157:H7 3.61 8.63 a 4.71 1.47 3.86 Salmonella Enteritidis 4.17 16.66 a 1.18 2.59 2.26 Fat Free Bacillus cereus 3.84 21.36 a 6.44 1.33 1.38 E. coli O157:H7 4.40 4.19 a 0.76 1.88 3.87 Salmonella Enteritidis 4.28 6.63 a 0.26 2.26 1.31 * M 0.05) + Negative control colony counts per experiment: m = 1 CFU and 1 CFU on separate plates. still be simultaneously extracted with high reliability using MNP. Little loss in capture capacity was experienced for simultaneous capture of Salmonella, E. co li and Bacillus using MNP - F#1 . Simultaneous capture in small volume capture (SVC, Table 6.4 ) showed high C I in PBS ( 268 ppt ) and vitamin D milk ( 67 ppt ) with each species at equal concentrations, and total bacterial concentrations of over 3 log CFU/mL. The C I for each of the species in the combined mixture compared well with their individual SVC for the corresponding matrix (Table 6.4) . MNP have a high surface area to volume ratio resulting in a high quantity of available ligand for capture. As described e arlier, an equilibrium is established for each type of bacteria between the supernatant and those attached to MNP. Therefore, the ligands are not saturated with 96 cell attachment for a particular microbe, causing a maximum amount of capture per cellular conc entration. Combined with the varied MNP - cell binding between different microbial species Table 6.3 Capture indexes (CI) of Salmonella, E. coli and Bacillus using suspended MNP - F#2 for large - volume capture (LVC) within three types of undiluted milk (25 each milk type following the standard five - fold serial dilution. + (n=3) Milk Bacteria Concentration at Capture (log CFU/mL) Capture Index Average (ppt) Standard Deviation ( ppt ) Positive Contro l Retained Cell Concentration (log CFU/mL) Increased Cell Concentration Factor in Milk Vitamin D Bacillus cereus m 2.93 39.30 a 19.00 - 0.18 1.03 E. coli O157:H7 4.46 4.48 a 2.33 1.92 2.09 Salmonella Enteritidis 4.25 6.41 a 1.47 1.93 3.42 2% Reduced Fat Bacillus cereus 3.96 76.63 b 40.47 1.82 0.94 E. coli O157:H7 3.61 6.71 a 0.86 1.58 3.86 Salmonella Enteritidis 4.09 16.63 a 1.42 2.49 2.26 Fat Free Bacillus cereus 3.84 5.36 a 1.37 1.42 1.38 E. coli O157:H7 4.40 3.42 a 0.l4 1.84 3.87 Salmonella Enteritidis 4.28 5.90 a 0.35 2.25 1.31 * M + Negative control colony counts per experiment: m = 4 CFU. discussed earlier, MNP can therefore simultaneously extract various bacterial species. As more s elective carbohydrate groups are identified in future work, these pathogens can then be reliably separated at lower costs than antibodies [75], [163] . Comparing values for CI in SVC were h igher than corresponding LVC by an order of magnitude for individual capture in vitamin D milk using MNP - F#1, except for Bacillus. Similarly, capture of Salmonella in PBS in using MNP - F#1 in LVC was 20.13 ppt (20.13 ppt +/ - 6.16 ppt), again an order of mag nitude lower than CI, 132 ppt, for SVC (Table 6.4). Capture in LVC was in 97 25 mL total volume, as opposed to 500 L for SVC which will affect the opportunity for interaction between the microbes and MNP. With 1 mL of MNP used in 25 mL LVC, or 4% of the volu me, this is an order of magnitude less than 200 L of MNP used in 500 L SVC, or 40% of the volume. Table 6.4 Microbial capture indexes using suspended MNP - F#1 for rapid extraction of three competitive bacteria, Salmonella, E. coli and Bacillus, either simultaneously (at 10 - 6 dilutions of the stock culture) or separately (at 10 - 5 dilutions) from PBS and vitamin D milk in small - volume capture (SVC) * . (n=3) Matrix Bacteria Concentration at Capture (log CFU/mL) Capture Index Ave rage ( ppt ) Standard Deviation ( ppt ) Positive Control Retained Cell Concentration (log CFU/mL) PBS Total Combined Bacteria 3.17 268 284 1.60 Salmonella Enteritidis fraction 2.71 163 156 1.22 E. coli O157:H7 fraction 2.91 342 416 0.52 Bacillus cereus fraction 2.19 143 - 1.30 Milk a Total Combined Bacteria 3.72 67 12 2.59 Salmonella Enteritidis fraction 3.20 59 8 2.14 E. coli O157:H7 fraction 3.50 70 23 2.34 Bacillus cereus fraction 2.73 113 97 1.52 PBS Salmonella Enteritidis alone 3.99 132 59 2.64 E. coli O157:H7 alone b 3.58 51 18 2.12 Bacillus cereus alone 3.75 624 444 3.06 Milk Salmonella Enteritidis alone 3.89 80 36 2.68 E. coli O157:H7 alone c 4.22 47 8 2.82 Bacillus cereus alone d 3.73 85 60 0.00 * Negative control colony counts: a 3 CFU E. coli , 1 CFU Bacillus , and 1 CFU fungi; b 1 CFU; c 3 CFU and 2 CFU on separate plates with 2 CFU, 1 CFU and 1 CFU fungi on separate plates; d 1 CFU and 6 CFU fungi on separate plates. 98 6 .3. 5 Milk Matrix E xposure E ffects on Suspended MNP C apture. Initially the goal of this project was the design of a nano - biosensor that may be inserted into individual food packaging to monitor pathogenic bacteria presence. Therefore, the effect that extended exposure of the complex matrix would have on the MNP was necessary to quantify. Binding between the MNP and targeted cells is the transduction element for this nano - biosensing method. Therefore, any noticeable reduction in attachment would likewise reduce detection responses. Capture indexes following nine - day long - term exposure of MNP to each milk type in LVC for Salmonella are shown in Table 6. 5 . Compared to Salmonella capture in fresh milk (Tables 6.2 and 6.3 ), C I were reduced to 1 ppt to 3 ppt, except for MNP - F#2 capture in fat free milk at 6 ppt . All reductions in C I were significant (p < 0.05), exce pt for MNP - F# 2 in fat free milk. Overall, vitamin D milk exposure caused an approximate 5 ppt loss in C I , 2% reduced fat milk caused up to a 1 5 ppt loss in capture, and fat free milk caused a 3 ppt de crease in MNP - F# 1 capture. Whereas, in SVC, long - term ex posure of MNP - F#1 suspended in vitamin D milk for five days actually showed an increase in Salmonella CI (100 ppt +/ - 22 ppt), while E. coli and Bacillus capture did decrease (37 ppt +/ - 8 ppt and 38 ppt +/ - 16 ppt , respectively) from individual capture values reported in Table 6.4 , all at comparable cell concentrations of approximately 4 log CFU/mL . Following long - term MNP exposure for LVC, a flocculated sludge residue on the MNP - F#1 was formed in vitamin D milk and fat free milk, and fat free milk samples did not magnetically separate as easily as in fresh milk. Re - suspension of the MNP - cell complexes into fat free milk was also difficult. Milk is a compl ex mixture of lipids and fat globules suspended in whey, along with phagocytes and casein micelles. The phagocytes and caseins act enzymatically against the lipids and fats, causing lipolytic hydrolysis of triacylglycerides and casein proteolysis over time . 99 This degradation of the milk components most likely contributed heavily to altered properties of the milk. Salmonella dilutions carried out in nine - day old milk for LVC, showed accelerated cell growth by factors of 17 to 22, depending upon the milk (Tab le 6. 5 ). These were an order of magnitude increase over those reported for fresh milk in Table s 6.2 & 6.3 , resulting in 1 to 2 log CFU/mL higher cell concentrations for MNP capture after long - term milk exposure. This increased accelerated growth over that in fresh milk is most likely due to the substantial amount of degraded fat and lipid products available for bacterial growth. These degraded products would require less microbial metabolic exertion. At over 1 ppt , MNP displayed reliable capture capacity i n various milk matrices, in both SVC and LVC, for individual or combined bacterial loads between 2 to 5 log CFU/mL, and even after MNP experienced extended milk exposure. Other researchers doing studies in milk reported accelerated bacterial growth [271], [2 72] , and MNP used in this study still captured at these higher concentrations. All milk types provided rich, nutritional environments for microbial growth. In an 8 oz serving, all three milk types had the same fraction of whey product and all contained 8 gm protein, but their fat content varied with 8 gm fat in vitamin D milk, 5 gm fat in 2% reduced fat milk, and 0 gm fat in fat free milk. Even though MNP functionalized with carbohydrate ligands suspend easier within milk matrices due to the similar chem ical composition, it was shown that the milk fats and proteins will not interfere with cellular binding. C I in milk was not adversely affected, even following extended MNP exposure to refrigerated milk over days. Since MNP capture is unaffected by matrix e xposure, this makes possible the development of a biosensor - patch using biocompatible 100 carbohydrate - functionalized MNP, that may be affixed within liquid food containers to signal contamination when integrated with an external magnetometer sensor. Table 6. 5 Capture indexes for Salmonella in large - volume capture following long - term suspended MNP - exposure+. (n=3) Milk Type MNP Type Concentration at Capture (log CFU/mL) Capture Index Average (ppt) Standard Deviation ( ppt ) Positive Control Retained Cell Concentration (log CFU/mL) Increased Cell Concentration Factor in Milk Vitamin D F#1 5.29 1.37 0.08 2.16 22 F#2 5.29 1.01 0.18 2.08 2% Reduced Fat F#1 5.23 1.20 0.14 2.52 19 F#2 5.23 2.16 0.39 2.76 Fat Free F#1 5.17 2.73 1.03 2.20 17 F#2 5.17 6.00 1.46 2.54 + Negative control colony counts were zero for all replicates. 6.3. 6 Maximum Capture Index due to Equilibrium By instinct, MNP should extract the same number of cells from liquids no matter the concentrations of cells present given otherwise the same sampling conditions. During capture, the supernatant consistently contain ed a fraction of targeted cells, and when extracted MNP - cell complexes were resuspended into fresh matrix, additio nal cells would be released into the fresh supernatant. There is a type of cellular he bond s between the microbial target surface epitopes and carbohydrate functional groups are formed , which has been se en in other detection methods [273] . This is similar to a chemical reaction that is at equilibrium, with the forward product formation reaction (MNP - cell bonding) matching the reverse reactant formation from the product (MNP and cell detachment). Multiple factors will define this equilibrium , or capture snapshot, between cells in MNP - cell complexes and those 101 remaining in the supernatant, particularly the matrix composition, MNP to sample volumes ratio, as well as the ligand - microbial - surface - epitope pair involved in bonding. Microbial captur e equilibrium could be seen i n SVC experiments where it was found that the ratio of captured cells to those in the super natant varied both by microbial target and matrix. In simple PBS, Salmonella , E. coli, and Bacillus had on average ratios of 0.25, 0.07 and 0.96, respectively, captured cells in MNP - cell complexes compared to those in the supernatant. These ratios were 0.10, 0.09 and 0.14, respectively, for these same microbial targets in vitamin D milk. The trends in these ratios of course mimicked the ch ange in CI reported in Table 6.4 for these same matrix - microbe pairs, which ranged from 47 ppt to 624 ppt. In a separate experiment looking at cell loss from MNP - cell complexes in fresh matrix , Salmonella captured in fat free milk using MNP - F#1 and MNP - F#2 showed an order of magnitude loss of the captured cells to the supernatant , shown in the resulting CI from 4.61 ppt to 0.99 ppt and 2.72 ppt to 0.54 ppt, respectively with just one change of fresh matrix. Binding between MNP carbohydrate groups and cells is through non - covalent, electrostatic forces that may create a dynamic environment, with multiple forces pulling at the cell from the MNP and matrix. Equilibrium within this dynamic environment of MNP extraction would then be defined as reaching the point where cellular detachment from MNP is balanced with MNP - cell re - attachment . Cell concentration, the ratio of sample volume to MNP volume, and the matrix composition will all affect the rate and possibility of forming an MNP - cell bond . At higher cell concentrations, MNP proximity to cells increases, but sufficient MNP presence may limit extraction . For example, Salmonella captured using LVC at three concentrations in 2% reduced fat milk ( 3 .3, 4.3 , and 5 .3 log CFU/mL) , showing higher captur e at lower cell concentrations ( 12.19 ppt, 7.86 ppt, and 1.09 ppt, respectively). When multiple microbial targets are present during 102 capture , as seen in the simultaneous multiple bacteria capture reported in Table 6.4, equilibrium is also established betwe en the MNP and each type of microbe present resulting . Decreases that were seen in microbial capture for LVC (Tables 6.2 and 6.3) , compared to SVC (Table 6.4) for the otherwise same conditions, were likely due to the reduced proximity of cells to MNP . Indi vidual bacterial capture on both scales was carried out at a standardized serial five - fold dilution of the stock culture, resulting in cellular concentrations of 3 to 5 log CFU/mL in their respective solution (25 mL for LVC and 500 L for SVC) . SVC had a 4 0% v/v MNP to matrix concentration (200 µL MNP into 500 µL total volume) whereas LVC w as one order of magnitude less at 4% v/v MNP to matrix concentration (1000 µL MNP into 25 mL total volume). But CI for capture in milk was only reduced at most by 15% C I for LVC for the same microbe - matrix - MNP combination , showing the capacity of MNP to capt ure in large volumes. Work in the Nano - Biosensors lab has shown that 1 mL of MNP suspension can capture bacteria with high C I and concentrate them from raw sewage, river water, and emulsified foods with volumes of 100 mL to 400 mL. 6 .3.7 Suspended MN P Capture in Beef Juices, Apple Cider and Homogenized Egg As discussed earlie r , complex matrices are rich in a variety of nutrients. A s shown earlier in Table 6.1 , i n a single serving (240 mL), milk has 28 gm of sugars, lipids and proteins in a whey serum, while, apple cider has 30 gm of sugars. Beef juices, though, contain a minimal amount of myoglobin proteins released from muscle tissue . In the same volume, homogenized eggs contain over 3 1 gm of carbohydrates, most as protein. In fact, it is this h igh protein content in homogenized egg that may have been the reason that MNP - cell complexes did not migrate to the magnet during separation. This prevented calculation of capture indexes in homogenized egg for Salmonella 103 extraction. Also, t his magnetic re sistance result was one driving force in the develop ment of an MNP - strip biosensing method , which will be introduced later . As reported above, microbial capture in milk was above 3 ppt, up to 120 ppt for Bacillus , for a variety of pathogenic bacteria, milk types and two MNP, even when the MNP were exposed to the aging fresh milk over days. In initial tests, though, suspended MNP - F#2 extraction of E. coli O157:H7 from beef juices and Listeria monocytogenes from appl e cider were at substantially lower amounts of 2.5 parts - per - million (ppm) and 11.4 ppm, respectively (n=3, data not shown). In Salmonella capture from homogenized eggs, t he minimal (< 10%) of MNP - cell complexes collected in magnetic separation due to thei r close proximity to the magnet even showed 0. 3 ppm C I on average (n=3, data not shown). Both milk and homogenized egg have high fractions of protein as indicated in Table 6.1, while beef juices and apple cider have essentially none. Further comparing the matrices used in microbial capture, milk has a more neutral pH, similar to PBS, whereas beef juices and apple cider are more acidic. Capture in PBS (20 ppt in LVC) and milk w as high er for all bacteria and MNP (Tables 6.2, 6.3 & 6.4) than those C I for beef juices and apple cider. During capture, the acidic environment may alter the binding properties of the glycan and amino acid ligands immobilized onto the MNP, reducing their ability for hydrogen bonding to the cell surface structure, resulting in lower ba cterial extraction. 6.3.8 Rapid D etection of B acteria in C omplex L iquids U sing C yclic V oltammetry Cyclic voltammetry detection was i nvestigat ed as a more rapid signaling method following rapid MNP extraction , even from complex liquid matrices. S uspended MNP - cell complexes from L VC were test ed following the method described for MNP - strips. Here, 100 µ L of the concentrated sample was resuspended into 150 µ L of chemical mediator, then magnetically separated from the 104 MNP - cell complexes. A volume of 75 µ L of t he chemical mediator was then tested for remaining strength using CV . Initial sampling found the sample - to - negative - control ratio ( S/N ) for E. coli O157:H7 detection in beef juices of 0.72, Salmonella detection in Vitamin D milk of 0.98, and Listeria detection in apple cider of 1.07. C apture of E. coli in beef juices showed significant separation between the sample and negative control responses (p < 0.05) , whereas Salmonella in milk and Listeria in cider were significantly separated from their respective negatives at p < 0.15 . These CV tests for sampling in complex matrices were the first to indicate matrix effects in CV testing with S/N > 1.00, as op posed to testing in PBS that is commonly carried out in the Nano - Biosensors lab which consistently show ed S/N < 1.00. Additionally, as mentioned above, MNP - Salmonella complexes were unable to be magnetically separated from the thicker homogenized egg matrix. This was a motivati ng factor to instead optimize resources and focus development on the MNP - strip technology with CV detection discussed next. 6.3. 