21" \\..
a. . . «Iii»; .
3: x 0.11}? ..

 

\oa.‘ .31., 1A....
1.5.... '35::

 

sv‘tflunnwbt.

! .3

3...? v):

 

 

LIBRARY
Michigan State

I 'n" Inrrdh
U. II'GIOI‘]

 

 

 

This is to certify that the
dissertation entitled

DEVELOPMENT AND VALIDATION OF A MULTIPLEX

Ph.D.

HAND-HELD GENE ANALYZER

presented by

ROBERT D. STEDTFELD

has been accepted towards fulfillment
of the requirements for the

degree in Environmental Engineeriml

 

 

x as..."

[law

 

 

 

 

 

Major Professor’s Signature

Sela la, 2coq

 

Date

MSU is an Affirmative Action/Equal Opportunity Employer

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Acc&Pres/ClRC/DateDue.indd

 

DEVELOPMENT AND VALIDATION OF A MULTIPLEX HAND-HELD
GENE ANALYZER

By

Robert D. Stedtfeld

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Environmental Engineering

2009

ABSTRACT

DEVELOPMENT AND VALIDATION OF A MULTIPLEX HAND-HELD
GENE ANALYZER

By

Robert D. Stedtfeld

The potential for microbial diagnostics to impact health care in developing
regions of the world is well recognized. However, there is a large gap in the use of such
technologies due to the unavailability of low cost point of care genetic screening tools.
The device developed within this project fills this gap, with the potential for providing
genetic screening at a cost that is well within reach of resource-limited regions (<
$1,000). This was accomplished in a synergistic manner by means of: i) highly parallel
validation of real time polymerase chain reaction (RT-PCR) amplification and
hybridization-based molecular approaches, ii) development of a hand-held real-time gene
analyzer, iii) validation of the device using isothermal loop mediated amplification
(LAMP), and iv) embedding the component into a hand-held platform.

Validation of primers and probes was accomplished using an existing high
throughput real time thermocycler for RT- PCR, and an in situ synthesized microarray for
hybridization-based studies. The limits of the hybridization-based approach for detection
of closely related fragments will be useful in developing assays for mutation detection.
In addition to validating 220 primers sets targeting 30 microbial pathogens, a predictive
equation for quantification of starting copies was experimentally established based on the

sequence characteristics of primers and target organisms (as an efficient alternative to

generating standard curves). Validated primers can be used in the gene analyzer afier it is

fully developed.

Constraints in the development of truly portable and low cost gene analyzers are
typically associated with the means of detection. Less costly components such as light
emitting diodes (LEDs) and photodiodes are promising alternatives to charge coupled
devices and lasers; however, simultaneous detection of multiple channels with spot
detectors typically requires movement or assigning independent detectors to each channel
on the device. To overcome this issue, an array of sixteen individually controlled LEDs
(one LED for each reaction well) was used with optical fibers sending fluorescent signal
to a single photodiode detector. Only one LED/well/optical fiber is turned “on” at a time
and identity is determined by the associated sequence and timing of the photodiode
reading. Optical cross talk between reaction wells was avoided using a thin-wall
(shelled) microfluidic chip. The disposable chips were pressed in 100-pm thin Zeonor
film by hot embossing with a male mold fabricated using stereolithography and a thermal
plastic counter tool.

Components and control was embedded into a hand-held footprint through design
of a custom printed circuit board, control with a microcontroller, and design of a cartridge
to package the device. Comparable results were observed between the gene analyzer and
a state-of-the-art real-time PCR instrument, demonstrating that this simple, small, low
component cost system has the potential to play a key role in propagating the use of gene

based screening devices in resource-limited settings.

Copyright by
Robert D. Stedtfled
2009

This thesis is dedicated to my family for their
support including: my Mother and Father, Brother
and Sister, and to my Wife and Son for their

patience and unconditional love.

