I‘V‘i-kl‘tn
‘1‘ HT“;

‘0 a; (rib—520

This is to certify that the
dissertation entitled

Design And Fabrication Of A Micro-Impedance Biosensor For
Detecting Pathogenic Bacteria

presented by
Stephen M. Radke

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Biosystems Engineering

2

Major Professor’s ignature

W§1M¢

Date

.
r I
;

 

MSU is an Affirmative Action/Equal Opportunity Institution

4AA.-.-.-.---.-o-.—o—-o--u-o-u-.---o-a—a--.-.-.—.—.-.-.—.—o-

..—.—.-.-.-.-.—.—.—.—.—.-.—.—.—.—._.—._.—._.--.-.-.-.-.—.a—.-

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DatoDue.p65~p.15

DESIGN AND FABRICATION OF A MICRO-IMPEDANCE BIOSENSOR
FOR DETECTING PATHOGENIC BACTERIA

By

Stephen M. Radke

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PI-HLOSOPHY
Department of Biosystems and Agricultural Engineering

2004

ABSTRACT

DESIGN AND FABRICATION OF A MICRO-IMPEDANCE BIOSENSOR
FOR DETECTING PATHOGENIC BACTERIA

By

Stephen M. Radke

A biosensor for bacterial detection was developed based on microelectromechanical
systems (MEMS), heterobifunctional crosslinkers and immobilized antibodies. The
sensor detected the change in impedance caused by the presence of bacteria immobilized
on interdigitated gold electrodes. Fabricated from (100) silicon with a 2pm layer of
thermal oxide as an insulating layer, the sensor active area was 9.6mm2 and consisted of
two interdigital gold electrode arrays each measuring 0.8mm x 6mm. Escherichia coli
specific antibodies were immobilized to the silicon oxide between the electrodes to create
a biological sensing surface. The electrical impedance across the interdigital electrodes
was measured at frequencies between lOOHz to lOMHz after immersing the biosensor in
a neutral buffer. Bacterial cells present in the sample solution attached to the antibodies
and became tethered to the sensor surface thereby causing a change in measured
impedance. The biosensor was tested using pathogenic and non-pathogenic E. coli strains
and was able to discriminate between different cellular concentrations from 105 - 107
CFU/mL (colony-forming units per milliliter) in pure culture. The design, fabrication
and testing of the biosensor is discussed along with the implications of these findings

towards developing a biosensor for the detection of foodbome pathogens.

© Copyright By
STEPHEN M. RADKE
2004

ACKNOWLEDGEMENTS

I would like to express my gratitude to my fellow coworkers Lisa Marie Kindschy,
Finny Papachan Mathew, Zarini Binti Muhammad-Tahir and Maria Ivelisse Rodriguez
and also to those undergraduates who have worked in our laboratory over the years. I
also wish to thank Dr. Evangelyn C. Alocilja for being a constant source of motivation,
encouragement and support. Lastly, I would like to thank the USDA for financial

support in the way of a National Needs Doctoral Fellowship. Thank you.

iv

TABLE OF CONTENTS

 

 

 

 

 

 

LIST OF TABLES - - vii
LIST OF FIGURES -- - viii
CHAPTER 1. INTRODUCTION 1
1.1 Objectives and Goals ............................................................................................... l

1.2 Foodbome Pathogens and Food Safety .................................................................... l

1.3 Market Analysis for Pathogen Detection ................................................................. 3

1.4 Biosecurity and Agroterrorism ................................................................................ 4

1.5 Routes of Infection ................................................................................................... 6

1.6 Use of Biosensors and Rapid Detection Methods .................................................... 7

1.7 Novelty of Research ............................................................................................... 14

1.8 Hypothesis and Specific Aims ............................................................................... 15
CHAPTER 2. LITERATURE REVIEW - -- 16
2.1 Overview of the Biosensor ..................................................................................... 16
2.2 Lipid Bilayer Membrane Theory ........................................................................... 18
2.3 Biological Recognition .......................................................................................... 20
2.4 Membrane Impedance Theory ............................................................................... 22
2.5 Impedance Measurements ...................................................................................... 25
2.6 Circuit Elements ..................................................................................................... 27

2.7 Electrode Spacing Model ....................................................................................... 33
2.8 Immobilization Methods ........................................................................................ 35
2.9 Microfabrication .................................................................................................... 37
CHAPTER 3. MATERIALS AND METHODS 41
3.1 Microfabrication of the Biosensor ......................................................................... 42
3.2 Functionalizing the Biosensor Surface .................................................................. 44

3.3 Validation of the Biosensor ................................................................................... 47
3.4 Signal Measurement ............................................................................................... 49

3.5 Statistical Analysis ................................................................................................. 49

3.6 Specificity Study .................................................................................................... 50

3.7 Testing in Complex Media ..................................................................................... 52

3.8 Facilities and Equipment ........................................................................................ 52
CHAPTER 4. RESULTS AND DISCUSSION 53
4.1 Simulation Results ................................................................................................. 53
4.2 Fabrication Results ................................................................................................. 60

4.3 Results and Discussion of Pure Culture Testing .................................................... 67
4.3.1 Frequency Dependence ........................................................................... 67

4.3.2 Effect of Non-Pathogenic and Pathogenic Bacteria ................................ 71

4.4 Results and Discussion of Specificity Study ........................................................ 77

 

4.5 Summary and Conclusions .................................................................................... 82
4.6.1 Summary of Lower Detection Limits ..................................................... 82

4.6.2 Innovation and Future Possibilities ......................................................... 83
APPENDICES - 84
Appendix A: US Foodbome Disease Outbreak Data .................................................. 85
Appendix B: Materials, Methods and Bill of Process Supplemental ........................... 86
Appendix C: Results and Discussion ........................................................................... 94
Appendix D.1: Testing in Complex Media ................................................................ 103
D.1.l Testing of Romaine Lettuce ................................................................. 103

D.1.2 Testing of Ground Beef ........................................................................ 104

D.1.3 Testing of Bovine Feces ....................................................................... 105

Appendix D.2: Results and Discussion of Testing in Complex Media ..................... 106
D.2.1 Results for Romaine Lettuce Testing ................................................... 106

D.2.2 Results for Ground Beef Testing ......................................................... 109

D.2.3 Results for Bovine Feces Testing ......................................................... 112

Appendix D.3: Summary and Conclusions ................................................................ 116
D.3.1 Summary of Lower Detection Limits .................................................. 116

Appendix E: Business Plan ....................................................................................... 118
BIBLIOGRAPHY 141

 

vi

LIST OF TABLES

Table 1.1 Food illness in the US caused by major foodborne pathogens ............................ 2
Table 1.2 Sample of recent product recalls due to pathogen contamination ....................... 3
Table 1.3 US food industry total microbial tests per sector ................................................. 3
Table 1.4 Novelty of research compared to existing electrochemical biosensors ............. 14
Table 3.1 Specificity testing matrix ................................................................................... 51
Table 4.1 Comparison matrix of simulation sizes ............................................................. 58

Table A.l Reported and estimated cases of foodborne illness by agent type

in the US .......................................................................................................... 84
Table A2 Reported and estimated cases of foodborne illness by surveillance type in

the US ............................................................................................................... 85
Table B.1 Bill of process for microfabrication of the biosensor ........................................ 86
Table 82 Bill of process for surface functionalization of the biosensor ........................... 87
Table 8.3 Bill of process for testing in ground beef samples ............................................ 87
Table 8.4 Bill of process for testing in romaine lettuce samples ...................................... 88
Table B.5 Bill of process for testing in bovine feces samples ........................................... 88
Table B.1 Bill of process for reagents used in the research ............................................... 89

vii

LIST OF FIGURES

Figure 1.1 A Schematic of the food supply chain showing potential entry points
for contamination of foodborne pathogens ........................................................ 8

Figure 2.1 Schematic of biosensor detection theory .......................................................... 16

Figure 2.2 Electric field between: (a) interdigital electrode array; (b) cross-section
of interdigital array with immobilized bacteria on the surface ....................... 18

Figure 2.3 Cross-section of lipid bilayer membrane showing charged hydrophilic

phospholipid head groups and hydrophobic fatty acid tail groups .................. 19
Figure 2.4 Antibody schematic ........................................................................................ 21
Figure 2.5 Schematic of dispersion in a cell at low (a) and high (b) frequencies .............. 24
Figure 2.6 Idealized spectrum of the dielectric properties of cell suspensions ................. 25

Figure 2.7 Impedance spectrum of ideal RC circuit showing reactive and resistive
components ...................................................................................................... 27

Figure 2.8 Schematic of the electrode-solution interface showing inner and outer
Helmholtz planes ............................................................................................. 29

Figure 2.9 Equivalent circuit of the impedance measurement system with electrodes
in solution ........................................................................................................ 30

Figure 2.10 Circuit model for the impedance of bacteria immobilized between two

interdigitated electrodes ................................................................................. 32
Figure 2.11 Calculated electric field lines (a) above interdigital electrodes; (B) as

a function of electrode spacing ratio .............................................................. 34
Figure 2.12 Antibody immobilization method ................................................................... 36
Figure 2.13 The sequence of a typical lift off process ....................................................... 39
Figure 3.1 (3-Mercaptopropyl) trimethyloxysilane (MT S) used for silanization .............. 45

Figure 3.2 N-y-maleimidobutyryloxy succinimide ester (GMBS) for crosslinking .......... 45

Figure 3.3 Process of antibody attachment to the silicon oxide ........................................ 46

viii

Figure 3.4 Biosensor test apparatus in raised and lowered position .................................. 48
Figure 4.1 Simulation results for different electrode widths and spacing: (3) lm x

lum array; (b) 6pm x 6pm array; (c) 5pm x l0um array and (d) 10pm x

Sum array” ... ... . ..........54

Figure 4.2 Electric field simulation of 4m x 3pm electrode array: (top) electric field
magnitude; (bottom) vector representation of electric field ............................ 56

Figure 4.3 Surface plot of the electric field distribution at 2pm from electrode array ...... 57
Figure 4.4 Percentage of electric field above sensor surface ............................................. 57

Figure 4.5 Simulation of 3m x 4pm array with immobilized bacteria: (top) electric

field magnitude; (bottom) vector representation of electric field .................... 59
Figure 4.6 Maxwell 3D simulation of the electric field on sensor surface ........................ 59
Figure 4.7 A 4" Silicon wafer before (left) and after (right) processing ........................... 60

Figure 4.8 CAD Drawing of Biosensor: (left) biochip layout; (top-right) close-up of
electrode arrays; (bottom-right) detail of electrode arrays .............................. 61

Figure 4.9 Die shot of biosensor; high density electrode arrays; close up of electrodes...62

Figure 4.10 Quality control diagram showing likely locations for damaged dies ............. 64
Figure 4.11 AFM of electrodes; bacteria immobilized to surface ..................................... 65
Figure 4.12 CLSM showing antibody immobilized to electrode and oxide ...................... 65

Figure 4.13 The effects of surface modification: (left) clean biosensor surface; (right)
silanized biosensor surface ............................................................................. 66

Figure 4.14 Impedance for a frequency distribution from 10Hz-10MHz for non-
pathogenic E. coli ............................................................................................ 68

Figure 4.15 Impedance for 3 Frequency Distribution from 10Hz-10MHz for 3 Pure
Culture of E. coli 01572H7 ............................................................................. 69

Figure 4.16 Comparison of E. coli and E. coli 0157:H7 at selected frequencies:
(top) lkHz; (middle) lOOkHz; (bottom) lOMHz ............................................ 72

Figure 4.17 Cole-Cole plot showing the real and imaginary impedance with time .......... 74

ix

Figure 4.18 SEM micrographs of bacteria bound to the sensor surface: (a) sample of 102
CFU/mL; (b) sample of 10° CFU/mL ............................................................. 76

Figure 4.19 Statistical significance of mean differences between concentrations for
pure culture ...................................................................................................... 77

Figure 4.20 Impedance for a frequency distribution from lOOHz - lOMHz for a pure
culture of S. infantis ....................................................................................... 78

Figure 4.21 Impedance for a frequency distribution from lOOHz - lOMHz for a mixed

culture of S. infantis and E. coli 0157:H7 ....................................................... 79
Figure 4.22 Comparison of E. coli 0157 :H7, S. infantis, and mixed culture at selected
frequencies: (top) lkHz; (middle) 100kHz; (bottom) lOMHz ........................ 80
Figure 4.23 Statistical significance of mean differences between concentrations for
specificity study ............................................................................................. 82
Figure B] Screen capture of user interface for LabVIEW 6.1 ......................................... 90
Figure B.2 Screen capture of circuit diagram for LabVIEW 6.1 ....................................... 90
Figure B.3 Clamp without biochip ..................................................................................... 91
Figure B.4 Clamp with biochip in locating chuck ............................................................. 91
Figure B.5 Apparatus setup ................................................................................................ 92
Figure B.6 Apparatus setup engaged in specimen testing ................................................. 92
Figure B? Detail of contact mating to biochip .................................................................. 93

Figure B.8 Wiring schematic showing HP 4192A impedance analyzer connected

to biosensor ...................................................................................................... 93
Figure C.1 Vector representation of electric field simulation of lum x 1m array .......... 94
Figure C.2 Surface plot of the electric field distribution at a 2m distance from a lum

x lum electrode array ...................................................................................... 95
Figure C.3 Vector representation of electric field simulation of 10m x Sum array ........ 96

Figure C.4 Surface plot of the electric field distribution at a 2pm distance from a 10m
x Sum electrode array ...................................................................................... 97

Figure C.5 Vector representation of electric field simulation of 5pm x 10m array ........ 98

Figure C.6 Surface plot of the electric field distribution at a 2pm distance from a 5pm
x 10m electrode array .................................................................................. .99

Figure C.7 Vector representation of electric field simulation of 6m x 6pm array ........ 100

Figure C.8 Surface plot of the electric field distribution at a 2m distance from a 6pm
x 6m electrode array .................................................................................... 101

Figure C.9 3-D rendering of CAD layout file depicting biosensor .................................. 102

Figure D.1 Representative samples of romaine lettuce, ground beef and
bovine feces ................................................................................................... 103

Figure D.2 Impedance for a frequency distribution from lOOHz - lOMHz for lettuce
samples inoculated with E. coli 01572H7 .................................................... 107

Figure D.3 Comparison of romaine lettuce samples inoculated with E. coli 01572H7
at selected frequencies: (top)1kHz; (middle)100kHz; (bottom)10MHz ..... 108

Figure D.4 Impedance for a frequency distribution from lOOHz - lOMHz for ground
beef samples inoculated with E. coli OlS7:H7 .............................................. 110

Figure D.5 Comparison of ground beef samples inoculated with E. coli 0157:H7 at
selected frequencies: (top) lkHz; (middle) 100kHz; (bottom) lOMHz ...... 111

Figure D.6 Impedance for a frequency distribution from lOOHz - lOMHz for a manure
samples inoculated with E. coli 0157:H7 .................................................... 114

Figure D.7 Comparison of bovine feces samples inoculated with E. coli 0157:H7 at
selected frequencies: (top) lkHz; (middle) 100kHz; (bottom) lOMHz ....... 115

Figure D.8 Statistical significance of mean differences between concentrations for
complex media study .................................................................................... 116

xi

Chapter 1. Introduction

1.1 Objectives and Goals

The long term goal of this research is to develop a field-deployable portable
biosensor for the real-time detection of Category B disease agents transmissible through
food and water. This is to be accomplished by using a new biosensor architecture, which
combines the use of microelectromechanical systems (MEMS) fabrication methods and a
biological sensing surface. The model disease agent for this research, aimed to
demonstrate proof of concept, is Escherichia coli OlS7:H7. The biosensor, designed to
detect whole cell organisms in a liquid volume, will consist of a reagent-coated MEMS
biochip for detecting the analyte. The biosensor is designed to enable health care
professionals, bioterrorism rapid—response teams, and food safety monitoring personnel to
quantify results in less than 5 minutes.

The short term goal of this research is to construct a prototype biosensor. The
biosensor will be evaluated for its ability to detect E. coli 0157:H7 in liquid media. It
represents an innovative approach to detecting infectious disease agents due to the
biosensor’s large sample size, minimal sample processing and 10 minute detection time

from sample application to results.

1.2 Foodbome Pathogens and Food Safety
Pathogenic bacteria and other microorganisms are ubiquitous in the environment.
Bacterial pathogens are found in soil, animal intestinal tracts and in fecal-contaminated

water. Human beings, on average, harbor more than 150 types of bacteria inside and

outside of the body (Madigan et al., 1997). Although many microorganisms are
harmless, some are known to be the causative agent of many different infectious diseases
including botulism, cholera, diarrhea, emesis, pneumonia and typhoid fever (Doyle et al.,
1997). More than 200 known diseases are transmitted through food and drink alone
(Mead et al., 1999). A table of outbreak incidents is included in Appendix A.

Although recent data suggests naturally occurring cases of foodborne disease
outbreaks are declining in the US (CDC, 2002), it is estimated that foodborne diseases
cause approximately 76 million illnesses, including 325,000 hospitalizations and 5,000
deaths in the US each year (Mead et al., 1999). Of these, known pathogens account for
an estimated 14 million illnesses, 60,000 hospitalizations, and 1,800 deaths indicating
that these pathogens are a substantial source of infectious disease. Outbreaks caused by
the four major foodborne pathogens, Campylobacter, Salmonella, Listeria

monocytogenes, and E. coli OlS7:H7 are characterized in Table 1.1.

 

 

we 1.1 Food illnesses in the US caused by ma'or foodborne pathogens (CDC, 1999)
Pathogen Number of Cases Hospitalizations Deaths
Campylobacter 1,963,141 10,539 99
E. coli 01572H7 62,458 1,843 52
L. monocytogenes 2,498 2,298 499
Salmonella 1,342,532 16,102 556

 

 

 

 

 

 

To demonstrate the scope of the contamination problem, selected recall data due to
contamination of pathogens is shown in Table 1.2. The United States Department of
Agriculture (USDA) estimates $2.9 billion to $6.7 billion is lost annually due to medical

costs and lost productivity caused by major food pathogens (Buzby et al., 1996).

Table 1.2 Examples of food recalls due to pathogen contamination (USDA-F818, 2002)
Company Product Recalled Contaminant Amount Recalled

 

 

Cargill Turkey, TX Poultry Products L. Monocytogenes 16.7 million pounds

Bar-S Foods, GA Meat & Poultry L. Monocytogenes 14.5 million pounds
Excel Corp, GA Ground Beef/Pork E. coli 0157:H7 190,000 pounds
American Food, WI Ground Beef E. coli 015 7:H7 530,000 pounds
Savoie's, LA Cajun Dressing Salmonella 500,000 pounds

 

 

 

 

 

Zartic, GA Chopped Beefsteak Salmonella 2,700,000pounds

 

1.3 Market Analysis for Pathogen Detection

The broad market for pathogen detection extends across a range of industries
including food processing companies, environmental monitoring agencies, healthcare
industries and the military. Combined, the total market size for pathogen detecting
biosensors is $563 million dollars and is growing at a compounded annual growth rate
(CAGR) of 4.5% (Radke and Alocilja, 2003). The food pathogen testing market alone is
expected to grow to $192 million and 34 million test units by 2005. The food processing
industry can be further segmented into the type of food product (meat, dairy, fruit,
vegetables, processed foods) and the target pathogen (bacteria, viruses, fungi and other
biohazardous agents). The total number of microbial tests performed by food industry

sectors is around 144 million tests per year and is shown in Table 1.3.

Table 1.3 US food industry microbial tests per sector. (Strategic Consulting, 1999)

 

 

 

Sector Number of Plants Total Tests Average/Plantlweek
eef and Poultry 1,679 32,212,471 369
airy 1,388 45,887,576 636
wit/Vegetables 652 13,981,305 412
essed foods 2,260 52,196,282 444
otal 5,979 144,277,634 464

 

 

 

 

 

Market data shows that a biosensor for the rapid detection of pathogens has an
excellent chance of success in the marketplace. Also, pathogen detecting biosensors are a
"disruptive" technology and tend to create their own markets. If a low cost and reliable
product were to be introduced into the marketplace, it is possible that the net number of
pathogen tests would increase simply because the technology would be available. Food
companies would perform more frequent product testing and new segments would also

Open up as restaurants and consumers seek to verify the safety of the food they eat.

1.4 Biosecurity and Agroterrorism

The deliberate introduction of a biological pathogen into US livestock, poultry or
crops would increase food prices, reduce food exports (costing billions of dollars in lost
revenue) and potentially increasing the number of illnesses associated with foodborne
pathogens. Indeed, biosecurity has become an increasingly important element in the
battle against terrorist acts. Biosecurity threats include disease causing agents of high
consequence, such as viruses, bacteria and toxins. Human exposure to pathogens may
occur through inhalation, skin exposure or ingestion of contaminated food or water.
Foodbome pathogens pose a risk to food safety and are a threat to the nation's food
supply chain.

One such danger to the nation’s food supply is the threat posed through agroterrorism.
Agroterrorism encompasses many aspects, including the destruction of cropland, the
intentional spread of livestock diseases, and the deliberate use of food pathogens to
disrupt the safety of the nation's food supply (Kohnen, 2000). The World Health

Organization (WHO) has indicated that terrorists may try to contaminate food supplies

and has urged countries to strengthen their surveillance. The WHO cites past examples
of intentional food attacks, including a Salmonella Typhimurium outbreak in Oregon
where more than 750 people became ill after members of a cult contaminated restaurant
salad bars (WHO, 2002).

The National Institute of Allergy and Infectious Diseases (NIAID), an institute of The
National Institutes of Health (NIH) and the Centers for Disease Control and Prevention
(CDC) categorize biological pathogens as either Category A, B or C. Category A agents
include organisms that pose a risk to national security because they can be easily
disseminated or transmitted from person to person; result in high mortality rates and have
the potential for major public health impact; might cause public panic and social
disruption; and require special action for public health preparedness (NIAID, 2004).
Category B agents include those that are moderately easy to disseminate; result in
moderate morbidity rates and low mortality rates; and require specific enhancements of
the nation's diagnostic capacity and enhanced disease surveillance. Category C agents
include emerging pathogens that could be engineered for mass dissemination in the future
because of availability; ease of production and dissemination; and potential for high
morbidity and mortality rates and major health impact. NIAID and the CDC have
identified foodborne pathogens such as Salmonella spp., L Monocytogenes, and E. coli
0157:H7 as Category B bioterrorism agents (NIAID, 2004; CDC, 2004). In particular, E.
coli 0157:H7 poses a significant threat to the nation's food supply as it has emerged as
one of the deadliest foodborne pathogens due to its combination of virulence and

pathogenicity (CDC, 2001).

l.5 Routes of Infection

E. coli are bacteria that naturally occur in the intestinal tracts of humans and warm-
blooded animals to help the body synthesize vitamins. A particularly dangerous type is
the enterohemorrhagic E. coli 0157:H7 or EHEC. In 2000, EI-IEC was the etiological
agent in 69 confirmed outbreaks (twice the number in 1999) involving 1564 people in 26
states (CDC, 2001). Of known vehicles, 69% were attributed to food sources, 11% to
animal contact, 11% to water exposures, and 8% to person-to-person transmission (CDC,
2001). Past outbreaks have also been traced to contaminated well water and improperly
disinfected swimming pools (Keane et al., 1994).

E. coli 0157:H7 produces toxins that damage the lining of the intestine, cause
anemia, stomach cramps and bloody diarrhea, and a serious complication called
hemolytic uremic syndrome (HUS) and thrombotic thrombocytopenic purpura (TTP)
(Doyle et al., 1997). In North America, HUS is the most common cause of acute kidney
failure in children, who are particularly susceptible to this complication. TI'P has a
mortality rate of as high as 50% among the elderly (FDA, 2004). Recent food safety data
indicates that cases of E. coli 0157:H7 are rising in both the US and other industrialized
nations (WHO, 2002).

Human infections with E. coli 0157:H7 have been traced back to individuals
having direct contact with food in situations involving food handling or food preparation.
In addition to human contamination, E. coli 0157:H7 may be introduced into food
through meat grinders, knives, cutting blocks and storage containers. Regardless of
source, E. coli 0157:H7 has been traced to a number of food products including meat and

meat products, apple juice or cider, milk, alfalfa sprouts, unpasteurized fruit juices, dry-

cured salami, lettuce, game meat, and cheese curds (Doyle et al., 1997; FDA, 2001).
Possible points of entry into the food supply chain include naturally occurring sources
from wild animals and ecosystems, infected livestock, contaminated processing
operations, and unsanitary food preparation practices, as illustrated in Figure 1.1.

The US government has significantly expanded its investment to ensure food safety.
One example is the implementation of a food safety initiative to help detect and respond
to outbreaks of foodborne illness (HHS, 2000). Key components of this initiative include
construction of the national Early Warning System, development of new methods for
monitoring the food supply, and improving awareness of safe food practices. Through
this initiative, the Food and Drug Administration (FDA), the Environmental Protection
Agency (EPA), the CDC, and the USDA are working to increase research in the
development of devices used for assessing the risk of the food supply. An example of
this research is the development of biosensors for quickly detecting bacterial

contamination in food.

1.6 Use of Biosensors and Rapid Detection Methods

The detection and identification of foodborne pathogens and other contaminants in
raw food materials, food products, processing and assembly lines, hospitals, ports of
entry, and drinking water supplies continue to rely on conventional culturing techniques.
Conventional methods involve enriching the sample and performing various media-based
metabolic tests (agar plates or slants). These are elaborate and typically require 2—7 days

to obtain results (FDA, 2000)-

.355on 0M3 mcowofiwm 05888 Mo :oumcgucoo L8 850m 95 33:88 953% £20 3&3. poem we ouafionom I: 95min

 

EEO bamsm coon

An increased demand for high-throughput screening, especially in the clinical and
pharmaceutical industries, has produced several technological developments for detecting
biomolecules. Some of these emerging technologies include enzyme linked
immunosorbent assay (ELISA), polymerase chain reaction (PCR) and hybridization, flow
cytometry, molecular cantilevers, matrix-assisted laser desorption/ionization,
immunomagnetics, artificial membranes, and spectroscopy (Food Manufacturing
Coalition, 1997). Pathogen detection utilizing ELISA methods for determining and
quantifying pathogens in food have been well established (Cohn, 1998). The PCR
method is extremely sensitive but requires pure samples and hours of processing along
with expertise in molecular biology (Meng et al., 1996, Sperveslage et al., 1996). Flow
cytometry is another highly effective means for rapid analysis of individual cells at rates
up to 1000 cells/sec (McClelland and Pinder, 1994), however, it has been used almost
exclusively for eukaryotic cells. These detection methods are relevant for laboratory use
but cannot adequately serve the needs of health practitioners and monitoring agencies in
the field. These systems are costly, require specialized training, have complicated
processing steps in order to culture or extract the pathogen from food samples, and are
time consuming. In comparison, a field-ready biosensor is inexpensive, easy to use,
portable and provides results in minutes.

Biosensors are analytical instruments possessing a bio-molecule as a reactive surface
in close proximity to a transducer, which converts the binding of an analyte to the
capturing bio-molecule into a measurable signal (Turner et al., 1978; D' Souza, 2001).

They often operate in a reagentless process enabling the creation of user friendly and

field ready devices. Biosensors are needed to quickly detect disease-causing agents in
food and water in order to ensure continued safety of the nation's food supply.