9 Confirmation of MNP - strip Bacterial Capture in Simple Buffer Simultaneous to suspended MNP testing, develop ment was ongoing to design tag - on biosensors for individual food packages . The goal in this work was to boost consumer confidence in food security along the suppl y chain using MNP - strips functionalized with carbohydrate ligands . These would provide a viable option for biocompatible detection as opposed to utilizing antibody - . C arbohydrate - function alized MNP - strip s would eliminate this concern, and using it for improved microbial extraction would eliminate the need for supernatant handling. During development of the MNP - strip, it was determined that MNP - F#2 affixed to plastic strips extracted on ave rage two - times more bacteria than MNP - F#1, using MNP - cell growth in 105 TSB as described earlier (data not shown) . T he amino acid and glycan carbohydrate ligands alone did not adhere to the plastic strips without being first immobilized onto the MNP surface . O nly when the ligands were immobilized onto the MNP did they adhere to the plastic strips in the economical method used. On average, 1.03 ± 0.51 mg of the MNP - F#2 were immobilized per MNP - strip, while the weight change of positive control strips, with no MN P, was - 0.12 ± 0.36 mg. This level of functionalization was slightly below the target of 1.25 mg of MNP - F#2 per strip for the method described above. Immobilization of the amino/glycan - functionalized MNP onto the plastic strips would presumedly provide hi gher capture ability for the ligands due to the high surface area to volume ratio for the 200 nm MNP. The amino/glycan ligand used in this study had shown preferential attachment to cells over complex matrix components as reported earlier , able to capture over 3 ppt of bacteria in different milks in suspended form . In addition, the biosensing method was able to distinguish between samples with cells against negative controls. Initial capture testing with MNP - strips was carried out in PBS . Salmonella cultures were diluted between 2 to 6 log CFU/mL in 25 mL of PBS and extracted using the MNP - strips. Capture was confirmed through subsequent Salmonella growth in TSB from the MNP - strip - cell complexes and was directly related to the concentration of available cells in PBS (R 2 = 0.9307, Fig. 6. 2 A ). Negative control MNP - strips that were exposed to PBS with no cells present did not grow bacteria when cultured in TSB. Separate MNP - strip capture experiments were run to investigate rapid bacterial load detection using cyclic voltammetry (CV) signaling. In these experiments both soft plastic (Fig. 6. 2 B ) and hard plastic (Fig. 6. 2 C ) strip types were used to capture Salmonella in PBS. In these preliminary tests, MNP - strips were exposed to 100 µ L of diluted culture in PBS. Upon capture, 106 Figure 6.2 Evidence of MNP - strip Salmonella capture in PBS through TSB culture (A) and detection using cyclic voltammetry for soft plastic MNP - strips (B) and hard plastic MNP - strips (C), which showed an inverse linear relation (R 2 = 0.7132) to cells available for capture (D). Inset plots (B & C) sh ow cathodic peaks in greater detail and the order of concentration listed in the legend (actual units log CFU/mL) follow the order of peak response. the MNP - strip - cell structures were then reacted with chemical mediator as described above . The remaining chemical mediator strength was then determined using CV, finding that normalized current response (S/N) decreased linearly as available cell presence increased (R 2 = 0.7132, Fig. 6. 2 D ). As cell presence on the MNP - strip increased, more of the chemical medi ator was reduced, thereby resulting in lower CV normalized current responses. High variance in the normalized CV response was most likely due to the variation seen in immobilized MNP weights on the MNP - strips. In future work, higher amounts of immobilized MNP - F#2 on the strips covering more of the plastic surface area sh ould reduce the amount of variability in MNP immobilization, improv ing test reproducibility . In addition, the volume of 107 chemical mediator used on the strip should then be optimized to allow for better distinction between bacterial loads. These changes would allow more significant separation between CV responses for 1 log CFU/mL differences in contamination. 6.3.10 Rapid D etection of B acteria in C omplex L iquids U sing MNP - strips with C yclic V oltammetry As described above, cyclic voltammetry (CV) allow ed quick detection of cell presence on the MNP - strips, when compared to a negative control. Following capture, a chemical mediator was applied to the MNP spots and r eaction between the chemical mediator and the MNP - strip contents reduced the mediator for bot h samples and negative controls. The resulting reduction in chemical reactivity for the mediator translated into a lower CV normalized peak current response (NPCR) for that MNP - strip. When complex matrix components were present on the MNP - strip, additional reduction in the chemical reactivity resulted in even lower NPCR. Matrix effects upon CV peak current responses can be investigated best by comparing the negative control to the chemical mediator responses. W hen s imple matrix components attach to the neg ative control MNP - strip from PBS (Fig. 6.3A ) and beef juices (Fig. 6.3B ) they deplete at most 10% of the chemical mediator reactivity , in addition to any mediator reaction with the amino - glycan carbohydrate ligands on the immobilized MNP and exposed plast ic stripping. Whereas, the effect of a complex matrix such as homogenized egg showed substantial reduction in the chemical mediator CV peak current response as the percent of the egg solution increased (Figs. 6.4A & B ). In fact, complex matrices reduced th e chemical mediator response to 58% for homogenized egg (Fig. 6.4A ), 75% for apple cider (Fig. 6.5B ), and 65% for vitamin D milk (Fig. 6.5C ). This 108 Figure 6. 3 Effect liquid matrix components upon CV detection for MNP - strip cell capture of Salmonella in PBS (6.19 log CFU/mL available cells, S/N = 0.85, p = 0.04) (A) and E. coli in beef juices (7.7 log CFU/mL available cells, S/N = 0.92, p = 0.08) (B). All peak c urrent responses here were normalized to the mediator reference run at the time of testing to compare matrix effects upon current responses. (error bars = standard deviation) contrasted with the matrix effects upon CV responses for suspended MNP testing d iscussed earlier where beef juices depleted over 20% of the mediator, vitamin D milk had no effect, and apple cider reduced it by 30%. Clearly differences between suspended and affixed MNP CV responses must be acknowledged in future selecting the optimal t est design. PBS had a larger effect on the peak current response compared to beef juices due to its available sodium content which can reduce the peak current response of the chemical mediator . Figure 6. 4 Effect of homogenized egg matrix component upon CV peak current response increased proportionally as the purity of the matrix increased (A) such that MNP - strips exposed homogenized egg diluted with PBS showed higher peak current responses for higher PBS dilutions (B). All peak current res ponses here were normalized to the mediator reference run at the time of testing to compare matrix effects upon current responses. (error bars = standard deviation) 109 Since the simple matrices used in this study did not contain appreciable carbohydrates, the sodium present may then bind to negative pockets within the MNP carbohydrate ligand and subsequently react with the chemical mediator . Beef has approximately 60 mg of sodium per serving, similar to the level in the PBS, but this sodium is mostly bound to the meat, and therefore may not dissolve into the water solution that is tested with the MNP - strip . In fact, the conductivity of PBS was 14.2 mS/cm, whereas beef juices conductivity was only 11.8 mS/cm. Table 6.6 Normalized peak current responses (NPCR) for MNP - strip negative controls in simple and complex matrices and the matrix composition used in linear regression statistical analysis. (n=3) Combination Normalized Peak Current Responses (NPCR) Sodium (mg) F ats (gm) C arbohydrates * (gm) P roteins (gm) PBS 0.87 100 0 0 0 0.93 100 0 0 0 0.90 100 0 0 0 Egg 0.65 470 0 5.3 26.5 0.53 470 0 5.3 26.5 0.56 470 0 5.3 26.5 Cider 0.71 135 0 30 0 0.73 135 0 30 0 0.83 135 0 30 0 Beef 0.94 60 0 0 1 1.01 60 0 0 1 1.07 60 0 0 1 Milk 0.64 105 8 12 8 0.68 105 8 12 8 0.64 105 8 12 8 *carbohydrates were determined to be have an insignificant effect upon NPCR; subsequent linear regression was run after carbohydrates being removed Overall, the matrix components that showed a significant effect upon the level of chemical mediator expended by the negative controls were sodium, fats and proteins ( Table 6.6, p < 0.05). 110 Figure 6.5 Effect of complex liquid matrix components upon CV detection for Salmonella in homogenized egg (12. 3 log CFU/mL available cells, S/N = 1.25, p = 0.04) (A), Listeria in apple cider (7.9 log CFU/mL cells available, S/N = 1.13, p = 0.06) (B), and E. coli in vitamin D milk (8.5 log CFU/mL cells available, S/N = 1.08, p = 0.04) (C) showed lower peak current responses for negative controls, while cell presence expended less of the chemical mediator, resulting in higher peak current responses. All peak current responses here were normalized to the mediator reference run at the time of testing to compare matrix effects upon current responses. (error bars = standard deviation) Sodium and fats had a negative effect upon the level of NPCR for the negative controls, while proteins showed a positive effect (R 2 = 0.9253). Carbohydrates showed no significant effect (p = 0.32) upon NPCR and were removed from the final linear regression. Finally, the presence of cells in the matrix has opposing effects depending upon the complexity of the matrix. When cells are present in simple matrices such as PBS (Fi g. 6.3A ) or beef juices (Fig. 6.3B ), the level of chemical mediator expended is increased above that of the negative controls, resulting in lower sample NPCR. For significant separation between sample (S) and negative controls (N) in simple matrices, S/N ratios would be less than 1.0. But, cell presence in complex matrices reduced the amount of expended chemical mediator, since the carbohydrate ligands on the MNP - strips preferentially bind to the cells over the matrix components. This results in samples NPCR that are h igher than corresponding negative controls NPCR as seen for capture in homogenized egg (Fig. 6.5A ), apple cider (Fig. 6.5B ), and milk (Fig. 6.5C ). Therefore, for significant separation, S/N ratios would be greater than 1.0 in complex matrices. For the bios ensing 111 tests in this pilot study, all averaged NPCR for samples were significantly different from corresponding negative controls (p < 0.05) except in beef juices. As noted in the captions of Figs. 6.3 and 6.5, the presence of any microbial nutrient cause d higher cell concentrations available for capture due to accelerated cell growth. All sampling was carried out at four - fold dilutions of the culture spiked into the appropriate matrix, but accelerated growth occurred in beef juices (Fig 6.3B), homogenize d egg (Fig. 6.5A), apple cider (Fig. 6.5B), and vitamin D milk (Fig. 6.5C) , for microbial concentrations at the time of capture of 7.7 to 12.3 log CFU/mL . 6.4 Conclusions Two pitfalls in most published MNP biosensing methods reviewed earlier were cross - r eactivity against non - target species and matrix interference with detection signaling. Even using selective antibody ligands, Listeria detection showed 50% cross reactivity [88] and a Staphylococcus assay showed 40% cross reactivity [87] , both against non - target pathogens. Another MNP - antibody method to detect Listeria cross - reacted with Staphylococcus [86] . Non - selective, carbohydrate surfactants show br oad spectrum capture, but vancomycin - PEG - MNP suffered only 20% cross reaction with Gram - negative bacteria [119] . When food matrices such as beef wash solution or pork extracts were introduced into the sampling method s , detection sensitivity was reduced with capture at 70% [89] . Carbohydrate - functionalized MNP produced in the Alocilja Nano - Biosensors Lab showed similar matrix interference, but matrix effects were factored in the final CV signal interpretation using controls. Glycan - and amino/glycan - functionalized MNP were used here to extract a broad spectrum of pathogenic bacteria from milk to demonstrate the ability of economical carbohydrate ligands to rapidly bind microbes in undiluted complex matrices. Many nano - biosensing methods are vetted using either simple b uffer solutions or following removal of food matrix components. Suspended 112 MNP - F#1 and - F#2 extracted over 3 ppt of three of the pathogenic bacteria (Tables 6.2 & 6.3), at concentrations of 2 to 5 log CFU/mL within minutes and concentrated them into 1 mL vo lumes . High C I occurred when multiple bacteria were present (Table 6.4) or even following MNP extended exposure to the milk matrix (Table 6. 5 ) . Through microbial concentration , electrochemical detection had show n increased sensitivity with initial CV dete ction tests in the Nano - Biosensors Lab using suspended MNP (data not shown) , even , as reported , at low levels of contamination [82], [92], [118] . In addition, although MNP can be use d directly in many detection steps, the non - covalent electrostatic binding of the MNP - cell complexes can be safely broken for further cell study. MNP - strip s were designed us ing these biocompatible amino/glycan ligands affixed to magnetic nanoparticles (MN P - F#2) . A r apid detection method was then successfully developed using cyclic voltammetry (CV) with f erricyanide chemical mediator. Overall, the MNP - strip capture and CV analysis nano - biosensing method allowed cell detection in under 30 mi n , without the ne cessity of first separating the target cells from the food matrix before extraction . - homogenized egg, as well as the other liquid food matrices, eliminati ng additional liquid supernatant handling experienced in suspended MNP methods. MNP - strip have additional benefits over more common methods. Gel electrophoresis with ethidium bromide detection and fluorescent ELISA methods require special handling and tech nical knowledge. Carbohydrate functional groups show a long shelf - life, without special storage conditions. But work is needed in improving detection limits at low cell concentrations. Improvements would include increased MNP - F#2 coverage along the plastic strip as well as optimizing the volume of chemical mediator. U sing a handheld potentiostat will allow this rapid 113 - - biosensing approach to be field portable for bacterial detection close to contamination sites thereby improving decision maki ng processes for implementing food safety mitigation strategies. In this work, cell capture was directly quantified using BAM plating methods for suspended MNP capture , in place of methods such as genomic DNA detection using PCR. Future work that would co rrelate MNP - cell quantification with BAM to that of PCR or electrochemical detection means would additionally decrease detection times. Thus, rapid extraction of total bacterial load from complex matrices is possible . Although antibody ligands may show higher specificity against targeted bacteria, their use in food packaging would create consumer concern about superbugs [274] . Therefore, as more selective carbohydrate ligands against pathogens are identified , greater selectivity will be possible [69], [75], [255] with higher detection sensitivity, and more economical assays to protect our food supply chain will be possible with greater selectivity. 