ACKNOWLEDGEMENTS

I would like to thank my lab-mates for their contribution to this research and also for
making the laboratory a fun and stimulating working environment. And thank Zheng
Chen, a graduate student from the Department of Electrical Engineering, who helped
me with basic electronics and control. I would also like to thank Scott Watson,
Jimmy Johnson, Daphne Stoner, and Paul Meakin, whom I worked with at the Idaho
National Laboratory for one year prior to attending graduate school at Michigan State
University. They shared in my first experiences in research and showed me the fun
side of science.

This work was supported by the National Institutes of Health (Grant R01
RR018625-01), Environmental Protection Agency (RD83162801-0 & RD83301001),
and 21St Century Michigan Economic Development Corporation (GR-476 PO

085P3000517).

vi

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................................ x
LIST OF FIGURES .............................................................................................................. xi

CHAPTER 1: INTRODUCTION

1.1 Impact of low cost point-of-care microbial diagnostic devices ................................... 1
1.2 Comparative evaluation of selected real time PCR devices ........................................ 3
1.3 Selection of parameters in developing the gene analyzer ............................................ 5
1.4 Overview of the project ............................................................................................... 8
1.5 Organization of this thesis ........................................................................................... 9
1.6 Objective of this research ........................................................................................... 14
1.7 References .................................................................................................................. 22

CHAPTER 2: DEVELOPMENT AND EXPERIMENTAL VALIDATION OF A PREDICTIVE
THRESHOLD CYCLE (CT) EQUATION FOR QUANTIFICATION 0F VIRULENCE AND
MARKER GENES USING HIGH-THROUGHPUT NANOLITER PCR ON OPENARRAYS

2.1 Abstract ...................................................................................................................... 25
2.2 Introduction ................................................................................................................ 26
2.3 Materials and Methods ............................................................................................... 27
2.3.1 DNA targets ......................................................................................................... 27
2.3.2 Complex background gDNA ............................................................................... 28
2.3.3 Design of PCR primers to establish the predictive Ct equation ........................... 29
2.3.4 Design of PCR primers to validate correlations used in development of
predictive Ct equation ................................................................................................... 29
2.3.5 PCR on BiOTrove OpenArrayTM plates ............................................................... 30
2.3.6 Design of sample mixtures for development of predictive Ct equation .............. 31
2.3.7 Design of validation sample mixtures for validation of predictive Ct equation ..31
2.3.8 Data analysis ........................................................................................................ 32
2.4 Results and Discussion .............................................................................................. 33
2.4.1 Sensitivity and specificity .................................................................................... 33
2.4.2 Development of predictive Ct equation ............................................................... 36
2.4.3 Validation of an organism’s sequence characteristics on primer success rate ..... 39
2.4.4 Validation of predictive Ct equation .................................................................... 40
2.5 Conclusion ................................................................................................................. 43
2.6 Acknowledgments ...................................................................................................... 43
2.7 Figures ........................................................................................................................ 48
2.8 References ................ 54

CHAPTER 3: INFLUENCE OF DANGLING END AND SURFACE PROXIMAL TAIL 0F
TARGETS ON PROBE-TARGET DUPLEX FORMATION IN 168 RRNA GENE-BASED

DIAGNOSTIC ARRAYS

3.1 Abstract ...................................................................................................................... 61
3.2 Introduction ................................................................................................................ 62
3.3 Materials and Methods ............................................................................................... 64

vii

3.3.1 Preparation Of target DNA ................................................................................... 65

3.3.2 Preparation of background DNA ......................................................................... 68
3.3.3 Probe design and microarray synthesis ................................................................ 70
3.3.4 Experimental approach ........................................................................................ 71
3.3.5 Hybridization and scanning ................................................................................. 73
3.3.6 Data analysis ........................................................................................................ 74
3.4 Results and Discussion .............................................................................................. 75
3.4.1 Klenow/random primer labeled target mixture vs. end-labeled 45-mer
synthetic target mixture ................................................................................................. 75
3.4.2 Synthetic targets of increasing dangling end length ............................................ 80
3.4.3 Synthetic targets with surface proximal tail sequence complementary to
background DNA .......................................................................................................... 82
3.5 Conclusions ................................................................................................................ 83
3.6 Acknowledgments ...................................................................................................... 84
3.7 Figures ........................................................................................................................ 88
3.8 References .................................................................................................................. 98