At the moment, biosensors that have been developed for the detection of pathogenic
bacteria in food and water may be classified into two groups: optical and electrochemical.
An integrated optical interferometer was developed for detecting S. Typhimurium in 10
minutes (Seo et al., 1999). The sensor involved the use of a planar waveguide with
antibody coated channels to make the channels immunochemically selective for antigen
molecules. The presence of antigen was detected by measuring the phase shift generated
by a change in the waveguide refractive index. The sensor was able to detect bacteria
concentrations of 105-107 CFU/mL in a sample flow rate of 50pIJmin. An optical
biosensor utilizing a fiber optic light guide and luminometer was used to detect E. coli
0157:H7 and Salmonella in inoculated samples of ground beef and fresh vegetables in a
time of 1 hour (Liu et al., 2003; Mathew and Alocilja, 2004). The sensor was based on
light (chemiluminescence) released by the reaction of HRP-labeled antibody-antigen
complexes and chemiluminescent reagents. The sensor was able to detect concentrations
of 102 to 105 CFU in a total sample size of 50uL. A surface plasmon resonance biosensor
was reported to detect E. coli 0157:H7 in meat and environmental samples (Meeusen et
al., 2003; DeMarco and Lim, 2002). The SPR biosensor worked by detecting the change
in refractive index of surfaces functionalized with antigen specific receptors. The SPR
sensor was able to detect concentrations of 105 CFU/mL in a sample size of lmL. A
portable evanescent wave fiber-optic biosensor was used to detect E. coli 0157:H7 in
samples of ground beef in 25 minutes and a concentration as low as 102 CFU/mL (Bao et

al., 1996).

10

Electrochemical biosensors, include amperometric, conductometric, impedimetric,
(and in some cases resonant, surface acoustic wave (SAW) and capacitive sensors).
These systems have the advantage of being highly sensitive, rapid, inexpensive and are
highly amenable towards microfabrication (Gau et al., 2001; Rishpon and Ivnitski, 1997).
They measure the change in electrical properties of electrode structures as cells become
entrapped or immobilized on or near the electrode. A flow injection amperometric
immunofiltration system was developed for the detection of E. coli and Salmonella in a
time of 35 minutes (Abdel—Hamid et al., 1998). That biosensor involved the use of
functionalized porous nylon membranes for antigen immobilization followed by
measuring the change in current of a working electrode. The sensitivity of the device was
50 CFU/mL for a sample volume of lmL. An enzyme linked amperometic
immunosensor was developed for the detection of Salmonella in a time of 4 hours
(Brooks et al., 1992). The biosensor measured the change in current of a platinum
working electrode functionalized with polyclonal antibody. The sensitivity of the device
was 104 CFU/mL in a sample size of 200uL. Using porous filter membranes, flow-
through conductometric immuno-filtration biosensors were developed for the detection of
E. coli 0157 :H7 in liquid media (Muhammad-Tahir and Alocilja 2003; Sergeyeva, 1996).
An immunoelectrochemical biosensor utilizing immunomagnetic separation was
developed for the detection of S. Typhimurium and E. coli 0157:H7 in chicken carcass
wash water in a time of 2.5 hours (Che et al., 2000). Sampling involved the separation of
antigen via magnetic beads coated with polyclonal antibody. After incubation, the
sample was processed through a flow injection analysis cell for amperometric detection.

The sensitivity of the sample was 103 CFU/mL in a sample flow rate of 500uIJmin. An

11

impedimetric biosensor was developed to detect E. coli 0157:H7 in pure culture in a time
of 5 minutes (Ruan et al., 2002). The biosensor utilized functionalized indium tin oxide
electrodes and detected the impedance change caused by the immobilization of bacteria
on the electrode surface. The sensitivity was 103 CFU/mL in a sample size of lOOuL.
Another impedimetric biosensor was developed for the detection of Trypanosoma cruzi
(causative agent of Chaga's disease) in a time of 1 hour (Diniz et al., 2003). Impedance
measurements were carried out in an electrochemical cell where the adsorption of
antigens caused a change in impedance. The biosensor was able to differentiate between
positive and negative in 20uL samples. A microfabricated amperometic biosensor was
created for E. coli detection in a time of 40 minutes (Gau et al., 2001). The device used
an array of independent square electrodes functionalized with a self assembled monolayer
(SAM) of streptavidin to capture rRNA for the bacteria. The detection system was
sensitive to 103 CFU (without PCR) in a sample size of SuL. A microfabricated
impedimetric biosensor was developed to detect L. monocytogenes in a time of 15
minutes (Gomez et al., 2001). The device involved antibodies bound to a pair of
electrodes enclosed in a rnicrofluidic chamber. The sensor was able to detect fewer than
10 cells in a 6nL volume (105 CFU/mL). The use of wireless electrochemical sensors
was reported for the detection of Bacillus subtilis, E. coli JM109, Pseudomonas putil and
Sachromyces cerevisiae (Ong et al., 2002). The resonant device involved the use of a
printed inductor-capacitor circuit placed in a liquid sample containing bacteria. The
concentration of bacteria present in the sample could be monitored by measuring the
system resonant frequency. The biosensor was not selective of bacteria species. SAW

devices and magnetoelastic thin film sensors were also been used to detect target analytes

12

remotely (Dutra et al., 2000; Grimes et al., 1999). These sensors involved detecting the
change in system resonant frequency in response to the presence of the analyte.
Interdigitated electrodes were used to detect the presence and measure the concentration
of a target analyte in fluids (Sergeyeva et al., 1996). Another impedimetric biosensor
was used successfully to detect the presence of glucose oxidase binding to interdigitated
electrodes deposited on silicon oxide (SiOz) surfaces (Van Gerwen et al., 1998). It was
demonstrated the possibility to detect urea concentrations as low as SOuM on
interdigitated electrodes immobilized with urease (Sheppard et al., 1995).

Sensors deve10ped for detecting bacteria and enzymes with interdigital electrode
arrays have had electrode widths and spacing ranging from 15pm to 80pm (Gomez et al.,
2001; Sergeyeva et al., 1996; Sheppard et al., 1995). This has the effect of detecting not
only the impedance change at the sensor surface, but also the environmental events taking
place significantly above the surface binding events. By using a novel electrode width
and spacing of 311m by 4pm, respectively, the sensor should be able to detect only the
event of bacteria binding to the surface, thus minimizing other events occurring at 10pm
or greater above the surface.

In general, biosensors experience difficulties detecting low levels of bacteria due
partly to the sample size. Interference of the food matrix represents a major challenge
when developing sensor systems. A novel biosensor is needed that is able to detect
bacteria in a liquid-food sample that requires no enrichment, minimal sample processing

and low environmental interference from food particulates.

l3

1.7 Novelty of Research

The biosensor to be presented in this dissertation is novel both in design and
application. The novelty of the design is that it is the first biosensor to incorporate a high
density, interdigitated microelectrode array to detect bacteria in solution by measuring the
impedance of cells bound to the sensor surface through antibody-antigen interaction.
Table 4 demonstrates the novelty of the biosensor versus similar existing biosensors,
which have not detected whole cell bacteria on interdigitated electrodes in a large sample
size. Furthermore, the electrode width and spacing (3pm and 4pm, respectively) is
unique only to this design and was selected specifically to detect for micron-sized
bacteria. The application is also novel in that the biosensor is tested in solutions

containing mixed particulates of ground beef, romaine lettuce or bovine feces.

'_l'__able 1.4 Novelty of this research compared to existing electrochemical biosensors

 

 

 

 

 

 

 

 

Biosensor . D . . 1 Sample Target Novelty of
Reference Biosensor I'lp tron S Size Analyte Researchw
High Density . Whole Cell k . _
(Radke, 2004) Microelectrode Array <10mrn 20mL Bacteria im 7 __*__ J
(Gomez, 2001) MEMS Microfluidic Device 15 min 6nL “0'6 CF" Larger.sa‘“1"°
Bacteria Size
Detects Whole
(Gau, 2001) MEMS Square Electrodes 40 min SuL RRNA Cell Bacteria,
Larger Sample
(Van Gerwen, Interdigitated Electrode ____ ____ Enz e Detects Whole
1998) Array ym Cell Bacteria
(Abdel-Hamid, Single Electrode with 35 min lmL Whole Cell Multiple
1998) Porous Nylon Membrane Bacteria Electrodes
(Sergeyeva, Whole Cell Electrodes,
l 99 6) Electrode Array 1 hour lmL Bacteria Sample, Time
. . Whole Cell Multiple
(Brooks, 1992) Single Platinum Electrode ---- ZOOuL Bacteria Electrodes

 

 

 

 

l4

1.8 Hypothesis and Specific Aims
Hypothesis: In this dissertation it is hypothesized that a biosensor incorporating an
interdigital microelectrode array and functionalized for recognition of pathogenic bacteria
in solution can be designed and fabricated.
To demonstrate proof of concept, the following specific aims are identified:
Specific Aim 1: To design and fabricate a biosensor that incorporates an interdigital
microelectrode array and functionalized surface for biological recognition.
Specific Aim 42; To employ the biosensor for detecting the presence of serially diluted E.
coli 0157:H7 bacteria in pure culture in a sample size of 20mL.
Specific Aim 3: To employ the biosensor for distinguishing the target bacteria in a liquid
sample size of 20mL containing mixed microflora.
Specific Aim 4: To provide initial research for employing the biosensor to detect E. coli
0157:H7 bacteria extracted from artificially contaminated samples of ground beef,

bovine feces and romaine lettuce.

15

Chapter 2. Literature Review

2.1 Overview of the Impedimetric Biosensor

This research aims to detail the design, fabrication and testing of an impedimetric
biosensor with MEMS technology integrated with biosensing methods to detect E. coli
0157:H7 cells. MEMS is an enabling technology allowing for micron—sized transducers.
MEMS technology makes it possible for the transducer to be integrated with electronics
and undergo batch fabrication in large quantities. The general objective is to develop a
biosensor to detect for whole bacteria cells in food and water.

A high density interdigital electrode array biosensor chip is used in the experiments
to detect different concentrations of E. coli 0157:H7 in solution. Figure 2.1 is a
schematic depicting the operating principles of the biosensor. First, a microelectrode
array is fabricated on a silicon substrate to serve as the electrical transducer (Figure 2.1a).
Second, the biosensor surface is functionalized by attaching analyte specific antibodies
via crosslinkers to form a biological transducer (Figure 2.1b). Finally, when the
biosensor is tested in solution, target analyte becomes bound to the antibodies
immobilized to the surface (Figure 2.1c). The presence of bacteria on the surface causes

the impedance measured across the electrodes to change. The impedance change can be

09

   
    

   

‘95»

Microelectrode Fabrication Antibody Immobilization Analyte Attachment
(6) (b) (C)

Figure 2.1 Schematic of detection theory (adapted from Gorschliiter et al., 2002).

16

measured and correlated to find the analyte concentration.

The biosensor chip is a thin silicon substrate with interdigitated electrodes and
immobilized antibodies patterned onto the surface. Because the electrodes are
interdigitated, opposing electrodes are connected to voltage sources of different polarity
creating a strong electric field (Figure 2.2a). When the interdigital array is immersed in a
sample solution, the active area is exposed to the bacteria. Surface antibodies
immobilized between the electrodes via heterobifunctional crosslinkers act as tethers,
which serve the purpose of holding the bacteria in place between the electrodes (Figure
2.2b). When bacteria bind to antibodies, a region of 2-4um (size of bacteria) above the
sensor surface becomes modified and the impedance created by the pathogenic bacteria
provides the sensing mechanism (Figure 2.2c). Different cellular concentrations of
bacteria bound to the sensor surface yield different changes in electrical impedance
between the electrodes. During testing, only minimal dissociation occurs between the
covalently bound amine crosslinkers and the silanized glass surface since the testing
solution has a neutral pH (Jung, 2001).

As outlined in chapter 1, there have been many sensors developed that detect the
change in impedance when bacteria are tethered in the vicinity of micro and nano-sized
interdigitated electrodes. Large electrode array (>10um) and small electrode arrays
(<1000nm) sensors have not optimized the electrode width and spacing and its effect on
the electric field near the sensor surface, which is the way the impedance change is
measured. The biosensor described in this research is novel because the width and
spacing of the interdigital electrode array is optimized to maximize the impedance change

at the surface of the electrode array and not throughout the test sample. This allows for

17

the biosensor to detect bacteria in a large, bulk solution with minimal environmental
effects, which is a novel testing method in comparison to some other interdigitated
electrode devices. Detecting for whole cell bacteria offers advantages over PCR, ELISA
and DNA based biosensors because these methods report positive results for samples
with dead or non-viable cells. A whole cell biosensor reports results based on detecting

live, viable bacteria.

 

 

 

 

(a) interdigitated electrode schematic (c) with bound bacteria

Figgre 2.2 Electric field between: (a) interdigital electrode array; (b) cross-section of
interdigital array, (c) cross-section of interdigital array with immobilized bacteria on the
surface (adapted from Van Gerwen et al., 1998).
2.2 Lipid Bilayer Membrane Theory

Cells consist of a lipid bilayer membrane surrounding an intracellular fluid
containing numerous organelles (mitochondrion, nucleus, lysosomes, etc.). The
membrane is the most significant portion of the cell in this research. Biological
membranes are constructed mainly from phospholipids. Phospholipids are molecules
containing long, hydrophobic fatty chains (tails) with a charged, hydrophilic phosphate
group (head) to make one end to be water soluble. The molecules spontaneously orient

themselves creating a self assembled double layer with the hydrophobic tails joining

together. As shown in Figure 2.3, this results in a layer with a hydrophobic fatty interior
and a hydrophilic charged exterior. The phospholipid bilayer membrane is about 10 nm

thick and folds around to enclose the entire cell, keeping cellular material on the inside

and the aqueous environment on the outside.

Aqueous

exterior

; Hydrophilic
E phospholipid
2 heads

fatty tails

Aqueous
interior

 

 

 

Figgre 2.3 Cross-section of lipid bilayer membrane showing charged hydrophilic
phospholipid head groups and hydrophobic fatty acid tail groups aligned to form a cell
membrane (Venable et al., 2000).

The phospholipid bilayer serves as the outer membrane of the bacterial cell and is the
basic structural building block of the cellular membrane. It is constructed so that the cell
may survive in an aqueous environment while maintaining the independent cellular
cytoplasm. While the fatty acid double layer serves well as a boundary layer, it does not
provide much physical strength. Cellular stability is achieved through a network of
proteins (along with cholesterol molecules) both inside and outside the cell membrane. It
is this network of proteins that constitutes the framework, giving the cell its shape and

ability to control motion, transport molecules, and adhere to surfaces.

2.3 Biological Recognition

Perhaps the biggest single difference between chemical and biological sensors is the
use of biological substances in biosensor devices. Biosensors incorporate a biological
recognition element to provide selective targeting for analyte(s) of interest. Biological
recognition elements are molecules that interact with the biochemical phenomena
occurring at the cellular level. The major biomolecules used in biosensor research are
enzymes, antibodies, DNA/RNA, and biomimetic polymers.

Enzymes for use in biosensor applications involve measuring the result of an
enzyme-catalyzed reaction involving the analyte. The enzyme is selected so that the
product of the reaction involves a measurable characteristic, such as change in pH, color
or electrochemical conductivity. Glucose oxidase is one popular enzyme due to the
commercial success of glucose biosensors for measuring glucose in blood, fermentations,
and food processing. One major limitation of the use of enzymes as a biological
recognition element is the long-term stability of the enzyme activity and sensitivity to
changes in pH and temperature.

Oligonucleotide strands are the most specific biological recognition molecules
known. The use of DNA probes for isolating and identifying gene sequences through
hybridization is common in biosensor research. One disadvantage of the use of DNA
biosensors is the long time required for hybridization and the slow binding step of target
oligonucleotides. The primary benefit to DNA/RN A is their high specificity.

Biomimetic molecules are molecular recognition elements engineered to have
synthetic receptors that mimic the receptors found on enzymes or antibodies. They are

typically synthesized from similar chemical molecules or are formed by molecular

20

imprinting. Using polymerization, a synthetic matrix with receptor-like recognition
characteristics can be created to mimic the selectivity of antibodies. Biomimetics is a
relatively new, but promising, field in biosensor research. The subject of this research,
however, is antibody-based biosensors.

Antibodies come in different classes including, IgA, IgD, IgE, IgG and IgM. Of
these, IgG is used almost exclusively in biosensors research and will be the subject of this
study. The IgG antibody is a protein with a molecular weight of about 150,000 Daltons.

A schematic of the IgG antibody showing the structural features is shown in Figure 2.4.

 

 

 

antigen
epitope
Fab Fab
light chaA /
heavy chain F0
S-S bond

 

 

 

Figgrg 2.4 Antibody schematic.

The antibody structure is represented as a "Y" shaped structure with a base and two
branches. It is considered a bifunctional receptor since it has twin-binding sites at the
branches (Fab) of the molecule. The base of the molecule is particularly important for
biosensor research because it allows the antibody to attach to other molecules or directly
on surfaces (Shriver—Lake et al., 1997). This can be accomplished either chemically or

through the use of a crosslinker and is discussed in detail in Section 2.4.

21

The recognition of the analyte and the receptor occurs when the binding site of the
antibody meets with a specific site in the cellular membrane called an antigenic
determinant, or epitope. Epitopes may be found on the cell membrane, the soma or the
flagella of bacteria and are denoted by O, H and F, respectively. For this work, however,
it is mainly important to understand that antibody attachment forms a bridge between the
cellular membrane and the substrate upon which the antibodies are immobilized. The
adhesion mechanism occurs when a cell comes into contact with a surface coated with
antibodies. The antibodies bind to specific epitopes on the cellular membrane and

flagella.

2.4 Membrane Impedance Theory
For any given homogenous conducting material, a bulk property called the resistivity,
p, can be defined as having the dimensions of Q-cm. Given this intrinsic property of the

material, the resistance, R, of any arbitrary shape may be determined (Nilsson and Riedel,

1996):

12:39
A

where A is the cross-sectional area in cm2 and L is the length of the material in cm. Thus,
if we consider the cells to be a cylinder of known length and cross-sectional area, it is
possible to estimate the resistance due to the cytoplasm inside the cell membrane. The
value of resistance for the entire cell will have to be adjusted to include such factors as
the membrane resistance, the availability of ions to pass through the cell membrane, the

action potential across the membrane, and the ionic content of the surrounding solution

22

(Borkholder, 1998). The overall membrane resistivity is large with estimates ranging
from 1MQ-m to lOOGQ-m.

The lipid bilayer membrane also acts as spherical capacitor since it serves as an
insulating layer separating two conducting solutions. The capacitance, C, is determined

by the permittivity of the material, 83, the area of the capacitor, A, and the thickness, d:

_ EREOA
d

 

C

where so is the permittivity of free space (8.85x10'12 C-V'l-m ’1) (Nilsson and Riedel,
1996). For most biological membranes, the total thickness, d, of the lipid bilayer is about
lOnm and results in a membrane capacitance of 0.01pF/um2 (T ien and Ottova, 2000).

When measuring the complex impedance characteristics (determination of both the
resistance and the capacitance), it is important to understand the dispersive behavior of
the cellular membrane. When an electric field is applied to a material, energy in the field
is either lost through heat (resistance) or stored by polarization of the material's
molecules. Polarization refers to the charge accumulation at the surfaces between
materials with different electrical properties within the electric field. The response of a
material to an applied electric field is described by its resistivity and permittivity. As
described above, the resistivity gives a measure of a material's ability to conduct (allow
charge to pass through it), whereas permittivity gives a measure of the polarizability of
the material (to store charge).

For most materials (including biological cells), the permittivity is only constant over
a limited frequency range. Perrnittivity decreases as the signal frequency increases. The

step changes in permittivity are called dispersions and reflect the reduction of

23

polarization at increasing frequencies. Biological materials show large dispersions at low
frequencies, mainly due to interfacial polarization at the cell membrane (Ciureanu et al.,
1997). At higher frequencies, the dispersion (and thus the polarization effect) is
minimized (Figure 2.5).

Cells in solution exhibit three different types of dispersions centered in the audio-,
radio-, and ultra-high frequency (UHF) ranges and are referred to as the a, B, y
dispersions, respectively (Figure 2.6). The a-dispersion, centered in the audio frequency
range (10-10‘ Hz) is mainly due to the polarization of the measuring electrodes and ions .
(pH dependent) of the liquid medium. The y-dispersion occurs at ultra-high frequencies
(GHz) and is mainly due to the polarization of small dipolar species, such as water
molecules. As a result, the y—dispersion range is not selective for cells in solution. The B-
dispersion range (10‘-107 Hz), on the other hand, is caused by the polarization of the
bilayer lipid membrane (which also causes membrane capacitance) of the whole bacteria
(Marks and Davey, 1999). Thus, impedance measurements for estimating the amount of

cells in solution is carried out in the lkHz-IOMI-Iz range.

 

Figure 2.5 Schematic of dispersion in a cell at low (a) and high (b) frequencies.

24

 

 

 

 

l ' l I l I r j l
3 _
t: L e ‘
2 . — -
lg - l

4 .—

22 _ a 1
§ 2 i— v -‘
p Y ..

O l r l r l r L r 1

2 4 6 8 10

LOG FREQUENCY (Hz)

Figure 2.6 Spectrum of the dielectric properties of cell suspensions.

2.5 Impedance Measurements

The sensing mechanism for the impedimetric biosensor is based on detecting the
change in impedance due to the presence of E. coli 0157:H7 cells bound to the sensor
surface. The antibodies bound to the surface between electrodes act to capture and
immobilize the cells on the surface of the biochip. The impedimetric biosensor utilizes
electrochemical methods and impedance spectroscopy to detect the target analyte.
Simplified, impedance spectroscopy is a technique used to measure the change in
electrical impedance, Z, over a wide range of signal frequencies.

Impedance is an expression of the amount of opposition an electrical circuit offers to
a flowing current. For alternating currents, there are three impedance elements: resistors,
R, capacitors, C, and inductors, L. The total impedance consists of the sum of the
transient component (inductance and capacitance) and the non-transient component
(resistance). In other words, the impedance due to capacitors and inductors is frequency
dependent, while the impedance due to resistors is constant, regardless of the frequency.

The impedance of a circuit due to capacitors and inductors is referred to as the reactance,

25

X. Inductance, however, is negligible in biological materials, since it is a measure of
energy storage in magnetic fields. For the purposes of this research, the inductance will
not be considered in making impedance measurements.

The impedance caused by the resistance in a circuit is also known simply as
resistance and the impedance caused by the capacitance is known as the reactance. The
magnitude of the impedance, then, can be expressed as the sum of the resistance and

reactance and is equal to:
|z| = JRZ + X 2

and,
X =XC :—

where a) is the angular frequency (co=2nf) of the circuit.

As described in the Section 2.4 above, cell suspensions exhibit dispersion
(polarization) due to cell membrane and cytoplasm biomass in the audio (or-dispersion)
and radio (Ii-dispersion) frequency range of 10-107 Hz. Impedance analysis techniques
can detect small changes in electrical current, on the order of 10'9A, which can be
translated into impedance, conductance, resistance, and capacitance.

The simplest model of a cell is an RC circuit in series where the capacitor and
resistor represent the cell membrane and cytoplasm. The equivalent impedance of the

system can be expressed as:

 

|z|=JRgn+_1__

wzCfim
and the idealized impedance spectrum is shown in Figure 2.7.

26

 

IMPEDANCE SPECTRA

 

’0?
E
.C
2
III
0
Z
<
D
I."
O.
E
100 1000 10000 100000 1000000 10000000
FREQUENCY (Hz)

 

 

 

Figu__re 2.7 Ideal impedance spectrum of RC circuit with reactive and resistive components.

As the signal frequency increases, the transient impedance decreases because of
decreased dispersion of the cell membrane. The dominant impedance element of cells in
solution is the dielectric capacitance and resistance at high frequencies. At low

frequencies, the reactance is the dominant portion of the impedance.

2.6 Circuit Elements

Simulation of the electric field was performed on different electrode widths and
spacing to determine the electric field strength near the sensor surface. The electrical
properties of the bacteria cell structures and testing media are incorporated into the
simulation to determine the optimum electrode width and spacing.

The membrane surrounding the cell has a lipid bilayer structure and is about 4-10nm

thick. The effect of proteins and water on the membrane dielectric constant is unclear but

27

reported permittivity values range typically between 2-10. Low-frequency alternating
current (AC) electric fields induce a large potential drop across the plasma membrane. If
the voltage is too large, dielectric breakdown can occur causing rupturing of the cell
membrane. The experiments in this research use a potential of 50mV to preserve the cell
membrane.

Cellular cytoplasm contains a complex mixture of salts, proteins, nucleic acids and
organelles, which contain individual membrane structures. The value of the inside
permittivity typically has a range of 50-100 (Gimsa et al., 1996). In most cases, however,
the cytoplasm can be approximated as a highly conducting salt solution with a large
concentration of organic material (Markx and Davey, 1999).

When measuring electrical properties in solution, chemical reactions result in the
formation of a space charge layer (layer of ions) near the electrode. Figure 2.8 shows an
illustration of the space charge layer near an electrode and depicts the resulting
capacitance caused by the ion layer. Looking at the negatively charged metal electrode,
the positive ends of the water molecule become aligned in a plane (inner Helmholtz
plane, IHP), forming the hydration sheath. Positively charged hydrated ions then align in
another plane (outer Helmholtz plane, OHP), creating what is referred to as a double-
layer. Under an alternating current, the voltage drop between the metal electrode and the

OHP act as a parallel plate capacitor with a gap of about 1 nm (Kovacs, 1998).

28

   

Ch d -: ® Electrolyte (D
(96%99
I, 1 ® 9 e 9
9 G)
’; ®e®
. 9 ®® @
@® ® 9CD
<98 «96%
©
to $9 ®
..... ® G)
0‘ e
W as:
® %®
es ca @9369
Hydration i \Vfl‘~

sheath IHP OHP Diffuse space charge

® (9

Solvated Water dipole Unsolvated
(+) ion (points to + charge) (-) ion

Figure 2.8 Schematic of the electrode-solution interface showing the inner and outer
Helmholtz planes (Bockris and Reddy, 1970).

The Helmholtz capacitance, often referred to as the double-layer capacitance, CDL,
serves as a constant source of noise in measuring the impedance. The theoretical
capacitance of the double-layer capacitance is given by the equation,

_ £0£RA
x

CDL

where eR is the relative dielectric pemrittivity of the medium between the two planes
(phosphate buffered saline in this case), A is the surface area of the metal electrode array
and x is the distance to the outer Helmholtz plane (a distance of about 10 A). The actual
value of the double layer capacitance is a function of ion concentration, temperature,
surface roughness of the metal electrode among other factors (McAdams et al., 1995).
Another source of noise in the system occurs on the backside of the electrodes and is

referred to as the parasitic capacitance. The gold electrodes are separated from the

29

silicon substrate by a layer of silicon oxide. While the silicon oxide serves as a dielectric
to insulate the electrodes, it also causes a capacitance to occur between the electrodes and
the silicon substrate. In addition to the Helmholtz capacitance, the oxide separation
capacitance is a constant source of noise when measuring the impedance. The parasitic

capacitance for a set of interdigitated electrodes, CPAR, is given by the equation,

flwsp
cos ——
( 2L )

Cm = nleoek
2sin(-”—WSi’-)
2L

where n is the number of electrodes, 1 is the electrode length, L is the sum of the electrode
width and spacing and Wsp is the length of the spacing between electrodes (Van Gerwen
et al., 1998). The relative dielectric permittivity, ER, is for the oxide layer in between the
metal electrode and the silicon substrate.