114 Chapter 7. Gold N anoparticle E nhancement in D ifferential P ulse V oltammetric and S pectrophotometric D etecti on of MNP E xtracted B acteria 7.1 Introduction Since the mid - government using chemical analysis. In 1906 the Pure Food and Drugs Act was passed, then the Federal Drug Administration (FDA) was created in 1930, to further the efforts of the US government to secure our food supplies, among many other human health goals. The FDA began publishing a set of accepted methods to detect pathogens in our food and cosmetics in 1965. Although these bacterio logical analytical methods (BAM) are held as the gold standard for pathogen detection, they require extensive supplies and entail time - consuming steps. In our fast - paced world, the food on our supper table may have been harvested just days earlier and fro m half - way around the world. Whereas the standardized methods published in the editions of BAM are more reactive to pathogen presence due to their time - consuming steps, food suppliers and consumers alike desire pro - active rapid microbial detection. Nano - bi osensors have been the purported answer for rapid microbial detection, but many of the sensors developed are either economically infeasible or require expensive highly - technical instruments in their design. Returning to more simplistic nano - biosensing methods that are easily adjusted to new targets, stable over long periods and utilize affordable instrumentation will more rapidly meet the needs of industry and consumers. Combining electrically active nanopart icles such as gold nanoparticles (AuNP) for enhanced detection signaling with proven rapid MNP microbial extraction which are both functionalized with biocompatible carbohydrate ligands will provide economically feasible nano - biosensing methods for both de veloped and under/developing countries to ensure a safe food supply. 115 AuNP find numerous uses in microbiological areas, from medical diagnostic applications to detecting pathogens for food and environmental safety [6], [30], [68] . Capping agents such as citrate or dextrin provide AuNP stability against electrostatic forces in solution, reducing aggregation, and provide support for subsequent functionalizat ion with various ligands such as DNA, antibodies, or microbiological components [6], [275] . Although applications using unmodified dextin - capped nanoparticles of carbon, selenium, and silver have been developed [141], [145], [276] , and dextrin - stabilized gold colloids [277] , none of these applications use dextrin - c oated AuNP (d - AuNP) in conjunction with spectrophotometric or differential pulse voltammetric (DPV) detection. In this work , d - AuNP standards we re first characterized with respect to their electrochemical signaling using spectrometry and DPV , including in the p resence of MNP and cells . Detection using anti - Salmonella antibody functionalization of d - AuNP is then investigated as a means to show specificity in pathogenic detection. Finally, a rapid nano - biosensing for microbial contamination in milk, a complex matr ix, is described. This method used carbohydrate - functionalized ( amino/ glycan) magnetic nanoparticles (MNP - F#2) and concentrated dextrin - coated gold nanoparticles (AuNP) for rapid detection of E. coli C3000 in undiluted vitamin D milk using both spectrometr y and electrochemical detection . These economical carbohydrate - based nano - biosensors currently show broad selectivity in pathogen detection for total load of bacterial contamination present. As more selective carbohydrate ligands are designed, they can eas ily be applied to this method with little increase in detection costs. 116 7.2 Materials and Methods 7.2.1 Method of Development for Carbohydrate - based MNP - cell - AuNP Nano - Biosensing Steps taken in investigating the ability of carbohydrate - based MNP and AuNP t o detect pathogenic bacteria in complex matrices were carried out in a logical progression. Initially, AuNP were functionalized with selective antibody in attempts to prove specificity of Salmonella Enteritidis capture by MNP - F#2. Subsequently, total carbo hydrate - based nano - biosensing was pursued , using E. coli C3000 for easier TEM imaging reported above . Improved signaling was first addressed by concentrating the dextrin - coated AuNP and removing excess dextrin solution. Spectral and differential pulse volt ammetry (DPV) responses to concentrated AuNP were tested . T hen the time - spans required to ensure sufficient AuNP coverage of MNP - cell complexes for sensitive spectral and DPV responses, including when complex matrix components are present. At this point, i step - to interpret scans in which signature peaks were masked by complex matrix components. Finally, the combined carbohydrate - based MNP - cell - AuNP with spectral and DPV signal ing was tested at two microbial cell concentrations. 7.2. 2 Materials Cell strains of Salmonella Enteritidis, E. coli O157:H7, E. coli C3000, Bacillus cereus, and Listeria monocytogenes - Biosensors Lab at Michigan State University (MSU). In - house, proprietary glycan - and amino/ glycan - functionalized magnetic nanoparticles (200 nm in diameter, MNP - F#1 and MNP - F#2, respectively) and dextrin - c oated gold nanoparticles (approximately 50 nm in diameter, d - AuNP ) were used as received from the Nano - Biosensors Lab, MSU [6], [143] . Phosphate buffer solution, pH 7.4 (Neogen, Lansing, 117 MI), was prepared as directed. Brilliant G reen agar (BGA), tryptic soy broth (TSB), tryptic soy agar (TSA), Luria Bertani broth (LBB), and Luria Bertani agar (LBA) were purchased from Sigma Aldrich (St. Louis, MO), CHROMagar TM E. coli was purchased from DRG Interna tional (Springfield, NJ), and Oxford Listeria Agar base and Modified Oxford Listeria Supplement for modified O xford agar (MOX) was purchased from Neogen Corp. (Lansing, MI), and prepared as directed. A Sphero FlexiMag Separator was purchased from Spherote ch Inc (Lake Forest, IL). Square 1 oz (30 mL) clear glass bottles with phenolic caps were purchased from Thomas Scientific (Swedesboro, NJ). Vitamin D, 2% reduced fat, fat free milk, apple cider, and homogenized eggs were purchased from a local commercial seller, while beef juices were procured from the MSU meat store, all stored at 4 o C, and used fresh. Before use, all containers were sterilized externally with 70% ethanol and the contents thoroughly mixed by inversion. 7 .2. 3 Bacterial Culture Salmonella Enteritidis, E. coli O157:H7 and E. coli C3000, Bacillus cereus, and Listeria monocytogenes were rejuvenated from frozen storage ( - 70 o C) frequently on BGA, CHROMagar TM E. coli , LBA following heat shock treatment, and MOX, respectively , as master plates. A single colony was isolated, inoculated into 10 mL media as follows. Salmonella Entertidis, E. coli O157:H7 and E. coli C3000 were cultured in 10 mL TSB, while Listeria monocytogenes was cultured in 10 mL LBB, in 15 mL Eppendorf tubes overnight at 37 o C. Bacillus cereus was heat - shocked at 75 o C for 15 min in 10 mL LBB in 50 mL flasks and grown with foam stoppers overnight at 37 o C and 100 rpm. A 10% spike was then grown with the same respective conditions for four hours to log phase immediately pr eceding all experiments. 118 Tenfold serial dilutions of the stock culture, from 10 - 1 to 10 - 7 , were prepared using PBS before each experiment. Similar tenfold dilutions were made in the appropriate liquid food from 10 - 1 up to 10 - 12 to account for the immediat e accelerated bacterial growth during dilutions and used as indicated below for cell capture. Viable cells for both cultures from PBS dilutions and liquid food were enumerated by microbial plating. Salmonella was plated on BGA while E. coli , Bacillus, and Listeria were plated on LBA and plates incubated at 37 o C for up to 16 hr. Target plate counts were between 25 and 300 colonies per plate. Diluted cultures were used as indicated below, with all testing done in triplicate and at room temperature. 7.2. 4 An tibody - functionalized AuNP Dextrin - c oated gold nanoparticles (d - AuNP) were used as received from the Nano - biosensors Lab [143] and functionalized with antibody using the method of Wang et. al. [6] . Volumes of 1000 L of the original d - AuNP suspension were transferred to 2 mL pinched - tip centrifuge tubes, sonicated for 10 min, and centrifuged at 15,000 rpm at 4 o C for 2 0 min. Following removal of the supernatant, the remaining 20 L AuNP pellet was sonicated for 15 min to evenly distribute the concentrated d - AuNP evenly. No additional sonication was performed. Protein A (from Staphylococcus aureus , 1000 uL, 0.25 mg/mL in PBS, Sigma - Aldrich, St. Louis, MO) was added and the AuNP hybridized on a shaker rotator at 350 rpm for 1 hr. Following each step that follows, the tube was centrifuged at 13,000 rpm at 4 0 C for 12 min and the supernatant removed. AuNP were washed with 1000 uL PBS, resuspended in 500 uL of Salmonella polyclonal antibody (1.25 mg/mL in PBS, Thermo Scientific, Rockford, IL), and hybridized as before. The resulting Ab - AuNP pellet was resuspended in 0.1% bovine serum albumin (BSA) in PBS as a blocking 119 agent and hybridized for 30 min. The final pellet was resuspended in 1000 uL 0.1% BSA , stored at 4 o C in aluminum foil and used within 24 hr. 7.2. 5 Specificity of Capture of Salmonella Monoclonal Antibody and Cross - reactivity Ab - AuNP were tested for their abil ity to prove target Salmonella Enteritidis capture using carbohydrate - functionalized MNP - F#2 , as well as their cross - reactivity against non - target E. coli O157:H7 , as MNP - cell - Ab - AuNP complexes using both spectrometry and differential pulse voltammetry (DPV) detection . In a 2 mL centrifuge tube, 80 uL of MNP (5 mg/mL in PBS) were exposed to 100 µ L of either diluted target at 4 or 5 log CFU/mL or non - target at 4 log CFU/mL. After capture for 5 min, 100 uL of 0.1% BSA was added and allowed to b lock open MNP carbohydrate ligands for 5 min. Samples were magnetically separated, resuspended in 200 uL of Ab - AuNP for DPV testing or 100 uL of Ab - AuNP for spectrometry samples, and labeling allowed for 5 min. The MNP - cell - Ab - AuNP complexes were magnetica lly separated and resuspended into 200 uL of 0.1% BSA and 800 uL of PBS. Samples and negative controls, prepared in the same manner without cells present, were run in triplicate, unless otherwise stated. All samples and negative controls were analyzed for electrochemical responses using either differential pulse voltammetry (DPV) or spectrophotometric signaling responses, as indicated later. 7.2. 6 Concentrated Dextrin - coated d - AuNP Dextrin - coated gold nanoparticles (d - AuNP) were analyzed for their ability to provide sufficient signaling shifts at various concentrations. Concentrated d - AuNP was prepared by removing excess dextrin and solution. Volumes ranging between 25 to 250 L of the original d - AuNP suspension were transferred to 2 mL pinched - tip centrifu ge tubes, sonicated for 10 min, and 120 centrifuged at 15,000 rpm at 4 o C for 20 min. Following supernatant removal, the remaining 20 L d - AuNP pellet was sonicated for 15 min to evenly distribute the concentrated d - AuNP evenly, then used as the starting aliqu ot. Tubes were stored at 4 o C and used within 24 h r of preparation. Concentrated d - AuNP are referred to using their initial volume, e.g. a 75 L concentrated d - AuNP as appropriate . 7.2. 7 Temporal A ttachment of Concentrated d - AuNP to MNP - cell Complexes Amino/glycan carbohydrate - functionalized magnetic nanoparticles (MNP - F#2, or MNP) , was used to capture either Salmonella or E. coli C3000 to study concentrated d - AuNP temporal attachment to the cell and MNP surface using spectrometry . This procedure was in conjunction with collaborative research, and therefore used minimized sample sizes to mimic design parameters. MNP (20 L of 5 mg/mL) were exposed to 20 L of cell culture di luted in PBS to 5 log CFU/mL plus 100 L PBS for 5 min, magnetically separated and re - suspended in AuNP250 plus 40 L PBS. AuNP250 attachment was allowed for 10, 20 , and 30 min incubation times. The MNP - cell - AuNP complexes were re - suspended in 1000 uL PBS. Negative controls followed the same procedure except without cells (n = 3) . 7.2.8 Carbohydrate - based MNP - cell - AuNP Nano - biosensing E. coli C3000 culture serially diluted in milk (50 L) was simultaneously exposed to MNP (50 L of 5 mg/mL prepared in buf fer) and AuNP250, plus 100 L PBS for 30 min, magnetically separated and re - suspended in 100 L PBS for electrochemical detection or 1000 L PBS for spectral detection , as indicated below. Negative controls followed the same procedure except 121 without cells. Sample cell concentrations at the time of capture were determined to be 3 log CFU/mL and 5 log CFU/mL through BAM plate counts . 7.2.9 Electrochemical Spectrophotometric or Differential Pu lse Voltammetric Detection Differential Pulse Voltammetry DPV were run using a Uniscan PG581 potentiostat with a 1 cm 2 screen - printed carbon electrode (SPCE), driven by UiEChem software. Samples and negative controls were oxidized for 2 min at 1.2 V then c urrent response was recorded for applied potentials between +1.2 V to - 1.2 V, with a pulse height and width of 0.5 V for 50 ms, at a scan rate of 33.3 mV/s. A DPV voltammogram was recorded for each sample. Sample (S) and negative control (N) responses were averaged and S/N ratio calculated (n = 3). Spectrophotometry Spectrophotometric scans were run from 900 nm to 200 nm wavelength (Shimadzu UV - 3101PC, UVProbe software), against a reference of 200 uL of 0.1% BSA diluted with 800 L PBS. Total spectral abso rbance w as calculated by step - integration between 220 nm to 900 nm. A spectral scan was recorded for each sample. Step - integration involves multiplying the 0.5 nm wavelength step - size by the resulting absorbance at that new wavelength and summing these are as. MNP - F#2 Specificity and Ab - AuNP Cross reactivity Both MNP - cell - Ab - AuNP samples and MNP - Ab - AuNP negative controls to determine MNP - F#2 specificity and Ab - AuNP cross - reactivity were prepared for differential pulse voltammetry (DPV) where a volume of 500 uL of the sample was magnetically separated and 122 reacted with 100 µ L of 1 M HCl for 5 min, and the entire contents transferred to the SPCE (n = 3) . Whereas, spectrophotometry was run on the samples and negative controls as prepared above (n = 3) . Temporal Attachment of AuNP250 Samples were tested as prepared above using spectrometry. Peak absorbances for MNP (620 nm) and AuNP (520 nm) were monitored for the 10, 20 and 30 min MNP - cell - AuNP and MNP - AuNP complexes. Both AuNP peak height and step - integrated sp ectral area were compared with respect to time using linear regression (n=3) . Carbohydrate - based MNP - cell - AuNP Nano - biosensing Carbohydrate - based MNP - E. coli - AuNP samples and corresponding negative controls were tested as prepared above . P eak absorbances for MNP (620 nm) and AuNP (520 nm) were monitored. All spectra were numerically step - integrated between 900 to 220 nm and a veraged MNP - cell - AuNP sample step - integrations were compared to averaged MNP - AuNP negative controls (n = 3). Samples and negative controls were magnetically separated and resuspended in 100 L 1 M HCl, reacted for 5 min, then the entire contents were applied to the SPCE. Maximum absolute AuNP current response was determined between 0.3 V to 0.5 V. Averaged sample peak curr ents were compared to averaged negative controls (n=3). Sample responses were compared to negative control responses using a t - test for unpaired samples and p - values reported below. 