CHAPTER 4: DEVELOPMENT OF EMBEDDED CONTROL SYSTEMS FOR GENE
ANALYZER

4.1 Introduction ................................................................................................................ 103
4.2 Temperature control using LabVIEW and breadboard .............................................. 104
4.2.1 Hardware for coupled PID/PWM implementation .............................................. 105
4.2.2 Software for coupled PID/PWM implementation ................................................ 108
4.2.3 Performance of coupled PID/PWM implementation ........................................... 112
4.3 Detection system control using LabVIEW and breadboard ...................................... 114
4.3.1 Hardware for detection system control ................................................................ 115
4.3.2 Software for detection system control ................................................................. 118
4.3.3 Performance of the detection system ................................................................... 123
4.4 Design of custom PCB to accommodate ARM microcontroller, ICs, and LED drivers
.......................................................................................................................................... 125

4.5 Replacement of USB-DAQS and PC with custom ICs and ARM microcontroller ...130
4.6 Development of circuit to establish communication between ARM microcontroller

and an iPod Touch ........................................................................................................... 136
4.7 Design of Housing ..................................................................................................... 137
4.8 Concluding Remarks .................................................................................................. 141
4.9 References .................................................................................................................. 142
CHAPTER 5: VALIDATION OF OPTICAL SYSTEM FOR GENE ANALYZER
5.1 Introduction ................................................................................................................ 143
5.2 Materials and Methods ............................................................................................... 146
5.2.1 Disposable amplification chip fabrication and ABS holder ................................ 146
5.2.2 Component setup ................................................................................................. 147
5.2.3 Bench-top control ................................................................................................. 148
5.2.4 Characterization of LAMP reaction using gene analyzer ..................................... 150
5.3 Results and Discussion .............................................................................................. 151
5.3.1 Microstructered amplification chip ...................................................................... 151

viii

5.3.2 Characterization of the gene analyzer .................................................................. 155

5.3.3 Refining the gene analyzer for RT-PCR .............................................................. 162
5.4 Conclusion ................................................................................................................. 165
5.5 References .................................................................................................................. 166
CHAPTER 6: ENGINEERING SIGNIFICANCE AND FUTURE PERSPECTIVES
6.1 Engineering significance ............................................................................................ 171
6.2 Future perspectives toward finalizing the gene analyzer prototype ........................... 174
6.3 References .................................................................................................................. 176

ix

LIST OF TABLES

Table 1.1. Comparison of parameters for representative RT-PCR machines .................. 5

Table 1.2. Comparison of user needs specified in the Customer Requirement
Document-Primary Care Level to the proposed capabilities of various devices in
this project ......................................................................................................................... 16

Table 2.1. Pathogens and VMGS targeted with OpenArrayTM plates and organisms
used for validation ............................................................................................................. 45

Table 3.1. Identity of pathogens, number of probes and range of Gibbs fiee energy
for the set of probes ........................................................................................................... 85

Table 3.2 Characteristics of various DNA sequences used to study the effect of dangling
end and surface proximal tail on signal intensity .............................................................. 86

Table 6.1 Component costs for Gene Analyzer prototype with iPod Touch and Software
........................................................................................................................................ 1 72

LIST OF FIGURES
Images in this dissertation are presented in color.