The circuit diagram used for measuring the impedance of electrodes in solution is
given in Figure 2.9 where CDL is the double layer capacitance between the electrode and
the electrolyte, CD, is the dielectric capacitance of the electrolyte, and R501. is the solution
resistance (Ehret et al., 1997).

——ll a»- ll—
CDL RSOL Cor.
l. e

Figr_1re 2.9 Equivalent circuit of the impedance measurement system with electrodes in
solution (adapted from Ehret et al., 1997).

 

 

 

 

30

The circuit model can be interpreted as having two parallel branches, the dielectric
capacitance branch (C01) and the impedance branch (CDL + Rsor. + C0,). In cases where
the frequency is sufficiently high (>1MHz), the current will tend to run through the
dielectric capacitance of the medium instead of the medium resistance. Therefore, the
dielectric capacitance of the medium dominates the total impedance, and the contribution
of the double layer capacitance and medium resistance to the total impedance is minimal.
At lower frequencies (<lMI-Iz), the current does not flow through the dielectric capacitor
and the effects of the electrode double layer capacitance and the solution resistance
dominate the total impedance.

Cells bound to antibodies immobilized to the biosensor surface add different
impedance elements in series to the impedance branch and a new model is needed. For
the new model, the electric field simulation uses a relative permittivity of 60, 10 and 80
for the cell cytoplasm, cell membrane and testing solution, respectively (W iegand et al.,
2002; Suehiro et al., 2003). The mean length and diameter of E. coli is 2.57pm and
0.49pm, respectively, and has a semi-log normal distribution (Koppes et al., 1978). The
resistivity of the bacterial cell membrane (106 Q-cmz), the resistivity of the bacteria
cytoplasm (ZOOQ-cm) and the capacitance of the lipid bilayer membrane (1 uF-cmz) cause
a change in impedance between interdigitated electrodes (Tien and Ottova, 2000). It
should be noted that the units for the resistivity are different for the cytoplasm and cell
membrane; this is because notation in the field of electrochemistry refers to membrane
resistance in terms of cell surface area, while the cytoplasmic resistance is in the standard

term of a cylinder.

31

Figure 2.10 is a modified circuit diagram for bacteria bound to antibodies
immobilized to the sensor surface. It includes the impedance of bacteria, which consists
of Rcyr, the resistance of the cytoplasm, Rum, the resistance of the cell membrane and
Cam, the capacitance of the cell membrane. The impedance elements also include CDL,
CD], and Rsor. from the original model. Additionally, CpAR, the parasitic capacitance,
represents the capacitance created from the oxide separation of the gold electrodes and

the silicon.

 

 

 

 

 

 

 

Figure 2.10 Circuit model for the impedance of bacteria immobilized between two
interdigitated electrodes.

The equivalent impedance for the circuit in Figure 2.10 is given by the equation,

 

(meg, + RSOLwCQL +1) + 1
aflDl (flwDL + RSOLwDL + 1) + wDL wPAR

where ,6 is a term used to simplify the expression as is equal to:

_ RBIM

_ —-————+R
,/1+(wR,,_M Cm, )2 C"

32

2.7 Electrode Spacing Model

The spacing of the interdigital array is determined by calculating the electric field
near the surface of an interdigitated electrode array. The electric field strength between
electrodes is geometrically dependent on both the width of the spacing between the
electrodes (Wsp) and the electrode width (WEL). The shorter the spacing between the
electrodes, the stronger the electric field will be between electrodes; the larger the
electrodes, the stronger the electric field will extend above the electrode surface. The
effect has been theoretically analyzed (Binns and Lawrenson, 1973; Jacobs et al., 1995)
by calculating the electric field between the interdigitated electrodes and is given by the

equation,

 

V
¢(x. y) = 7M
2K [sin J)
2L

' r r - V
...(aijl
L

Re F asin ,
. (my) 2L
sm —

\ K 7-L J 11

where the electric potential (p changes along the x-y axis, the characteristic length L is the

 

 

 

 

 

 

 

 

L.

sum of both, L=Wsp+ W151. and V is the applied voltage differential between electrodes of

different polarity (+V/2 is applied to the positive electrodes and -Vl2 to the negative
electrode). The electric field geometry is governed by the first order ellipse F(a,B) where
a is the amplitude, 13 is the modular angle and K(k) is the modulus. The theoretical
solution is calculated in Figure 2.11, which shows the curves under which a certain

amount of current is flowing.

33

 

 

21: “ . .1" rev-5 v.‘ 32.
l. ,m r ‘3 . k .
-. ~ 'y'f‘jiii-Mii g,

. - _.4. ‘ _ r ‘) ,x'v

a

. .‘ .. “
”1....r- ,.

80% ' '
70%

 

60% 1 *, " -
500/0 LI . . A' ' "T i j

40% ‘ , ‘. ~ , -_ -,

 

 

 

 

 

0 0.1 02 0.3 0.4 0.5 0.0 0.7 08 0.9 l
Spacing ratio (Wsp/L)

 

Figure 2.11 Calculated electric field lines (a) above interdigital electrodes; (b) as a
function of electrode spacing ratio wsp/L (Van Gerwen et al. 1998; used with permission).

For example, electrode width and spacing of Sum (L=10p.m) results in 80% of the
total electric field flowing in a layer not higher than 5pm above the biosensor surface.
Figure 2.1 lb shows the amount of current flowing in a layer of particular thickness. For
a spacing ratio of 0.5, the layer with a thickness of 0.4L carries 92% of the current. It is
shown here that if the spacing is too small (with respect to the electrode width) there are
very high electric fields at the comers of the electrode. This has the effect of a reduced
current in the middle of the electrode, minimizing the influence of the center electrode
area on overall impedance.

This model points to the advantage of using micron sized electrodes when detecting
for whole cell bacteria. It is expected that this will translate into better sensor sensitivity
as well. It allows for selection of the optimal electrode geometry based on maximizing
the percentage of electric field through a specified layer of thickness. In other words, the
thickness of the analyte (E. coli 0157:H7 bacteria in our case) will determine the optimal

electrode width and spacing of the interdigitated array. This system, however, is more

34

complicated than electrode geometry in space. The effect of the lipid bilayer membrane,
cellular cytoplasm, and testing solution all must be factored into the model. For this,
Finite Element Analysis (FEA) is used to incorporate the permittivities of all elements in
the system, including the silicon substrate, the silicon dioxide, the gold electrodes, the
testing solution, the lipid bilayer membrane, and the cell cytoplasm. The results of the
model are used to design the electrode structure of the biosensor and can be found in the

Results and Discussion section with further detail in Appendix C.

2.8 Immobilization Methods

As mentioned in Section 2.3 (Biological Recognition), the key feature that
distinguishes biosensors from other types of sensors is the use of a biomolecule on the
surface layer of the sensor. The surface layer may be as simple as a biomolecule bound
directly to the surface or, in more complex situations, the biomolecule is attached through
multiple layers of chemical and biological interactions. The transducer must be
chemically modified in order to immobilize the active molecules (active molecules refer
to both chemical and biomolecules) to the surface.

There are five major immobilization methods in biosensor research: covalent binding,
entrapment, cross-linking, adsorption and biological binding. Briefly, covalent binding
refers to the attachment of the active molecule to the transducer surface using a chemical
reaction such as silanization, peptide bond formation, or linkage to activated surface
groups (thiol, epoxy, amino, etc). Entrapment refers to physical trapping of the active
molecule into a thin film or coating, such as a polymer or sol gel (Taylor and Schultz,

1996). Cross-linking is similar to entrapment but only a crosslinker (such as

35

gluteraldehyde) is used to provide a chemical linkage to the active molecule and the
transducer's activated surface, film or coating. Adsorption involves the association of the
active molecule with the transducer's activated surface, film or coating through
hydrophobic, hydrophilic, and/or ionic interactions. Biological binding is the direct
attachment of the active biomolecule to the transducer's activated surface, film or coating
via biochemical binding. Other methods, such as proteins immobilized by adsorption
tend to suffer partial denaturation and leaching off the surface while in solution.
Similarly, capture molecules immobilized through entrapment of surface polymers tend
to attract residues and form multiprotein complexes, the effects of which are likely to
interfere with the antibody function.

For this research, a combination of covalent binding and cross-linking methods are
used to attach the antibody to the biosensor. (An argument can be made that biological
binding also occurs as the bacteria bind to the immobilized antibody, but this section
serves only to review immobilization methods of biomolecules and not the analyte
attachment itself.) For this research, the general method used to attach the antibody is to
first activate the oxide surface of the sensor through silanization. The activated surface

allows for hydroxyl groups on the silica surface to serve as binding sites for the covalent

?CH3 0 o o
“ti-w: + W;) +
00“a o

Silanized Crosslinker IgG Antibody
Surface

 

Figgre 2.12 Antibody immobilization method.

36

attachment of organic molecules. One end of the crosslinker is covalently bound to the
hydroxyl group of the silanized surface while the other end is reactive with the antibody
(Figure 2.12). When selecting crosslinkers for use in biosensor applications, it is
important to note that the crosslinker is reactive with the base (Fc region) of the antibody
and that the antibody binding sites (Fab regions) are facing outward, enhancing the
binding capability of the antibody. This is the preferred method of antibody
immobilization when the substrate is subjected to fluid flow or extended times in
solution. The antibody immobilization procedure used in this research is outlined in

detail in the Methods and Materials section.

2.9 Microfabrication
Electrochemical biosensors involving microfabricated electrodes are an integral
part of biosensor research. Microfabrication is based on electronic integrated circuit (IC)
and thin-film manufacturing methods. The processes used in microfabricated devices
include photolithography, wet and dry etching, wet and dry oxidation, doping, physical
vapor deposition, chemical vapor deposition, evaporation and sputtering among others.
There are too many processes to discuss here but comprehensive reviews can be found in
reference textbooks (Kovacs, 1998; Madou, 2002; Van Zant, 2000) used to teach
graduate and undergraduate microfabrication classes. The main processes utilized in this
research are photolithography and evaporation and are discussed below. Figure 2.13
shows the lift-off fabrication sequence used in this study.
Microelectronic fabrication begins with photolithography, the technique used to

transfer copies of a master pattern onto the surface of a solid material, usually a

37

semiconductor such as silicon. The photolithography process involves the use of a
photomask to block ultraviolet radiation directed at a substrate coated with photoresist.
The photomask is a glass or quartz plate with a master pattern made from a 0.1 pm thick
layer of chromium, which serves to absorb the UV radiation while the glass or quartz is
transparent. The photomask is placed in direct contact (hard contact) with a photoresist
coated surface and then exposed to UV radiation for a prescribed time. This results in a
1:1 image transfer of the entire mask onto the photoresist.

The first step in photolithography, when using Si as a substrate, is to grow a thin layer
of oxide on the surface. This can be accomplished by either dry or wet oxidation of the
wafer at temperatures in the range of 900-1150°C. After oxide growth, a thin layer of
organic polymer, sensitive to ultraviolet radiation, is deposited on the surface. Before the
coated wafers are exposed to the UV radiation they undergo a mild bake (soft bake) in
order to remove the solvent of the resist and to anneal, reducing surface stress. Once they
are soft baked, they then are transferred to a mask aligner where they are exposed to UV
radiation. Resist exposure is controlled by using the proper intensity, direction, exposure
time and wavelength, which is typically in the near UV range of 350-500nm. The mask
serves to both block UV light and to pass UV light where desired onto the resist coated
wafers. Resist that is exposed to the UV light undergoes a chemical reaction making it
susceptible to development in an organic solvent.

Development transforms the pattern of the mask onto the resist, where it serves to
mask further downstream processes. Development occurs by immersing the wafer in a
solvent selective for chemically modified (by the UV) resist. This is also known as wet

resist stripping and results in a wafer with a pattern of bare and photoresist (PR) covered

38

 

     

oxide isolation layer
silicon substrate

photoresist layer

UV light

patterned chromium
masks UV light

 

 

 

 

 

 

 

 

 

 

 

patterned resist

metal layer

patterned metal

 

 

 

F'gure 2.13 The sequence of the typical lift-off process.

39

oxide on the wafer surface. After PR development, the wafer is baked again (hard bake)
at a high temperature to harden the PR and increase the adhesion of the PR to the wafer.

For this research, the next step after development and hard bake is the thin film
metallization of the wafer surface. Deposition is accomplished by the thermal
evaporation of metals, such as platinum, gold, silver, aluminum, copper, titanium, and
chromium, among others. Electrochemical biosensor electrodes are generally made of
noble metals (platinum, silver, gold) because their catalytic properties make them
compatible with biological materials. Indium tin oxide, palladium and iridium have also
been used as electrodes in electrochemical biosensors. Depositing a thin film of metal
below the selected metallic film can enhance the adhesion of metals to the substrate. For
example, the deposition of gold electrodes on oxide can be enhanced by first depositing a
thin layer of titanium onto the substrate.

Etching away the photoresist is completed by a process known as ‘lift-off’ and results
in patterning the metal. Because the metal is deposited over the entire wafer, there is
metal on both the resist and the exposed substrate. The removal of the metal covered
photoresist is accomplished with wet chemical etching selective to the resist. The result
is that the metal on top of the photoresist is dissolved away while the metal pattern on the
oxide is left intact. After lift-off, the wafer is inspected and subjected to a variety of post
fabrication processing including wafer dicing, packaging, and surface functionalization
for further modification. A detailed procedure of the microfabrication process is included

in the Bill of Process in Appendix B.

40

Chapter 3. Methods and Materials

The following steps and experiments were conducted for this research:
1. Biosensor Fabrication (Specific Aim 1)
a. Microfabrication of Biosensor Device
b. Functionalization of Biosensor Device
3. Validation Testing (Specific Aim 2)
a. Pure culture of non-pathogenic generic E. coli
b. Pure culture of pathogenic E. coli 0157:H7
4. Specificity Testing (Specific Aim 3)
a. Pure culture of S. infantis

b. Mixed culture of E. coli 0157:H7 and S. infantis

41

3.1 Microfabrication of the Biosensor

The biosensor was fabricated from 4" (100) p-type silicon wafers, thickness 500-
550m. The wafers were supplied with a 2pm thick layer of thermal oxide grown over
the silicon to serve as an insulator between the electrodes and the substrate. Prior to
fabrication, the wafer was polished to create a smooth surface. The polished wafers were
cleaned in isopropyl alcohol (Spectrum Chemical; New Brunswick, NJ) and dried under a
stream of nitrogen gas inside a glove box.

Photolithography was used to pattern 81805 photoresist (PR), which was purchased
from Shipley (Marlboro, MA). Photolithographic patterning was accomplished by first
placing the clean wafer onto the resist spinner and pipetting 800uL of PR onto the center
of the wafer. The resist spinner was then engaged to spin at 4000rpm for 30 seconds
resulting in a PR coating thickness of about 500nm. After spinning, the wafer was
visually examined to ensure that uniform coverage of PR over the wafer surface was
achieved. Uniform coverage is important because missing PR will result in the absence
of the electrode pattern in the final device. After inspection of PR coverage, the wafer
was placed in an oven at 90°C for 45 minutes to soft bake the PR to ensure dryness, a
requirement prior to UV exposure.

After the soft bake was completed, the wafer was taken to a mask aligner (AB-M, San
Jose, CA). The photomask (Adtek Photomask, Montreal, QC) was placed into the mask
aligner and centered over the wafer. The mask aligner was then engaged to expose the
wafer to 2.2 seconds of UV light at 440nm. The exposure time of 2.2 seconds was based
on the time required for the UV radiation to penetrate and react with a 500nm thick

coating of PR. Different photoresist materials and thickness may require different

42

exposure times. After exposure, the wafer was placed in a PR developer solution for 1
minute to dissolve away exposed areas and then dried under a stream of nitrogen gas.
Next, the PR mask was inspected by a metallurgical microscope to ensure the pattern was
developed properly. After inspection, the wafer was placed in an oven at 135°C for 1
hour to hard bake the PR mask.

After the hard bake, metal deposition occurred in an Edwards Auto306 thermal
evaporator (BOC Edwards, West Sussex, UK). The wafer, along with titanium and gold
pellets, were loaded into the evaporator. The evaporator was pumped down to a pressure
of 4 x 10’6 Pa and a current (2.2A for titanium and 1.6A for gold) was applied to
evaporate the metal in the chamber. First, a 3-5nm layer of titanium was deposited to
ensure strong adhesion of the metal to the silicon oxide surface. This was followed by
the deposition of a 50nm layer of gold. The metallization process deposited metal over
the entire wafer, which had a PR mask on the surface.

After metal deposition, a lift-off process was used to form the MEMS electrode
arrays. The wafer was immersed into a crystallizing dish filled with acetone (J .T. Baker;
Phillipsburg, NJ). The dish was sonicated for 2 minutes to dissolve the PR mask. Metal
was removed with the PR resulting in the lift-off of patterned metal areas. After
sonication, the wafer was cleaned in isopropyl alcohol and distilled water and dried under
a stream of nitrogen. After lift-off, the wafer was inspected with a metallurgical
microscope, an atomic force microscope, and a surface profilometer to inspect the quality
of the electrode array.

After lift-off, the wafer was diced into 68 individual 12mm x 8mm dies for use as

biosensors. First, a coating of PR was applied to protect the surface from the harsh

43

environments of the wafer dicing band saw. (The protective PR coating was applied
using the resist spinner in the same process described above.) The dicing band saw was
engaged to dice the wafer dies to a depth of 450um allowing for individual dies to be
broken apart by hand. At this stage, microfabrication of the sensor chips was completed.

The complete bill of process for microfabrication can be found in Appendix B.

3.2 Functionalizing the Sensor Surface

After microfabrication of the electrode array in the cleanroom, the chip surface was
functionalized for the attachment of the antibody. First, the chips were immersed in
acetone in a crystallizing dish to dissolve away the protective PR layer. The chips were
then cleaned in a mixture of methanol (Sigma; St. Louis, MS) and hydrochloric acid
(CCI; Columbus, WI) for 30 minutes followed by immersion in boiling distilled water for
30 minutes. The chip surfaces were allowed to air dry completely. The cleaning and
drying of the chips allowed for a fresh, activated surface for silanization.

Silanization of the clean chip surfaces occurred in an anaerobic glove box (Coy;
Lansing, MI). Inside the glove box, the chips were immersed in a crystallizing dish
containing a solution of [3-Mercaptopropyl] trimethyloxysilane (MT S) for 2 hours
(Sigma; St. Louis, MS). The MTS solution was diluted in toluene to a concentration of
2%. The silanizing agent MT S is shown in Figure 3.1. The chips were then rinsed in
toluene and allowed to dry completely in a glove box under anaerobic conditions. The

chips were removed from the glove box.

OCH3
l
CH30 — st —CH2CHQCHQSH
OCH3

Figure 3.1 (3-Mercaptopropyl) trimethyloxysilane (MT S) used for silanization (courtesy
of Sigma-Aldrich).

After silanization, crosslinkers were added to the sensor surface. The crosslinker
used was N-y-maleimidobutyryloxy succinimide ester (GMBS) (Sigma; St. Louis, MA)
dissolved in dimethylforrnamide (DMF) (Spectrum; New Brunswick, NJ). The solution
was made from 25mg of GMBS dissolved in a minimum amount of DMF and diluted to
2mM in ethanol (Pharmco; Brookfield, CT). The chemical structure of the crosslinker
GMBS is shown in Figure 3.2. Enough crosslinker solution was pipetted to cover the
electrode array of each individual die and left for 1 hour. After crosslinking, the

biosensor was rinsed in phosphate buffered saline (PBS, pH 7.4).

0
ll
N—o—c—CH2CH2c1-i2 —N |

O 0

F'ggre 3.2 N-y-maleimidobutyryloxy succinirrride ester (GMBS) used for crosslinking
(courtesy of Si gma-Aldrich).

After application of the crosslinker, polyclonal antibody was immobilized to the
sensor surface. Purified polyclonal antibodies specific to generic E. coli and E. coli
0157:H7 were used in this research. The purified goat IgG for E. coli 0157:H7

(Kirkegaard & Perry Laboratories; Gaithersburg, MA) has low cross-reactivity to other E.

45

 

OCH3
l
=OH + ,o-sr-crgcrrzcmsu

OCH3
oicu3
O

MTS
Crosslinker
ecu3 q 0 H N
- o-sli ‘CHQ QN—O—C—CH2CH2CH2 -N@ + %
OCH“ 0 0 I G

Silane

 

7*:
if/

\\\\\V\\\\\\\\\\

g

[1

II
0

 

 

 

 

Figgre 3.3 Process of antibody attachment to the silicon oxide (adapted from Bhatia
et al., 1989).

46

coli. The polyclonal antibody was rehydrated and diluted with PBS to a concentration of
lSOug/mL and 250,11. of it was placed on each individual crosslinker coated sensor chip.
The chips were placed in a petri dish, sealed with parafilm (Pechiney, Chicago, IL) and
allowed to incubate at 37°C for 1 hour. An antibody concentration of 150ug/mL was
used based on the results reported in the antibody immobilization procedure (Bhatia et
al., 1989). After incubation, the surface was rinsed with PBS (pH 7.4) and allowed to air
dry. At this stage, the biosensor (or chip) preparation was completed. The biosensors
were then refrigerated at 4°C until needed for sample testing. The antibody
immobilization process (Shriver-Lake et al., 1997) is outlined in Figure 3.3. A complete
bill of process for surface functionalization and formulations for the reagents used can be

found in Appendix B.

3.3 Validation of the Biosensor

Specific Aim 2 reads: To employ the biosensor for detecting the presence of serially
diluted bacteria in pure culture in a sample size of 20mL.

To validate the biosensor against pure culture, separate experiments were conducted
using non-pathogenic E. coli (ATCC#25922) and pathogenic E. coli 0157:H7
(ATCC#43895) obtained from the Biosystems Engineering collection, Michigan State
University.

Nutrient broth (Difco; Detroit, MI) was used for bacterial enrichment. A 10uL loop
of each bacteria isolate was cultured in 10mL of Nutrient Broth and incubated for 24
hours at 37°C to make a stock culture. The stock culture was serially diluted in 0.1% of

peptone water (Sigma; St. Louis, MS) to obtain logarithmic concentrations of the

47

organism from 100 CFU/mL to 107 CFU/mL. The different concentrations were added to
PBS (pH 7.4) creating test samples of 20mL with concentrations ranging from 100
CFU/mL to 107 CFU/mL. The sample concentrations were then determined by the
standard plating method according to the FDA Bacteriological Analytical Manual (FDA,
1998). For generic E. coli confirmation, MacConkey agar (Difco, Detroit, MI) was used
to plate lOOuL of each serial dilution. The colonies were counted after 24 hours of
incubation at 37°C. For E. coli 0157:H7 confirmation, Sorbitol MacConkey agar (Difco,
Detroit, MI) was used to plate lOOuL of each serial dilution. (E. coli 0157:H7 can be
distinguished by its inability to ferment sorbitol.) The colonies were counted after 24

hours of incubation at 37°C.

 

(a) (b)

 

test fixture

biochip

sample

 

 

 

 

Figure 3.4 Biosensor test apparatus in (a) raised and (b) lowered position.

48

After sample preparation, a biosensor was obtained from refrigerated storage and
inserted into the apparatus test fixture. The biosensor was then lowered into the sample
so that the electrode array was immersed into solution. Figure 3.4 shows the biosensor
test apparatus and fixture in raised and lowered position (Appendix B contains detailed
drawings of the biosensor apparatus and test fixture.) After being lowered into solution,
the biosensor was left for 5 minutes to allow antigen to bind to the biosensor. The

impedance signal was then measured.

3.4 Signal Measurement

A potential of 50mV was applied across the electrodes with a 0V bias and the
impedance magnitude of the biosensor was measured from lOOHz-lOMHz with an HP
4192A Impedance Analyzer (Agilent Technologies, Palo Alto, CA), while the biosensor
was immersed in the sample. For each frequency step, the impedance and phase were
measured at 120 different frequencies (20 logarithmically spaced frequencies per decade)
over a range of lOOHz to lOMHz. The data was acquired through a general purpose
interface bus (GPIB) connecting the HP 4192A to a computer. LabVIEW 6.1 was used
to write the program controlling the data acquisition device. (A general circuit diagram
and snapshot of the LabVIEW user interface can be found in Appendix B.) The sample

testing process, including data acquisition, required 10 minutes.

3.5 Statistical Analysis
Three replications of each serially diluted concentration (100 CFU/mL to 107
CFU/mL) were tested against a sterile blank solution to assess the sensitivity of the assay.

The randomized trials were performed for several consecutive days to nullify the effect of

49

daily variation in bacterial counts. All biosensors were assumed to have the same
physical properties. The means and standard deviations of the impedance magnitude and
phase angle were calculated for frequencies between lOOHz-lOMHz. The significance of
the differences between means was determined and analyzed based on a one-way
ANOVA followed by a Tukey's W test to a significance of 95% (p < 0.05). The lower
detection limit of the biosensor (sensitivity) was determined as the lowest bacteria

concentration with a mean impedance value significantly different from the blank.

3.6 Specificity Study

Specific Aim 3 reads: To employ the biosensor for distinguishing the target bacteria
in a liquid sample of mixed rnicroflora.

For the specificity testing, characterized strains of pathogenic E. coli 0157:H7
(ATCC #43895) and S. infantis (ATCC #51741) were obtained from the Biosystems
Engineering collection, Michigan State University. Nutrient broth (Difco; Detroit, MI)
was used for bacterial enrichment. A lOuL loop of each bacteria isolate was cultured in
10mL of nutrient broth and incubated for 24 hours at 37°C to make stock cultures of each
bacteria strain. The stock cultures were serially diluted in 0.1% of peptone water (Sigma;
St. Louis, MS) to obtain logarithmic concentrations of the organisms from 100 CFU/mL
to 107 CFU/mL. The different concentrations were added to PBS (pH 7.4) creating test
samples of 20mL with both E. coli 0157:H7 and S. infantis concentrations ranging from
100 CFU/mL to 107 CFU/mL. The sample concentrations were then determined by the
standard plating method according to the FDA Bacteriological Analytical Manual (FDA,

1998). For E. coli 0157:H7, Sorbitol MacConkey agar (Difco, Detroit, MI) was used to

50

plate lOOuL of each serial dilution. The colonies were counted after 24 hours of
incubation at 37°C. For S. infantis, Bismuth Sulfite agar (Difco; Detroit, MI) was used to
plate lOOuL of each serial dilution.

After sample preparation, a biosensor was obtained from refrigerated storage and
inserted into the apparatus test fixture. The biosensor was then lowered into the sample
so that the electrode array was immersed into the solution, as shown previously in Figure
3.4. The biosensor was left for 5 minutes to allow antigen to bind to the biosensor
surface. The impedance signal was then measured as described in Section 3.4.