123 7.3 Results and Discussion 7.3.1 Electrochemical S ignature R esponses for MNP - cell - AuNP B iosensing Gold nanoparticle (AuNP) labeling is an effective method to intensify electrochemical responses in nano - biosensing applications [6], [135] . In combination with rapid MNP extraction, microbial presence within liquid matrices can be identified within 45 min. Both MNP and AuNP have distinctive signature responses in both spectrophotometric detection and differential pulse voltammetry (DPV) signaling. This ensured that the AuNP response would not be masked by MNP presence, eliminating the need to separate MNP from the cell - AuNP complexes before detection f or DPV . Complex matrix components, though, may attach and mask spectral signaling, necessitating development of a novel data analysis technique presented below. Binding dynamics between carbohydrate - based MNP and AuNP, and targeted cells, were presented ea rlier in TEM images. The AuNP were of diameters smaller than 20 nm and seemingly aggregated in small 50 nm clumps when binding cell surfaces (Fig. 5.5B) but saturated the surface of the MNP (Fig. 5.6B). These binding properties were evident in both the spe ctral detection or DPV signaling , as will be discussed below . Maximum spectral responses for the concentrated dextrin - coated AuNP used in this work occurred at approximately 520 nm (AuNP250 Fig. 7.1A), whereas MNP peaks were located at approximately 620 n m (MNP Fig. 7.1A). When the two nanoparticles were prepared in a control mix in simple PBS , the stronger AuNP response persisted, while the MNP presence only shifted total absorbance to higher levels in an additive manner (AuNP250 & MNP mix Fig. 7.1A). But a s can be seen in Figure 7.1B, spectral scans of carbohydrate - based MNP - cell - AuNP and MNP - AuNP complexes following capture and labeling in PBS show lower absorbance levels, even though the same initial quantities of MNP and AuNP were used. This is the res ult of non - 124 complexed unattached nanoparticle s loss during supernatant removal. Non - complexed MNP residue may not be separated from the matrix due to insufficient vessel wall space next to the magnet. This loss account ed for at least 50% signal loss in MNP absorbance in Figure 7.1B. Substantial loss in AuNP signal is also apparent due to the limits of attachment between AuNP and the MNP or cell target. As noted in the prior chapter, a dynamic chemical equilibrium is established between attached nanoparticle s and the surrounding supernatant , causing limited attachment . Figure 7.1 Spectral scans of AuNP250 controls combined with MNP (A) and fo llowing attachment through the cell capture and labeling method (B). Although substantial nanoparticle loss was experienced in MNP - cell - AuNP nano - biosensing in Figure 7.1B, sufficient AuNP labeling occurs to preserve its signature peak at 520 nm in both the MNP - AuNP negative control and MNP - cell - AuNP scans. As seen in Figure 5.5B, aggregated AuNP were selectively attached to the cell surface molecular structures and its presence minimized the MNP attachment, which results in the lower MNP - cell - AuNP absorb ance seen in Figure 7.1B. In contrast, AuNP was seen saturating the surface of the MNP in Figure 5.6B, which 125 translated to higher spectral absorbance for MNP - AuNP in Figure 7.1B , along with a more prominent AuNP peak at 520 nm . DPV analysis of MNP - cell - AuN P complexes show similar distinctive signature current responses for the individual components present. Figure 7.2 shows current responses for these components both individually and combined in complex. The activator chemical, HCl, has a strong response at approximately 0.8 V, while dextrin - coated AuNP or antibody - functionalized Ab - AuNP show a response at 0.4 V , and MNP responds at +0.1 V. Each type of sample and control is first reacted with HCl to activate the sample for electrochemical detection. Any res idual, unreacted, HCl presence in the sample then shows a secondary peak at - 0.5 V. Additionally, a secondary AuNP response occurs at - 0.1 V. Finally, all DPV scans are normalized with respect to their baseline between 0.9 V to 0.6 V which has no HCl peak present when standards, samples with cells or negative controls are present. In normalization, the current responses between 0.9 V and 0.6 V were averaged, then all current responses in the scan, from +1.2 V to - 1.2 V , were Figure 7.2 Signature responses for MNP - cell - AuNP biosensing controls with differential pulse voltammetry. (Se= Salmonella Enteritidis) 126 divided by this averaged value. This would allow all scans results to evaluated on a level basis. DPV current responses for samples (MNP - cell - AuNP complexes) and negative controls (MNP - AuNP complexes) result in convoluted AuNP and MNP peaks, with AuNP commonly appearing as a shoulder at 0.4 V on the MNP peak at +0.1 V (Fig. 7.2). MNP - cell - AuNP/DPV nano - biosensing methods compare the A uNP current responses between the sample complexes that have pathogenic bacteria present to those of the negative controls prepared from uncontaminated standards of the same matrix. During analysis, raw DPV data is corrected with respect to the baseline, a s stated above, then the potential that results in the maximum average AuNP current response is located between 0.3 V to 0.5 V . An example of corrected DPV responses is presented in Figure S 8.2 , where the potential for maximum averaged AuNP current respons e is at 0.38 V (arrow). Finally, replicates of sample (S) and negative control (N) responses are averaged and the S/N ratio calculated. As the complexity of the DPV sample increases, the AuNP peak height decreases due to HCl mediator reaction in proportion to all components present. As shown in Figure 7.3 when pure the largest AuNP response is recorded, increasing in response as volume of concentrated AuNP increases . This indicates that HCl is not limiting in the activation step. The presence of MNP in mediator reacts with all components present , including the carbohydrate liga nds . Finally, the well as milk matrix components which will be discussed later. Method optimization for MNP, AuNP and HCl quantities would be a major factor in designi ng a reliable test for pathogen detection. 127 Figure 7.3 Effect of sample complexity upon AuNP current response in DPV biosensing. 7.3.2 Specificity of Salmonella D etection U sing A ntibody - functionalized AuNP Validation of target capture is an important aspect of nano - biosensor development, particularly when dealing with detection of pathogenic bacteria in our food supply. Antibody ligands are one of the selective methods used to ensure target capture, but they may still result in false positive and false negative results. In addition, antibody functionalization results in higher test costs, requires special test kit handling and shortens shelf life. Wang et. al designed a nano - biosensing test targeting E. coli O157:H7 using antibody ligands for both MNP extraction and AuNP labeling [6] . Using DPV signaling, they were able to detect target bacteria as low as 1 log CFU/mL target in brot h in an 1 h , although testing against a non - target bacteria for false positive results was not reported . Applying this same technique, the dextrin carbohydrate coating on the AuNP was replaced with antibody ligand functionalization (Ab - AuNP). This specifi c label was used in attempts to validate specificity of Salmonella Enteritidis detection using carbohydrate - functionalized MNP - 128 F#2 for microbial extraction. Results consistently showed unreliable, non - reproducible DPV signaling responses for samples consis ting of MNP - cell - Ab - AuNP complexes (Tables 7.1 & S 8.2 ). Table 7.1 Specificity of Salmonella detection using antibody - functionalized AuNP . Trial Capture Matrix Pathogen * n Samples Cells Available for Capture (log CFU/mL) MNP Capture Index ( ppt ) @ Post Ab - AuNP Capture Index ( ppt ) t - test + p - value AuNP Peak S/N # ave . 1 PBS Se Target 3 3.12 170 40 0. 11 0.87 2 PBS Se Target 6 2.76 520 76 0.0 2 0.83 3 Milk Se Target 6 3.40 763 21 0. 39 0.96 4 Milk Se Target 6 3.30 935 34 6 0. 49 0.99 5 Milk Se Target 6 5.14 14.4 0. 9 0. 49 1.00 6 Milk Se Target 3 9.08 1.48 0.01 0.0 4 0.60 Se Target 3 8.08 0.62 0.01 0.0 2 0.45 Se Target 3 7.08 1.45 0.03 0. 36 0.90 7 Milk Se Target 3 6.95 0.23 0.00 0. 16 0.86 Se Target 3 5.95 0.99 0.04 0. 06 1.12 Se Target 3 4.95 1.33 0.00 0.0 4 0.82 8 Milk Se Target 3 5.57 76 40 135 0.1 0 1.32 Se Target 3 4.57 8108 62 2 0.0 5 1.72 Se Target 3 3.57 8108 751 0. 44 0.97 9 PBS Se Target 3 5.69 NA 1 2 0.01 0.74 Se Target 3 4.69 NA 14 0.0 1 0.70 Ec Non - target 3 5.94 NA 1 0. 08 0.79 10 PBS Se Target 3 4.90 NA 103 0. 27 0.90 Ec Non - target 3 5.63 NA 0.08 0. 39 0.97 * Se = Salmonella Enteritidis, Ec = E. coli O157:H7; @ standard deviations for MNP Capture Index are not shown for brevity of data comparison (see Table S7. 2 ) ; + reported one - tail p - values for t - test ran for unpaired samples and negative controls within same trial test, assuming equal variances ; # S/N = AuNP current response ratio for Samples to Negative controls. Numerous test parameters were adjusted throughout development of the test, including capture volume (180 µL 25 mL), MNP - F#2 quantity (80 µL 1000 µL), cellular concentration (2.76 log CFU/mL 9.08 log CFU/mL), liquid matrix (PBS & milk), Ab - AuNP quantity (100 µL 129 to 250 µL), total volume during Ab - AuNP attachment (100 µL to 800 µL), stock culture dilution (10 - 2 10 - 6 ), diluted culture spike volume (50 µL 250 µ L), and total test volume reacted with HCl (100 µL to 700 µL) used in DPV signaling. Attempted optimization of the various parameters listed above, in efforts to develop a sensitive nano - biosensing method, focused on improving the separation between sampl e and negative control AuNP current responses at 0.4 V in DPV signaling in. This involved reducing the MNP - F#2 response at +0.1 V, while still maintaining an initially high MNP capture index . The biosensing method, though, was not able to reliably detect S almonella pathogen presence, even though a specific anti - Salmonella antibody was used. In fact, resulting S/N values from the AuNP current responses were not significantly affected by MNP - F#2 quantity, capture volume, diluted culture spike volume, cellular concentration, MNP - captured cell counts, Ab - AuNP quantity, total volume during Ab - AuNP attachment, nor total test volume reacted with HCl. Subsequent analysis of the data using principle component analysis (PCA) and multi - variate analysis of variance (MAN OVA) was also unable to identify any of these parameters as significant in distinguishing samples from negative controls. Applying the earlier description s of MNP and AuNP binding, it appeared that the antibody ligand on the AuNP were not binding only the targeted Salmonella cells. In most of the trials, the AuNP current responses for the negative controls were stronger than those of the samples that contain ed MNP - Salmonella - Ab - AuNP complexes (Table S7. 2 ). Stronger AuNP current response is indicative of hig her AuNP presence, which is shown in Table S7. 2 for most of the negative controls. In addition, the Wang et. al method applied here for antibody functionalization uses an ionic PBS matrix, which was found post - project to cause AuNP aggregation. This aggreg ation possibly limited Protein A and antibody exposure to the AuNP surface during functionalization. 130 Non - selective binding of any remaining dextrin coating would also increase negative control MNP - Ab - dextrin - AuNP signaling. This resulted in S/N values in T able 7.1 that are less than 1.0. Seemingly the anti - Salmonella polyclonal antibody bound the MNP amino/glycan ligands at a higher concentration than the selective binding occurring on the MNP - extracted Salmonella. In the final trials 9 and 10, non - target E. coli O157:H7 were included to further investigate attachment of the Ab - AuNP. As can be seen in Figure 7.4, the AuNP current responses for both target and non - target MNP - cell - Ab - 7.4 represent those from trial 9 listed in Table 7.1, which were each shown to be significantly different from the negative control (p < 0.05), but neither target sample were significantly different from the non - target E. coli sample. Also, the negative c ontrol, again, shows greater AuNP current response than any of the samples. Similar results were seen for the samples in trial 10 (Table 7.1), with the negative control showing greater AuNP current response, although not at a significant difference from th e target and non - target samples. Clearly the Ab - AuNP were Figure 7.4 Specificity testing using anti - Salmonella - antibody - functionalized AuNP to label MNP - Salmonella (Target Se1 at 5.69 log CFU/mL and Se2 at 4.69 log CFU/mL) and MNP - E. coli (Non - Target Ec at 5.94 log CFU/mL) complexes, as well as negative controls, for DPV detection of the AuNP label. 131 equally binding both Salmonella and E. coli surface markers, and h eavily saturating the MNP - F#1 carbohydrate ligands. Another potential aspect causing the insignificant biosensing separation concerns the optimal non - covalent binding properties of carbohydrate ligands. As discussed in chapter 6, the electrostatic bonds b etween MNP - F#2 and the microbial surface molecular structures are non - covalent , similar to antibody - antigen binding, but at significantly lower specificity and strength. When the MNP - cell complexes are suspended into fresh matrix, fractional levels of the cells release and come to a new equilibrium between the MNP - cell complexes and the fresh matrix, causing a fractional loss of cells to the supernatant. During the multiple rinse steps required in the antibody - based specificity method, a large fraction of t he MNP - captured cells was lost, as indicated in Table 7.1. Comparing the MNP c apture index and p ost Ab - AuNP c apture index data it can be seen that capture indexes drop by an approximate one to two orders of magnitude across trials 1 through 8. Since the Ab - AuNP seemingly saturate the MNP surface, as more cells are lost through rinsing, the final MNP - cell - AuNP complexes will have similar AuNP current responses as the negative control MNP - AuNP responses. Some minor success was accomplished with the later tria ls. The method used in trial 9 was able to retain a sufficient fraction of the cells, with capture indexes over 10 ppt for the target MNP - cell - AuNP complexes. This method minimized all volumes and matrix changes, using only 80 µL of MNP combined with 100 µL of diluted culture spike volume. The MNP - cell complexes were directly resuspended into 200 µL of Ab - AuNP, with the s eparated MNP - cell - Ab - AuNP complexes diluted directly into 1000 µL for spectrometry. Less rinsing was utilized since the realization that the goal of the method was to detect pathogen bacteria, not just stronger bound microbes. DPV tests were run from spect ral samples, where for trial 9 a volume of 500 µL of the diluted sample 132 was separated and directly reacted with HCl and significantly distinguished between Salmonella samples and the negative controls. But this method was still not able to distinguish from non - target E. coli O157:H7. Given the cross - reactivity of the anti - Salmonella antibody ligand with both the MNP ligands and non - target E. coli , the cost, limited shelf - life, special storage, and even superbug concerns inherent with antibody ligands seem exclusionary as a nano - biosensing mechanism. In the drive to design a biocompatible user - friendly biosensing method to detect pathogenic bacteria in our food supply, efforts would be best placed in incorporating emerging carbohydrate - based methods. Althou gh the electrostatic binding forces of the carbohydrate ligands is less than that of antibody ligands, the limitations of these inexpensive ligands can be built into the method to provide robust initial detection methods for broader industry and consumer r apid testing application. 7. 3.3 Dextrin - coated AuNP B iosensing with DPV and S pectrometry As seen in TEM images Figures 5.5A - D, amino/glycan and dextrin carbohydrate ligands attach to the cell surfaces, which enables rapid MNP extraction and increased AuNP labeling of bacteria for en ha nced nano - biosensing electrochemical responses. MNP were shown to extract pathogens with capture indexes of over 3 parts - per - thousand (ppt) in milk but only over 2.5 parts - per - million (ppm) in beef juices and apple cider. AuNP labeling increases signaling, since their dextrin coating binds readily to the microbial surface molecular structures and their approximate 20 µL diameter allows for extensive coverage , especially when excess dextrin solution was removed . The high quantit y of the electrically - active AuNP coverage then afforded stronger 133 electrochemical responses . In addition, c oncentrated dextrin - coated AuNP signaling was linearly related to the DPV current response at 0.4 V and spectral peak absorbance at 520 nm. DPV sign aling of AuNP presence was found to be linearly related to the resulting current response. As identified in the specificity Ab - AuNP research in the previous section, minimized volumes of the MNP - cell - AuNP components actually improved biosensing signals sin ce cell proximity to the nanoparticles was increased . Therefore, standards were prepared using only 20 µL of MNP - F#2, 10 µL of cell culture diluted 10 - 5 in PBS, 10 µL milk matrix, and varying volumes of concentrated AuNP, from 5 µL to 20 µL. The relation b etween AuNP current response at 0.4 V and the volume of AuNP showed a linear relation ( R 2 = 0.96, Fig. 7.5A). These responses were taken from DPV current responses shown in Figure 7.5B. Figure 7.5 DPV current responses for concentrated AuNP responses (A) showing a linear relation between current response at 0.4 V and AuNP volume where (B) AuNP responses are indicated with an arrow. Likewise, concentrated AuNP responses in spectral signaling showed linear responses between absorbance and the volume of AuNP. Figure 7.6A shows sharp peaks for AuNP standards that were concentrated as described above then resuspended into 1000 µL PBS for spectrophotometric scanning. To ensure that the concentrated AuNP do not aggregate when the excess dextrin is removed, concentrated AuNP75 spectral absorbance was compared to an 134 unconcentrated sample, both diluted to 1000 µL in PBS. The concentrated AuNP does not show a red shift to higher wavelength, but shares the same spectral characteristics, except for the excess Figure 7.6 Spectral absorbance of concentrated AuNP (A) showing an extinction peak at approximately 520 nm; (B) AuNP concentration is line arly related to both the peak height and the step - integrated spectral absorbance between 220 nm to 900 nm (n = 3). dextrin absorbance between 200 nm to 250 nm. The peak extinction absorbance of the concentrated