Figure 1.1. Depiction of the options for various parameters related to an analytical
device ............................................................................................................................... 8

Figure 1.2. Major considerations for development of the device .................................. 9

Figure 2.1. Impact of gDNA from various environmental water samples and primer

stability on specificity and sensitivity .............................................................................. 48
Figure 2.2. Experimental versus predicted Ct using predictive Ct equation ................... 50
Figure 2.3. Success rate (%) of all assays targeting an organism ................................... 51
Figure 2.4. Distribution of predicted starting copies numbers using predictive Ct
equation and standard curves ........................................................................................... 52
Figure 2.5. Factors potentially influencing quantitative inaccuracies as observed with
primers and samples designed to establish the predictive Ct equation ............................ 53
Figure 3.1. Klenow/random primer labeled target mixture vs. end-labeled 45-mer
synthetic target mixture .................................................................................................... 88
Figure 3.2. Synthetic targets of increasing dangling end length. .................................... 91
Figure 3.3. Synthetic targets with surface proximal tail sequence complementary to
background DN ................................................................................................................ 94
Figure 4.1. Schematic depicting the various stages during the development of gene
analyzer ............................................................................................................................ 104
Figure 4.2. Schematic of temperature and detection system control for the gene analyzer
.......................................................................................................................................... 105
Figure 4.3. Detailed schematic of circuit used for feedback control of the heater on the
bench-top device including. ............................................................................................. 107
Figure 4.4. The temperature control VI developed to drive the amplification reaction.
......................................................................................... 110
Figure 4.5. Front panel of temperature control VI. ......................................................... 1 12

Figure 4.6. Demonstration of a thermocycle temperature profile using PID/PWM
feedback control and PID control without PWM (tested using doped silicon heater). . . .1 13

xi

Figure 4.7. Detailed schematic of circuit used for the optical detection on the bench-top
device ............................................................................................................................... 1 18

Figure 4.8. Front panel of Optical component control VI ............................................... 120

Figure 4.9. Block diagram of optical control VI developed to detect the amplification
reaction ............................................................................................................................. 122

Figure 4.10. Coefficient of variation (%) for 4 replicate readings taken within the same
reaction channel versus the amount of time required for a one cycle measuring all 16

channels ............................................................................................................................ 124
Figure 4.11. Schematic of circuit layout drawn in ExpressSCH .................................... 127
Figure 4.12. Screenshot of PCB layouts for the LED array and for ARM

microcontroller and ICs ................................................................................................... 128
Figure 4.13. Picture of constructed circuit used to control and test components. .......... 129

Figure 4.14. Picture of the ARM evaluation board used with the LabVIEW
embedded control module. ............................................................................................... 133

Figure 4.15. Block diagram of VI constructed to test PID-based feedback control of
the heater with the LabVIEW embedded module ............................................................ 133

Figure 4.16. Testing control of LEDs, PD, and temperature sensor with ARM
microcontroller ................................................................................................................. 134

Figure 4.17. Temperature measured at three different set-points using a thermocouple
connected to the National Instruments conditioning USB-DAQ and Analog Devices IC
.......................................................................................................................................... 1 36

Figure 4.18. Component layout ....................................................................................... 139

Figure 4.19. Schematic of the cartridge for packaging the LAMP based hand-held device
.......................................................................................................................................... 140

Figure 5.1. Illustration of the low temperature plastic-assisted microforrning and process
for enclosing reaction wells and photograph of a microshelled lOO-um thick COP sheet.
.......................................................................................................................................... 147

Figure 5.2. Picture of nickel-plated stereolithography mold pressed into a piece of low
glass transition temperature thermoplastic (AB S) used as a counter-tool for
microstructuring channels in ZeonorFiIm COP ............................................................... 152

xii

Figure 5.3. Cross sectional picture taken of 0.75 um wide microstructured channels
pressed in 100 um ZeonorFilm with rubber and low glass transition temperature thermal
plastic counter tools. ........................................................................................................ 155

Figure 5.4. Difference in baseline signal between reaction wells and systems
prompted calculation of the inflection point) for determination of Tt, circumventing
potential normalization bias. ............................................................................................ 156

Figure 5.5. LAMP in all sixteen-reaction wells Spiked with 10,000 genomic copies of
DNA and positive control with calcein (solid line) and without calcein (dotted line) ....157

Figure 5.6. Standard curves observed with the gene analyzer and standard laboratory
real-time thermocycler ..................................................................................................... 159

Figure 5.7. Results of experiments to examine optical cross talk in the transparent
amplification chips ........................................................................................................... 161