In the specificity study, the sample testing sequence is shown in Table 3.1. Assays of

pure culture of the target bacteria were first conducted to determine the overall specificity

T=able 3.1 Specificity testing matrix

 

 

 

 

 

Test Sample Species Target Antigen Non-Target Antigen

E. coli 0157:H7 . .

I (Pure Culture) Ecol: 0157.H7 -------

S. infantis . .

H (Pure Culture) """" 3' mf‘m’”
E. coli 0157:H7

III and S. infantis E.coli 0157:H7 S. infantis
(Mixed Culture)

 

 

 

 

 

of the biosensor as described in Section 3.1. Test I was on pure culture of the target
antigen E. coli 0157:H7. Test H was on pure culture of S. infantis with no target antigen.
Test 111 was on a mixture of E. coli 0157:H7 and S. infantis to determine the effects of S.
infantis on the ability of the biosensor to detect for E. coli 0157:H7, the target antigen.
The biosensor was functionalized with E. coli 0157:H7 specific antibodies for all three
tests. Assays on dilution series of target and non-target bacteria in the same sample were

conducted to assess the specificity of the biosensor in mixed cultures. All microbial

51

analysis, signal measurement and statistical analysis were conducted as described in

Section 3.4 and 3.5.

3.7 Testing in Complex Media
Experiments were conducted on complex food matrices to explore the performance of
the biosensor in these substrates. The Methods and Materials and Results and Discussion

for the complex media study are included in Appendix D.

3.8 Facilities and Equipment

Fabrication of the biosensor chip was completed in the WM Keck Microfabrication
Facility. The WM Keck Microfabrication Facility, MSU, is a full service clean room for
electronic device fabrication. The clean room has capability for deep UV
photolithography, wafer dicing and bonding, and metal evaporation. All biosensor
testing was conducted in the Biosensors laboratory. The laboratory is equipped to
handle Biosafety Level 2 organisms according to the MSU biosafety standard procedures
(MSU, 1998). All microbial analysis was conducted in the Biosensors Laboratory. All
laboratory and biohazard wastes were labeled, handled, and disposed of according to the
MSU standard procedures for handling bio-hazardous waste (MSU, 1998). The
laboratory is under supervision by the Office of Radiation, Chemical, and Biological

Safety (ORCBS) of MSU.

52

Chapter 4. Results and Discussion

4.1 Simulation Results

Simulation of the electric field was performed for different interdigitated electrode
widths and spacing. The simulations were performed using Maxwell 2D and Maxwell
3D Finite Element Method (FEM) software (Ansoft Corporation; Pittsburgh, PA) over a
variety of width and spacing combinations. The results for lum x 1pm, 6m x 6pm, 5pm
x 10pm and 10pm x Sum arrays are shown in Figure 4.1. These simulation values are
representative of electrode size ranges found in the literature and have been used to
optimize the electrode width and spacing (Gomez et al., 2001; Sergeyeva et al., 1996;
Sheppard et al., 1995; Van Gerwen et al., 1998).

The results show that evenly spaced electrodes cause an evenly distributed electric
field resulting in uniform coverage across the sensor surface. When the electrode width
is larger than the electrode spacing, the electric field tends to favor the area near the
middle of the electrode. When the electrode width is smaller than the electrode spacing,
the electric field is disproportionately larger in the spaces and at the edges of the
electrodes. Furthermore, the electric field is not uniform when width and spacing are
different.

Figure 4.1a shows the simulation results for a 1pm x 1m electrode array. The
electric field distribution is quite uniform across the surface of the sensor. The small
width and spacing of the array results in the electric field being concentrated within 2pm
of the sensor surface. This distance is closer than needed for applications involving

bacteria in the 2pm-4um range. The 6m x 6pm (Figure 4.1b) also shows uniform

53

 

Figure 4.1 Simulation results for different electrode widths and spacing: (a) lum x 1pm
array; (b) 6m x 6pm array; (c) 51m x 10pm array and (d) lOum x 511m array.

54

distribution across the surface. The large size of the array, however, allows for the
electric field to permeate deep into the sample, about 15m above the surface. Contrary
to the 1pm x 1m array, this distance is larger than needed for applications involving
bacteria. The 5m x 10m array (Figure 4.1c) shows the effects of having larger spacing
with respect to the electrode width. The result is that the field is not uniform across the
sensor surface. In fact, the field is disproportionately strong around the electrode and
weaker in the spacing. This is a problem since bacteria bound on or near the electrode
will impact impedance more than bacteria bound to the oxide spacing. The10um x Sum
array (Figure 4.1d) also results in a non-uniform distribution but differs from the 5m x
10m array in that the field is disproportionately concentrated near the spacing. Because
of this, bacteria bound to the oxide spacing will have a larger impact on the measured
impedance than bacteria bound on or near the electrode. The Sum x 10pm and 10m x
Sum arrays both contribute to a large characteristic length (L=Wsp+ WEL) resulting in the
electric field permeating through the solution farther away from the sensor surface.

In selecting for the optimum electrode width and spacing, the goal was for a uniform
electric field with a characteristic length resulting in most of the electric field being
within Sum of the surface. Based on the simulation, the optimum electrode width and
spacing was determined to be 3m and 4pm, respectively, based on a mean E. coli length
of 2.5m and a target of 90% of total electric field strength below a distance of 511m from
the sensor surface. Figure 4.2 shows the simulation results for an applied potential of
lOOmV and interdigitated electrodes with a width of 311m and spacing of 411m. The

resulting electric field distribution is nearly as uniform for configurations with electrodes

55

I . 62 92 .9005

 

 

 

 

 

 

 

 

Figgre 4.2 Electric field simulation of 4m x 3m electrode array: (top) electric field
magnitude; (bottom) vector representation of electric field.

that are evenly spaced. The electric field uniformity is demonstrated by the nearly
horizontal line shown surface field plot of Figure 4.3. The 3m x 4m configuration is
also optimal since the electric field is concentrated within Sum above the sensor surface
as shown in Figure 4.4. For example, if there is a large particle floating at a distance of
10m above the sensor surface the effect on sensor impedance will be minimal since
98% of the electric field is below 10pm. Because the application of the sensor is to test
for bacteria in a complex food matrix (such as ground beef), the change in impedance due

to foreign particles near the sensor surface will need to be minimized.

56

 

Electric Field Distribution of a Distance of 2 microns Above Surface
30000

 

27000
2401)
Elmo
norm
ISCID

IPCII)

Electric Field! V / nl

 

0 IIAAAAAAJIAAALJ‘l‘llLLAAIllllllALALlAL‘luLlALJLIE LLLLLLLLL

 

Distance Along Cross-Section (pm)

 

 

Figure 4.3 Surface plot of the electric field distribution at 2pm from the electrode array.

 

100
90
80

32 70

'2 60

.9-3 50

7330
$20
10

 

O 1 2 3 4 5 6 7 8 9 10
Distance above sensor suface (m’crons)

 

 

 

Figgre 4.4 Percentage of electric field above sensor surface.

57

Table 4.1 summarizes the electric field uniformity, bias, and penetration depth for all

the simulated array sizes.

Table 4.1 Comparison matrix of simulation sizes

 

 

 

 

 

 

 

 

 

 

Array Slze 1pm x 1pm 6pm x 6pm 5pm x 10pm 101m x 5pm 3pm x 4pm
Field Uniformity Uniform Uniform Non Uniform Non Uniform Near Uniform
Field Bias None None Electrode Spacing None

Field Depth 2pm 12pm 12-15pm 12-15um 5-7pm

 

 

The simulation also included immobilized bacteria of different diameters to evaluate
how the electric field distribution will change in the presence of target analyte. Figure
4.5 displays a cross-section of the biosensor suspended in solution with simulated
bacteria immobilized on the surface and shows that most of the electric field is relatively
uniform across the sensor surface. The simulation also shows that the electric field on the
surface permeates through the bacteria. The presence of the immobilized bacteria results
in an increase in the impedance measured across the two electrodes. This simulation
validates the choice for using a 3m x 4m array since the bacteria is well within the
range of the electric field, yet the field does not permeate deeper than necessary into the
solution above the biosensor surface.

After selecting the optimum electrode width and spacing, the next goal was to select
an appropriate active area. The active area was determined based on an estimate of the
surface area required to accommodate as many as 107 CFU of immobilized E. coli cells
on the sensor surface. Assuming the bacteria is lum or smaller in diameter, the minimum
surface area needed is 107ij2 (lOmmz) to capture all the bacteria in a lmL sample.

There is no existing data to support how large the active area of the device needs to be.

58

E(V/-)

l
l
|

y.

E
E
is.

EEV/njrl

 

Figure 4.5 Simulation of 311m x 4m array with immobilized bacteria: (top) electric field
magnitude; (bottom) vector representation of electric field.

 

 

        
 

[[V/nlzz

l . 000021415
9 . I13 Illa-HM
I . 0020904

I . 110110902

 

 

 

 

 

 

Figgre 4.6 Maxwell 3D simulation of the electric field on sensor surface.

59

An area of 10mm2 was selected to allow for adequate binding sites for both lower
concentrations and high concentrations of bacteria.

After the electrode width, spacing, length and active area were selected, a final
simulation was performed to visualize the electric field magnitude in 3D. As shown in
Figure 4.6, the electric field is densely concentrated in the electrode array. This is ideal
because this is the region where the target analyte is expected to be bound to the
biosensor. Furthermore, areas outside of the array will have minimal effect on the
impedance measurement. It should be noted also that there may be some undesired
electric field presence caused by the circuit traces and contact pads, though the effect

again is minimal.

4.2 Fabrication Results

In fabricating the biosensor, the first step needed was to design the pattern needed to
fabricate the photomask for use in photolithography. The photo mask pattern looks
similar to the patterned silicon wafer in Figure 4.7 (right). Figure 4.8 shows the final
CAD layout drawing used to manufacture the photo mask. The final device features 4

rectangular contact pads, with the three pads leading to electrode arrays each measuring

 

Figure 4.7 A 4" Silicon wafer before (left) and after (right) processing.

60

dag 38820 mo :83 9.3.7.8595 3.35 39820 no gnome—o Onwtdoc snob: 92v

£8585 ho MEBSQ Q<O melanoma

 

{0:0 mmcmwmemd so .\.m.o
. ,, _ ;

ll H

 

 

 

 

 

 

 

 

 

 

 

ooodfi

 

 

 

 

61

 

devotee—o we a: 820 cnwteoaoc ”manta secede—o bacon :mE anudoga ”82085 we 85. 0% 92V .3» Bani.

 

 

62

2.5mm x 1mm and leading to circuit traces. The left contact pad leads to a large
metallized area that surrounds the outside of the electrode array for possible use as a
ground electrode (not needed/used in this research). The second and fourth contact pads
terminate to the left and right ends, respectively, of the interdigital array. The third pad
terminates to the middle branch and allows the electrodes to be interdigitated. The
middle branch is set to a different polarity than the left and right branches.

For ease of fabrication, two small electrode arrays were fabricated to reduce the
electrode aspect ratio as compared to one large array. The arrays each measured 6.0mm
x 0.8mm and were selected to achieve the required active area of the sensor (9.6mm2).
Each sensor chip has a total of 1700 electrodes where each electrode had a length of
750nm, a thickness of 30—50nm and a width of 3pm with an in-between spacing of 411m.
After fabrication, each sensor was diced to a dimension of 12mm x 8mm.

During fabrication, the biosensors were inspected in order to validate the physical
properties. A metallurgical rnicrosc0pe was used to examine the electrode array for
complete deposition of metal and to ensure that the high aspect ratio electrodes (250:1)
were within tolerance (Figure 4.9). It was determined that sensors located at the outer
edge of the wafer were often damaged or incomplete. This was to be expected since the
edge of the wafers were in frequent contact with tweezers and chucks as part of the wafer
handling process required for fabrication. Also, when the photoresist (PR) was spun onto
the wafers, the coating was naturally thinner at the edges due to centrifugal force
spreading the liquid over the surface. Of the 68 dies on each wafer, anywhere from 12-16
dies (all at the edges) were not usable. This was expected and quite normal for a non-

automated fabrication facility. The dies in the middle of the wafer were consistently

63

uniform and of high quality. Figure 4.10 is a quality diagram showing where damaged

dies are often located.

  
      
 
  
  
   

    

   

   
  
 
     

{filial—'51":

"1111111111111 'l'li'1'li1'
'1'fll‘l'll'fl'il'lil'l'il'll‘fll'l'll'fll'f
'l'fli'l'll‘l'll'fll'l'l'il'll 1'l'1‘1'll'1'l'1'i
'1'iiil'll'i'l'il'll'l'l'i'fli'l'll'fl'i‘fl'll'
'il'll‘j‘l'il'll 1'll 1'll‘1'l'1'l'1'1'll'1'l'11'
. l*"l'1"1'll'fll'flll"1“ll111'le

MMMM

Figure 4.10 Quality control diagram showing likely locations for damaged dies.

 
 

     
   
 
  

  

1"!
ll"!

 

 

    

Atomic force microscopy (AFM), scanning electron microscopy (SEM) and confocal
scanning laser microscopy (CLSM) were used to validate the fabrication of the biosensor.
AFM was used to measure the length, thickness, width and spacing of the electrode on a
clean sensor surface. AFM images show that the sensor is within tolerance having an
electrode width of 3m, spacing of 411m and a thickness of 35nm (Figure 4.11). The
CLSM was used to validate antibody immobilization and determine where it occurred
(Figure 4.12). CLSM images show that the antibody immobilization is occurring on both
the oxide area between the electrodes and the electrodes themselves. The antibodies have
a natural affinity to the thiol group of the gold electrodes, thus it can be concluded that

the antibody covers the entire electrode array. Actual antibody concentration on the

surface was not determined, though a concentration of lSOuymL was used in the

immobilization procedure.

 

 

 

 

‘4’"

 

Figure 4.11 (left) Atomic Force Microscopy image of the electrodes; (right) Scanning
Electron Microscopy image of bacteria immobilized to the surface.

 

Figure 4.12 Confocal Laser Scanning Microscopy image of electrodes showing antibody
immobilized to electrode and oxide surfaces.

SEM was used to measure the physical properties (length and width) of clean
biosensors, to verify binding of bacteria to antibodies immobilized on the sensor surface
and to observe the surface properties of functionalized biosensors. When measuring for

the electrode length and width, the clean sensors were not silanized, and thus were

65

without a biological sensing surface, and showed the state of the sensor immediately after
the microfabrication process. The biosensors exposed to bacteria were immersed in
solutions containing non—pathogenic E. coli followed by a gentle rinse in PBS. Figure
4.11 shows the binding of non-pathogenic E. coli to the surface of the biosensor.

A comparison was made between clean and functionalized biosensors. In observing
the functionalized biosensors, it was found that modification of the sensor surface with
antibodies caused an irreversible thin film of silanes to build up on the sensor surface.
Though attempts were made to clean the sensor surface after each use, the repeated
treatment of the oxide surface with oxysilanes resulted in a permanent change to the
sensor surface between trials. Figure 4.13 shows the difference between a clean sensor
and a surface modified with crosslinkers and antibodies via silanization. As a result of

this finding, a new biosensor was used for each trial and then was sterilized and discarded

after testing.

 

Figure 4.13 The effects of surface modification: (left) clean biosensor surface; (right)
silanized biosensor surface.

66

4.3 Results and Discussion of Pure Culture Testing
4. 3.1 Frequency Dependence

The change in impedance of the biosensor is directly proportional to the number of
bacteria immobilized on the sensor surface. The impedance spectra for different
concentrations of non-pathogenic E. coli and pathogenic E. coli 0157:H7 bacteria in pure
culture are shown in Figures 4.14 and 4.15. The measured impedance for all
concentrations of bacteria from 100t0107 CFU/mL and the blank for non-pathogenic E.
coli and E. coli 0157:H7 are plotted over a frequency range of 10Hz-10MHz.

The impedance is dependent on both frequency and bacteria concentration. At high
frequencies (>1MHz) there is little change in impedance with respect to bacteria
concentration because the impedance is largely dominated by dielectric capacitance of
the sample media. At these high frequencies, the effect of bacteria bound to the
biosensor surface is minimized due to the relaxation of small dipole species (water
molecules). Also, it is suspected that the impedance caused by the double layer
capacitance (C91) and the parasitic capacitance (CpAR) is minimized at high frequencies
resulting in a convergence of the impedance toward the resistance of the testing solution
(R301). At low frequencies (<1kHz), the difference in impedance is shown to increase
with increasing bacteria concentration. At these low frequencies, the impedance caused
by bacteria is found to increase linearly with the number of cells present in solution.
However, this linear increase in impedance does not register for low cell concentrations
and only begins to take effect at 103 CFU/mL and greater. This is because a solution with
low bacteria concentration (<104CFU/mL) results in fewer bacteria being immobilized on

the biosensor surface while solutions with high cell concentrations result in bacteria

67

..aeo M Snowofienéoc cow N522 9 £3 Soc grape—ammo Sconce: a e8 ooze—Some: v.-1~..v.||=w8 E.

 

 

 

 

 

 

 

 

 

 

 

 

.320 PS x ..EBuo 22 1 ..ESuo «<2 0 ..ESuo mg: a
.3an 28 + ..ESuo ms: 0 ..ESuo PS x ..ESHB E: a
ANIV 5:33.“.
ooooooor oooooop ooooor coco. coca oo. o—
— _ F — b CF
o2
- 82 12;
O
U-
w
S
a 88 F (
A
cocoop

 

 

 

 

oooooo —

 

 

68

.2320 :8 .m as £22 2 is sec 895.5% 5:895 e 8c 85.8%: lwlfle 2: a

 

 

 

 

 

 

..ESuo 92 x aESuo 2o. 1 ..ESuo ma: 0 ..ESao P9 a
..ESuo 22 + ..ESuo ms: 0 ..ESuo 9.2 x ..EBao a: n.
ANIV 3:909...
8882 888, 882 802 82 8F 3
p _ b F L o F

 

..F

i .. .C‘.

 

  

 

 

     

 

 

- . . - . .... 02 .w
$3....,.................. a
...-0. pr ..:
a c on w .u. ... p
. B
U
3
8o F a
)
O
U.
w
3
(
oooo_

 

 

 

ooooo w

 

 

69

covering the sensor surface. In fact, there is no statistically significant detectable
difference in impedance between bacteria concentrations from 100 to 103CFU/mL at a
frequency of lkHz. It is suspected that the polarization effect of the bacteria on the
biosensor surface only begins to change the impedance at concentrations greater than 103
CFU/mL because of the sufficient number of bacteria required to change the impedance
above the base impedance of the biosensor. The inability of the biosensor to detect
impedance changes for low bacteria concentrations may be due to the effect of double
layer capacitance and parasitic capacitance found in the biosensor, which act
independently of whether bacteria are present in solution.

For high frequencies, the current passes through the dielectric capacitance instead of
the cellular impedance and medium resistance (Ehret et al., 1997; Van Gerwen et al.,
1998). Therefore, the dielectric capacitance of the medium dominated the total
impedance, and the cellular impedance, double layer capacitance and medium resistance
can be largely ignored. Since the dielectric capacitance is the only contribution to the
impedance at high frequencies, the impedance value is inversely proportional to the
frequency.

At low frequencies, the effect of cellular impedance is dominant (Yang et al., 2004).
The total impedance has contributions from the double layer capacitance, the cellular
impedance, and the solution resistance. There is a frequency region (lkHz-lMHz),
however, where the impedance is controlled by a combination of all the impedance
elements. The change in double layer capacitance and cellular impedance is more

significant compared to the change in medium resistance, implying that the decrease in

70

impedance value due to the bacterial growth is dominated by the increase in cellular
impedance and double layer capacitance.

Surface chemistry issues related to the silanization process allowed antibodies to bind
to the gold electrodes. While having antibodies on the gold electrodes increases that
amount of bacteria immobilized on the biosensor surface it also increased the base level
of noise in the system, particularly for samples with low bacteria concentrations. At high
bacteria concentrations, increased antibody immobilization to the gold electrodes actually
increases biosensor performance since the high number of bacteria present in the sample
will allow for more binding, translating into a higher impedance measurement. At lower
frequencies, however, the increased antibody immobilization to the gold electrodes
provides more background noise to the system reducing the effect of bacteria on

impedance change at low concentrations.

4.3.2 Effect of Non-pathogenic and Pathogenic Bacteria

For both non-pathogenic and pathogenic species, there is no statistically significant
detectable difference in impedance between the blank and bacteria concentrations from
10° to 103CFU/mL at a frequency of lkHz. This demonstrates that for both generic E.
coli and E. coli 0157:H7, the impedance measured by the biosensor is heavily influenced
by the double layer capacitance and parasitic capacitance at low frequencies. In the case
of Helmholtz capacitance, it is interesting to note that as crosslinkers and antibodies
adsorb to the surface, the effective area available for ion exchange is reduced. And even
more so as the electrodes become covered with bacteria. With increasing frequency, the

presence of bacteria has a decreasing effect on overall biosensor impedance resulting in a

71

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    

 

 

 

 

00000
E 10000
9.
8
C
i
g ‘000
‘00
Bacteria Concentration (CFU/n1.)
[II Ecoli 0157:H7 El Ecoli generic]
11000
A 917 396
E 000
Q
8
C
(U
8
Q on .
E
n .
BLANK pro m1 0‘2 1m 11M ws me 1w
Bacteria Concentration (CFU/m L)
l Ecoli 0157:H7 El Ecoli generic]
10000
A1000.
E
.C
9
E 100
(U
‘U
o
D.
g 10 4
1 . ‘ ‘ ~
BLANK mm mm 10‘2 10«3 mm was we 1047
Bacteria Concentration (CFU/mL)
[lacoii 0157:H7 1:1 Ecoli genencj

 

 

 

Figure 4.16 Comparison of E. coli and E. coli 0157:H7 at selected frequencies: (top)
lkHz; (middle) 100kHz; (bottom) lOMHz.

72

convergence of the impedance. This could be due to the impedance being dominated by
the dielectric behavior of the medium at high frequencies.

Comparing pathogenic bacteria and nonpathogenic bacteria, it is shown that E. coli
0157:H7 yields lower impedance than E. coli. This effect is most pronounced at low
frequencies. Figure 4.16 shows the biosensor responses at lkHz, 100kHz, and lOMHz.
Note that the impedance difference at lkHz is pronounced as the impedance is largely
due to the number of bound bacteria. At IONHIZ, there is little difference between
generic E. coli and E. coli 0157:H7 since the impedance is due to the dielectric
capacitance of the testing medium. High frequency measurements yield little measurable
difference between bacteria concentrations for either species.

It is suspected that the pathogenic bacteria do not bind as well as the nonpathogenic
strain as observed from the impedance measurements. The specificity of their respective
antibodies are different, though non-specific binding of interferant bacteria species with
specific antibodies is not a problem for this study because the samples were artificially
inoculated only with the target organisms and nothing more. Another possible reason for
the difference in impedance between pathogenic and non-pathogenic species is that the
physical properties of the cells are different. For this study (including the electric field
simulation), the properties of the lipid bilayer membrane and cellular cytoplasm were
assumed to be the same for both generic E. coli and E. coli 0157:H7. It may be that the
physical properties of different E. coli species are different and have an effect of the
measured impedance. Currently, there has not been any research regarding the actual
values of the membrane and cellular impedance of different E. coli species. Experiments

to determine specificity will focus on the effects of multiple species present in the testing

73

solution as described in the specificity testing.

After measuring the impedance data against time at low frequencies, it was found that
the impedance increased the longer the biosensor remained in solution. Figure 4.17
shows Cole-Cole plots of the impedance for 0, 15, 30 and 60 minutes after insertion in
the test solution. The increase in impedance could be due to an increased number of
bacteria cells binding to the sensor surface over time. Based on the data, the increased
impedance is most pronounced within the first 20 minutes after insertion into solution.
The Cole-Cole plots show that after 35 minutes, the rate at which new cells bind to the
surface slows down considerably. Also, it is intuitive that given enough elapsed time,
dense packing of bacteria may eventually occur on the surface. By taking impedance
measurements at specific time intervals, the effect of bacteria concentration is shown. At
high frequencies, the effect of cells bound to the biosensor surface have little effect on
impedance thus there is no change in impedance with an increase in time. The attached

bacterial cells acted as impedance elements in series with medium resistance.

 

~25000

 

 

 

 

 

 

Imag Impedance (Ohm)

 

 

 

Real Impedance (Ohm)

 

[ no min A15min +30min 060min ]

 

 

 

 

Figure 4.17 Cole-Cole plot showing the real and imaginary impedance with time.

74

The impedance caused by bacteria was found to increase linearly with the number of
cells present in solution. Figure 4.18 shows SEM (1600X) micrographs of a solution
with a concentration of 102 CFU/mL resulting in few bacteria being immobilized on the
sensor surface and a solution with a concentration of 106 CFU/mL resulting in bacteria
completely covering the sensor surface. The difference in the number of immobilized
bacteria for high and low concentrations supports the impedance results that a high
number of bacteria is required to change the measured impedance above the base
impedance of the biosensor. The membrane resistance of attached bacterial cells affect
the biosensor impedance. These attached cells act as elements connected in series and
block the current flow from the electrodes in a passive way causing the impedance to
increase. The larger the number of attached cells, the larger the magnitude of the
resulting increase in impedance.

For nonpathogenic E. coli in pure culture, the statistical analysis shows that the
biosensor has a lower detection limit of 10S CFU/mL with respect to the blank when
using a log transformation of impedance data at a frequency of lkHz (Figure 4.19). For
pathogenic E. coli 0157:H7 in pure culture, a lower detection limit of 104 CFU/mL is

obtained with respect to the blank.

75

1011111

Figure 4.18 Scanning Electron Microscopy images of bacteria bound to the biosensor
surface: (top) sample containing 102 CFU/mL; (bottom) sample containing 106 CFU/mL.

 

 

 

 

Mean i SD and Significancel
Concentration E. coli E. coli 0157:H7
(CFU/mL) Impedance (log of Impedance (log of

BLANK 2.42 i 0.13 a 2.29 i: 0.32 a
2 x 10° 2.45 i 0.04 a 2.25 i 0.05 a
2 x 10‘ 2.51 :t: 0.32 a 2.32 :I: 0.10 a
2 x 102 2.32 :I: 0.04 a 2.50 :1: 0.33 a

5 x 103 2.52 1. 0.03 a 2.70 i 0.16 a,b

1 x 10‘ 2.73 :l: 0.07 a,b 2.87 :I: 0.02 b,c

1 x 105 3.20 i 0.16 b 3.28 :l: 0.09 c,d
1 x 106 4.40 i 0.16 c 3.50 :l: 0.16 d

1 x 107 4.67 :I: 0.31 c 3.34 i 0.09 c,d

 

 

 

[1] Means with same letter are not significantly different (p>0.05)
[2] Log transform of E. coli impedance data from a frequency of lkHz
[3] Log transform of E. coli 0157:H7 impedance data from a frequency of lkl-lz

Figu_r_e 4.19 Statistical significance of mean differences between concentrations in pure
culture.
4.4 Results and Discussion of Specificity Study

In this study, the biosensor was functionalized for E. coli 0157:H7 but was tested for
the presence of S. infantis in pure culture. The impedance spectra for different
concentrations of S. infantis in pure culture is shown in Figure 4.20. As expected, there
was no significant difference in impedance with respect to different concentrations of S.
infantis bacteria, as shown in the data where the impedance spectra are close together.
Further, the polyclonal antibody has low cross-reactivity with Salmonella spp.
(Kirkegaard and Perry, 1992) and any non-specific binding that did occur was not large
enough to significantly increase the impedance with respect to the blank. This suggests
that S. infantis did not attach to the E. coli 0157:H7 specific antibodies immobilized on
the biosensor surface, demonstrating specificity of the biosensor in the presence of non-
target organisms.