AuNP at 520 nm were linearly related to the ir original volume (R 2 = 0.94), but when samples are more complex, the AuNP peak may not be as readily visible, as shown in Figure 7.1B. It was found that when the spectrum were step - integrated between 220 nm to 900 nm, the resulting area was also linearly related to the original AuNP volume (R 2 = 0.99) In spectral signaling for carbohydrate - based MNP - cell - AuNP complexes, the presence of MNP can mask the AuNP peak, as discussed earlier and seen in Figures 7.1B and 7.7A. Optimal concentrated AuNP attachment to extracted MNP - cell complexes were determined over 10 to 30 min. Spectral absorbance of the resulting complexes showed a linear relation between the shifted AuNP peak and attachment time (R 2 = 0.99). The AuNP peak shifted due to the presence of cells and MNP, as well as possible AuNP aggregation that was seen in Figure 5.5B. AuNP extinction peaks were at 548 nm, 535 nm, 542 nm and 532.5 nm for 10, 20, and 30 min for the samples and 135 30 min for the negative control, respectively. As discussed for Figure 7.1 B, spectral step - integration between 220 nm to 900 nm for the samples was also linearly related to the attachment Figure 7.7 Spectral signaling for carbohydrate - based MNP - cell - AuNP complexes showing (A) shifted AuNP peak from 520 nm slightly masked with the MNP peak at approximately 620 nm where (B) both the AuNP peak height and step - integrated spectral area 220 900 nm are linearly related to the time allotted for AuNP attachment. time (R 2 = 0.99). Binding between the dextrin coating on AuNP and amino/glycan ligands of MNP in the negative controls again showed saturation levels in Figure 7.7A. This binding dynamic results in negative controls with higher AuNP signaling than for the sample MNP - cell - AuNP complexes, seen in the 30 - min sample respons e. Both the AuNP peak height and step - integrated spectral area for the negative controls showed greater response at 30 min than for the samples at 30 min (Fig. 7.7B). Although this dynamic can create an inherent detection limit for bacterial presence in li quid food samples, these limitations may be overcome using optimized levels of MNP and AuNP in electrochemical nano - biosen s ing. 136 7.3.4 Electrochemical Q uantification of MNP - cell - AuNP Economical carbohydrate - based nano - biosensors currently show broad select ivity in pathogen detection for total load of bacterial contamination present. As more selective carbohydrate ligands are identified, though, they can easily be applied to a proof - of - concept method described here. This method uses amino/glycan - functionaliz ed magnetic nanoparticles (MNP - F#2) for rapid microbial extraction and concentrated dextrin - coated gold nanoparticles (AuNP) for electrically - active labeling, coupled with electrochemical signaling. Testing was carried out in undiluted vitamin D milk to de monstrate the ability of the biosensor to perform rapid detection of bacterial presence in a complex matrix. In conjunction with another project, non - pathogenic E. coli C3000 was used as a model microbe strain to allow easier transfer of samples between c olleagues without pathogenic contamination concerns. Applying the properties of minimized volumetric sampling discussed earlier for trial 9 in Table 7.1, 50 L of serially diluted c ell culture was simultaneously exposed to 50 L of MNP and 250 L of concentrated AuNP µ L PBS . Negative controls followed the same procedure except without cells. The resulting MNP - cell - AuNP or MNP - AuNP complexes were magnetically sep arated and re - suspended in 100 L PBS for DPV detection or 1000 L PBS for spectral detection. Sample complexes were significantly distinguished from negative controls in both electrochemical DPV ( Fig. 7.8A for 5 log CFU/mL) and spectral signaling ( Fig. 7 .8B for 3 log CFU/mL) (p < 0.2) using this one - step simultaneous capture - n - label process in undiluted milk. As covered earlier, binding dynamics showed more seemingly selective AuNP attachment to the cell surface as seen in Figure 5.5B while saturating the MNP surface (Fig . 5.6 B) . Meanwhile, c ell attachment to the MNP limits the AuNP exposure to the MNP surface. Th ese binding pattern s 137 result ed in higher signaling for negative controls (MNP - AuNP) than for samples - e5Ec - - e3 Ec - & B, respectively. Figure 7.8 Electrochemical nano - biosensing for simultaneous carbohydrate - based MNP - E. coli - AuNP rapid extraction from milk and electrically - active labeling followed by (A) DPV or (B) spectral analysis of samples and E. coli E. coli . For bo th signaling methods, DPV and spectral, current responses and spectral absorbances were inversely related to cell concentration, due to the selective binding of AuNP to cell surface markers. Even though spectral signaling was able to detect at 3 log CFU/mL , samples that contained 5 log CFU/mL E. coli showed similar step - integrated spectral absorbance (Fig. 7.8B). This was due to a high response of one replicate, which also caused insignificant separation from negative controls (p = 0.28). Meanwhile samples that contained 3 log CFU/mL E. coli in DPV sampling showed insignificant separation (p = 0.37) from the negative control. DPV biosensing was less sensitive for the current method parameters most likely due to insufficient HCl to react 138 with the AuNP present in the negative controls, reducing separation (Fig. 7.8B). Separation between the two DPV samples at the different concentrations was significant though (p < 0.2). Reproducibility within the sample replicates is lower (greater standard deviation) than for negative controls in both detection methods, as can be seen in Figures 7.8A & B. In addition, the current sampling method resulted in minimal separation in current response or spectral absorbance between samples and negative control responses (Figs. 7.9A & B). Only 0.4 E - 6 amps separated DPV samples at 5 log CFU/mL from the negative controls for the AuNP current response. Spectral scans between both samples and the negative control were separated by an absorbance of 0.05 and step - integrated spectral absorb ance of approximately 40. This aspect would need to be addressed in future method optimization to both improve detection sensitivity and increase significance in sample and negative control separation. Particularly for samples, though, cell sizes are much greater than nanoparticle sizes, showing the robust MNP extraction properties. But in the current minimized volume method, as pathogenic bacteria cell concentrations decrease, too few cells may Figure 7.9 Graphs for electrochemical nano - biosensing for simultaneous carbohydrate - based MNP - E. coli - AuNP rapid extraction and electrically - active labeling followed by (A) DPV or (B) spectral analysis of samples and negative controls for microbial detection in milk (n=3). Arrows identify (A) 0.4 V and (B) 520 nm for the AuNP response E. coli E. coli . 139 be extracted to prevent saturation binding of AuNP to the exposed MNP surface to distinguish from negative controls. The spectral method here may be more sensitive than DPV, detecting 3 log CFU/mL, since spectral scans do not require activation of the AuNP with HCl, but detect total sample contents. DPV biosensing was not able to distinguish between 3 log CFU/mL and zero cells most likely due to the expense of the MNP and milk component reactions with HCl, reducing the overall AuNP activation. In addition, the amount of bound AuNP onto MNP in negative controls may exceed the reactive character of HCl f or complete activation, arbitrarily reducing the negative control response itself. Future work w ould fo cus on optimizing the assay to improve sensitivity of the tests in microbial detection focusing initially on improving negative control signaling separat ion from sample signaling . During early development work in PBS, the AuNP peak was clearly visible in spectrometry (Fig. 7.7A) and DPV detection (Fig. 7.5B). But in milk, the MNP - cell - AuNP peaks are masked since milk components also attached to both MNP an d AuNP (Figures 7.9A & B) . Optimization work would also include determining the volume of HCl required to react with all components present in the MNP - cell - AuNP complex. As more selective carbohydrate ligands are identified, such as described earlier in th e literature review, greater specificity will also improve limits of detection for targeted pathogens. Finally, identifying common complex matrix component effects upon sampling would allow faster application to new liquid or homogenized foods. This rapid, economical MNP - cell - AuNP biosensing method c ould then be easily applied to other liquid foods which would make our food s upply more secure through affordable rapid detection . 140 7.4 Conclusions Combining carbohydrate - functionalized MNP cell capture , MNP - cell concentration, and concentrated dextin - coated AuNP labeling resulted in promising nano - biosensing development towards the goal of feasible, less complicated tests. Much work in nano - biosensing developm ent has focused heavily on designing highly specific, one - time - substantial technology to produce. These are not cost - effective for many worldwide food suppliers, though. Even many of similarly suspended MNP/EANP biosensing assays that utilized antibodies or DNA ligands are still out of reach for many users, requiring special storage or highly technical detection equipment. Carbohydrate ligands are easily incorporated onto nanoparticles, and require little special handling, showing a long shelf life at room temperature. As more selective carbohydrate ligands are identified and incorporated into MNP - cell - AuNP assays, consumers can be assured of safer food supplies free of targeted pathogens at affordable costs. While the pro of - of - concept test presented here requires optimization for detection at lower cell concentrations, it was successful in detecting E. coli presence directly from milk within 45 min , whereas many reported tests reviewed earlier were studied only in simple b uffer solution , or required extensive sample handling . The future in nano - biosensing is not in developing more technically complex sensors, but applying technological know - how to develop inexpensive, user - friendly microbial as says. 141 Chapter 8. Conclusions Frequent and reliable detection of foodborne pathogens would provide both consumer food security and supplier economic safety. As cautioned earlier though, biosensing must FIRST address the broad goals of any biosensor: Fiel d operability, Inexpensive, with Real - time detection that is both Sensitive and Specific to target, while being as Trouble - free as possible (FIRST) . Throughout development of these biosensing assays focus was placed on economical solutions to microbial ass ay s that minimize sample handling and user - required technology while making use of stable carbohydrate - functionalized nanoparticles to provide broad - based detection . MNP - strip and MNP - cell - AuNP biosensing methods developed here use versatile magnetic nanop article (MNP) extraction, electrically active nanoparticles (EANP) labeling, and carbohydrate - based ligand chemistry. Nano - biosensing can be an economical method to both detect and quantify the presence of pathogenic bacteria, when developed with the mindset tha t s ometimes the beauty of a test is not in the amount of technology that is incorporated into the biosensing method or biosensor design that only one person can interpret, but that this technology is used to design a simple biosensing method and result that can be understood by all. Inexpensive food carbohydrate functional groups can reliably attach to pathogenic bacteria, and when immobilized onto MNP, easily extract, and concentrate, them from liquid/liquified food samples using a common ma gnet. Further, MNP - strips functionalized with carbohydrate ligands provide a viable option when faced with homogenized food products that are too thick and resist magnetic pull during separation . The amino/glycan ligand used in this study preferentially bo und cells over complex matrix components. Gold nanoparticles coated with dextrin, another carbohydrate, also attach to captured bacteria and provide an inexpensive, quick method to substantially increase detection signal 142 strength , whether by electrochemica l or spectrophotometric means. Nano - biosensing using these three options: MNP - cell, MNP - strip - cell or MNP - cell - AuNP, was then able to quickly distinguish between samples with cells against negative controls without cells present. Electrochemical detection , although complicated in theory, can easily be reduced to positive and negative results. CV and DPV detection require a simple hand - held potentiostat unit that utilizes cheap, disposable printed chips to quantify the captured bacterial response, with or w ithout the presence of AuNP. Spectrophotometers are more commonly available in even smaller research and food industry testing laboratories. AuNP presence provided enhanced signaling in electrochemical DPV detection as well as spectrophotometric methods. M NP - strip with CV could detect E. coli in milk at 8.5 log CFU/mL (S/N = 1.08, p < 0.05), whereas MNP - cell - AuNP with DPV could detect 5 log CFU/mL (S/N = 0.79, p < 0.20) or with spectrophotometry could detect 3 log CFU/mL (S/N = 0.88, p < 0.20). A nano - biose nsor that is not designed as a self - contained unit for a single test, but takes advantage of reusable technology, along with taking only minutes to prepare, would greatly assist countries in ensuring safe food supplies. Even though TEM imaging of MNP - cell complexes could identify differences in attachment for MNP - F#1 and MNP - F#2 to the cell surface, both MNP showed no significant difference in suspended capture in either simple or complex matrices. As development continued though, it was necessary to ident ify one preferred MNP to minimize expended resources. During capture in homogenized egg, it was apparent that suspended MNP capture had its limits, as the MNP - Salmonella complexes were unable to migrate to the magnet for separation. MNP - strips were prepare d using both MNP - F#1 and MNP - F#2 and compared for their ability to capture bacteria. MNP - F#2 - strips outperformed all other types (non - functionalized and amino or glycan 143 alone without MNP), specifically capturing over two times the cells as MNP - F#1 - strips. Therefore, research was carried forward focusing only on MNP - F#2. Just as homogenized egg prevented suspended MNP - cell migration during magnetic separation, complex matrix components were found to effect CV responses. The liquid food matrices evaluated we re milk, beef juices, apple cider and homogenized egg, with varying amounts of carbohydrates, fats, proteins and sodium. Although carbohydrates showed no significant effect upon CV signaling, sodium and fats had a negative effect, while proteins had a posi tive effect. Most of all, complexity showed the greatest effect, where PBS and beef juices, containing minimal salts and proteins, respectively, resulted in negative control signals higher than corresponding samples. These matrices were mostly composed of water , with beef juices having minimal proteins to bind and PBS containing minimal phosphates, sodium - and potassium - chloride to interfere with signaling . Whereas, milk, apple cider, and of course homogenized eggs, all rich in sodium among other components , caused negative control signaling that was lower than corresponding samples since attached matrix components expended significantly more of the chemical mediator . Electrically active AuNP were applied as a means to intensify electrochemical signaling. A nti - Salmonella antibody were immobilized onto d - AuNP (Ab - AuNP) in attempts to use MNP - cell - Ab - AuNP complexes as a means to specificity of MNP capture of Salmonella microbes in a complex milk matrix. DPV results, though, were unreliable across a range of pa rameters. Princip al component analysis (PCA) of the data across ten trials showed that there was a closer relationship between samples and negative controls from a particular trial, than even samples of different trials prepared at the same microbial conce ntration. Eventually, it was determined that several parameters created the unreliable performance. Although the electrostatic bonds between the carbohydrate ligands and cell surface molecules are similar to antibody binding, the current 144 amino/glycan carbohydrate ligands bind at weaker strength. A fraction of these bonds also break - cell complexes and cells suspended in the matrix, reducing cell presence with each ri nse. Additionally, the antibody was found to equally bind both target Salmonella and non - target E. coli cells, as well as saturating the MNP carbohydrate ligands for negative controls . Although carbohydrate ligands will show less selectivity and lower b i n ding strength than antibody