The impedance spectra for a mixed culture of E. coli 0157:H7 and S. infantis is

shown in Figure 4.21. For the mixed culture, the increase in impedance is less

77

.8883 .m. me 85:8 83 a com maze - #83 Bob :oaBEmE meson—cob a com 8588:: omé E

 

 

 

 

 

 

..ESuo ~29 D ..ESuo 9‘9 x ..ESuo m<ow o ..EBuo 39 +
.855 92 4 .853 «<9 0 ..ESuo 22 r ..ESuo 9,9 u
3.: 8:88.“.
8882 882 88. 88’ 83 8F

 

 

 

 

 

 

 

 

8
(sumo) eouEpadrul

g

88..

 

 

78

.2320 :8 .m nee eases a co 23.8 ease e as E22 - N82 88.. 5858 05:8 a as 8:885 8.4 a

 

 

 

 

 

 

 

 

 

 

..ESmO 99. u ...—{Duo For 1 ..EBuO «<9. 0 ..ESuO m<o_. 4
..ESLO gr + ..ESuO m<ow O ..ESuO 96.. X ..ESuO ~29. D
3.: 5:082“.
0000009 000000.. 00003 80°F 0°C.. 8..
. 1 _ e . cop
..........w...w.vubukc+¢+¢+¢+c+chWshcbch¢hsw¢W¢ww». a. a w* 51.1.1111? .
.-....HHooHHmHIIII w
a .. u
c C c c C m
I C 1 gm
.. . . w
p e ......e. IS\
0...... e H e 0 0%"~
II.I.I. .eeee “
I.I.I.I.e e e“"

 

 

 

 

79

 

 

Impedance (Ohm)

 

 

 

‘0‘2 0‘3 TIM 0‘5 07
Bacteria Concentration (CFU/ml.)

I Is. infantis lMxed L1Ecoli0157zl-I7

 

 

 

 

1000

 

Impedance (Ohm)
'8‘

 

 

 

1000 1041 102 1043 10M 10*!) we 1007
Bacteria Concentration (CFU/mL)

 

l
IS. infantis lMixed DEcoli 0157:H7 J

 

 

 

1000

 

Impedance (Ohm)
8
O

 

 

 

10% son 1002 1043 10M ms 10"6
Bacteria Concentration (CFU/mL)

10‘7

 

1 I S. infantis Mixed El E.coIi 0157:H7

1

 

 

 

Fime 4.22 Comparison of E. coli 0157:H7, S. infantis, and mixed culture at selected
frequencies: (top) lkHz; (middle) 100kHz; (bottom) lOMHz.

8O

pronounced than in pure culture, but the impedance spectra are much more spread out
when compared to the S. infantis pure culture. In examining frequency dependence, the
impedance difference between concentrations become significant at 106 CFU/mL for low
frequencies (Figure 4.22). The mixed culture impedance is greater than the S. infantis
impedance whether at low or high frequencies or at low or high bacteria concentrations.
This suggests that in mixed culture samples, target E. coli 0157 :H7 are bound by the
antibodies on the surface. At low frequencies, the mixed culture impedance is higher
than both S. infantis and E. coli 0157:H7. It is suspected that the impedance is higher
than S. infantis due to the presence of E. coli 0157:H7 bound on the surface of the
biosensor. The E. coli 0157:H7 impedance value is less than the value for both mixed
culture and pure culture of S. infantis. This may potentially be due to S. infantis having
different conductive properties than E. coli 0157:H7. While large numbers of S. infantis
did not bind to the biosensor surface, the presence of the bacteria in the solution
potentially changes the value of the solution resistance and the double layer capacitance
of the solution. This would explain the resulting higher impedance values (at low
frequencies) when S. infantis is in solution, both in the presence and absence of E. coli
0157:H7.

At low concentrations of mixed culture, the impedance measurement is significantly
higher than the single species culture of S. infantis. It should be noted that the serially
diluted mixed culture contains twice the number of bacteria than either of the single
species cultures, suggesting non-specific binding in the presence of large amounts of

bacteria.

81

For the specificity study, the statistical analysis shows that the biosensor has a lower
detection limit of 104 CFU/mL when using a log transformation of impedance data at a

frequency of lkHz (Figure 4.23). For mixed culture, a lower detection limit of 10°

 

 

 

CFU/mL is obtained.
Mean i SD and Significancel
Concentration E. coli 0157:H7 Mixed Culture S. infantis
(CFU/mL) Impedance (log S2)2 Impedance (log Q)3 Impedance (log £2)4
2 x 100 2.25 i 0.05 a 3.14 i 0.09 a 3.04 i 0.05 a
2 x 101 2.32 i 0.10 a 3.14 i 0.03 a 3.03 i 0.03 a
2 x 102 2.50 i 0.33 a,b 3.14 i 0.04 a 3.06 i 0.05 a
5 x 103 2.70 i 0.16 b,c 3.12 i 0.11 a 3.03 i 0.01 a
1 x 104 2.87 e 0.02 c 3.23 i 0.05 a,b 3.01 :t 0.02 a
1 x 105 3.28 i 0.09 d 3.25 3: 0.12 a,b 2.95 i 0.10 a
1 x 106 3.50 a. 0.16 d . 3.36 :t 0.08 b 3.10 a: 0.04 a
1x107 3.34i0.09d 3.38 i0.16b 3.10:0.08 a

 

 

 

[1] Means with same letter are not significantly different (p>0.05)

[2] Log transform of E. coli 0157:H7 impedance data from a frequency of lkl-Iz
[3] Log transform of mixed culture impedance data from a frequency of lkHz
[4] Log transform of S. infantis impedance data from a frequency of lkl-lz

Figure 4.23 Statistical significance of mean differences between concentrations for the
specificity study.

Experiments were conducted on complex food matrices to explore the performance of
the biosensor in these substrates. Results are preliminary and are included in Appendix D

for reference.

4.5 Summary and Conclusions
4.5.1 Summary of Lower Detection Limits

In testing the biosensor, the best results occurred under pure culture conditions. The
lower detection limit for detecting E. coli 0157:H7 in pure culture was 104 CFU/m]... In
terms of specificity, the biosensor was specific to E. coli 0157:H7 when functionalized

With E. coli 0157:H7 polyclonal antibodies. When testing for S. infantis with sensors

82

functionalized for E. coli 0157:H7, the biosensor yielded no response between different
S. infantis concentrations suggesting that surface binding of non-specific bacteria did not
occur. In the presence of mixed culture between S. infantis and E. coli 0157:H7, the
biosensor had a lower detection limit of 106 CFU/mL, largely because of the interferant

effects of S. infantis bacteria on the target organism.

4.5.2 Limitations and Future Possibilities

The goal was to develop a portable device to enable health care professionals,
bioterrorism rapid-response teams, and food safety monitoring personnel to quantify
results in less than 10 minutes for both clinical detection and point-of-care use.
Innovation of the biosensor comes in the form of targeting pathogenic bacteria in a large
sample volume whilst requiring only a 10 minute detection time.

The biosensor utilized an antibody concentration of ISOug/mL based on the
procedure used to immobilize antibodies to the biosensor surface (Bhatia et al., 1989;
Shriver—Lake et al., 1997). Increasing the antibody concentration would allow for a
greater number of target organisms to bind to the biosensor at a given concentration.
Increasing the biosensor active area would have the same effect. It may also be possible
to bind different antibody species onto the same biosensor making it into a multi-analyte
detector.

The price of microfabricated devices, as in the semiconductor industry, diminishes to
pennies per biosensor when produced in sufficient numbers. The biosensor platform
would fare well in the market place if commercialized to address the growing need for

food pathogen testing (Alocilja and Radke, 2003).

83

APPENDIX A

Table A.l Reported and estimated cases of foodborne illness by agent type in the US

(Mead, 1998).

 

 

 

Illnesses Hospitalizations Deaths

Disease or Agent Total Foodbome Total Foodbome Total Foodbome
Bacterial

Bacillus cereus 27,360 27,360 8 8 0 O
Botulism, foodborne 58 58 46 46 4 4
Brucella spp. 1,554 777 122 61 11 6
Campylobacter spp. 2,453,926 1,963,141 13,174 10,5 39 124 99
Clostridium perfringens 248,520 248,520 41 41 7 7
Escherichia coli 73,480 62,458 2,168 1,843 61 52
E. coli, non-015 36,740 31,229 1,084 921 30 26
E. coli, enterotoigenic 79,420 55,594 21 15 0 0
E. coli, other diarrheogenic 79,420 23,826 21 6 0 0
Listeria monocytogenes 2,518 2,493 2,322 2,298 504 499
Salmonella Typhi 824 659 618 494 3 3
Salmonella, nontyphoidal 1,412,498 1,341,873 16,430 15,608 582 553
Shigella spp. 448,240 89,648 6,231 1,246 70 14
Staphylococcus food 185,060 185,060 1,753 1,753 2 2
Streptococcus, foodborne 50,920 50,920 358 358 0 0
Vibrio cholerae, toxigenic 54 49 18 17 0 0
V. vulnificus 94 47 86 43 37 18
Vibrio, other 7,880 5,122 99 65 20 13
Y ersinia enterocolitica 96,368 86,731 1,228 1,105 3 2
Subtotal 5,204,934 4,175,565 45,826 36,466 1,458 1,297
Parasitic

Cryptosporidium parvum 300,000 30,000 1,989 199 66 7
Cyclospora cayetanensis 16,264 14,638 17 15 0 0
Giardia lamblia 2,000,000 200,000 5,000 500 10 1
Toxoplasma gondii 225,000 112,500 5,000 2,500 750 375
Trichinella spiralis 52 52 4 4 0 0
Subtotal 2,541,316 357,190 12,010 3,219 827 383
Viral

Norwalk-like virus 23,000,000 9,200,000 50,000 20,000 310 124
Rotavirus 3,900,000 39,000 50,000 500 30 0
Astrovirus 3,900,000 39,000 12,500 125 10 0
Hepatitis A 83,391 4,170 10,841 90 83 4
Subtotal 30,883,391 9,282,170 123,341 21,167 433 129
Grand Total 38,629,641 13,814,924 181,177 60,854 2,718 1,809

 

84

Table A2 Reported and estimated cases of foodborne illness by surveillance type in the
US (Mead, 1998).

 

 

 

Estimated Reported C3568 % Hosipit- Case
Total BY Surveillance Type Foodbome alization Fatality

Disease or Agent C8868 Active Passive Outbreak Origins Rate Rate
Bacterial
Bacillus cereus 27,360 720 72 100 0.006 0.0000
Botulism, foodborne 58 29 100 0.800 0.0769
Brucella spp. 1,554 1 ll 50 0.550 0.0500
Campylobacter spp. 2,453,926 64,577 37,496 146 80 0.102 0.0010
Clostridium perfringens 248,520 6,540 654 100 0.003 0.0005
Escherichia coli 73,480 3,674 2,725 500 85 0.295 0.0083
E. coli, non-015 36,740 1,837 85 0.295 0.0083
E. coli, enterotoigenic 79,420 2,090 209 70 0.005 0.0001
E. coli, diarrheogenic 79,420 2,090 30 0.005 0.0001
Listeria monocytogenes 2,518 1,259 373 99 0.922 0.2000
Salmonella Typhi 824 412 80 0.750 0.0040
Salmonella, nontyphoidal 1,412,498 37,171 37,842 3,640 95 0.221 0.0078
Shigella spp. 448,240 22,412 17,324 1,476 20 0.139 0.0016
Staphylococcus food 185,060 4,870 487 100 0.180 0.0002
Streptococcus, foodborne 50,920 1,340 134 100 0. 133 0.0000
Vibrio cholerae, toxigenic 54 27 90 0.340 0.006
V. vulnificus 94 47 50 0.910 0.3900
Vibrio, other 7,880 393 112 65 0.126 0.0250
Yersinia enterocolitica 96,368 2,536 90 0.242 0.0005
Subtotal 5,204,934
Parasitic
Cryptosporidium parvum 300,000 6,630 2,788 10 0.150 0.005
Cyclospora cayetanensis 16,264 428 98 90 0.020 0.001
Giardia lamblia 2,000,000 107,000 22,907 10 nla nla
Toxoplasma gondii 225,000 15,000 50 nla nla
Trichinella spiralis 52 26 100 0.081 0.003
Subtotal 2,541,316
Viral
Norwalk-like virus 23,000,000 40 n/a nla
Rotavirus 3,900,000 1 nla nla
Astrovirus 3,900,000 1 nla nla
Hepatitis A 83,391 27,797 5 0.130 0.003
Subtotal 30,883,391
Grand Total 38,629,641

 

85

APPENDIX B

Table B.1 Bill of process for rrricrofabrication of the biochip.
Bill Of Process-(Microfabrication)

 

ep #1 Process Descrlptlon

 

1 btain 4" (100) silicon waters from supplier. Wafers should contain 3 2pm thick layer of
hermally grown silicon oxide. The wafer surface should be polished.

 

21W ater cleaned in isopropyl alcohol solution followed by distilled water

 

3|Wafer dried with nitrogen stream

 

4ilNafer placed on chuck of photoresist (PRLSpinner

 

5A volume of 800uL of S1805 PR is pipetted onto the center of the wafer

 

PB spinner is actuated to spin wafer at 4000rpm for 30 seconds

 

 

After spin, wafer is visually examined to ensure complete PR coverage over entire wafer

 

ater transferred to oven for soft bake @ 90°C for 45 minutes

 

9Mater removed from oven

 

10Mater placed in photomask aligner

 

11lPhotomask is aligned over the center of the wafer in the photomask aligner

 

1 Photomask aligner is actuated and wafer is exposed to 2.2 seconds of UV light. The
UV light was set to emit at a wavelength of 440nm

 

13'Water is removed from photomask aligner

 

14Wafer immersed in PR developer solution for a time of 1-2 minutes

 

15Wafer removed from PR developer solution and dried under nitrogen stream

 

1 ater inspected with a metallurgical microscope to examine quality of PR mask

 

17Water is transferred to oven for hard bake @ 135°C for 1 hour

 

 

181W ater removed from oven

 

1 ater loaded into metal evaporation chamber. Evaporation chamber includes titanium
nd gold pellets for use in the evaporation procedure

 

20|Vacuum unit pumped down to 4x10’° Pa and current is applied to evaporate metal

 

21IA 3-5nm thick layer of Titanium is evaporated over entire surface of the water

 

22IA 50nm thick layer of Gold is evaporated over entire surface of the wafer

 

231Vacuum unit is disengaged and wafer is removed

 

2 ater is immersed into crystalliziqu dish filled with acetone

 

25 he crystallizingdish (with water and acetone) is sonicated for 60 seconds for lift-oft

 

26'Wafer is cleaned with distilled wafer followed by isopropyl solution

 

 

27 Water is dried under a stream of nitrogen

 

2 ater surface is inspected by metallurgical microscope, atomic force microscope and a
urface profilometer to ensure proper lift-off was achieved

 

29llf satisfied with water, a protective layer of PR is applied as described in process 4-6

 

SOMafer loaded onto dicigg band saw

 

31|Dicing band saw set to dice water into 68 separate 8 x 12mm dies to a depth of 450nm

 

 

 

32 After dicing, water has been microfabricated and clean room work is complete

 

 

86

Table B.2 Bill of process for surface functionalization of the biochip.
Bill Of Process-(Functionalization)

 

ep if Process Description

 

1 Diced wafer is obtained after microfabrication in the clean room

 

 

2 ater is immersed in acetone to remove protective PR coating

 

3Mafer is cleaned by immersing in a 50:50 mixture of HCI and methanol for 30 minutes

 

afer is immersed into boiling distilled water for 30 minutes

 

5 ater is allowed to air dry completely

 

 

6Mafer is placed in an anaerobic chamber (glove box under inert conditions)

 

7ilnside the glove box, the wafer is immersed in a 2% MDS solution for 2 hours

 

Bilnside the glove box, the wafer is rinsed with dry toluene

 

911' he wafer is removed from glove box

 

1dWafer is immersed in GMBS crosslinking solution for 1 hour

 

11IThe wafer is washed with PBS

 

12IAntionr is carefully pipetted onto each electrode array region

 

13Mafer with antibody is sealed with paratilm and incubated for 1 hour at 37°C

 

1 afer is removed from incubator and rinsed with PBS

 

1 Wafer is placed under refrigerated conditions 014°C until use

 

 

 

1 urface functionalization process is complete

 

 

Table B.3 Bill of process for testing in ground beef samples.

 

Bill Of Process-(Ground Beet Testigg)

 

Step it I Process Description

 

throw bacteria (such as E. coli0157zH7) overnight in nutrient broth (NB) at 37°C

 

2]Weigh and separate 259 samples of ground beef

 

alPlace ground beef samples in sterile stomacher bag

 

4ilnoculate ground beef with 2mL of pure culture

 

SILet sit at 40 under refrigerated conditions for 1 hour

 

6|Add 225ml of 0.1 % peptone water

 

7|Stomach the sample for 2 minutes

 

BLSerially dilute the stomached sample to make 20mL volumes of sample

 

Elnsert fresh biochip into the apparatus test fixture

 

1 Lower fixture into the sample so the biochip array is in the solution

 

11 Wait 5 minutes for bacteria to bind to sensor surface

 

 

12|Begin taking impedance measurements with HP 4192A Impedance Analyzer

 

13[lmpedance measurements take about 4 minutes to record from 100Hz - 13MHz

 

14lFiemove sensor from sample and discard biochip

 

 

1§lSterilize sample and biochip usingan autoclave

 

 

87

Table B.4 Bill of process for testing in romaine lettuce samples.

 

Bill Of Process-(Romaine Lettuce Testing)

 

lStep it] Process Description

 

ilGrow bacteria (such as E. coIiO157zHfloverMht in nutrient broth (NB) at 37°C

 

ZlPerform serial dilution on pure culture

 

3|Weigh and separate 3g samples of romaine lettuce

 

4[P|ace lettuce samples in sterile stomacher bgg

 

Sllnoculate samples with 1mL of corresponding serially diluted bacteria concentration

 

SILet sit at room temperature for 1 hour

 

SiAdd 30ml of 0.1% peptone water

 

7

Stomach the sample for 30 seconds

 

 

8|Use the liquid portion to make 20mL volumes of samples

 

9[lnsert fresh biochip into the apparatus test fixture

 

1

Lower fixture into the sample so the biochip array is in the solution

 

11

ait 5 minutes for bacteria to bind to sensor surface

 

 

12|Begin takinLimpedance measurements with HP 4192A Impedance Analyzer

 

131lmpedance measurements take about 4 minutes to record from 100Hz - 13MHz

 

14lRemove sensor from sample and discard biochip

 

 

 

1

Sterilize sample and biochip using an autoclave

 

Table B.5 Bill of process for testing in samples of bovine feces.

 

 

Bill Of Process-(Bovine Feces Testing)

 

teptt

Process Description

 

1

Grow E. coli 0157:H7 ovemight in nutrient broth (NB) at 37°C

 

 

Perform serial dilution on pure culture

 

3|Weigh and separate 209 samples of bovine feces

 

4IPIace manure samples in sterile stomacher bag

 

5Ilnoculate samples with 1mL of correspondinggerially diluted bacteria concentration

 

SILet sit at room temperature for 1 hour

 

6|Add 50ml of 0.1% peptone water

 

7iStomach the sample for 30 seconds

 

BJUse the liquid portion to make 20mL volumes of samples

 

9||nsert fresh biochip into the apparatus test fixture

 

1

Lower fixture into the sample so the biochip array is in the solution

 

11

 

Wait 5 minutes for bacteria to bind to sensor surface

 

1angin taking impedance measurements with HP 4192A Impedance Analyzer

 

tallmpedance measurements take about 4 minutes to record from 100Hz - 13MHz

 

141Bemove sensor from sample and discard biochip

 

 

15Eterilize sample and biochip using an autoclave

 

88

 

Table B.6 Bill of process for reagents used in fabrication and surface functionalization.
I Bill Of Process-(Reagents)

 

 

[Reagent Name Process Description

 

as prepared from 89 of NB mixed with 1000mL of distilled water and

Nutrient Broth (NB) was obtained from Ditco Labs (Detroit, MI). The solution
Nutrient Broth
utoclaved at 121 F for 20 minutes.

 

Phos hate Phosphate Buffered Saline (PBS) was made from 7.659 of NaCl, 0.7249 of
Buffereg Saline a2HPO4. 0.21 g of KH2PO4 and 1000mL distilled water. After preparation
he solution pH was adjusted to 7.4 with NaOH.

 

0 1‘7 P 0 tone Peptone Water (PW) was purchased from Sigma Labs (St. Louis, MS). The
' W at Sr olution was prepared from 19 of PW added to 1L of distilled water. The
olution was autoclaved at 121 F for 20 minutes.

 

Polycl on al | G Ecoli generic and 0157:H7 specific polyclonal antibody were purchased from
Antibodyg PL Labs (Fiockville. MA). The IgG was rehydrated in 1mL of PBS and diluted
0 a concentration of 150ug/mL in PBS.

 

Methanol-HCI The 50:50 mixture contains equal parts of methanol purchased from CCI
' (Columbus, WI) and 1.0M HCI purchased from Sigma (St. Louis, MS).

 

2% MTS (3-Mercaptopropyl) trimethyloxysilane (MTS) purchased from Sigma (St. Louis,
(Silane) W8) was diluted in toluene purchased from J.T. Baker (Phillipsburg, NJ).

 

4-maleimidobutyric acid N-hydroxysuccinimide (GMBS) was purchased from
GMBS Sigma (St. Louis, MS) and diluted in N,N-Dimethylformamide purchased from
(Crosslinker) Spectrum (New Brunswick, NJ). The solution was diluted to 2mM in ethanol
purchased from Phan'nco (Brookfield, CT).

 

Acetone Acetone was purchased from J.T. Baker (Phillipsburg, NJ).

 

Toluene Toluene was purchased from J.T. Baker (Phillipsburg, NJ).

 

 

liqggggl'fl [sopropyl Alcohol was purchased from Spectrum (New Brunswick, NJ).

 

Sorbitol orbitol Mac Conkey Agar (SMAC) was purchased from Ditco Labs (Detroit,
MacConkey I). The solution was prepared from 509 of SMAC and 1000mL of distilled

 

 

Agar ater followed by autoclaving at 121 F for 20 minutes.
Bismuth Sulfite ismuth Sultite Agar (BS) was purchased from Ditco Labs (Detroit, MI). The
A ar olution was prepared from 529 of BS and 1000mL of distilled water and
9 eated to boiling.

 

89

 

 

 

 

    

 

 

:% GPIB was! I

 

 

 

m

Figure B.2 Screen capture of circuit diagram for LabVIEW 6.1 data acquisition software.

 

 

 

 

 

Figure 8.3 Clamp without biochip. (Detail 1 upper chuck; Detail 2 contact plate (includes
4 separate leaf contacts); Detail 3 locating chuck; Detail 4 wire harness).

 

Figure B.4 Clamp with biochip in locating chuck. (Detail 5 biochip in locating chuck).

91

 

 

 

 

 

 

 

 

 

 

 

E

Fi ggre B.5 Apparatus setup. (Detail 6 slide assembly; Detail 7 specimen sample).

 

 

 

 

 

 

 

 

 

 

 

 

13% Apparatus setup engaged in specimen testing.

92

 

 

A |

| \ l

\ Receptacle Contact Point

 

 

 

 

 

 

 

 

Figure B.7 Detail of contact mating to biosensor (4 separate leaf contacts for each pad).

 

Clamp Assembly
Contact Points

 

+0 0—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GPIB [l_l_l_|'|
HP4192A
COMPUTER Impedance Analyzer
(LabVIEW 6.1)
Biochip Schematic ti:
Figure B.8 Wiring schematic showing HP 4192A impedance analyzer connected to
biochip.

93

 

$3.8 208820 E: g x E: “.0 cone—=85 Boa 2.820 .8 8385858 eouoo> “.0 Emmi

 

..i... . 1hr». . b . ...»I.
4

 

 

 

«it../.. ..

I
.. . it . .. . . . .. e .. . 1 . _ r
.u .31.... .1 .. 1. . 1...... .. H.111 174.-.. 242.11.... 114:1...1318141 ..r ... .4... .. 441...... .. a
II . \- n V ’ . a L I- . ‘1. O. , \. I .m . . . n 1 turn.

.... . . t r . e . a i. p . I .
.. . . . a . r . _ . . . .. . .. .
I Wu. .ee.9”.graces—”"1?“lathe? ”1%.“... ....e ...... \ h~....1.4m~.. .5

 

4.

.\~ 4... e ..

.___
e

u 1a....

.11...

G . .
{...-... .4
a e . 4

rl.

 

 

 

O N—DZm—mm<

94

.35 3958—0 :5: x E: fl a 80.: 853.6 Eim a an .8955”? 20m 0583 05 .6 83 8856 g

 

 

:53 5.88.320 9.0? 3:399

0m av Om cm

9

 

d d d d q q q u — 1 d 4 q d d d u u — 4 A 1 d d d d d u — q d d u d 4 d u - — 4 d d d n d d u

§>§i§§§<§

.4
L o

lllllllllllllLl

l

 

llllllllllIllllllllllllllllllllLllllllLl

 

 

3:

95mm.

gm

gm

80:5 Lomcom o>8< 282:. m .o 8565 E SE EEEE: 23... 252m

3

 

 

95

Saba 30.383 61m x 812 .«o 852586 20c 0.58? we cows—homage“ “88> fig

 

 

 

 

 

 

 

 

 

Saba 88820 81m x :39 a Eat 853.6 Eim a 3 consafimfi 20: 058—0 05 mo SE ooatsm v dam

 

 

:53 5:03.320 9.0? 3:820
B B ow on am 9 o

dfidddd—ud—qdu-Junan—.11qqfidfififid-qdddufid—q4-unun-u—qq-qquu-.D

 

cam

2.50

mg

a8m_

can.

Exam

DBSN

26$

 

 

5%

323 32$ 2% gm .0 35% o *0 5:52.05 2...: 038m

{‘1'/MB

 

 

97

Saba 38820 :53 x 51m we nova—58v. 20¢ 0583 .«o couflcomoumou .588, gum

 

 

 

 

 

 

 

 

 

98

Saba 38320 812 x Em a Set 853.6 Sim a a 5:92:36 Boa Shoo—o 05 «o 83 335m 00'8st

 

 

SEE 5.63-320 use? 3535
Q. B 8 ow 8 mm o.

du-d-.dde—-ded-4dd—d-d«HH41flu-dfid4-qd—dfidddd-q-l—HJ-de4dfi114A—d1d-4ddfiq

 

§ 3 E °
(“l/A13

E

g

LlLLlllllllllllllllllIllllllllllll llllllLLlllllllllllIJLLL1o

 

 

 

§

823 o>on< 8825 m .o 8865 B costma 2m: 252m

 

 

99

Saba 3880—0 810 x 810 no coca—25m Bow 0583 Mo couauaomoEB “28> g

 

   

 

    

. .5
. .

for. .. u .—
. .m... ...ca—lal-ul. ~12!

Lfllulvu __ _ ”V7.
c ... ...