ligands, nano - biosensing using carbohydrate - based nanoparticles such as with MNP - F#2 - E. coli - AuNP showed the ability to detect microbial presence with an economical method . Having identified capture concern s due weak electrostat ic binding allowing cell loss with rinsing steps in early experimental work , the MNP capture and AuNP labeling steps were thereafter carried out simultaneously , or as a - reduce cell loss to the s urrounding matrix and increase cell - nanoparticle vicinity . Additionally, excess dextrin coating was removed from the AuNP, concentrating them while allowing enhanced labeling of samples. Resulting DPV and spectrometer signaling successfully allowed detecti on at lower microbial concentrations than MNP - cell signaling with CV. S pecifically addressing the hypotheses of this research, hypothesis one dealt with testing that m agnetic nanoparticles (MNP) coated with carbohydrates , without the presence of antibodies , will still rapidly bind, extract , and concentrate microbial pathogens from complex liquid food matrices to reduce the rate of false negatives without initial sample preparation whether as suspended MNP or immobilized onto plastic strips. MNP functionaliz ed with glycan (MNP - F#1) or amino/glycan (MNP - F#2) ligands were able to bind pathogenic bacteria Salmonella , E. coli , and Bacillus in both simple PBS and complex milk matrices and concentrate them into 1 mL ( for larger volume capture, LVC) and 500 uL ( for small volume capture, SVC) volumes. LVC was 145 used to simulate BAM methods requiring sample sizes of 25 gm, which was approximated w ith 25 mL of liquid matrix. In LVC, both MNP were able to capture over 3 ppt and up to 120 ppt of the bacteria present, while even capturing up to 600 ppt in SVC with single or multiple bacteria present . In more acidic matrices such as beef juices or appl e cider, capture was approximately 10 ppm . Immobilized MNP - F#2 onto plastic strips (MNP - strip ) captured bacteria in PBS and reliably report ed their concentration from 3 to 5 log CFU/mL using a 4 - hr liquid media growth and plating detection method. Second, it was hypothesized that electrochemical instrumentation will simplify testing methods for accelerated pathogen detection of MNP - extracted microbial pathogens from complex food matrices despite the presence of food components. Electrochemical cyclic volta mmetry ( CV ) detection takes only 30 min from assay start and eliminates the time required to grow the captured bacteria on agar media for suspended MNP capture or in liquid media for MNP - strip capture. Suspended MNP capture showed significant separation be tween samples and negative control s for E. coli capture in beef juices (p < 0.05), and capture of Salmonella in milk and Listeria in apple cider (p < 0.15). CV also similarly expedited MNP - strip detection of Salmonella presence in PBS, showing a linear res ponse between 2 to 6 log CFU/mL ( R 2 = 0.71) with a limit of detection for true positive response (S/N < 1.0) at 3.6 log CFU/mL (S/N = 0.92). MNP - strip capture in complex matrices clearly showed the effect of matrix composition on the resulting CV signaling . This was due to the binding of matrix components to the MNP - strip ligands, with simple matrices PBS or beef juices resulting in S/N < 1, while milk, apple cider or homogenized egg components caused S/N > 1. CV biosensing was able to distinguish between sample capture and negative control responses for capture of Salmonella in PBS , E. coli in beef juices , Salmonella in homogenized egg, Listeria in apple cider, and E. coli in vitamin D milk (p < 0.10). These re sults in the initial 146 MNP - strip design were for microbial concentrations between 6.19 to 12.3 log CFU/mL, all above industry testing standards , but show good progress in developing a rapid, accurate microbial assay requiring minimal sample preparation and l ittle liquid handling. Additional advance in the assay design was made in identifying the effect that the sodium, fats and proteins present in the liquid matrices had upon the normalized peak current response (NPCR). Finally, it was shown that concentrated, d extrin - c oated gold nanoparticles , without the presence of antibodies , will rapidly label MNP - extracted microbial pathogens to reduce false positives by better separation from negative controls even in the presence of complex liquid food matrices through enhanced electrochemical instrumental . This biosensing method was most successful when carried out in smaller volumes and in a simultaneous MNP - capture with AuNP - label procedure. Dextrin - coated AuNP labeling was foun d to saturate the surface of MNP and selectively bind the microbial surfaces more selectively. This resulted in electrochemical signaling that showed higher signaling for negative controls (MNP/AuNP) than samples (MNP - cell - AuNP) which creates a limit to si gnificantly detecting a microbial load when too few are present. The current design was able to detect E. coli in vitamin D milk at 3 log CFU/mL using spectral signaling, but only 5 log CFU/mL using differential pulse voltammetry (DPV) (p < 0.20). Although these limits meet testing needs within the milk industry in providing a rapid assay without sample preparation, improving this assay to detect at 1 log CFU/mL, or even 1 CFU/mL would require optimizing the amount of MNP and AuNP present. Nano - biosensing microbial contamination in liquid foods using MNP - strip - cell/CV or MNP - cell - AuNP/DPV /spectral methods can provide the food industry with economical, rapid assays and under/developing countries with simple, affordable assays to minimize microbial foodborne outbreak events. A cost - benefit analysis, comparing assay cost with the technology 147 required for detect ion shown in Fig. 8.1 identifies the niche that these methods fill. Although the current carbohydrate ligands target broad - based microbial presence, they have proven their ability to detect microbes without the necessity of first removing the food componen ts or the need to increase the microbial load, reducing both testing time and supplies cost. They are also user - friendly, requiring little training to apply, while using more common spectral instrumentation or inexpensive potentiostat technology. MNP - strip /CV detection allows minimal sample handling when suspected bacterial load s are high er than 6 log CFU/mL . Significant s eparation between the negative control and sample NPCR currently requires a higher bacterial load, but entire assay time from extraction to detection is under 30 min and allows a yes/no response. If minimal sample handling is still desired for suspected lower microbial load between 3 to 5 log CFU/mL , the cells captured on the MNP - strip may be grown in liquid media for 4 h r , recaptured with a fresh MNP - strip and then tested for microbial presence with CV. More sensitive direct detection is possible using the one - step MNP/AuNP capture and labeling method, currently for loads between 3 to 5 log CFU/mL, using samples even under 500 L , with dete ction possible by either handheld potentiostat for DPV or spectrometry. DPV detection does allow for immediate testing in the field using a simple magnet to separate the prepared sample from the matrix and a handheld potentiostat . W hereas when bulk sample processing is preferred, numerous samples may be simultaneously handled, and spectral detection completed in minimal time, such as 30 samples tested in 1 hr. Food safety for people in both the US and worldwide has repeatedly proven elusive as we experience multiple microbial foodborne outbreaks each year that threaten the safety of our food supplies . In the US, the CDC monitors foodborne outbreaks and noted that between 1996 and 2010 microbial outbreaks did decrease but still reported over 19,000 infections and 71 deaths in 2014 from the ten top microbial pathogens [278] . W orl dwide, the WHO reports on the status of the 148 human population, including the occurrence of over 400,000 deaths in 2010 from microbial, viral and protozoan sources [1] . The United Nations officials assert that food safety is entwined with food security and that ensuring the securing safe food supplies will also improve human economic growth [279] through reduced medical costs due to disease and increased work hours for a healthy populace . Frequent testing along the food supply chain, from source to store, will more readily identify contaminated product before it reaches th e consumer, increasing consumer confidence that they are purchasing a safe food product . The affordable, user - friendly nano - biosensor assays developed in this research will advance this goal of a more secure food supply since it combines emerging robust ca rbohydrate - ligands with user - friendly nanoparticle technology and rapid electrochemical detection , as summarized earlier. Simply d etermining whether a food product is free from a microbial load, or below a specific tolerance level, using these broad - based detection Figure 8.1 Cost - benefit analysis of microbial contamination detection in the food supply chain including (nano - ) biosensing methods. (see text for acronym definitions) 149 assays will allow this product to continue along the food supply chain. Fu ture improvements in these MNP - based nano - biosensing methods will require focus on the FIRST considerations identified earlier . Of utmost importance is improving the level of capture experienced by the MNP above the current 100 ppt and 10 ppm levels. This will in turn improve the level of detection and reliability of the test , which must be addressed in the design of final assay for pathogen ic bacteria in a liquid food sample. False negative results can be reduced by minimizing the quantity of MNP required to capture microbial loads below 5 log CFU/mL, thereby eliminating the masking effect of excessive MNP in electrochemical testing. I denti fication of more effective carbohydrate ligands which will both bind more readily and securely to the targeted pathogens will further increase secure capture and allow additional reduction in MNP . The search for tighter - binding, yet still economically - feas ible, carbohydrate ligands can be achieved through exclusion assays comparing novel carbohydrate - functionalized MNP extraction levels in selected complex matrices for statistically significant improvement. Broad - based capture is still desirable since these assays, as a n initial test for possible microbial presence, keep testing costs lower in comparison to selective antibodies, while improving detection at lower levels of food contamination. For food suppliers that handle both small and large quantities of perishable liquid product, these simpler nano - biosensing methods also allow high throughput product movement since the assay will provide a response within 30 min . Reliability and reproducibility go hand - in - hand in assay design, but first reliability must be addressed in future MNP/AuNP nano - biosensing improvements. Current assay reliability has only been tested at microbial levels that targeted 3 log CFU/mL concentrations in PBS with respect to diluting log - growth culture at approximately 8 log CFU/mL by 10 - 5 . Within complex matrices rich in nutrients, though, the actual level of bacterial presence during assay conditions was between 3 150 to 5 log CFU/mL for milk , and as high as 6 log CFU/mL in beef juices and apple cider, due to accelerated bacterial growth . In these early stages of MNP/AUNP assay development, reliable capture across a broad range of microbial concentrations was not investigated , therefore reliability at different concentrations will need to be established . In addition, MNP capture showed min imal reproducibility, with only 2 out of 3 replicates for LVC within one standard deviation of the mean. Both accuracy and precision of the assays would need to be improved beyond this first design stage, again through increased extraction capability at lo wer microbial contamination levels more pertinent to the food industry and optimization of the MNP quantity to sample volume ratio per matrix type. MNP capture and AuNP labeling of bacteria takes advantage of the drive of pathogenic microbes to survive. Mi crobial cell surfaces provide numerous molecular structures for their attachment to other cells or surfaces so that they may extract nutrients for survival from the surrounding environment. As more selective carbohydrates are identified that attach to targ eted bacteria, their application as ligands for MNP - cell - AuNP nano - biosensing will substantially increase selectivity and specificity of the assay . Optimization of the significant parameters must be considered for particular matrices, especially the comple xity of the matrix components and their effect upon voltammetric signaling. In addition, capture volume, MNP and AuNP quantities, and HCl volume will require optimization to further improve both reliability and sensitivity of this biosensing method. Used i n conjunction with optical or electrical detection of gold nanoparticles (AuNP), carbohydrate - functionalized MNP - cell - AuNP nano - biosensing advances the goal of being the FIRST biosensor of choice in detecting microbial pathogens throughout our food supply chain. 151 APPENDIX 152 Supplementary Data Figure S 8.2 Example of adjusted DPV current response for MNP - F#2 - cell - Ab - AuNP nano - biosensing, showing corrected baseline with respect to average current response between 0.9 V to 0.6 V; identifying maximum (average) AuNP response potential (arrow, 0.38 here); and po rtion of a sample signal for clarity during visual data analysis. 153 Table S7. 2 Specificity of Salmonella detection using antibody - functionalized AuNP. Trial Capture Matrix Pathogen * n Samples Cells Available for Capture (log CFU/mL) MNP Capture Index +/ - St. Dev. (ppt) Post Ab - AuNP Capture Index (ppt) @ Sample AuNP Peak Ht. Ave. (amps) Sample AuNP Peak Ht. St. Dev. (amps) Negative AuNP Peak Ht. Ave. (amps) Negative AuNP Peak Ht. St. Dev. (amps) t - test + p - value AuNP Peak S/N Ave. 1 PBS Se Target 3 3.12 170 +/ - 28 40 1.17E - 05 8.47E - 13 1.35E - 05 3.12E - 12 0.23 0.87 2 PBS Se Target 6 2.76 520 +/ - 214 76 1.05E - 05 7.04E - 14 1.26E - 05 2.42E - 12 0.04 0.83 3 Milk Se Target 6 3.40 763 +/ - 285 21 6.32E - 06 1.23E - 12 6.57E - 06 3.76E - 13 0.79 0.96 4 Milk Se Target 6 3.30 935 +/ - 159 346 5.87E - 06 2.79E - 12 5.92E - 06 2.65E - 12 0.97 0.99 5 Milk Se Target 6 5.14 14.4 +/ - 3 0.9 4.76E - 06 7.44E - 13 4.74E - 06 5.14E - 13 0.98 1.00 6 Milk Se Target 3 9.08 1.48 +/ - 0.21 0.01 3.83E - 06 1.89E - 12 6.36E - 06 1.51E - 12 0.08 0.60 Se Target 3 8.08 0.62 +/ - 0.09 0.01 2.88E - 06 2.71E - 12 6.36E - 06 1.51E - 12 0.04 0.45 Se Target 3 7.08 1.45 +/ - 0.22 0.03 5.72E - 06 6.76E - 12 6.36E - 06 1.51E - 12 0.72 0.90 7 Milk Se Target 3 6.95 0.23 +/ - 0.13 0.00 7.57E - 06 2.63E - 12 8.81E - 06 8.23E - 13 0.31 0.86 Se Target 3 5.95 0.99 +/ - 0.22 0.04 9.86E - 06 2.04E - 14 8.81E - 06 8.23E - 13 0.12 1.12 Se Target 3 4.95 1.33 +/ - 0.71 0.00 7.24E - 06 4.56E - 13 8.81E - 06 8.23E - 13 0.07 0.82 8 Milk Se Target 3 5.57 7640 +/ - 1282 135 5.43E - 06 1.41E - 13 4.12E - 06 1.98E - 12 0.19 1.32 Se Target 3 4.57 8108 +/ - 8108 622 7.09E - 06 3.46E - 12 4.12E - 06 1.98E - 12 0.09 1.72 Se Target 3 3.57 8108 +/ - 8108 751 3.99E - 06 1.49E - 13 4.12E - 06 1.98E - 12 0.88 0.97 9 PBS Se Target 3 5.69 NA 12 4.78E - 06 7.75E - 14 6.50E - 06 3.76E - 13 0.01 0.74 Se Target 3 4.69 NA 14 4.58E - 06 3.68E - 13 6.50E - 06 3.76E - 13 0.02 0.70 Ec Non - target 3 5.94 NA 1 5.12E - 06 1.58E - 12 6.50E - 06 3.76E - 13 0.16 0.79 10 PBS Se Target 3 4.90 NA 103 3.32E - 06 3.52E - 13 3.69E - 06 5.49E - 13 0.54 0.90 Ec Non - target 3 5.63 NA 0.08 3.56E - 06 1.41E - 14 3.69E - 06 5.49E - 13 0.79 0.97 * Se = Salmonella Enteritidis, Ec = E. coli O157:H7; + reported two - tail p - values for t - test ran for unpaired samples and negative controls within same trial test, assuming equal variances; # S/N = AuNP current response ratio for Samples to Negative controls. 154 BIBLIOGRAPHY 155 BIBLIOGRAPHY [1] Food Safety , 2015. [Online]. Available: http://www.who.int/foodsafety/areas_work/foodborne - diseases/ferg/en/. [2] - RF Sensor: An innovative application of RF technology for sensing pathogens electrically in liquids (SPEL) within Biosens. Bioelectron. , p. 12, 2018. [3] L. L. M atta and E. C. A locilja Centrifuge - J Food Prot , vol. 00, no. 00, p . 10, 2018. [4] - functionalized Nano - Biosensor for Rapid Extraction of Pathogenic Bacteria Directly from Complex Liquids with Ieee Trans. Nanote chnol. , vol. Special Issue 2018, p. 9, 2018. [5] Nanoparticle - IEEE Sens. J. , vol. 15, no. 8, pp. 4692 4699, Au g. 2015. [6] - labeled biosensor for rapid and sensitive J. Biol. Eng. , vol. 9, no. 16, pp. 1 7, Dec. 2015. [7] Biosensors for Medical Diagnostic Biosensors for Medical Diagnostics, Michigan State University, 12 - Jan - 2015. [8] U.S. Department of Health and Human Services , 2003. [Online]. Available: https://www.fda.gov/Food/FoodScienceResearch/LaboratoryMethods/ucm2006949.htm. [Accessed: 18 - Apr - 2017]. [9] M. Adams, M. O. Moss, and P. McClure, Food Microbiology . 