    

 

 

 

 

 

 

100

Saba 38620 :56 x 810 a Soak ooaaummv Sim a “a sauna—Ema 22m 0583 05 mo 33 ooatam Iloalammd E

 

 

S83 5.63.320 9.0? 85320

E. on om 0.. cm cm 0. o

dqdddfiddd—q4fidd-.d‘i—‘d-dqqddqd—dddqdqu‘ddwdfid‘dd-anu—uq-ud4441—-ud1¢uuu4.o

 

96m.

95m.

86.

89m

gum

gm

 

 

82% «>22 2822 m 8 8:32.20 22... 2.52m

(“l/A13

 

 

101

Q8585 manage—u 2c BOSE Q<O we manage 01m a0 Emmi.

 

102

APPENDIX D

D.1 Testing in Complex Media

Specific Aim 4 reads: To employ the biosensor for detecting E. coli 0157:H7 bacteria
extracted from artificially contaminated samples of romaine lettuce, ground beef, and
bovine feces.

To validate the biosensor in samples of complex media, separate experiments were
conducted using pathogenic E. coli 0157:H7 (ATCC#43895). Ground beef, bovine feces

and romaine lettuce samples were inoculated with stock cultures of E. coli 0157:H7 cells

and a comparison was made between the plate count and the biosensor for each sample.

 

Figure D.1 Representative samples of: (left) romaine lettuce; (center) ground beef; (right)
bovine feces used for test samples.

D.1.I Testing of Romaine Lettuce

Romaine lettuce test samples were prepared from varying concentrations of a stock
culture of E. coli 0157:H7 (procedure based on Bennett and Beuchat, 2001). Lettuce
samples were obtained from a local grocery store (Goodrich's Shop-Rite, East Lansing,
MD. The pure culture was grown overnight in nutrient broth at 37°C. Samples were
made by separating and weighing 3-gram samples of lettuce and placed into a sterile

whirl pak bag (Nasco; Modesto, CA). Figure D.1 (left) is a representative sample of

103

romaine lettuce used in the experiments. The lettuce samples were inoculated with 1mL
of serially diluted concentrations of E. coli 0157:H7 and left for 60 minutes at room
temperature. A volume of 30mL of 0.1% peptone water was added to each whirl pak bag
to make a 1:10 dilution. The sample was processed by stomaching for 1 minute.
Samples of 20mL were made from concentrations of 101 to 105 CFU/mL out of the
processed samples. After sample preparation, biosensors were used to determine the
presence of bacteria. Prior to inoculation, lettuce samples were tested for generic E. coli
and E. coli 0157:H7. The testing procedure, signal measurement and analysis were
performed as described in Sections 3.3, 3.4 and 3.5. A detailed Bill of Process for sample

preparation is included in Appendix B.

D.1.2 Testing of Ground Beef

Ground beef test samples were prepared from varying concentrations of a stock
culture of E. coli 0157:H7 (procedure based on FDA, 2001). The pure culture was
grown overnight in nutrient broth at 37°C. Ground beef samples were obtained from a
local grocery store (Kroger; Cincinnati, OH). Meat samples of 25g were separated,
weighed and placed in a sterile whirl pak bag (Nasco; Modesto, CA). Figure D.1 (center)
is a representative sample of ground beef used in the experiments. The ground beef was
inoculated with 2mL of pure culture and left for 60 minutes under refrigerated conditions.
A volume of 225ml of 0.1% peptone water was added to each whirl pak bag and
processed by stomaching for 1-2 minutes. Samples of 20mL were made from
concentrations of 10l to 106 CFU/mL out of the processed samples. After sample

preparation, biosensors were used to determine the presence of the inoculants. Samples

104

were plated for continuation. The testing procedure, signal measurement and analysis

were conducted as described in Sections 3.3, 3.4 and 3.5.

D.1.3 Testing of Bovine Feces

Bovine feces (manure) test samples were prepared from varying concentrations of a
stock culture of E. coli 0157:H7 (procedure based on Sanderson et al., 1994). Manure
samples were obtained from mature lactating dairy cows at the MSU Dairy Farm. Prior
to inoculation with E. coli 0157:H7, fecal samples were tested to confirm sample
negativity for E. coli 0157:H7. This was accomplished by weighing l-gram samples of
manure and incubating at 37°C for 24 hours in nutrient broth. The manure samples were
then serially diluted and tested for E. coli 0157:H7.

For sample preparation using the biosensor, manure samples of 20g were separated,
weighed and placed in a sterile whirl pak bag (Nasco; Modesto, CA). Figure D.l (right)
is a representative sample of manure used in the experiments. The manure samples were
inoculated with 1mL of serially diluted concentrations of E. coli 0157:H7 and left for 60
minutes at room temperature. A volume of 50ml of 0.1% peptone water was added to
each whirl pak bag and processed by stomaching for l-rninute. Samples of 20mL were
made from concentrations of 101 to 105 CFU/mL out of the processed samples. After
sample preparation, biosensors were used to measure the presence of the inoculants. The
testing procedure, signal measurement and analysis used were as described in Sections

3.3, 3.4 and 3.5.

105

D.2 Results and Discussion of Testing in Complex Media
0.2.] Results for Romaine Lettuce Testing

Romaine lettuce samples inoculated with E. coli 0157:H7 exhibit a similar response
to pure culture in that the change in impedance of the biosensor is proportional to the
bacteria concentration. Figures D.2 and D3 show the impedance for different
concentrations of romaine lettuce samples inoculated with E. coli 0157:H7. As with
pure culture, the impedance is dependent on frequency and concentration. As frequency
increases the presence of bacteria has a diminished effect on overall biosensor
impedance, irrespective of concentration. High frequencies result in the near
convergence of the impedance. For the lettuce wash water samples, low frequencies also
show an increase in impedance with bacteria concentration, but the effect is less
pronounced.

In proposing the specific aims for this study, one goal was to test the biosensors in
samples having rrrinimal sample processing requirements. Sample processing is time
consuming since it often involves acquiring, separating and filtering of the sample. As
described in the Methods and Materials section, the samples were not processed to reduce
the amount of particulate in solution. For the inoculated lettuce, the sample solution was

mainly free of particulates but the solution had a light green color, suggesting micro-sized
particles and pigments of lettuce were present.

In addition to the effects of the double layer and parasitic capacitances, testing of the
lettuce wash water yields reduced sensitivity due to the interference from food

particulates in the liquid sample, and is one factor the rrricroelectrodes were designed to

106

 

HEKEO :8 ....w 53, Baa—=85 moafiam 83.2 oEaEE ..8 5.957 aloe 89¢ coasnEmE S2532.“ a ..8 3539.5 3

 

 

 

 

 

..ESmO 96— x ..ESuO 9.9 o ..ESmO 39 ... ..ESmO «<9 4
..ESmO N<ow o ..ESmO F3 I ..ESmo 9‘3 x xzfim a
ANIV Socoacoi
8889 Soccer 088.. 809 coop cop

 

oo—

.......
.....................

   

 

DrhoD-qu. 1111111111
I O O o ......
0.01.9.9..V‘ O O

 

 

(wuo) eouepedwl

 

 

 

 

 

 

107

 

 

      

 

 

 

 

 

 

 

 

 

 

 

 

 

10000
E
.C
Q
o
‘c’
a 2473
E 2067
1973
1877
E 1750 1727 1590 1753 1697
1000 -
BLANK roro 10'“ 1m 1m 1w we we 1M7
Bacteria Concentration (CFU/mL)
I Inoculated Lettuce
1000
E .3.
412 420
5 387 373 387 361 394 380
§
if
D.
E
100.
BLANK 100 row 1m 1m 1w tore 1M3 1w
Bacteria Concentration (CFU/mL)
I Inoculated Lettuce
1000
E 183 196 134 189 191 195 192 2‘5 199
Q
8
c 100
I
'0
a
O.
E
10<
BLANK 1M) 1M we we 10M we we 1w
Bacteria Concentration (CFU/mL)
I Inoculated Lettuce

 

 

 

Figgg D.3 Comparison of romaine lettuce samples inoculated with E. coli 0157:H7 at
selected frequencies: (top) lkHz; (middle) 100kHz; (bottom) lOMHz.

108

minimize. The lower detection limit of E. coli 0157:H7 in lettuce wash water is 107

CFU/mL.

D. 2.2 Results for Ground Beef Testing

Ground beef samples inoculated with E. coli 0157 :H7 exhibit a similar response to
romaine lettuce samples in that the change in impedance of the biosensor is proportional
to the bacteria concentration, but less so than for samples in pure culture. Figures DA
and D5 show the impedance for different concentrations of ground beef samples
inoculated with E. coli 0157:H7.

At low frequencies, an increase in impedance is shown with bacteria concentration,
but the effect is less pronounced than exhibited with pure culture. As with pure culture,
the impedance is dependent on frequency and concentration but the ground beef samples
behave differently for high frequencies. As frequency increases the presence of bacteria
has a diminished effect on overall biosensor impedance, but the impedance between
samples is considerably different. This could potentially be due to the effects of ground
beef sample on the dielectric capacitance of the testing solution. High frequencies do still
result in the convergence of the impedance, but there is more variability between sample
Concentrations. In addition to the effects of the double layer and parasitic capacitances,
the ground beef samples has reduced sensitivity due potentially to the contribution of fats
and proteins in the sample.

The ground beef samples were visibly full of fats, oils and proteins and had a light red

Color. The goal of the biosensor was to detect for bacteria in minimally processed

Samples. It is suspected that poor detachment of bacteria from the ground beef

109

 

5.3.20 :8 ..m :33 toga—:85 moEEam hE3 95on a8 awn—22 - #52 Bob cementum. Sufi—268% a c8 coca—cogs: adieu. mam

 

 

 

 

 

 

 

 

 

 

 

 

xcflm a ..ESmO 9A: x ..ESmO PS: I ._E\D..._0 NS: 9
...EBmO gr 4 ..EBmO For + ..ESuO 9S: 0 4.595 99 x
3.: Socoaaoi
89509 00809 00009 oooow coop cop
. . _ . . 8..
x.xxx.x¢ ...» a a a a I
XXXXXX¢MA-1 w
, . .....- m.
_..._........ 82 a
, . ....”JAJL- L. L m
. ”fun-.... .c u-
. anyway... . [M
o o o...W........ .. .
. uni... .. ..
. .S.
Fauna... ...
o o‘eWeIWWrUWN.’
oooop

 

 

 

110

 

 

   

   

 

 

 

 

 

 

 

 

 

 

 

 

10000
13
E
.C
Q 2897 331° 3°37
0
3 269° 2497 236-, 2510 2657
i
E
1000-
BLANK 1m 1049 10A2 we 10M 107.5 we
Bacteria Concentration (CFU/mL)
Ilnoculated Beef
1000
E 436 434 494
413
g 403 385 366 349
d)
O
c
as
E
1001
BLANK 1w 10M we we 10M 1o~5 role
Bacteria Concentration (CFU/mL)
IlnoculatedBeet
1000
’8‘ .99
5 153 146 163 139 182 167 165
0
g 100
a
8
D.
E
10‘
BLANK we 10M we we 10M we we
Bacteria Concentration (CFU/mL)
I Inoculated Beef

 

 

 

Figu_re D.5 Comparison of ground beef samples inoculated with E. coli 0157:H7 at
selected frequencies: (top) lkHz; (middle) 100kHz; (bottom) lOMHz.

lll

particulates resulted in fewer bacteria binding to the surface thereby further reducing the
sensitivity. For ground beef, the sample matrix had a strong effect on the biosensor lower
detection limit resulting in the inability of the biosensor to significantly distinguish
between different bacteria concentrations.

The biosensor has a lower detection limit of 105 CFU/mL with respect to a sample
concentration of 101 CFU/mL, however, all concentrations are insignificant with respect
to the blank. In all data throughout this study, whether pure culture or complex media
testing, the impedance of concentrations between 100 to 103 CFU/mL cannot be
distinguished by the biosensor. Both 100 CFU/mL and blank measurements were
insignificant due to a high degree of variability in impedance measurements due to the
unreliable measurements at low concentrations. This suggests that the biosensor is
"almost" significant for ground beef testing when comparing against the blank since low

concentrations of bacteria are not able to be distinguished from one another.

D-2.3 Results for Bovine Feces Testing
Bovine feces (manure) samples inoculated with E. coli 0157:H7 exhibited a similar
response to ground beef and romaine lettuce samples in that the change in impedance of
the biosensor is proportional to the bacteria concentration, but less so than for samples in
pure culture. Figures D.6 and D7 show the impedance for different concentrations of
ground beef samples inoculated with E. coli 0157:H7.
For the bovine feces, the samples were visibly full of partially digested fiber and
organic matter and resulted in a dark brown, smelly sample. Bovine feces samples were

dense with particulate in comparison to the lettuce and ground beef and were the most

112

challenging test for the biosensor. Also, the bovine feces samples contained a large
amount of non-target E. coli bacteria. The sample concentrations used for the manure
testing were 101, 102, 103, 104, and 105 CFU/mL, but were not able to be confirmed
through plating due to the high concentration of E. coli 0157:H7. Instead, the
concentrations of the serially diluted pure culture used prior to inoculation are used to
denote the concentration of E. coli used in the sample testing.

For the bovine feces samples, there is little uniformity between impedance
measurements at high frequencies. The presence of bacteria still has a diminished effect
on biosensor impedance at high frequencies, but the impedance between samples is
considerably different due to the complexity sample matrix. The manure sample matrix
also changes the dielectric capacitance of the testing solution, which is the dominant
impedance element at high frequencies. At low frequencies, the impedance increases
with increasing bacteria concentration. The data suggests the impedance is dependent on
frequency and concentration as already discussed earlier. As with the lettuce and ground
beef samples, it is suspected that poor detachment of bacteria from the manure
particulates resulted in fewer target bacteria binding to the surface. It is also likely that
there was non-specific binding due to the high numbers of non-target generic E. coli
present in the sample.

For the manure testing, the biosensor was unable to significantly distinguish between
different bacteria concentrations, as defined in the statistical analysis of the Methods and
Materials. However, when lowering the significance level to 90% (D: 0.1), the lower

detection limit is 104 CFU/mL with respect to the blank.

113

.hmkhfio .-QU .m :33 Guam—592$ wO—ng mecca 0Cm>g MOM EA: 1 NEDA: 80H“— COmuSDEmm—v cacao—408M a hon“ Qua—NEE ©.fl— ”Haw—"m

 

 

 

 

..E5u_0 m<ow o ..ESmO .29 + ._E\Du.0 meow 4 ._E\Du_0 «<0? 0 1.85....0 For 1 v_Z<._m n

 

 

SNIV S8259“.

0000009 000009 00000.. 080.. 009. 00..
— L — 8’

000—.

 

 

(WHO) e<>UIBF>0dw|I

 

 

oooor

 

 

 

114

 

 

10000

 

 

 

 

 

 

 

 

’E‘
.1: 1767 1753
O 1417 1347 1553 1513
8
c 1000 .
i
E
100 4
BLANK w~1 we we we we
Bacteria Concentration (CFU/mL)
I Inoculated Manure
1000
’E‘
.C
Q 368
g 297
E, 253 251
E 216 224
E
100 -
BLANK w~1 we we we we

Bacteria Concentration (CFU/mL)

I Inoculated Manure

 

 

 

 

 

1000
E
.C
Q 131
a 93 96 99 10°
0 100 79
C
i
E
10 4
BLANK 1M1 1W 1W 1W 1W

Bacteria Concentration (CFU/mL)

I Inoculated Manure

 

 

 

Figure D.7 Comparison of Bovine Feces samples inoculated with E. coli 0157:H7 at
selected frequencies: (top) lkHz; (middle) 100kHz; (bottom) lOMHz.

115

For the complex media, the statistical analysis of romaine lettuce samples shows that
the biosensor has a lower detection limit of 107 CFU/mL with respect to the blank (Figure
D.8). Inoculated ground beef samples did not have a lower detection limit as samples
were not statistically significant from each other, though there was a trend of increasing
impedance with concentration among impedance means. For inoculated manure samples,
the lower detection limit was 104 CFU/mL with respect to the blank (when lowering the

significance level from 95% to 90%).

 

 

 

Mean i SD and Significancei
Concentration Romaine Lettuce Ground Beef Bovine Feces
(CFU/mL) Impedance (16g of Impedance (log of Impedance (log n)‘
BLANK 3.24 i 0.04 a 3.43 i 0.05 a,b 3.14 i 0.09 a
1 x 10" 3.24 a 0.04 a 3.39 .1: 0.09 a,b
l x 101 3.21 :l: 0.03 a 3.37 a 0.07 a 3.13 d: 0.04 a
1 x 102 3.24 i 0.06 a 3.40 a 0.06 a 3.19 :1: 0.09 a,b
1 x 103 3.23 :l: 0.04 a 3.46 :I: 0.02 a,b 3.18 i 0.03 a,b
1 x 104 3.29 a 0.05 a,b 3.42 a 0.06 a,b 3.25 1: 0.04 b
1 x 105 3.27 :I: 0.09 a,b 3.50 a 0.08 a 3.24 1 0.03 b
1 x 106 3.31 a 0.03 a,b 3.48 a 0.05 b
1 x 107 3.40 1 0.07 b

 

 

 

[1] Means with same letter are not significantly different (p>0.05)

[2] Log transform of romaine lettuce inoculated with E. coli 0157:H7 at lkHz

[3] Log transform of ground beef inoculated with E. coli 0157:H7 at lkHz

[4] Log transform of bovine feces inoculated with E. coli 0157:H7 at lkHz (p>0.10)

Figgr_e D.8 Statistical significance of mean differences between concentrations for
complex media study.
D.3 Summary and Conclusions
03.! Summary of Lower Detection Limits

When testing the biosensor in complex sample media, the biosensor performance
decreased. In testing for romaine lettuce, the lower detection limit was 107 CFU/mL. For

ground beef, the sample media proved to cause too much variability in testing and as a

116

result, the biosensor is unable to distinguish between any concentration of bacteria with
respect to the blank. (Though ground beef has a lower detection limit of 105 CFU/mL
when compared to 101 CFU/mL.) Bovine feces was also affected by the sample matrix
and had a lower detection limit of 10s CFU/mL with respect to the blank (with a
significance of 90%).

In addressing the biosensor performance, it is noted that the presence of interferants
(fats, oils, proteins and organic matter) changes the electrical properties of the sample
media. It may be possible to improve signal processing to filter out the effects of the
interferants on impedance. Though sample processing steps add time to the testing
process, it might be interesting to try innovative sample processing techniques to reduce
the presence of interferants in the sample. Perhaps the biosensor could be coupled in-line

with a processing unit to provide real time data of filtered samples.

117

APPENDIX E
(BUSINESS PLAN)

  
   
 

  

AGEN BIOSENSE

RAPID PATHOG EN DETECTION

AGEN BIOSENSE TEAM INFORMATION PAGE

TEAM ID#:
1051

BUSINESS CONCEPT NAME:
AGEN BIOSENSE

TEAM MEMBER INFORMATION:

DR. TODD ZAHN
zahnt@michigan.org
517.241.0368

MR. MAHENDRA RAMSINGHANI

ramsinghanim@michigan.org
517.241.4180

DR. EVAN GELYN ALOCILJA
alocih§@egr.m_§u.edu
517.355.0083

MR. STEPHEN RADKE

steveradke@hotmail.com
517.485.3770

118

TABLE OF CONTENTS

 

§ETION PAGEfiNUM§_E_
EXECUTIVE SUMMARY ............................................................................................... 121
AGEN BIOSENSE OFFERING ...................................................................................... 122
PRODUCT DEVELOPMENT ......................................................................................... 123
MARKET ANALYSIS ..................................................................................................... 124
COMPETITIVE ANALYSIS ............................................................................................ 126
MARKETING ................................................................................................................. 127
SALES ........................................................................................................................... 129
MANAGEMENT TEAM .................................................................................................. 131
GROWTH PLAN ............................................................................................................ 135
RISKS AND COUNTERMEASURES ............................................................................ 136
FUTURE GROWTH ....................................................................................................... 137
FINANCIAL PLANNING ................................................................................................. 139

 

119

GREAT LAKES VENT QRE QUEST-PHASE II BU§INE§ PLAN

*This Business Concept Overview is prepared in response to the Great Lakes Venture Quest
Phase II requirement and is submitted on behalf of Agen BioSense, Team #1051.

EXECUTIVE SUMMARY

PROBLEM STATEMENT
In the meat and food processing industry, undetected pathogens cause widespread diseases and
lead to product recall. Pathogen testing is mandatory and regulated by USDA/ FDA.

Public concern regarding food safety has increased markedly over the past decade. From farm to
table, there are numerous opportunities for the pathogenic contamination of food, which results in
food industry recalls, lost productivity, increased insurance costs, unnecessary illness and
thousands of fatalities per year.

Escherichia coli are bacteria that naturally occur in the intestinal tracts of humans and warm-
blooded animals to help the body synthesize vitamins. A particularly dangerous type is referred
to as enterohemorrhagic E. coli, or EHEC. EHEC strains has been associated with foodborne
outbreaks traced to undercooked meats, apple juice or cider, salad, salami, and milk. EHEC
produces toxins that can cause anemia, stomach cramps and bloody diarrhea, and a serious
complication called hemolytic uremic syndrome (HUS), which can lead to kidney failure. In North
America, HUS is the most common cause of acute kidney failure in children, who are particularly
susceptible to this complication.

The Centers for Disease Control and Prevention estimates that 76 million foodborne illnesses
occur each year in the United States accounting for 325,000 hospitalizations and 5,000 deaths
annually. The four major foodborne pathogens, Salmonella, Listeria monocytogenes,
Campy/obacter, and Escherichia coli0157:l-l7, are characterized in Table 1.

of Cases
141

coli 0157:H7 458
498
1 1 02

 

Table 1. Food Illnesses and Deaths in the United States Caused by Major Foodbome Pathogens in 1999.
Source: Centers for Disease Control and Prevention (Atlanta).

According to the United States Department of Agriculture (USDA), medical costs and lost
productivity resulting from food-home illnesses is estimated to range between $5 and $6 billion
annually. Due to the recent trend in Food & Drug Administration (FDA) and United States
Department of Agriculture (USDA) regulations with the Hazard Analysis and Critical Control Point
(HACCP) program, pathogen testing is mandatory in all meat processing, diary, food, fruit and
vegetable processing plants. Recent food security data indicates that cases of EHEC and other
foodborne pathogen infections are rising in both the US. and in other nations.

SHORTCOMINGS OF EXISTING PATHOGEN DETECTION TECHNOLOGIES

The detection and identification of food borne pathogens continue to rely on conventional
culturing techniques. These are very elaborate, time-consuming and expensive. Typical tests
take a minimum of 24 hours for culture followed by 20 minutes of detection. The existing test
methods are completed in a microbiology laboratory and are not suitable for on-Site monitoring.
As a result, the food and beverage industry needs real time pathogen detection sensors with
higher sensitivity. According to The lntemational Society of Optical Engineering (SPIE), existing
pathogen detection methods, culture techniques and bioassays such as enzyme-linked

120

immunosorbent assay (ELISA) for determining pathogens in food are elaborate, time consuming
and expensive.

AGEN BIOSENSE OFFERING

REAL-TIME PATHOGEN DETECTION WITH HIGHER SENSITIVITY

Agen BioSense has developed a patent protected, portable, real time pathogen detection
biosensor. The sensors will enable the food processing industry to conduct real time microbial
tests with higher sensitivity.

 

 

 

 

 

Company Technologies Sensltlvlty (cfulmL) Time
Neogen(Lansin&MI) Lateral Flow 10,000 8 hours
Molecular Circuitry (King of Prussia, PA) lmmunobiosensor 100,000 24 hours
BioMerieux (France) Immunoassay 100,000 25 hours
AGEN BIOSENSE IMMUNOSENSOR 500 <5 mln.

 

 

 

 

 

 

Table 2. Sensor Performance Comparison of Three Industry Leaders.

An estimated 25,000 US based food processors perform 144 million tests annually. Current
pathogen detection tests conducted in the microbiology laboratory take an average of 8 ~ 25
hours. Delayed detection of pathogens has led to product recall resulting in losses of several
million dollars for the meat processing industry.

Rapid, simple, and accurate on-site testing will provide considerable value to the food and beverage
industry. Agen Biosense is developing user-friendly biological analysis systems that will be targeted at food
processing lines and inventories. This is expected to be a significant advantage over existing testing
methods conducted in the laboratory, which lead to production delays and product recalls.

VALUE PROPOSITION

Agen Biosense rapid pathogen detection will enable higher efficiencies in food industry by the
following:

0 Higher sensitivity ensures high product quality

0 Faster on-site testing results in greater yields
0 Reduction / Elimination of product recalls

0 Reduced liability litigation cases

RAPID AND GROWING MARKET

Recent trends in FDA and USDA regulations suggest testing for pathogenic bacteria will be on the rise.
Information extracted from Strategic Consulting, Keen Solutions and Business Communications Company
data have estimated the 2001 pathogen testing market at over $230 million (Table 2 in full plan). This
represents approximately a 30% share of the total world market. The product design and manufacturing
aspects of Agen’s technology are being accomplished with a low cost, high quality mindset to permit at
least a 5% savings per test for our customers. This translates into an annual savings of $7.2 million for the
overall industry. Additionally, Agen Biosense believes that the versatile and easy to use pathogen detection
biosensor will have first mover advantage into the home healthcare market of $6.3B.

121

FINANCING REQUIREMENT S 81 RETURN ON INVESTMENT

 

 

 

 

 

 

 

 

 

Year 1 Year 2 Year 4
Complete proto-type & third party Hire CEO and sales Develop Sales 81
Uses of Cash testirg team Dist. Channels
Capital Required $ 500,000 $ 1.5 million $ 4 ~ $5 million
Founders, Grants Angel 8. Seed Venture Cap.
Potential Sources (SBIR, MLSC) Stage Investors

 

Table 3. Finance Requirements.

Agen is in the final prototype development stages of the single test pathogen detection biosensor.
Production of the first 5,000 units (estimate first year production capacity of 30,000 units) is
anticipated by August of this year. Our initial capital outlay is estimated at $500 K (Year 1
expenditures). The first $100 K will be solely financed through founder investments, while the
remaining $400 K will be raised through federal and state granting programs and other external
funding sources. Emphasis will be on research and development of new and improved product
lines and will require an additional $1 million (covers overhead R & D expenditures) infusion by
the third quarter of this year. A final growth stage capital infusion is expected between year 4 and
5. Agen anticipates a breakeven in Year 3 and generate a healthy ROI via an acquisition after
Year 5. Current acquisitions in the microbial testing sector are between 5X ~ 10X of actual
revenues.

PRQDQCT DEVELOPMENT

Agen has three biosensors in the product development pipeline: the single array, multiple array
and a Microsystems based sensor. The single array biosensor (detects a specific pathogen at a
time) is in the final prototype development stage and will be ready for production by September
2002. The biosensor consists of an electronic detection docking unit and disposable pathogen
specific electrochemical cartridge. When the bacteria (pathogen) are present in a sample, a
reaction occurs sending an electronic signal to a multi-meter that is used to quantify the bacterial
count.

PRODUCT FUNCTIONALITY

Agen Biosensors are designed with the input and feedback from the end user. In 3 easy steps,
detection of harmful pathogens can be achieved within 5 minutes. Additionally, the inexpensive
disposable cartridges will minimize cross contamination and maximize results.