2015. [10] Anal. Chim. Acta , vol. 599, no. 1, pp. 7 15, Sep. 2007. [11] - Nano Today , vol. 4, no. 4, pp. 318 334, Aug. 2009. 156 [12] J. W. - F. Law, N. - S. Ab Mutalib, K. - G. Chan, and L. - Front. Microbiol. , vol. 5, Jan. 2015. [13] M. Pumera, S. Sánchez, I. Ichin Sens. Actuators B Chem. , vol. 123, no. 2, pp. 1195 1205, May 2007. [14] Biosensors , vol. 6, no. 1, p . 5, Feb. 2016. [15] of a highly sensitive and selective electrochemical sensor based on chitosan - coated Fe3O4 magnetic nanoparticle for determination of antibiotic c iprofloxacin and its application in Can. J. Chem. , vol. 94, no. 10, pp. 803 811, Oct. 2016. [16] sensitive and selective sensor for morphine using chi tosan coated Fe3O4 magnetic Mater. Sci. Eng. C , vol. 58, pp. 53 59, Jan. 2016. [17] A. M. Fernandes et al. on multiwalled carbon nanotubes - chitosan - bismuth and lead sulfide nanoparticles for the Biosens. Bioelectron. , vol. 63, pp. 399 406, Jan. 2015. [18] Bitesize Bio , 09 - Sep - 2014. [19] Appl. Environ. Microbiol. , vol. 66, no. 3, pp. 1228 1232, Mar. 2000. [20] A. V. Hadjinicolaou, V. L. Demetriou, J. Hezka, W. Beyer, T. L. Hadfield, and L. G. - allelic real - time PCR for detection of and J. Microbiol. Methods , vol. 78, no. 1, pp. 45 53, Jul. 2009. [21] immunosensor based on double - Sens. Actuators B - Chem. , vol. 177, pp. 1010 1016, Feb. 2013. [22] Z. Li, Y. Li, X. - F. Qian, J. Yin, and Z. - Appl. Surf. Sci. , vol. 250, no. 1 4, pp. 109 116, Aug. 2005. [23] Y. Sun et al. - modified plasti cs by UV Anal. Bioanal. Chem. , vol. 402, no. 2, pp. 741 748, Oct. 2011. 157 [24] diaminopimelic J. Biol. Chem. , vol. 247, no. 23, pp. 7684 7694, 1972. [25] Int. J. Agric. Biol. Eng. , vol. 6, no. 1, pp. 28 35, Mar . 2013. [26] K. C. de Souza, G. F. Andrade, I. Vasconcelos, I. M. de Oliveira Viana, C. Fernandes, and - phase extraction based on mesoporous silica - coated Mater. Sci. Eng. C , vol. 40, pp. 275 280, Jul. 2014. [27] E. Lesman - Movshovich and N. Gilboa - - IIL as a J. Dairy Sci. , vol. 86, no. 7, pp. 2276 2282, 2003. [28] Mol. Microbiol. , vol. 30, no. 2, pp. 285 293, 1998. [29] - biosensin g with suspended MNP microbial Biosens. Bioelectron. , vol. 117, pp. 781 793, Oct. 2018. [30] Acta Naturae , vol. 3, no. 2, pp. 34 55, 2011. [31] gold - Rsc Adv. , vol. 6, no. 54, pp. 48445 48448, 2016. [32] Y. - L. Kuo et al. - based antibacterial agents for nosocomial and antibiotic - Nanomed. , vol. 11, no. 19, pp. 2497 2510, Oct. 2016. [33] Electroanalysis , vol. 25, no. 4, pp. 815 831, Apr. 2013. [34] N. Sattarahmad Biochem. Eng. J. , vol. 97, pp. 1 7, May 2015. [35] rimetric biosensing of pathogens Biotechnol. Adv. , vol. 33, no. 6, Part 1, pp. 666 680, Nov. 2015. [36] - Time 012. 158 [37] J. Dairy Sci. , vol. 94, no. 7, pp. 3279 3287, Jul. 2011. [38] Eur - Food Safety - European Food Safety , 2017. [Online]. Available: /food/safety/general_food_law/principles_en. [39] ic - Guidance for Industry: Dairy Farms, Bulk Milk Transporters, Bulk Milk Transfer Stations Available: http://www.fda.gov/Food/GuidanceRegulation/GuidanceDocumentsRegu latoryInformation/ ucm083049.htm. [Accessed: 29 - Oct - 2015]. [40] World Health Organization and Foodborne Disease Burden Epidemiology Reference Group, WHO estimates of the global burden of foodborne diseases . 2015. [41] - Food Safety - Food Safety , 2017. [Online]. Available: /food/safety/general_food_law_en. [42] foodmanufacture.co.uk , 21 - Nov - 201 7. [Online]. Available: https://www.foodmanufacture.co.uk/Article/2017/11/21/Food - supply - chain - guide - to - protect - businesses - from - attack. [Accessed: 31 - Jul - 2018]. [43] http://www.usda.gov/wp s/portal/usda/usdahome?navid=food - safety. [Accessed: 23 - May - 2016]. [44] US Food & Drug Administration: FSMA Final Rule on Produce Safety , 26 - Jan - 2016. [On line]. Available: https://www.fda.gov/food/guidanceregulation/fsma/ucm334114.htm. [Accessed: 26 - Oct - 2018]. [45] BMC Immunol. , vol. 16, no. 1, p. 13, 2015. [46] - Jul - 2018. [Online]. Available: https://www.fda.gov/food/recallsoutbreaksemergencies/outbreaks/default.htm. [Accessed: 31 - Jul - 2018]. 159 [47] itical Control Point (HAACP); Procedures Fed. Regist. , vol. 66, no. 13, Jan. 2001. [48] REGULATIONS VERSUS PRACTICE - http://www.nmconline.org/articles/qualityeuro.pdf. [Accessed: 23 - May - 2016]. [49] Food Safety News , 19 - May - 2017. [Online]. Available: http://www.foodsafetynews.com/2017/05/fda - asked - to - allow - interstate - sales - of - raw - milk - with - warnings/. [Accessed: 30 - Jun - 2017]. [50] Philos. Trans. R. Soc. B Biol. Sci. , vol. 363, no. 1492, pp. 849 861, Feb. 2008. [51] - Apr - 2008. [52] V. Deperrois - for the identification of spoilage and pathogenic bact eria species in milk and milk products: Lett. Appl. Microbiol. , vol. 55, no. 2, pp. 99 108, Aug. 2012. [53] http://www.fda.g ov/Food/FoodScienceResearch/LaboratoryMethods/ucm2006949.htm. [Accessed: 29 - Oct - 2015]. [54] Am. J. Food Technol. , vol. 6, no. 2, pp. 87 102, Feb. 2011. [55] R. S. Bejhed, T. Z. G. De la Torre, M. Donolato, M. F. Hansen, P. Svedlindh, and M. - on optomagnetic bacterial DNA sequence detection using volume - Biosens. Bioelectron. , vol. 66, pp. 405 411, Apr. 2015. [56] A. M. Fernandes et al. on multiwalled carbon nanotubes - chitosan - bismuth and lead sulfide nanoparticles for the Biosens. Bioelectron. , vol. 63, pp. 399 40 6, Jan. 2015. [57] - Assembled Monolayers for Rapid Detection of Escherichia coli O157:H7 with Signal Sensors , vol. 15, no. 8, pp. 19212 192 24, Aug. 2015. [58] food - borne pathogenic bacteria using multicolor quantum dot probes based on multiplex 160 LWT - Food Sci. Techno l. , vol. 61, no. 2, pp. 368 376, May 2015. [59] A. Petrosova et al. Sens. Actuators B Chem. , vol. 122, no. 2, pp. 578 586, Mar. 2007. [60] - free somatic cell Cytometry A , vol. 81A, no. 12, pp. 1076 1083, Dec. 2012. [61] cations, Future Trends and Standardization and Quality Assurance in Fluorescence Measurements II , vol. 6, U. Resch - Genger, Ed. Berlin, Heidelberg: Springer Berlin Heidelberg, 2008, pp. 307 342. [62] E. Lesman - Movshovich and N. Gilboa - Garb - IIL as a J. Dairy Sci. , vol. 86, no. 7, pp. 2276 2282, 2003. [63] Nanoparticle - ba sed Sensor for the Detection of Dengue Virus - Int. J. Sci. Basic Appl. Res. IJSBAR , vol. 26, no. 3, pp. 138 157, Apr. 2016. [64] electrocatalysis of heme prot eins immobilised in carbon - coated nickel magnetic nanoparticle - chitosan - dimethylformamide composite films in room - Bioelectrochemistry , vol. 94, pp. 94 99, Dec. 2013. [65] C. H. Gayathri, P. Mayuri, K. Sankaran, and A. S. Kumar, immunosensor for efficient detection of uropathogenic E. coli based on thionine dye immobilized chitosan/functionalized - Biosens. Bioelectron. , vol. 82, pp. 71 77, Aug. 2016. [66] H. Liu, F. Zhan, F. Liu, M. viable pathogenic bacteria by sensing of RNA markers in gold nanoparticles based paper Biosens. Bioelectron. , vol. 62, pp. 38 46, Dec. 2014. [67] E. B. Setterington and E. C. Alocilj Biosensors , vol. 2, no. 4, pp. 15 31, Jan. 2012. [68] - PCR Amplified S. enteritidis Genomic DNA from Food Matrices Using a Gold - Nanoparticle DNA Biosensor: A Proof - of - Sensors , vol. 12, no. 12, pp. 10487 10499, Aug. 2012. 161 [69] I detection of E. coli in spinach using the strong affinity of derivatized mannose with fimbrial Biosens. Bioelectron. , vol. 61, pp. 266 273, Nov. 2014. [70] M. S. Cheng, S. H. Lau, V. T. Chow, and C. - - Based Electrochemical Environ. Sci. Technol. , vol. 45, no. 15, pp. 6453 6459, Aug. 2011. [71] M. Kimura, H. Fukushima, Y. Potentiostat With an Electrochemical Cell Using Thin - IEEE Trans. Electron Devices , vol. 56, no. 9, pp. 2114 2119, Sep. 2009. [72] M. S. Webster et al. abolites for the discrimination of bacteria Sens. Actuators B - Chem. , vol. 220, pp. 895 902, Dec. 2015. [73] - free NIR - SERS discrimination and d J. Nanobiotechnology , vol. 13, p. 45, Jun. 2015. [74] L. Malic et al. - based microfluidic chip for rapid and efficient immunomagnetic capture and release of Listeria monocytogen Lab Chip , vol. 15, no. 20, pp. 3994 4007, 2015. [75] X. - Z. Shi and X. - - terminal carbohydrate - recognition domain of Manduca sexta immulectin - Amino Acids , vol. 42, no. 6, pp. 2383 2391, Jun. 2012. [76] N. Duan et al. - co lor upconversion fluorescence and aptamer - functionalized magnetic nanoparticles - based bioassay for the simultaneous detection of Salmonella Typhimurium and Anal. Chim. Acta , vol. 723, pp. 1 6, Apr. 2012. [77] D. Wang et al. ient separation and quantitative detection of Listeria monocytogenes based on screen - Food Control , vol. 73, pp. 555 561, Mar. 2017. [78] Y. Bai et al. of foodborne pathogens by extraction of a trace amount of DNA from raw milk based on amino - modified silica - coated magnetic Anal. Chim. Acta , vol. 787, pp. 93 101, Jul. 2013. [79] H. Huang et al. noparticle Based Magnetophoresis for Efficient Separation of Trans. ASABE , vol. 54, no. 3, pp. 1015 1024, Jun. 2011. 162 [80] immunosensor for fast, highly sensit ive, and in situ detection of the magnetic nanoparticles - Sens. Actuators B Chem. , vol. 230, pp. 191 198, Jul. 2016. [81] Y. Luo et al. for the De Ieee Trans. Nanotechnol. , vol. 11, no. 4, pp. 676 681, Jul. 2012. [82] S. Mun and S. - - Biochip J. , vol. 9, no. 1, pp. 10 15, Mar. 2015. [83] Salmonella typhimurium Based on Peroxidase - Like Activity of Magnetic Nanoparticles with J. Nanomater. , p. Aricle ID 527126, 2015. [84] W. - B. Shim, J. - E. Song, H. Mun, D. - H. Chung, and M. - detection of Salmonella typhimurium using a selective filtration technique combined with antibody - Anal. Bioanal. Chem. , vol. 406, no. 3, pp. 859 866, 2014. [85] L. Zhiyang et al. unological Acta Chim. Sin. , vol. 68, no. 3, pp. 251 256, Feb. 2010. [86] W. - B. Shim, C. - W. Lee, M. - G. Kim, and D. - - magnetic nanoparticle conjugate - based selective filtrat ion method for the rapid colorimetric detection of Listeria Anal. Methods , vol. 6, no. 22, pp. 9129 9135, 2014. [87] Y. J. Sung, H. - J. Suk, H. Y. Sung, T. Li, H. Poo, and M. - nanoparticle/magnetic nanoparticle n anocomposites for immunomagnetic separation and Biosens. Bioelectron. , vol. 43, pp. 432 439, May 2013. [88] Y. Zhao et al. ized Food Control , vol. 71, pp. 110 116, 2017. [89] D. A. Kanayeva et al. monocytogenes using an impedance immunosensor based on magnet ic nanoparticles, a J. Food Prot. , vol. 75, no. 11, pp. 1951 1959, Nov. 2012. [90] A. - Functionalization, and App Angew. Chem. Int. Ed. , vol. 46, no. 8, pp. 1222 1244, Feb. 2007. 163 [91] gold nanoclusters as effective peroxidase mimics and their application for colorimetr ic Biointerphases , vol. 12, no. 1, p. 401, Mar. 2017. [92] sensitive and selective sensor for morphine using chitosan coated Fe 3 O 4 magnetic nanop Mater. Sci. Eng. C , vol. 58, pp. 53 59, Jan. 2016. [93] J. Biomed. Nanote chnol. , vol. 9, no. 7, pp. 1254 1260, Jul. 2013. [94] C. - C. Lin et al. - Encapsulated Gold Nanoparticles to Type 1 Pili in Escherichia coli J. Am. Chem. Soc. , vol. 124, no. 14, pp. 3508 3509, Apr. 2002. [95] Sens. Bio - Sens. Res. , vol. 5, pp. 33 36, Sep. 2015. [96] - solid - Food Chem. , vol. 204, pp. 135 140, Aug. 2016. [97] Acta Chim. Slov. , vol. 61, no. 1, 2014. [98] K. L. B. Chang, G. Tsai, J. Lee, and W. - rogeneous N - deacetylation of chitin in Carbohydr. Res. , vol. 303, no. 3, pp. 327 332, Sep. 1997. [99] A. Galhoum et al. - Functionalized Chitosan Magnetic Nano - Based Particles for the Recovery of Light and Heavy Rare Earth Met als: Uptake Kinetics and Sorption Nanomaterials , vol. 5, no. 1, pp. 154 179, Feb. 2015. [100] Anal. Bioanal. Chem. , vol. 408, n o. 7, pp. 1783 1803, Mar. 2016. [101] G. Basina et al. - Fe2O3 and Fe3O4) MRS Proc. , vol. 1256, Jan. 2010. [102] M. Mahdavi et al. , Molecules , vol. 18, no. 7, pp. 7533 7548, Jun. 2013. 164 [103] N. K. Prasad, D. Panda, S. Singh, M. D. Mukadam, S. M. Yusuf , and D. Bahadur, J. Appl. Phys. , vol. 97, no. 10, p. 10Q903, May 2005. [104] - coated magnetite nanoparticles as MRI contrast agent: Effect of pH and cysteine addition on particles size Mater. Sci. - Pol. , vol. 30, no. 4, pp. 382 389, Dec. 2012. [105] nanoparticles as M Nano - Micro Lett. , vol. 4, no. 3, pp. 180 183, 2012. [106] D. Horák et al. - mannose - Bioconjug. Chem. , vol. 18, no. 3, pp. 635 644, 2007. [107] N. Sharon, Y. Eshdat, F. J. Si Ciba Found. Symp. , vol. 80, pp. 119 141, 1981. [108] nanoparticles functionalized by mannose - rhodam ine conjugate for detection of E. coli J. Appl. Polym. Sci. , vol. 131, no. 6, pp. 1 9, Mar. 2014. [109] Int. J. Food Micr obiol. , vol. 144, no. 1, pp. 51 63, Nov. 2010. [110] chitosan coated iron oxide - PAN hollow fiber membrane using Gram - positive and Gram - Mater. Sci. Eng. C , vol. 75, pp. 133 148, Jun. 2017. [111] S. Shi et al. chitosan - Int. J. Nanomedicine , vol. Volume 11, pp. 6499 6506, Dec. 2016. [112] scence Modified Chitosan - Coated Magnetic Nanoparticles for High - Nanoscale Res. Lett. , vol. 4, no. 4, pp. 287 295, Apr. 2009. [113] Prope Int. J. Mol. Sci. , vol. 14, no. 11, pp. 21266 21305, Oct. 2013. [114] Nanomaterials , vol. 5, no. 1, p p. 63 89, Jan. 2015. 165 [115] Campylobacter DNA using magnetic nanoparticles coupled with PCR and a colorimetric end - Food Sci. Biotechnol. , vol. 25, no. 1, pp. 1 93 198, Feb. 2016. [116] Staphylococcus aureus and Salmonella typhimurium using multicolor time - resolved Int. J. Food Microbiol. , vol. 237, pp. 172 179, Nov. 2016. [117] acid) J. Electroanal. Chem. , vol. 719, pp. 72 76, Apr. 2014. [118 ] O157:H7 with a wireless magnetoelastic - sensing device amplified by using chitosan - Sens. Actuators B Chem. , vol. 147, no. 1, pp . 343 349, May 2010. [119] - magnetic nanoparticles combined with PCR for efficient enrichment and detection of Listeria Sens. Actuators B Chem. , vol. 247, pp. 546 55 5, Aug. 2017. [120] M. Unni et al. ACS Nano , vol. 11, no. 2, pp. 2284 2303, Feb. 2017. [121] J. Joo et al. sensitive detection of pathogenic bacteria using magnetic The Analyst , vol. 137, no. 16, p. 3609, 2012. [122] J. Sanchez - Ramirez et al. - coated magnetic nanoparticles: Bioprocess Biosyst. Eng. , vol. 40, no. 1, pp. 9 22, Jan. 2017. [123] Y. G. Abou El - - glutaraldehyde as J. Environ. Chem. Eng. , vol. 4, no. 4, pp. 3835 3847, Dec. 2016. [124] J. Castelló, M. Gallardo, M. A. Busquets, and J. Est - coated Colloids Surf. Physicochem. Eng. Asp. , vol. 468, pp. 151 158, Mar. 2015. [125] and characterization of chitosan - coated iron oxide nanoparticles produced for biomedical J. Nanoparticle Res. , vol. 14, no. 11, pp. 1 13, Nov. 2012. 166 [126] P. I. P. Soares et al. - iron oxide nanoparti Carbohydr. Polym. , vol. 149, pp. 382 390, Sep. 2016. [127] - Biosens. Bioelectron. , vol. 55, p p. 289 293, May 2014. [128] Biosens. Bioelectron. , vol. 26, no. 4, pp. 1624 1630, Dec. 2010. [129] Biosens. Bioelectron. , vol. 55, pp. 289 293, May 2014. [130] D. Lim et al. Philipp. J. Pathol. , vol. 2, no. 2, p. 47, Nov. 2017. [131] Antibody - mimics for t he Expedient Extraction of 5 Salmonella Enteritidis, E. coli O157:H7 Manuscr. Prep. Publ. , Dec. 2017. [132] sensitive and selec tive sensor for morphine using chitosan coated Fe3O4 magnetic Mater. Sci. Eng. C - Mater. Biol. Appl. , vol. 58, pp. 53 59, Jan. 2016. [133] Nanotechnology , vol. 21, no. 16, p. 165102, Apr. 2010. [134] - Induced Temporary Superhydrophobic Coatings from One - Pot Synthesized Hydrophobic Magnetic ACS Appl. Mater. Interfaces , vol. 2, no. 5, pp. 1449 1455, May 2010. [135] - based Aptasensor Using Anal. Sci. , vol. 32, no. 4, pp. 431 436, Apr. 2016. [136] H. Kuang et al. - Step Homogeneous Sandwich Immunosensor for Salmonella Int. J. Mol. Sci. , vol. 14, no. 4, pp. 8603 8610, Apr. 20 13. [137] detection of Salmonella pullorum and Salmonella gallinarum based on immune blue silica Sens. Actuators B Chem. , vol. 226, pp. 69 7 5, Apr. 2016. 167 [138] M. R. Hormozi - determination of glutathione and cysteine based on aggregation of colloidal gold Sci. Iran. , vol. 19, no. 3, pp. 958 963, Jun. 2012. [139] G . Liu et al. exploiting hydrogen bond - induced aggregation of melamine - Microchim. Acta , vol. 182, no. 11 12, pp. 1983 1989, Aug. 2015. [140] M. Lepoitevin, M. Lemouel, M. Bechelany, J. - for the bare - eye based and spectrophotometric detection of proteins, polynucleotides and Microchim. Acta , vol. 182, no. 5 6, pp. 1223 1229, Apr. 2015. [141] K. Afshinnia, M. S J. Colloid Interface Sci. , vol. 487, pp. 192 200, Feb. 2017. [142] fects of gold Toxicol. Ind. Health , vol. 27, no. 6, pp. 547 554, Jul. 2011. [143] M. J. Anderson, E. Torres - synthesis of biocompatible gold n J. Nanoparticle Res. , vol. 13, no. 7, pp. 2843 2851, Jul. 2011. [144] K. Bankura et al. - mediated synthesis of Ag NPs for colorimetric assays of Cu 2+ Int. J. B iol. Macromol. , vol. 80, pp. 309 316, Sep. 2015. [145] In vitro and in vivo antioxidant, cytotoxic, and anti - chronic inflammatory arthritic effect of selenium nanoparticles: In vitro and in vivo eff J. Biomed. Mater. Res. B Appl. Biomater. , vol. 104, no. 5, pp. 993 1003, Jul. 2016. [146] Anal. Chim. Acta , vol. 457, no. 1, pp. 39 45, 2002. [147] SURFACE LECTINS OF Escherichia coli, Klebsiella pneumoniae, AND Salmonella Carbohydr. Res. , vol. 120, pp. 235 249, 1983. [148] A. Muñoz et al. ular multivalent glycofullerenes as potent Nat. Chem. , vol. 8, no. 1, pp. 50 57, Jan. 2016. 