Steps
1) Place Pathogen Specific cartridge in Docking station
2) Apply liquefied sample (100 uL) to application window and wait 5 minutes. Compare
reading to the bacterial estimation chart on back of docking station (Indicates quantifiable
presence of bacteria)
3) Remove and dispose of cartridge

The multiple array biosensor is anticipated to be completed by Q1 2003. As it is based on the
same architecture and detection technology as the single test biosensor, thus will require less
development time. It will offer the customers the ability to test several (~ 5-10) different
pathogens per disposable card. Both the single and multiple arrays will use the same electronic
docking display device.

122

 

Year 1 Year 2 Year 3 Year 4
02IO3 I O4 OIJ ozios I O4 O1 I oz I 03 I O4 01 I oz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SINGLE PRODUCTION _' ‘ Ewes"7?“1-=*"=r-‘--~tr .3.._,,_.]

SINGLE PROTOTYPE _fi____l

[MULTI PRODUCTION tea a,
bliULTI PROTOTYPE F

IMULTIR&D ;_ . _ 1A;

 

 

 

 

Microsystems
PRODUCTION

Microsystems . '-
PROTOTYPE ' . __ _ __ _ I; ‘=-.'~::.--'::;:-:-;;-_r;--=:_--- see-J

Microsystems R&D I ' "flufi
Figure 1. Product Development Timeline.

   

 

r2538": P: f..."._' B.2-n Z

 

 

The Microsystems (MEMS) based biosensor is presently in the initial stages of R&D. It has the
potential to offer the customer 10005 of tests all on a single silicon wafer. The MEMS test will not
be disposable but it will be reusable. The customer would have to return the MEMS chip to Agen
Biosense after use for chemical reapplication. The MEMS based biosensor is anticipated to be
ready for production by late 2004.

There are two key technological points of interest that need to be considered: sensitivity and
speed. The sensitivity refers to the concentration of bacteria present (colony forming units, cfu)
in a sample needed for detection by the biosensor. Most biosensors being produced by research
efforts have a sensitivity of 1000-10,000 cfu/mL. This is especially troubling since only 1 Ecoli
cell is necessary to cause an infection. Speed is the other critical technological variable. Culture
based tests are sensitive, however, they take days to produce results. This is a problem since
even a few hours is enough time for several people to be exposed to a contaminant. The current
Agen biosensor is in the process of being validated by a third party with a sensitivity of 1-100
cfu/mL (10 fold decrease in sensitivity compared to competitor product lines) and a detection time
of 5 minutes (Closest competitor is 4 hours).

MARKET ANALYSIS

LARGE AND A GROWING MARKET

The overall food products testing industry is growing steadily. According to Business
Communications Company, Inc., study titled The Growing Food Testing Business: Pathogens,
Pesticides, Genetically Modified Organisms (GMOs), sales in the US. for food-testing products
will grow at an AAGR (average annual growth rate) of 9.9% between 1998 and 2005.

300 also forecasts that the larger share (82%) of sales will be for tests to detect pathogens and
will grow at an AAGR of 9.4%. (From $122.6 million in 2000 to $192.5 million in 2005.) 800
forecasts that sales for pesticide-residue tests will increase at an AAGR of 7.7% from $8.9 million
in 2000 to $12.9 million in 2005

Companies in the pathogen detection and microbial testing industry are growing at a rate of 30 ~

40 % in revenues each year with average Gross Profit margins in the range of 25- 50%. (Data
obtained from Hoovers, Corp Tech, Dun & Bradstreet and Company Financial reports)

123

 

Food Testing Products Growth Estimates

 

- J {fl 1; V -
1998 1 999 2000 2001 2002 2003 2004 2005

 

lPathogentests (in m'llions) 13.02 23.5 27.53 23.33 30.13 31.43 32.73 34.03
lerketSize($m'llions) F I 149.5 162.955 177.621 193.6068 211.0315 230.0243

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 2. Food Testing Products Growth. Source: Business Communications Company, Inc., study titled
1719 Growing Food Testing Business: Pathogens, Pesticides, Genetically Modified Organisms (GMOs)

NO CLEAR MARKET LEADER

While several companies are in the pathogen detection space, there is no single market leader in
this industry. This has been validated by analysis from a study conducted by University of
Michigan Business School Team.

INCREASED PATHOGEN TESTING

The Food and Drug Administration’s Center for Food Safety and Applied Nutrition has adopted a
food safety program known as Hazard Analysis and Critical Control Point, or HACCP. HACCP
implementation is intended to be a proactive approach to prevent hazards that could cause food-
borne illnesses. HAACP includes toxins, chemicals, and biological pathogens. Under the
oversight of the USDA, all meat and poultry processing plants (~5,530) in the US. were required
to comply with HACCP regulation effective January 2000. In 1995, the FDA established the
HACCP regulation within the seafood industry that includes almost 4100 processing plants.
Effective January 20'", 2004 all US. juice processing facilities are required to comply with
HACCP standards that include pathogen testing prevention. The FDA is now considering
developing regulations that would establish HACCP as the food safety standard throughout other
areas of the food industry, including both domestic and imported food products.

BIOTERRORISM at NATIONAL SAFETY CONCERNS

Recent threats on national security and the events of September 11th have created an impetus for
ensuring the protection of food and water supply. Companies involved in the food and beverage
industries are aware of the possibilities for the biological contamination of their products and are
evaluating low cost solutions that ensure their products are free of pathogenic material.

WORLD TRADE AND CORPORATE LIABILITY

The size of the food industry and the diversity of products and processes have grown
tremendously in the amount of domestic food manufactured along with the number of foods
imported. This trend has been associated with the increased number of new food pathogens and
increased incidence of contamination. Not surprisingly, the FDA has noticed an increase in
corporate liability settlements from food-bome illness cases. Reducing the litigation expenses
incurred by food and beverage processing facilities is a primary objective for the industry leaders.

Technological advances in pathogen detection are being developed, and the companies that
keep abreast of the technology and understand the value of such advances are positioning

124

themselves to increase consumer confidence, reduce corporate liability expenses, and build
brand quality.

These trends are an indication that the pathogen testing market will continue to expand over the
next few years.

COMPETITIVE ANALYSIS

There are a number of companies currently involved in the detection of food pathogens and
several other entrants are expected to enter as the market segments grow. A few of the largest
competitors include companies like Neogen (NEOG), bioMerieux and a start-up Molecular
Circuitry.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Neogen Corp. (NEOG) Molecular Circuitry, Inc. bioMerieux Industry
Lansing, MI King of Prussia, PA France Average
Year
Established 1981 1992 1963
2001 Revenues $23 million $ 1 million $560 million'
Public: NEOG Privately Held Privately Held
Gross Profit 54% (Not shown profitsyet) NA. 62.20%
Pre Tax Profit 14.52% - NA. 4.65%
12 Month
Revenue 38.80% - NA 49.50%
Sales I
Employee $106,400 N.A. $120,000
Employees 230 48 4000
Management Established Developing Established
Competitive
Products Reveal 8 8 Alert Detex System MC-18 VIDAS ECO
9710109 $ 9.50 per test $ 10 per test $ 8 per lGSi
Market Food Food Food,
Segments Pharmaceuticals and
Cosmetics
Distribution Direct Sales Direct Sales Direct Sales
Follow sales conducted 2 Sales Managers for USA 2 Sales Managers for
on phone USA
Strengths 1) Market leader 1) Solid Board and 1) Well established in
2) Well established executive team. EU 8. lntemational
Isales 81 marketing 2) Product mature I tested IMarkets
channels 2) Strong R & D
3) Fiscal strength
Weaknesses 1) No lntemational 1) Losses for past eight I” Limited presence
presence years. an US
2) Limited R 8 D focus 2) Poor market penetration
_ acquisition approach
Other Remarks 1) Acquisition related 1) 12 Employees laid off in 1) 23% of Sales is
growth 2001 from North America
2) About 25% Sales are ) Company has not 2) 38% of total Sales
from lntemational Sales hown profit yet. revenues from
Ilmmunoassays.
Notes: 1) ' bioMerieux Sales figures are for year 2000.
2) + Industry Average is based on Diagnostics Industry/Source: Hoovers Information. Other

 

 

information is from annual reports.
TQIg 4, Competitive Analysis of Three Industry Leaders.

125

MARK IN

MARKET STRATEGY
Agen’s strategic marketing plan has been developed in detail to address the products, customers
and the target market. Progress has been made on several of the following areas:

0 Market Research: Agen has extensively evaluated data (annual financial reports of
competitors, market research sources such as BBC, Hoovers, Dunn & Bradstreet) over
the past five years to conclude that the market has been growing at a healthy rate. As
indicated in the initial market analysis section, the market growth trends and size are
healthy. The unit sales are growing at a CAGR of 9.9 °/o and the dollar sales are growing
at a CAGR of 9.4%.

o Ememl 8r lntemal audit: After analyzing the tactics in distribution and pricing methods of
various competitors, Agen has completed an extensive audit that assists in developing a
market entry strategy. The strategy is explained in the following Market Entry section.

0 Customer Resegm A University of Michigan Business School team of five students is
currently assisting Agen to conduct customer research. The research is targeted towards
top diary companies and top meat producers in North America. The survey focuses
extensive interviews with target customers to gather data related to product performance,
price, testing procedures, distribution channels. The findings will be available by end of
Q1 2002.

o Bela Testinq and Initial product anajjgis: Agen has contacted Kraft Foods to conduct
beta tests of the product. The Agen team has had several discussions with Koegel Meat
and is in the process of identifying meat processing sites for beta tests. The beta testing
will be monitored closely by Agen. Results of the beta tests are expected to arrive by 02
2002.

SEGMENTATION

The processed food sector accounts for the largest number of tests, with over 52.2 million
performed annually (Table 5). This represents over 36% (Figure 3) of total tests performed, most
likely driven by the larger number of plants (almost 38%). The dairy sector has the highest
testing rate per plant, averaging over 630 tests per plant per week, while the beef and poultry
sector performs the least number of tests per plant averaging 369 tests per plant per week. As a
result, the beef and poultry sector accounts for only 22.3% of all testing in the industry. The fruit
and vegetable sector is currently the smallest of the four sectors accounting for only 9.7% of
testing. However, the fruit and vegetable sector is becoming more of a focus by the USDA food
safety inspection service and is expected to result in a substantial increase in the next few years.

 

 

 

 

 

 

 

 

 

 

 

US Food Industry Sector Review

Sector Number of Plants Total Tests AverageIPlantlweek

Beef and Poultry 1,679 32,212,471 369

Dairy 1 .388 45,887,576 636

Fruit/Veg 652 13,981 .305 412

Processed foods 2,260 52,196,282 444

Total r 5,979 144,277,634 464

TM; US Food Industry Sector Review. (Source:Strategic Consulting—Pathogen Testing in Food Industry,

1999)

126

The food testing industry is comprised of the following sectors (US figures):

 

_ Meat
22%

  
 
  
 
 
 

Processed

36% \

Dairy
32%

FruiWng/,//
1 0%

 

 

 

I Meat Dairy I FruiWeg I Processed

 

 

Figure 3. US Food Industry Tests per Sector.

MARKET ENTRY

Entry Methods & Launch timing: Agen expects to penetrate the diary and meat processing market
steadily by building relationships with key potential customers. The initial approach for year I will
include leveraging upon the relationships of the Board members. After the product has been
tested by a third party laboratory an aggressive market penetration strategy will be adopted. This
is expected in year 2 and the low-cost approach will include the following:

0 Creating customer awareness by making presentations at key events such as the
Association of Analytical Communities (AOAC) annual exposition held in
September each year, the International Dairy Food Association (IDFA) worldwide
expo held in October every year. A list of all target events and their impact has
been developed by Agen.

o Using a public relations strategy to generate media in key industry publications
such as Food Quality, Food testing and Analysis and several other target
publications.

Presence at exhibitions, collateral material development and full-blown campaign will be executed
after initial funding in year 2.5 / year 3.

MARKETING MIX: THE FOUR P's
PRODUCT PLAN: Agen’s product will be positioned as a “Faster and Accurate” product with the
ability to conduct tests for several pathogens on one strip.

PLACE: As most of the diary companies are based in Midwest (primarily Wisconsin) along with
several meat processing companies, the customer focus will remain in the Midwest region of
North America during the beta testing and initial launch stages. The regional focus will shift to
North America during the growth phase.

 

PROMOTIONAL PLAN: The company's promotion plan primarily focuses upon events (technical
workshops, seminars) during year one with media strategy to broadcast the significant
milestones. Development of an aggressive promotional plan, including development of collateral
marketing material will be subsequent to third party testing of the product.

PRICING PLAN: Agen will follow the industry price levels of $12 per test with sales emphasis

upon speed and sensitivity. A discount structure for key customers has been developed
depending upon the volumes of purchase.

127

0 ~ 12 Direct Sales to testing Third

12~ 24 Direct Sales to processing In house

 

Table 6. Marketing Entry Strategy for Targeted Industries.

 

BUDGET: The Marketing 81 Sales budget is expected to be as high as 45 ~ 50% of the revenues
in first 2 years and will be driven down gradually to 35% by year 5.

SALES

IDENTIFYING THE CUSTOMER
0 To determine the appropriate target customer within each sector, one must understand
where pathogen testing is being performed. The market report, Pathogen Testing in the
US. Food Industry (Strategic Consulting, 2000) analyzed where microbiological testing
practices occurred most often, and discovered that testing took place in one of three
areas:

    

Figure 4. Point of Pathogen Testing in Various Sectors. (Source: Keen Solutions, 1999 Report)

128

0 Processing plant — Microbiology Laboratory: Average 40% Tests
o External / Third Party outside Reference laboratories: Average 41% Tests
0 Centralized corporate laboratories: Average 19% Tests

While aggressive sales are not expected to start till the end of year 1, a target customer list of the
top 100 meat producers in North America has been acquired. Also, Agen is currently in advanced
stages of gathering details of various laboratories in the Midwest that conduct pathogen testing
for the dairy industry.

SALES STRATEGY: HOW TO GET THERE

 

Year1 I Year2 I Year3 I Year4
I02I03I04LOII02I03IO4IO1 IQZIQ3IO4I01I02

 

 

 

Figure 5. Sales Strategy and Yearly Milestones.

PROJECTED SALES & MARKET SHARE

Strategic Consulting, Keen solutions and Business Communications Company data anticipate a
2005 pathogen testing market estimate over $230 million (Table 7), representing approximately
30% of the total world market. Not surprisingly, the processed foods sector represents the largest
pathogen testing segment with total US consumable sales of $75.9 million, representing 33%
overall. The dairy sector comprises 29% ($66.7million) of total US consumable sales in the
pathogen testing market, while the beef and poultry sector accounts for 21% of total sales. The
fruit and vegetable. and seafood sectors combine for the remaining 17% (9% and 8%

respectively).
Food Industry Seafood Beet & Dairy Frult 8t Processed Total
Market Poultry vegetable foods

Consumable
Sales (US) (million) $18.4M $48.3M $66.7M $20.7M $75.9M $230M

 

 

 

Overall Percentage 8% 21% 29% 9% 33% 100%

 

Agen Market Share
‘(5yr Min.
Estimate: $ 1.8M $7.24M $6.6M $1 .65M $3.79M $21.08M

10%Market share)

Agen Market Share 5 3.6M $ 14.48M $132M $3.3M $7.58M $42.16M
‘(5yr Max.
Estimate: 20%

Market share)

Table 7. Market Share Estimates and Target Sales Revenues from Food Industry Sectors.

Notes:
1) Agen Penetration is expected to be higher in Beef & Poultry (15%) with about 5% -10% in other markets.

 

 

 

 

 

 

 

 

 

 

129

2) Beef and poultry data was collected from only large 1,679 processing plants. Additional 3248 small plants
are not included in the estimates.

3) HACCP pilot studies are currently being conducted in the fruit and vegetable sector. The expected date
of HACCP implementation for this sector is January 20'", 2004, which could easily translate into a 3-5 fold
increase in pathogen detection within this sector translating into potential returns of $9-$14 million for Agen
Biosense.

MANAGEMENT TEAM
Dr. Evangelyn AloclllaI Ph.D.: Chief Scientific Adviser

Dr. Alocilja is a professor of Biosystems Engineering, Michigan State University. She also holds
an adjunct position at the National Food Safety and Toxicology Center at MSU. She is dedicated
to the teaching profession and has been the proud recipient of the 1995 Withrow Teaching
Excellence award.

Dr. Alocilja holds a patent portfolio related to pathogen detection biosensors, with one patent
issued and three currently pending. She is very well connected in the industry as a biosensors
consultant and is nationally recognized for her cutting edge research in biosensors for pathogen
detection. An invited speaker at several national conferences, including the prestigious
Knowledge Foundation's conference on “Electronic Nose Technologies”, Dr. Alocilja is the
elected chair of the 2002 biosensors committee for the American Society of Agricultural
Engineers. Dr. Alocilja will lead the efforts of scientific discovery & new product development.

SLeghen Radke 3.5.: Product Development
Mr. Radke will be an integral part of the biosensor product development team focusing on product

improvements and quality control issues. Steve received his bachelor degree in engineering and
is currently pursuing a doctorate in bio systems engineering from Michigan State University. His
research focus is on biosensor development for rapid pathogen detection.

Mr. Radke is on leave from the General Motors Company as a product engineer where he was
responsible for managing the design, build and installation of million dollar projects involving
manufacturing integration equipment. Prior to this Steve was awarded a National Science
Foundation fellowship at the Virginia Polytechnic Institute and State University where he focused
on biosystems engineering and water quality research.

Todd Zahn Ph.D.I MBA: Financial and Strategy

Dr. Zahn will be responsible for overseeing global corporate strategy and early stage corporate
development. With his science and business background, Todd brings a unique perspective to
the Agen team. He has a significant understanding of technology development and industry
awareness. Additionally, he has overseen the activities for a $1 billion fund dedicated to building
the life sciences industry in the State of Michigan. As an integral part of a small team, Todd has
contributed extensively to the strategic design and implementation of the Michigan life science
corridor, which has become a benchmark and model for life science initiatives in other States
including New York, Missouri, Pennsylvania, and Texas.

Prior to this, Todd was a corporate strategy consultant for a technology commercialization and
licensing incubator that commercialized medical device technology discovered at Los Alamos
National Labs. His research in anti-cancer agents led to the discovery of a handful of potent anti-
tumor agents (patent pending) and provided useful insight into the mechanism of the most
common link to human cancer. He was also successful in scaling up the manufacturing of
anticancer agents that were sold to the largest chemical supply company in the US. Todd has
published his research findings in highly respected international journals that include the Journal
of the American Chemical Society and Journal of Medicinal Chemistry. Todd is an active member
in the Society of Competitive Intelligence Professionals, and is a member and award recipient of
the American Chemical Society.

130

Mahendra RamsinghaniI B.E.I MBA: Sales & Marketing
Mr. Ramsinghani will lead sales and marketing efforts for Agen. Mahendra brings a wealth of

resources and contacts to Agen. including a broad skill set in sales, marketing and business
development. Currently, Mahendra serves as the Director of Venture Capital Initiatives for
Michigan Economic Development Corporation and is responsible for the creation and utilization of
incentives that help grow the venture capital resources. He has led development of a $75 million
venture capital incentive plan (under approval at legislature) expected to create upwards of $ 300
million in venture capital for Michigan over time.

Mr. Ramsinghani received his MBA in 1996 from the University of Pune, India focusing on
marketing and finance. Shortly thereafter, he led the growth & market penetration of Aluminum
Company of America (ALCOA) in India. He successfully grew Alcoa's installation exceeding
150% of target goals in 18 months. In 1997, Mahendra was head for business development for
Kemtec Technologies, Singapore where he led corporate sales & market growth from 0 to SS 2
million in 18 months. Subsequently, Kemtec Technologies went public In year 2000
(SGX2ISOFTEL).

 

 

 

 

   

ORGANIZATION CHART
Board of Directors
Scientific ,
Advisory Board I
-------------------- ’ —
V

 

 

Needs
Year 2 Year 3 Year 4 Year 5

 

Tablg 8. Projected Human Resources Requirements.

131

BOARD MEMBERS
We are currently in the process of recruiting five individuals to serve on the executive board. We
have identified the following individuals to act as board members:

Dr. Michael H. Brodsky is President of Brodsky Consultants, in Thomhill, Ontario, Canada and
is immediate past president of the Association of Analytical Communities (AOAC)
INTERNATIONAL, the internationally recognized analytical test validation and approval agency
for foods and agriculture. Mr. Brodsky began his career as a research scientist in environmental
bacteriology, for the Laboratory Services Branch of the Ontario Ministry of Health. In 1982 he
became Chief of Environmental Microbiology and Microbiological Support Services for the
Ontario Ministry of Health, a position he held for 17 years. In 1999 Mr. Brodsky retired from the
Ontario Ministry of Health and accepted a one-year appointment as General Manager of Silliker
Laboratories of Canada, and subsequently founded his own consulting firm. In addition to his
many years of service to AOAC as a training course instructor, Board member, and President,
Mr. Brodsky serves as a Technical Assessor for ISO under the auspices of the Standards Council
of Canada, and is active in a number of several professional associations.

Dr. Tom ske is chief executive officer and president of Cogene Biotech Ventures, Ltd, a
private equity fund that focuses on intermediate stage and start-up biotech companies. He
served as senior vice president, human genetics and vaccines discovery at Merck Research
Laboratories, West Point, Pa. and president of the Merck Genome Research Institute. He serves
as an adjunct professor at Baylor College of Medicine, Houston, Texas. Dr. Caskey earned his
medical doctorate from Duke University, Durham, NC. He has received numerous academic and
industry-related honors. He is a member of the National Academy of Sciences and Institute of
Medicine. He is past president of American Society of Human Genetics and the Human Genome
Organization. He served as Chair, Advisory Panel on Forensic Uses of DNA Tests, US.
Congress Office of Technology Assessment from 1989-1990. He was a Committee Member on
DNA Technology in Forensic Science, National Research Council, National Academy of
Sciences, 1989-1991 .

Dr. ev Henl is the past president of the National Food Processors Association. He currently
serves as Senior VP of Technology and Innovation for Ocean Spray Cranberries, Inc and is past
Senior Vice President of Technology and Marketing Services for the Hunt-Wesson Company. He
joined Hunt-Wesson in 1983 as Vice President of Research and Development. His previous
position was Vice President of Corporate Research & Development and Corporate Engineering
for Land O'Lakes, Inc., in Minneapolis, Minnesota. Prior to that he held positions with Pillsbury
Company and General Foods Corporation. Dr. Henig received his Bachelor Science Degree in
Chemical Engineering and a Master of Science Degree in Food and Biotechnology from
Technion-lsrael Institute of Technology. He earned his Ph.D. Degree in Food Science from
Rutgers University, New Brunswick, New Jersey. Steve is a member of the Board of Directors for
Bionutrics Inc. and served as director of LipoGenics from August 1992 until October 1996.

Dr. Paul Hall is on the Executive Board of the lntemational Association for Food Protection. He
serves as Vice President to the Kraft Foods Corporation. We are waiting for an e-mail
confirmation from him.

MANAGERIAL RISKS] WEAKNESSES AND COUNTERMEASURES
o Start-up Management and Growth Experience: The basic team of four founders has
minimal experience in the start-up environment.

@untermeasure in Phasel: While no founder has “started" a company, the team has

identified two experienced business mentors. This will ensure that the first few steps in
structuring the business and technology follow best practices.

132

@untermeagure in Phase 2: As soon as Agen completes building the product and beta tests,
Agen will hire a seasoned sales and marketing expert with the necessary domain expertise.
Agen will also actively seek the help of the Board to identify and recruit a Chief Executive
Officer who can continue the rapid growth of the company by attracting customers and
investors. Subsequent hiring of executives will be per the decision of the CEO.

OVERALL EXISTING MANAGERIAL STRENGTHS
0 Strong technical and product development expertise
o Aggressive growth plan with milestones being executed per planned timelines
0 Strong Board (5 Members)
0 Strong Technical Advisory Board (7 Members)
0 Extremely high motivation, drive, and determination of founders

BUSINESS SYSTEMS AND ORGANIZATION

Research gnd Development: The MEMS technology utilized in the Agen Biosensor is rapidly
evolving. The research and development involved in further perfecting the speed, sensitivity, and
selectivity are critical elements for the company and should be a source of competitive
advantage. In addition to the technology, the development of biological material must be
maintained as it Is critical to have access to the latest antibodies. Agen believes that its current
strengths are very strong in the research area.

Human Resogrces: Agen BioSense is focused on a high caliber of talent. As the company grows,
focus will be on both the research and development team as well as the management team with
the necessary skills to capture the company's market niche.

Marketing and Sales: Agen BioSense will employ a direct sales force to actively seek out new
opportunities. We estimate that initially the company will employ 3 sales resources to focus on
the target markets. Agen will focus the inbound logistics, manufacturing and outbound logistics
subsequent to product development. In the initial phase, it is expected that all these activities will
be outsourced.

Inbound ngistics of Supplies and Eguigment: Inbound logistics of supplies and equipment is the
process to obtain the raw materials from suppliers — electrical circuitry, cartridge housing, reader,

and biological material. This process can range from creating contracts with suppliers and
shipping companies to developing the inbound logistics in house.

Manufacturing: Manufacturing is the process of applying the antigen solution to the cartridge.
This process is currently time consuming, but could be a source of competitive advantage in the
future. The antigen solution is brushed on to the electrical circuitry board and then allowed to dry
for 8 hours. Following the drying period, the coated circuit board is placed in the housing
cartridge and set to outbound logistics.

QMund ngistics: Outbound logistics will pair the cartridge housing and the number of

requested detection readers and prepare them for shipping. Once packaged and ready to ship,
options exist for Agen BioSense to outsource the activity or keep outbound logistics inhouse.

133

OVERALL GROWTH PLAN

 

 

 

 

 

 

 

 

 

 

 

 

Stage Start-up Growth Rapid Growth Continuous
Growth
lPhase Phase 1 Phase 2 Phase 3 Phase 4
0 ~ 2 years 2 ~ 4 years 4 ~ 6 years Year 7
Timeline onwards
Goals Develop Product Drive Sales Lead the market Dominate the
industry
Objectives
Technical LComplete Proof of -Ensure consistent -Complete MEMS -Prototype
concept product performance prototype nanoscale
-Complete 3rd party -Complete multi-sensor -Diversify product sensors
testing -ldentify IP for portfolio -Diversify
-Complete beta acquisition product
testing with portfolio
customers
IFinanclal -Raise Federal grants -Raise angel round and -Raise VC rounds 1-3 -Target
-Build relationships build strategic alliances. potential
Erith angels 8 VCs -Ensure revenues are buyout or
Finalize product- per target. acquisition
ricing -Control Ace. Rec’ble. parent
lManagerlaiI-ldentify industrial -Attract CEO & VP, -Continue building -Build on
partnerships Sales industry specific sales international
-Develop -Add legal expertise force in processed sales force
manufacturing and -Build sales in dairy and foods industry and product
testing vendors meat markets -Enter fruit and diversification
-Build R 8 D Group ~Add Additional R 8 D vegetable market team
staff
Essential -Cash Required: $ -Capital Infusion: $ -Capital Infusion:$4-5M -Human
Resources 100,000 1 .5M Resources,
Required -Existing team to Experienced sales and manufacturin
Isupport growth executive team. 9, and
managerial
talent

 

 

13mg, Growth Plan Milestones for Company Sectors.