168 [149] - fimbriated bacteria to understand and combat bacterial Carbohydrate Chemistry: Chemical and Biological Approaches, Vol 37 , vol. 37, A. P. Rauter and T. K. Lindhorst, Eds. Cambridge: Royal Soc Chemistry, 201 1, pp. 194 207. [150] Biochem. Cell. Arch. , vol. 5, no. 2, pp. 137 143, Oct. 2005. [151] J. Collins et al. Nature , vol. 553, no. 7688, pp. 291 294, Jan. 2018. [152] Anal. Methods , vol. 8, no. 6, pp. 1235 1238, 2016. [153] M. - G. Baek, R. C. Stevens, and D. H. Charych, Glycopolythiophene Assemblies for Colorimetric Detection of Influenza Virus and E. coli Bioconjug. Chem. , vol. 11, no. 6, pp. 777 788, Nov. 2000. [154] s to Study Carbohydrate - Chem. Biol. , vol. 11, no. 12, pp. 1701 1707, Dec. 2004. [155] S. Park et al. - Surface Carbohydrate Binding Proteins with Dual - Modal Glycan - J. Am. Chem. Soc. , vol. 137, no. 18, pp. 5961 5968, May 2015. [156] - injection assay of the pathogenic bacteria using lectin - Talanta , vol. 77, no. 2, pp. 468 472, Dec. 2 008. [157] Food Chem. , vol. 136, no. 3 4, pp. 1303 1308, Feb. 2013. [158] E. Uribe, T. J. Steele, R. C. R of the Atlantic salmon serum C - Biochim. Biophys. Acta BBA - Gen. Subj. , vol. 1830, no. 1, pp. 2129 2138, Jan. 2013. [159] N. Sinha, S. Mohan, C. A. Lipschultz, and S. J. Smith - Gill, Properties at Antibody - Antigen Binding Sites: Implications for Specificity and Cross - Biophys. J. , vol. 83, no. 6, pp. 2946 2968, Dec. 2002. 169 [160] V. A. Kulshin, U. Zahringer, B. Linkner, K. - E. Jager, B. A. Dmitri ev, and E. T. Rietschel, - type Eur. J. Biochem. , vol. 198, no. 3, pp. 697 704, 1991. [161] S. Galdiero et al. - host interac tions: structure and role of Gram - negative bacterial Curr. Protein Pept. Sci. , vol. 13, no. 8, pp. 843 854, 2012. [162] T. Candela et al. - dependent Changes in a Bacillus cereus J. B iol. Chem. , vol. 286, no. 36, pp. 31250 31262, Sep. 2011. [163] R. Sommer et al. - Mannose as Inhibitors of the Bacterial ChemistryOpen , vol. 4, no. 6, pp. 756 767, Dec. 2015. [164] - 1 Flagellar Filament Protein of J. Biol. Chem. , vol. 260, no. 29, pp. 16758 16761, 1985. [165] Bacillus cereus Potential Cell Wall Peptidase Implicated in Adhesion, Biofilm Formation, J. Bacteriol. , vol. 192, no. 10, pp. 2638 2642, May 2010. [166] A . Muller et al. - binding cassette - type cysteine transporter in Campylobacter Mol. Microbiol. , vol. 57, no. 1, pp. 143 155, Jul. 2005. [167] Microbiol. Mol. Biol. Rev. MMBR , vol. 67, no. 4, pp. 593 656, Dec. 2003. [168] A. Diaz - Quiñonez, N. Martin - Orozco, A. Isibasi, and V. Ortiz - Salmonella OmpC Kb - Restricted Epitopes for CD8+ - T - Infect. Immun. , vol. 72, no. 5, pp. 3059 3062, May 2004. [169] C. Gil - Cruz et al. typhoidal Salmonella is a key target for a Proc. Natl. Acad. Sci. , vol. 106, no. 24, pp. 9803 9808, Jun. 2009. [170] membrane p Mol. Microbiol. , vol. 37, no. 2, pp. 239 253, Jul. 2000. [171] protein determinants for recognition of extracellular matrix componen ts: ECM recognition Cell. Microbiol. , vol. 14, no. 11, pp. 1687 1696, Nov. 2012. 170 [172] M. Gohar et al. PLoS One , vol. 3, no. 7, p. e2793, 2008. [173] S. P. Singh, S. R. Singh, Y. U Infect. Immun. , vol. 63, no. 12, pp. 4600 4605, Dec. 1995. [174] lar J Mol Biol , vol. 12, pp. 342 355, 1965. [175] Appl. Environ. Microbiol. , vol. 57, no. 6, pp. 1635 1639, 1991. [176] S. K. Collinson [177] as influenced by J. Food Prot. , vol. 70, no. 11, pp. 2473 2479, 2007. [178] M. Kukkonen et al. - binding fimbriae, the type 1 fimbria of Salmonella enterica serovar typhimuriu m and the G fimbria of Escherichia coli, as Infect. Immun. , vol. 66, no. 10, pp. 4965 4970, 1998. [179] S. G. Stahlhut et al. - Function Analysis of Mannose - Specific J. Bacteriol. , vol. 191, no. 21, pp. 6592 6601, Nov. 2009. [180] - terminal FEMS Microbiol. Lett. , vol. 16, pp. 149 151, 1983. [181] cereus J. Bacteriol. , vol. 145, no. 1, pp. 613 619, 1981. [182] J. Bacteriol. , vol. 144, no. 2, pp. 800 805, 1980. [183] J. - R. Neeser, B. - type glycopeptides inhibiting adhesion of Escherichia coli strains mediated by type 1 pili: preparation of potent Infect. Immun. , vol. 52, no. 2, pp. 428 436, 1986. [184 ] V. Oxaran et al. PLoS ONE , vol. 7, no. 12, p. e50989, Dec. 2012. 171 [185] H. Xu et al. - Mediated Adhesion of Biofilm - Forming E. coli to Langmuir , vol. 29, no. 9, pp. 3000 3011, Mar. 2013. [186] P. Burguiere, S. Auger, M. F. Hullo, A. Danchin, and I. Martin - systems participate in L - c J. Bacteriol. , vol. 186, no. 15, pp. 4875 4884, Aug. 2004. [187] multifunctional cysteine transport - associated protein required Mol. Microbiol. , vol. 74, no. 4, pp. 956 973, Nov. 2009. [188] M. Pokorná et al. - Driven Affinity of Chromobacterium violaceum Lectin CV - IIL toward Fucose and Mannose , Biochemistry , vol. 45, no. 24, pp. 7501 7510, Jun. 2006. [189] I. M. Helander, E. - L. Nurmiaho - Lassila, R. Ahvenainen, J. Rhoades, and S. Roller, - Int. J. Food Microbiol. , vol. 71, no. 2, pp. 235 244, 2001. [190] Int. J. Food Microbiol. , vol. 149, no. 3, pp. 218 225, Oct. 2011. [191] J. C. Fernandes, P. Eaton the antibacterial effects of chitosans on Bacillus cereus (and its spores) by atomic force Ultramicroscopy , vol. 109, no. 8, pp. 854 860, Jul. 2009. [1 92] L. Manni, O. Ghorbel - Characterization of Chitin, Chitosan, and Protein Hydrolysates Prepared from Shrimp Waste Appl. Biochem. Bio technol. , vol. 162, no. 2, pp. 345 357, Sep. 2010. [193] E. Fournier et al. Bioelectrochemistry , vol. 111, pp. 109 114, Oct. 2016. [194] A. Motaharian, F. Motaharian, K. Abnous, M. R. M. Hosseini, a nd M. Hassanzadeh - - based electrochemical sensor for Anal. Bioanal. Chem. , vol. 408, no. 24, pp. 6769 6779, Sep. 2016. [195] N. voltammetric acetylcholinesterase biosensor for the detection of captan in apple samples with Anal. Bioanal. Chem. , vol. 407, no. 16, pp. 4863 4868, Jun. 20 15. 172 [196] - cast organogel to form a liquid - gel microinterface array for electrochemical detection of Anal. Chim. Acta , vol. 893, pp. 34 40, Sep. 2015. [197] of Horseradish Peroxidase Immobilized in a DNA/Chitosan - Fe3O4 Magnetic Nanoparticle Bio - Materials , vol. 7, no. 2, pp. 1069 1083, Feb. 2014. [198] A. S aptasensor for immunoglobulin E detection based on sandwich assay using enzyme - linked Anal. Biochem. , vol. 466, pp. 89 97, Dec. 2014. [199] A. Ravalli and G. Marrazz - Based Electrochemical J. Nanosci. Nanotechnol. , vol. 15, no. 5, pp. 3307 3319, May 2015. [200] lmonella pullorum and Salmonella gallinarum based on a screen - printed carbon electrode modified Microchim. Acta , vol. 182, no. 13 14, pp. 2267 2275, Oct. 2015. [201] R. Das et al. chemical genosensor for Salmonella typhi on gold nanoparticles - J. Biotechnol. , vol. 188, pp. 9 16, Oct. 2014. [202] A. N. Kozitsina et al. - free electrochemical method for the determination of E - J. Anal. Chem. , vol. 70, no. 5, pp. 540 545, May 2015. [203] in a selective medium by monit oring oxygen consumption with electrochemical cyclic J. Electroanal. Chem. , vol. 519, no. 1 2, pp. 33 38, Feb. 2002. [204] - based immunosensor for electrochemical detectio Anal. Biochem. , vol. 441, no. 1, pp. 1 7, Oct. 2013. [205] S. P. Kounaves, Voltammetric techniques . Prentice Hall, Upper Saddle River, NJ, USA, 1997. [206] M. M. Radhi, E. A. Jaffar Al - Res. Chem. Intermed. , vol. 40, no. 1, pp. 179 192, Jan. 2014. 173 [207] UMass Boston, [208] T. E. Meyer et al. a basis for sy Proc. Natl. Acad. Sci. , vol. 80, no. 22, pp. 6740 6744, Nov. 1983. [209] cations, anions a nd neutral guest species by redox - Coord. Chem. Rev. , vol. 185 186, pp. 3 36, May 1999. [210] Anal. Chim. Acta , vol. 598, no. 2, pp. 181 192 , Aug. 2007. [211] Anal. Chem. , vol. 76, no. 4, pp. 1083 1088, Feb. 2004. [212] F. Li, J. Han, L. Jiang, Y. Wan g, Y. Li, Y. Dong, and Q. Wei, - type electrochemical immunosensor based on signal amplification strategy of gold nanoparticles functionalized magnetic multi - Biosens. Bioelectron. , vol. 68, pp. 626 632, Jun. 2015. [213] nanoparticles based electrochemical sensor for the measurement of Gram - negative bacterial quorum signaling molecules (N - acyl - homoserine - Biosens. Bioelectron. , vol. 75, pp. 411 419, Jan. 2016. [214] C. Cheng et al. - assembled Sens. Actuators B - Chem. , vol. 190, pp. 900 906, Jan. 2014. [21 5] - Active Amino Acids on Compound I Stability, Substrate Oxidation, and Protein Cross - Linking in Yeast Biochemistry , vol. 40, no. 49, pp. 14942 14951, Dec. 200 1. [216] production by the synergistic action of Bacillus tequilensis DMR - 5 and Pseudomonas aeruginosa DMR - Am. J. Environ. Sci. , vol. 9, no. 5, pp. 424 430, 2013. [217] Y. - M. Huang, H. - Y. Hsu, and C. - J. Taiwan Inst. Chem. Eng. , vol. 62, pp. 39 44, May 2016. 174 [218 ] Electrochemical Interaction between Shewanella Nano Lett. , vol. 14, no. 6, pp. 3688 3693, Jun. 2014. [219] T. Kinoshita, Y. Hatsuoka, D. Q. Nguyen, R. Iwata, H. Shiigi, and T. Nagaoka, Electrochemistry , vol. 84, no. 5, pp. 334 337, 2016. [220] microbia l cell viability using nanostructured electrodes modified with Graphene/Alumina Biosens. Bioelectron. , vol. 91, pp. 857 862, May 2017. [221] Orient. J. Chem. , vol. 31, no. 4, pp. 2259 2263, 2015. [222] J. H. Chien et al. Na no/Micro , 2006, pp. 278 282. [223] M. A. Hoppens et al. J. Colloid In terface Sci. , vol. 413, pp. 167 174, Jan. 2014. [224] inactivated Francisella tularensis bacteria by using a quartz crystal microbalance with A nal. Bioanal. Chem. , vol. 404, no. 3, pp. 843 851, Aug. 2012. [225] P. M. Reddy, K. - C. Chang, Z. - J. Liu, C. - T. Chen, and Y. - iron oxide (Fe3O4) nanoparticles for capturing gram - positive and gram - J. Biomed. Nanotechnol. , vol. 10, no. 8, pp. 1429 1439, Aug. 2014. [226] H. S sensor: A case study on E. coli O157:H7 as a model for Gram - Sens. Actuators B Chem. , vol. 161, no. 1, pp. 298 303, Jan. 2012. [227] S. B. Ghosh, K. Bhat tacharya, S. Nayak, P. Mukherjee, D. Salaskar, and S. P. Kale, FTIR, UV - Spectrochim. Acta Part - Mol. Biomol. Spectrosc. , vol. 148, pp. 420 426, Sep. 2015. [228] J. K. Streit, S. M. Bachilo, S. Ghosh, C. - Optical Absorption Cross Sections for Structure - Selected Single - Nano Lett. , vol. 14, no. 3, pp. 1530 1536, Mar. 2014. 175 [229] Y. Wan et al. Sci. Rep. , vol. 6, no. 1, Sep. 2016. [230] UV/VIS spectrometer determination of caffeine in green coffee beans from Haraghe, Ethiopia, using Beer - Lambert's Law and integrated absorption coefficient techniques," Sci. Study Res. Chem. Chem. Eng. Biotechnol. Food Ind. , vol. 17, no. 2, p. 109, 2016. [231] WH World Health Organization , 31 - Oct - 2017. [Online]. Available: http://www.who.int/news - room/fact - sheets/detail/food - safety. [Accessed: 20 - Nov - 2018]. [232] Pew Research Center , 08 - Jul - 2015. [Online]. Available: http://www.pewglobal.org/interactives/global - population - by - income/. [Accessed: 20 - Nov - 2018]. [233] Y. - M. Bae, L. Zheng, J. - E. Hyun, K. - S. Jung, S. Heu, and S. - Characteristics and Biofilm Formation of Various Spoilage Bacteria Isolated from Fresh J. Food Sci. , vol. 79, no. 10, pp. M2072 M2080, Oct. 2014. [234] regulatory sanitization process in Egyptian dairy plants in regard to t he adherence of some food - Int. J. Food Microbiol. , vol. 158, no. 3, pp. 225 231, Sep. 2012. [235] N. - Y. Choi, B. - R. Kim, Y. - M. Bae, and S. - J. Korean Soc. Appl. Biol. Chem. , vol. 56, no. 2, pp. 207 220, Apr. 2013. [236] S. Marc hand, J. De Block, V. De Jonghe, A. Coorevits, M. Heyndrickx, and L. Herman, Compr. Rev. Food Sci. Food Saf. , vol. 11, no. 2, pp. 133 147, Mar. 2012. [237] J. Appl. Microbiol. , vol. 90, pp. 901 908, 2001. [238] S. Galdiero et al. - host interactions: structure and role of Gram - negative bacterial Curr. Protein Pept. Sci. , vol. 13, no. 8, pp. 843 854, 2012. [239] virulence factors deployed by Listeria monocytogenes to Virulence , vol. 2, no. 5, pp. 379 394, Sep. 2011. 176 [240] P. Cossart, J. Pizarro - Trends Cell Biol. , vol. 13, no. 1, pp. 23 31, 2003. [241] B. - Consumption Broken Down by Demographics, 1994 - - 2016. [Online]. Available: http://www.ers.usda.gov/media/2053482/err - 206_s ummary.pdf. [Accessed: 23 - May - 2016]. [242] E. E. Jackson et al. J. Food Prot. , vol. 75, no. 8, pp. 1382 1393, Aug. 2012. [243] L. H. Ledenbach and Compendium of the Microbiological Spoilage of Foods and Beverages , W. H. Sperber and M. P. Doyle, Eds. New York, NY: Springer New York, 2009, pp. 41 67. [244] Administration, 2015. [245] - temperature, short - J. Dairy Sci . , vol. 92, no. 10, pp. 4833 4840, Oct. 2009. [246] J. Dairy Sci. , vol. 93, no. 11, pp. 4999 5009, Nov. 2010. [247] N. Silanikove, Int. Dairy J. , vol. 16, no. 6, pp. 533 545, Jun. 2006. [248] A. J. Langer, T. Ayers, J. Grass, M. Lynch, F. J. Angulo, and B. E. Mahon, teurized Dairy Products, Disease Outbreaks, and State Laws United States, 1993 Emerg. Infect. Dis. , vol. 18, no. 3, pp. 385 391, Mar. 2012. [249] 2005 - 2006, Sep - 2010. [250] - Nov - 2011. [Online]. Available: https://www.fda.gov/Food/FoodborneIllnessContaminants/BuyStoreServeSafeFood/ucm24 7991.htm. [Accessed: 21 - Apr - 2017]. [251] Curr. Opin. Biotechnol. , vol. 34, pp. 242 250, Aug. 2015. 177 [252] Special - Purpose Modifications and Immobilized Immobilisation of DNA on Chips II , C. Wittmann, Ed. Springer Berlin Heidelberg, 2005, pp. 131 168. [253] S. - Y. Lee, J. - H. Lee, J. - H. Chang, and J. - H. - based BMB Rep. , vol. 44, no. 2, pp. 77 86, Feb. 2011. [254] - J. Aquat. Food Prod. Technol. , vol. 24, no. 5, pp. 533 543, Jul. 2015 . [255] Carbohydrate - Chem. Biol. , vol. 11, no. 12, pp. 1701 1707, Dec. 2004. [256] M. Leitgeb, K. Herzic, G. H. Podrepsek, A. Hojs 2014. [257] S. Park et al. - Surface Carbohydrate Binding Proteins with Dual - Modal Glycan - Conjugated Nanoparticle J. Am. Chem. Soc. , vol. 137, no. 18, pp. 5961 5968, May 2015. [258] D. Wang et al. based on screen - Food Control , vol. 73, no. Part B, pp. 555 561, 2017. [259] M. Q. Carter et al. PLoS ONE , vol. 7, no. 9, p. e44186, Sep. 2012. [260] - associating autotransporters, SAATs: Int. J. Med. Microbiol. , vol. 296, no. 4 5, pp. 187 195, Aug. 2006. [261] J. D. Barak, C. E. Jahn, D. L. Gibson, and A. O. O - Mol. Plant. Microbe Interact. , vol. 20, no. 9, pp. 1083 1091, 2007. [262] T. Candela et al. - dependent Changes in a Bacillus cereus J. Biol. Chem. , vol. 286, no. 36, pp. 31250 31262, Sep. 2011. [263] I. - - assisted surface enhanced Raman spectroscopy for the rapid detection of E - Biosens. Bioelectron. , vol. 64, pp. 171 176, Feb. 2015. 178 [264] P. - T. Chu, M. - F. Hs ieh, S. - Y. Yin, and H. - immunomagnetic - Eur. Food Res. Technol. , vol. 229, no. 1, pp. 73 81, May 2009. [265] D. C. Vanegas, Y. Rong, N. Schwalb, K. D. Hil detection of listeria spp. using an internalin A aptasensor based on carbon - metal nanohybrid - 1 to 7. [266] for Bioconjug. Chem. , vol. 18, no. 2, pp. 297 301, Apr. 2007. [267] spectroscopy with asymmetric - Microfluid. Nanofluidics , vol. 1, no. 2, pp. 161 167, May 2005. [268] of Special Research Skills, CFSAN, USFDA, 1998. [269] \ \ Influencing its Quality for , vol. 13, pp. 633 651, 2011. [270] and characterization of psychrotolerant coliform bacteria isolated from pas teurized fluid J. Dairy Sci. , vol. 99, no. 1, pp. 130 140, Jan. 2016. [271] S. Migeemanathan, R. Bhat, L. Min - Tze, and W. - N. Wan - Temperature Abuse on the Survival, Growth, and Inactivation of Salmonella typhimurium in Goat Mil Foodborne Pathog. Dis. , vol. 8, no. 11, pp. 1235 1240, Nov. 2011. [272] L. Necidová et al. Acta Vet. Brno , vol. 83, no. 10, pp. S3 S8, 2015. [273] B. I. Escudero - Ab arca, S. H. Suh, M. D. Moore, H. P. Dwivedi, and L. - A. Jaykus, PLoS ONE , vol. 9, no. 9, p. e106805, Sep. 2014. [274] p the Spread of Superbugs Help Fight Drug - NIH News in Health , Feb - 2014. [Online]. Available: https://newsinhealth.nih.gov/issue/feb2014/feature1. [Accessed: 22 - Jul - 2016]. [275] ptake of gold nanoparticles: what J. Nanoparticle Res. , vol. 12, no. 7, pp. 2313 2333, Sep. 2010. [276] of carbon nanoparticles - modified ch iral separation systems using dextrin as chiral selector 179 ELECTROPHORESIS , vol. 34, no. 13, pp. 1901 1907, Jul. 2013. [277] J. Mol. Catal. Chem. , vol. 204 205, pp. 553 559, Sep. 2003. [278] Foodborne Outbreaks , 2015. [Online]. Available: http://www.cdc.gov /foodsafety/outbreaks/multistate - outbreaks/index.html. [Accessed: 31 - Oct - 2015]. [279] Food Safety News , 09 - Jun - 2016. [Online]. Available: http://www.foodsafetynews.com/2016/06/u - n - officials - say - food - safety - is - entwined - with - food - security/. [Accessed: 28 - Jun - 2017].