INDUSTRIAL PARTNERSHIPS

Agen is in the process of identifying industrial partners to help with the manufacturing and testing
of the Agen single test biosensor. We have initial interest from a well known manufacturing
company who specializes in biosensor and MEMS based manufacturing. We are looking to form
a partnership with them to increase the production scale-up of the single test biosensor.
Additionally, we have received interest from the Kraft Foods corporate testing facility to help with
our third party testing efforts and validation.

These deals are tentative and may involve equity, sublicense agreements, service contracts, or
other financial arrangements in the final negotiations.

134

 

RI KS AND C UNTERMEASURES

EXTERNAL RISKS

Developing Technologies: Several new technologies are being developed in the rapid
pathogen detection space. Some of these include the use of fiber optics while other
techniques use silicon I DNA based technologies.

Countermeasure: Agen will aggressively continue to develop the first product for E. coli
testing, ensure that beta customers are lined up and the third party testing is completed
quickly.

Entry of larger players I Better products from existing players: Several companies like
bioMerieux have been aggressively investing upwards of 12 to 15 % of their revenues in
research and development. It is inevitable that large companies will come up with new and
improved technologies.

Countermeasure: Agen will negotiate strategic alliances with the most appropriate partners to
ensure that its products are accepted rapidly in the market place. Focus will be on leveraging
our resources with others that have existing distribution and market channels. A significant
emphasis will be on new product development as we attempt to continue the technological
advantage our products currently offer.

Regulation: The food industry is regulated by the Food & Drug Administration (FDA), which
has implemented the Hazard Analysis and Critical Control Point (HACCP) guidelines. In
1998, the US. Department of Agriculture established HACCP for meat and poultry
processing plants. Most of these establishments were required to start using HACCP by
January 1999. Very small plants had until Jan. 25, 2000. (USDA regulates meat and poultry,
FDA all other foods) FDA now is considering developing regulations that would establish
HACCP as the food safety standard throughout other areas of the food industry, including
both domestic and imported food products.

Countermeasure: While the FDA has established the HACCP guidelines, the food processing
companies have lobbied extensively to ensure that the industry is not unduly regulated.
Should the regulation change, Agen will have to react rapidly to ensure that its market
position is not affected and product development strategy is in line with regulation.

INTERNAL RISKS
The success of the company depends on effective attraction and growth of financial, managerial
and technical resources:

Financial I Funding: In the initial product development stages, the progress of the company
depends on availability of financial resources. Lack of funding can delay the implementation
strategy and negatively affect our milestone achievements.

Quntermeasure: Agen will pursue a multiple pronged strategy to attract federal and state
grant funding in the early stages. Several such funding sources have been identified
(SBlR/STTR, ATP, MLSC) and serious efforts are being made to maximize the probability of
being funded through these programs. Additionally, as we meet our product manufacturing
goals and obtain beta customers, we will be seeking Angel and V0 investors to help us
continue our aggressive growth potential.

Technical / Product Development: The current sensor Is In advanced stages of

development and testing. Several risks exist at this stage In terms of stable product
performance under severe test conditions. Third Party testing Is essential for

135

validating the reliability of the tests. If the product falls to perform at any of the testing
stages, design efforts will have to be revamped.

quntermeasure: Agen has confidence in the high profile scientific advisory board and
believe we are allocating adequate financial and technical resources to ensure that product
development is not compromised. However, we understand that validation testing does not
always go as planned, thus we will continue product improvements and new product
development to emphasize a solid product pipeline.

0 Managerial I Attraction of key personnel: The existing team of Agen is in a position to
grow the company to the beta test and initial market acceptance stage. However, it will be
necessary to recruit highly experienced management talent when we begin an aggressive
growth stage. In particular, we will need to attract a high profile marketing and sales
professional as our current team is lacking the necessary skills.

Countermeasure: After crossing the initial milestones, the company will explore the possibility
of hiring senior executives from the industry. Stock options, high growth industry and the
challenge of the start-up environment may be some of the attraction tools. Most importantly,
we believe the board of directors we are recruiting will help minimize the risk of failure and act
as an incentive for recruiting executive talent.

FL! I QRE GRQWTH

Agen's growth strategy will utilize a portfolio of technologies that would encompass the following:

o Pathogen Prevention Technologies
0 Pathogen Detection
o Pathogen Elimination

The company will initially focus on their competitive strengths to market the pathogen detection
biosensors product line. As significant progress is accomplished, efforts will be made to help
offset the risk of competing companies and technologies. We will continue to improve existing
biosensor technology and develop new biosensor capabilities until we have built significant brand
awareness in this space. Ultimately we aim to offer products for the complete spectrum of
pathogen specific needs.

FOOD SAFETY

The market potential for detection and identification of bacterial and viral pathogens in the food
safety area is estimated at around $100 million per year. Applications include detecting
contaminants in food raw materials, food products, processing and assembly lines, and water
supplies continue to rely on conventional culturing techniques. Currently, detection techniques
typically require 2-3 days, and thus do not alert industrial producers to quality control problems
until well after the fact. Real time testing will provide value to food producers through the
elimination of product recalls and reduced treatment costs.

ENVIRONMENTAL QUALITY

Pathogens present in the environment is becoming crucial to a wide range of industries, including
food, pulp and paper, cosmetics, metals, plastics, petrochemical and power generation. With
greater pressure to recycle water, minimize the use of antibacterial agents, and maintain quality
discharges, manufacturers in a wide variety of industries are seeking technologies to rapidly
identify contamination problems at the source.

For example, Cryptosporidium parvum is a waterborne pathogen infective at a dose of a single
organism. It is responsible for frequent widespread outbreaks of intestinal disease that can be
life-threatening for individuals with compromised immune systems. To detect the presence of
such organisms, there is a need for rapid biological testing systems that can concentrate the

136

organisms from several gallons of water. Real time, on-site testing systems will play an important
role in further enabling the detection of environmental pathogens.

BIODEFENSE
As the threat of domestic and international bioterrorism continues to grow, so does the need for

rapid, automated, field-based tests for pathogenic agents, as well as faster, more specific
laboratory bioanalysis and detection systems. Military units facing an enemy with the potential for
an arsenal of biological weapons require the ability to monitor the environment and provide at-risk
troops with the means to rapidly identify contaminated air, water, food, and equipment. Testing
may also be helpful in guiding cleanup after an attack with spore-forming agents such as anthrax,
which can persist in the environment for years.

Field-ready systems are being deployed to enable environmental surveillance because the
biological agents most likely to be used in a terrorist attack do not immediately produce effects.
Currently, samples taken from the environment, such as soil and water, and most clinical samples
must be cultured for reliable identification, typically requiring 4—48 hours before a result is
available. Real time, highly sensitive on-site testing systems will play an important role in
enabling timely detection of these types of pathogens.

137

FINANCIAL PLANNING é FINANCING

A gen BioSense

 

 

 

 

 

 

 

Income Statement (3)

2002 2003 2004 2005 2006
Revenue
All Cartridges $300,000 $2,400,000 $13,440,000 $16,406,000 $21,089,900
Base unit $2,500 $72,500 $103,675 $134378 $191,139
Total Revenue $302,500 $2,472,500 $13,543,675 $16,540,778 $21,281,039
Cost of Goods Sold $56,250 $612,875 $2,510,176 $3,264,860 $4,496,729
Gross Margin $246,250 $1,859,625 $11,033,499 $13,275,917 $16,784,310
% of Revenue 81% 75% 81% 80% 79%
Operating Expenses
Engineering $1 17.600 $400,949 $856,910 $1,039,787 $1 .081 .607
% of Revenue 39% 16% 6% 6% 5%
Marketing/Sales $196,253 $1,145,366 $4,986,861 $6,175,866 $7,377,540
% of Revenue 65% 46% 37% 37% 35%
Administration $89,288 $313,783 $685,535 $839,287 $988,742
% of Revenue 30% 13% 5% 5% 5%
Total Operating
Expenses $403,141 $1,860,097 $6,529,306 $8,054,941 $9,447,889
% of Revenue 133% 75% 48% 49% 44%
Income Before Int 8t
Taxes ($156,891) ($472) $4,504,193 $5,220,977 $7,336,421
% of Revenue -52% 0% 33% 32% 34%
Income Before Taxes ($156,891) ($472) $4,504,193 $5,220,977 $7,336,421
Tax Exp $0 $0 $1,735,714 $2,088,391 $2,934,568
Net Income ($156,891) ($472) $2,768,480 $3,132,586 $4,401 .853
% of Revenue -52% 0% 20% 19% 21%

138

 

A gen BioSense

 

Balance Sheet (3)
2002 2003 2004 2005 2006
ASSETS
Current Assets
Cash ($385,674) ($310,323) $2,042,885 $5,067,686 $9,372,121
Net Accounts Rec $299,475 $210,375 $1,117,353 $1,364,614 $1,755,686
Inventory (15 days) $22,500 $101,082 $132,428 $183,653 $237,099
$1 1,364.90
Total Current Assets ($63,699) $1,135 $3,292,666 $6,615,953 6
Gross Fixed Assets $32,500 $88,500 $142,500 $179,500 $174,000
Less Accum Depreciation $5,750 $25,458 $72,958 $121,958 $150,458
Net Fixed Assets $26,750 $63,042 $69,542 $57,542 $23,542
$1 1,388.44
TOTAL ASSETS ($36,949) $64,176 $3,362,208 $6,673,495 8
LIABILITIES
Short Term Liabilities
Accounts Payable (30 days) $95,150 $173,257 $246,293 $321,158 $423,044
Salaries Payable (15 days) $14,792 $38,282 $60,869 $76,536 $76,206
Taxes Payable (90 days) $0 $0 $433,928 $522,098 $733,642
Line of Credit (10% of net AIR) $0 $0 $0 $0 $0
Current Portion of Capital
Equipment Lease $0 $0 $0 $0 $0
Current Portion of Long Term
Debt $0 $0 $0 $0 $0
Total Short Term LIabllltles $109,942 $211,539 $741,091 $919,792 $1,232,892
TOTAL LIABILITIES $109,942 $211,539 $741,091 $919,792 $1,232,892

 

 

 

 

 

139

 

BEGINNING CASH

Sources of Cash

Net Income

Add Depr/Amort

Plus Changes In:

Accounts Payable (30 days)
Salaries Payable (15 days)
Taxes Payable (90 days)
Total Sources of Cash

Uses of Cash

Less Changes In:
Net Accounts Rec
Inventory (15 days)
Gross Fixed Assets
Total Uses

CHANGES IN CASH

ENDING CASH

A gen BioSense

Statement of Sources & Uses (3)

2002

2003

$10,000 ($385,674)

2004 2005 2006

($310,323) $2,042,885 $5,067,686

 

($156,891) ($472) $2,768,480 $3,132,586 $4,401 .853
$5,750 $19,708 $47,500 $49,000 $28,500
$95,150 $78,107 $73,036 $74,865 $101 .886
$14,792 $23,490 $22,587 $1 5.667 ($330)
$0 $0 $433,928 $88,169 $21 1 .544

($41 .199) $120,834 $3,345,532 $3,360,287 $4,743,453

$299,475 ($89.1 00)

$906,978 $247,261 $391,072

 

 

$22,500 $78,582 $31 .346 $51 .225 $53,446
$32,500 $56,000 $54,000 $37,000 ($5.50Q
$354,475 $45,482 $992,324 $335,486 $439.01 7
($395,674) $75,351 $2,353,208 $3,024,801 $4,304,435

 

($385,674) ($310,323)

140

$2,042,885 $5,067,686 $9,372,121

BIBLIOGRAPHY

References

l. Abdel-Hamid, 1., Ivnitski, D., Atanasov, P., Wilkins, E., 1998. Flow-through
immunofiltration assay system for rapid detection of E. coli 0157:H7. Biosensors
and Bioelectronics 14:3:309-316.

2. Alocilja, E., and Radke S., 2003. Market analysis of biosensors for food safety.
Biosensors and Bioelectronics 18:841-846.

3. Bao, L., Deng, L., Nie, L., Yao, S., Wei, W., 1996. Determination of
microorganisms with a quartz crystal microbalance sensor. Analytical Chemistry
Acta 319:97-101.

4. Bard, A.,]. and Faulkner, L. R., 1980. Electrochemical Methods: Fundamentals
and Applications. J. Wiley, New York, NY USA.

5. Bhatia, S., Shriver-Lake, L., Prior, K., Georger, J., Calvert, J., Bredehorst,
R., Ligler, F., 1989. Use of thiol-terminal silanes and heterobifunctional
crosslinkers for immobilization of antibodies on silica surfaces. Analytical
Biochemistry 178:408—413.

6. Binns, K. and Lawrenson, P., 1973. Analysis and computation of electric and
magnetic field problems. Pergamon (Oxford), 25-29 and 151-187.

7. Bockris, J. and Reddy, A., 1970. Modern Electrochemistry. Plenum Press, New
York, NY USA.

8. Borkholder, D., 1998. Cell based biosensors using microelectrodes. Dissertation.
Stanford University, Stanford, CA. USA.

9. Brooks, J., Mirhabibollahi, B., Kroll, R., 1992. Experimental enzyme linked
amperometric immunosensors for the detection of Salmonella in foods. Journal of
Applied Bacteriology 73: 189—196.

10. Burnett, A., Beuchat, L., 2001. Comparison of sample preparation methods for
recovering Salmonella from raw fruits, vegetables. and herbs. Journal of Food
Protection 64:10: 1459-1465.

11. Buzby, J., Roberts, T., Lin, C., MacDonald, J., 1996. Bacterial foodborne

disease: medical costs and productivity losses. Agricultural Economics Report
No. 741:100. USDA Economic Research Service. Washington DC USA.

141

12. CDC-Centers for Disease Control and Prevention, 2002. Report on the Decline
of Foodbome Illness. Atlanta. GA.

13. CDC-Centers for Disease Control and Prevention, 2001. Outbreaks caused by
Shiga toxin-producing Escherichia coli. Summary of 2000 Surveillance Data.
Atlanta, GA.

14. CDC-Centers for Disease Control and Prevention, 2004. Bioterrorism
agents/diseases, http://www.bt.cdc.gov/agentlagentlist-category.asp.

15. Che, Y., Li, Y., Slavik, M., Paul, D., 2000. Rapid detection of Salmonella
Typhimurium in chicken carcass wash water using an immunoelectrochemical
method. Journal of Food Protection 13:8: 1043-1048.

16. Cohn, G.E., 1998. Systems and Technologies for Clinical Diagnostics and Drug
Discovery. SPIE Proceedings Vol. 3259. Bellingham, WA: The lntemational
Society for Optical Engineering.

17. Ciureanu, M., Levadoux, W., Goldstein, S., 1997. Electrical impedance studies
on a culture of a newly discovered strain of Steptomyces. Enzyme and Microbial
Technology 21:441-449.

18. DeMarco, D. and Lim, D., 2002. Detection of Escherichia coli 0157:H7 in 10-
and 25-gram ground beef samples with an evanescent-wave biosensor with silica
and polystyrene waveguides. Journal of Food Protection 65:4:596-602.

19. Doyle, M.P., Zhao, T., Meng. J., Zhao, S., 1997. Escherichia coli 0157:H7. In
Food Microbiology Fundamentals and Frontiers, 171-191. M. Doyle, L Beuchat,
T. Montville eds. Washington, DC: American Society for Microbiology.

20. D'Souza, S., 2001. Microbial biosensors. Biosensors and Bioelectronics 16:337-
353.

21. Diniz, F., Ueta, R., Pedrosa, A., Areias, M., Pereira, V., Silva, E., da Silva, J .,
Ferreira, A., Comes, Y., 2003. Impedimetric evaluation for diagnosis of Chaga's
disease: antigen-antibody interaction on metallic electrodes. Biosensors and
Bioelectronics 19:79-84.

22. Dutra, R., Castro, C., Azevedo, C., Vinhas, E., Malagueno, E., Melo, E.,
Lima-Filho, J ., Kennedy, J ., 2000. Immobilization of pneumococcal
polysaccharide vaccine on silicon oxide wafer for an acoustical biosensor.
Biosensors and Bioelectronics 15:511-514.

23. Ehret, R., Baumann, W., Brishwein, M., Schwinde, A., Stegbauer, K., Wolf,

B., 1997. Monitoring of cellular behavior by impedance measurements on
interdigitated electrode structures. Biosensors and Bioelectronics 12:1:29-41.

142

24. FDA-Food and Drug Administration, 1998. FDA/CFSAN Bad Bug Book.
Rockville, MD USA.

25. FDA-Food and Drug Administration, 2001. Bacteriological Analytical Manual.
Rockville, MD USA.

26. Food Manufacturing Coalition, 1997. Real Time Monitoring of Foodbome
Pathogens: State-of—the-Art Report. PO. Box 741. Great Falls. Virginia 22066.

27. Fratamico, P., Strobaugh, T., Medina, M., Gehring, A., 1998. Detection of
Escherichia coli 0157:H7 using a surface plasmon resonance biosensor.
Biotechnology Techniques 1227:571-576.

28. Gau, J., Lan, E., Dunn, B., Ho, C., Woo, J., 2001. A MEMS based
amperometric detector for E. coli bacteria using self-assembled monolayers.
Biosensors and Bioelectronics 16: 745-755.

29. Gomez, R., Bashir, R., Sarikaya, A., Ladisch, M., SturgiS, J ., Robinson, T.,
Gong, A., Bhunia, H. Apple, 2003. Microfluidic biochip for impedance
spectroscopy of biological species. Biomedical Microdevices 3:201-209.

30. Gorschliiter, A., Sundermeier, C., Rob, B., Knoll, M., 2002. Microparticle
detector for biosensor applications. Sensors and Actuators B, 85: 158-165.

31. Ghindilis, A., Atanasov, P., Wilkins, M., Wilkins, E., 1998. Irnmunosensors:
electrochemical sensing and other engineering approaches. Biosensors and
Bioelectronics 13:113-131.

32. Gimsa, J., Muller, T., Schnelle T., Fuhr, G., 1996. Dielectric spectroscopy of
single human erythrocytes at physiological ionic strength: dispersion of the
cytoplasm. Biophysics Journal 71:495-506.

33. Grimes, C., Kouzoudis, D., Orig, K., Crump, R., 1999. Thin-film
magnetoelastic microsensors for remote query biomedical monitoring. Biomedical
Microdevices 2: 1 :51-60.

34. HHS-US Department of Health and Human Services, 2000. HHS Fact Sheet:
HHS Initiative to Reduce Foodbome Illness. March 16, 2000. Washington, DC.
Available at http://www.hhs.gov/news/press/2000pres/20000316a.html

35. Hoyle, B. 2001. High-tech biosensor speeds bacteria detection. ASM News
67:9:434-435.

143

36. Jacobs, P., Varlan, A., Sansen, W., 1995. Design optimization of planar
electrolytic conductivity sensors. Medical Biological Engineering Computation
33:802-810.

37. Jung, A., Stemmler, I., Brecht, A., Gauglitz, G., 2001. Covalent strategy for
immobilization of DNA-microspots suitable for rnicroarrays with label-free and
time-resolved optical dectection of hydridization. Fresenius Journal of Analytical
Chemistry 371:128-136.

38. Keene, W.E., McAnulty, J-M., Hoesly, F.C., Williams, L.P., Hedberg, K.,
Oxman, G.L., Barrett, TJ., Pfaller, M.A., Fleming, D.W., 1994. A swimming
associated outbreak of hemorrhagic colitis caused by Escherichia coli 0157:H7
and Shigella sonnei. New England Journal of Medicine 331:579-584.

39. KPL-Kirkegaard and Perry Laboratories, Inc., 1992. Polyclonal affinity
purified goat antibody to E. coli 0157:H7 product sheet. Gaithersburg, MD USA.

40. Kohnen, A. 2000. Responding to the threat of agroterrorism: specific
recommendations for the USDA. BCSIA Discussion Paper 2000-29, ESDP
Discussion Paper ESDP-2000-04, John F. Kennedy School of Government.
Harvard University.

41. Koppes, L., Woldringh, C., Nanninga, N., 1978. Size variations and correlation
of different cell cycle events in slow-growing Escherichia coli. Journal of
Bacteriology 134:423-433.

42. Kovacs, G, 1998. Micromachined transducers sourcebook. McGraw Hill. New
York, NY. USA.

43. Liu, Y., Ye, J., Li, Y., 2003. Rapid detection of Escherichia coli 0157:H7
inoculated in ground beef, chicken carcass and lettuce samples with an

immunomagnetic chemiluminescence fiber optic biosensor. Journal of Food
Protection 66:3:512-517.

44. Madlgan, M., Martinko, J., Parker, J., 1997. Brock Biology of
Microorganisms. 8th edition. Upper Saddle River, NJ: Viacom.

45. Madou, M., 2002. Fundamentals of microfabrication: the science of
miniaturization, 2nd edition. CRC Press, Boca Raton, FL, USA.

46. Markx, G. and Davey, C., 1999. The dielectric properties of biological cells at
radiofrequencieszApplications in biotechnology. Enzyme and Microbial
Technology 25:161-171.

47. Mathew, F., Alagesan, D., Alocilja, E., 2004. Chemiluminescence detection of
Escherichia coli in fresh produce obtained from different sources. Luminescence

144

19:4:193-198.

48. McAdams, E., Lackermeier, A., McLaughlin, J ., Macken, D., 1995. The linear
and non-linear properties of the electroe-electrolyte interface. Biosensors and

Bioelectronics 10:67-74.

49. McClelland, R. and Pinder, A., 1994. Detection of Salmonella typhimurium in
dairy products with flow cytometry and monoclonal antibodies. Applied
Environmental Microbiology 60:4255-4262.

50. Mead, P.S., Slutsker, L., Dietz, V., McCaig, L., Bresee, J., Shapiro., Griffin,
P., and Tauxe, R., 1999. Food-related illness and death in the United States.
Emerging Infectious Diseases 5:607-625.

51. Meng, J., Zhao, 8., Doyle, M., Kresovich, S., 1996. Polymerase chain reaction
for detection E. coli 0157:H7. lntemational Journal of Food Microbiology 32:1-
2:103-1 13.

52. Meeusen, C., Alocilja, E.C., Osburn, W., 2001. “Detection of E. coli 0157 :H7
using a surface plasmon resonance biosensor.” Transactions of the ASAE,
(submitted for review 2003).

53. MSU-Michigan State University, 1998. Biosafety Manual. Office of Radiation.
Chemical, and Biological Safety (ORCBS).

54. Muhammad-Tahir, Z. and Alocilja, E., 2003. A conductometric immunosensor
for biosecurity. Biosensors and Bioelectronics 18:813-819.

55. Muhammad-Tahir, Z and Alocilja, E., 2003. Fabrication of a disposable
biosensor for Escherichia coli 0157:H7 detection. IEEE Sensors Journal 3:4:345-
351.

56. NIAID-National Institute of Allergy and Infectious Disease, 2004.
NIAID Category A, B & C Priority Pathogens,
http://www.niaid.nih.gov/biodefense/bandc_priority.htm.

57. Nilsson, J. and Riedel, S., 1996. Electric circuits, 5th Edition. Addison-Wesley
Publishing Company, Reading, MA, USA.

58. Ong, K., Bitler, J ., Grimes, C., Puckett, L., Bachas, L., 2002. Remote query
resonant-circuit sensors for monitoring of bacteria growth: application to food

quality control. Sensors 2:219-232.

59. Radke, S., and Alocilja, E., 2004. Design and Fabrication of an Impedimetric
Biosensor. IEEE Sensors Journal 4:434-440.

145

60. Rishpon, J., Ivnitski, D., 1997. An amperometric enzyme—channeling
immunosensor. Biosensors and Bioelectronics 12:3: 195-204.

61. Ruan, C., Wang, H., Li, Y., 2002. A bienzyme electrochemical biosensor
coupled with immunomagnetic separation for rapid detection of E. coli 0157:H7
in food samples. Transactions of the ASAE, 45:249-255.

62. Sanderson, M., Gay, J., Hancock, D., Gay, C., Fox, L., Besser, T., 1994.
Sensitivity of bacteriologic culture for detection of Escherichia coli 0157:H7 in
bovine feces. Journal of Clinical Microbiology 33:10:2616-2619.

63. Seo, K.H., Brackett, R.E., Hartman, N.F., Campbell, D.P., 1999. Development
of a rapid response biosensor for detection of Salmonella Typhimurium. Journal
of Food Protection 62:5:431-437.

64. Sergeyeva, T., Piletskii, S.A., Rachkov, A.E., El’Skaya, A.V.. 1996.
Polyaniline label-based conductometric sensor for IgG detection. Sensors and
Actuators B 34:283-288.

65. Sheppard, N., Mears, D., Guiseppi-Elie, A., 1996. Model of an immobilized
enzyme conductimetric urea biosensor. Biosensors and Bioelectronics 11:10:967-
979.

66. Shriver-Lake, L., Donner, B., Edelstein, R., Breslin, K., Bhatia, S., Ligler, F.,
1997. Antibody immobilization using heterobifunctional crosslinkers. Biosensors
and Bioelectronics 12:11:1101-1106.

67. Sperveslage, J., Stackebrandt, E., Lembke, F., Koch, C., 1996. Detection of
bacterial contamination, including bacillus spores, in dry grth media and in
milk by identification of their 168 RDNA by polymerase chain reaction. Journal
of Microbiolog. Methods 26:3:219-224.

68. Suehiro, J., Hamada, R., Noutomi, D., Shutou, M., Hara, M., 2003. Selective
detection of viable bacteria using dielectrophoretic impedance measurement
method. Journal of Electrostatics 57:157-168.

69. Taylor, R. and Schultz, J., 1996. Handbook of chemical and biological sensors.
IOP Publishing, London, UK.

70. Tien, H. and Ottova-Leitmannova, A., 2000. Membrane biophysics as viewed
from experimental bilayer lipid membranes. Elsevier Science. Amsterdam, The
Netherlands.

71. Turner, A.P.F., Karurbe, 1., Wilson, G.S. (Eds.), 1987. Biosensors
Fundamentals and Applications. Oxford University Press. Oxford, UK.

146

72. Van Gerwen, P., Laureyn, W., Laureys, W., Huyberechts, G., De Beeck, M.,
Baert, K., Suls, J., Sansen, W., Jacobs, P., Hermans, L., Martens, R., 1998.
Nanoscaled interdigitated electrode arrays for biochemical sensors. Sensors and
Actuators B 49:73-80.

73. Van Zant, P., 2000. Microchip fabrication. McGraw Hill, New York, NY, USA.

74. Venable, R., Brooks, B., Pastor, R., 2000. Molecular dynamics simulations of
gel (L-beta-I) phase lipid bilayers in constant pressure and constant surface area
ensembles. Journal of Chemical Physics 11221024822-4832.

75. WHO-World Health Organisation, 2002. Terrorist threats to food: guidance for
establishing and strengthening prevention and response systems. Geneva,
Switzerland.

76. Wiegand, G., Arribas-Layton, N., Hillebrandt, H., Sackmann, E., Wagner,
P., 2002. Electrical properties of supported bilayer membranes. Journal of
Physical Chemistry B 106:4245-4254.

77. Yang, L., Li, Y., Griffis, C., Johnson, M., 2004. Interdigital microelectrode

(IME) impedance sensor for the detection of viable Salmonella Typhimurium.
Biosensors and Bioelectronics 19:1139-1147.

147

   

IIIIIIIIIIIIIIIIIIIIIIIIIIIII‘II