Figure 5.8. Dilution series of fluorescein with the gene analyzer, change in signal
with varying concentrations of fluorescein is show for three separate reaction wells ..... 164

Figure 5.9 Ratio of fluorescein dilution signal over signal of wells without
fluorescein (water) tested with Bio-Rad and gene analyzer. ........................................... 164

Figure 6.1 Conceptual representation of the screens for analysis software to be used

to control the Gene Analyzer, Visualize the amplification curves, and report the final
results. .............................................................................................................................. 175

xiii

CHAPTER ONE

INTRODUCTION

1.1 IMPACT OF Low COST POINT-OF-CARE MICROBIAL DIAGNOSTIC DEVICES

Microbial agents are responsible for more than 75,000 deaths annually in the US
(11). Current screening practices consist of culture-based assays, smear tests, drug
sensitivity testing, or real time PCR (RT-PCR). Culture based assays can take two to four
days in most cases. For very slow growing organisms, e.g., multidrug or extensively drug
resistance Mycobacterium tuberculosis, culture-based assay may take much longer (from
weeks to months) because specimens must be shipped to a central testing facility. Genetic
assays based on the presence/absence of quantification of a given gene or those capable
of identifying specific mutations in a given gene have the potential to be much faster and
more specific because: i) genes can be amplified by either polymerase chain reaction
(PCR) with temperature cycling or by loop mediated isothermal amplification (LAMP).
Based on the advancements taking place in the area of microfluidics, there are a number
of efforts around the world to bring genetic assays into smaller packages and increase its
ruggedness so that genetic marker-based assays can be deployed at the point-of-care. This
is true for both amplification-based and hybridization-based assays. Combined with a
suitable technique to break the microbial cells and expose the genomic DNA for
amplification and/or hybridization, such devices have the potential to have a large impact
on healtlrcare.

Reducing the time of genetic assays is expected to have a profound effect
universally. Reducing the assay time from days to 15-20 minutes with a device that is not

expensive (e.g., less than $1000), enable screening in clinical settings rather than

centralized labs. When the assay takes weeks to months because of its complex nature
(e.g., detection of multiple mutations) and slow growth of the organisms (such as
multiple or extensively drug-resistant tuberculosis MDR—XDR-TB), reducing the time
and cost is equally critical. It is important to note that reducing the time has an indirect
long-term (and perhaps intangible) benefit of minimizing the spread of antibiotic
resistance. But often the immediate benefit alone is enough to justify the development of
such approaches as evidenced by the outbreak of MDR—XDR TB in KwaZulu Natal
during which 70 percent of those inflicted died within 30 days of initial sputum collection
(12).

Essential characteristics of an ideal fieldable screening tool include: i) rapid time
to completion, ii) rugged and user fiiendliness, iii) low cost, iv) hand-held or portable, v)
high Specificity, vi) superior detection sensitivity, Vii) quantitative, and viii) simultaneous
detection of multiple gene or mutation Signatures. Individually, these characteristics are
available in molecular diagnostics devices; however, no such device has been developed
that implements all of these characteristics. Simultaneous detection of multiple
signatures is important for two reasons, one for the added ability of incorporating
redundant targets and controls, and the other for the need to diagnose mutations. Even for
point-of-care devices, which are supposed to be inherently simple, the level of
multiplexing of Signatures will vary based on the application and/or organisms screened.
For example, screening for MDR-XDR TB requires an assay that targets hundreds of
genetic mutation associated with resistance against anti-TB drugs to determine the
suitable means of treatment. On the contrary, an assay for methicillin resistant

Staphylococcus aureus (MRSA) will require half a dozen assays with specificity at the

gene level. It is evident that to develop powerful point-of-care devices, amplification,
hybridization, and even ligation based mutation detection approaches may need to be
incorporated individually or in sequence to address various diagnostic needs. It is evident
that the availability of portable and low cost devices at a point-of-care (POC), such as
emergency first responders, healthcare centers, or physician’s will aid in the realization of
a more rapid means of suitable treatment, reduce fatality, and decrease emergence of
strains with increasing resistance.

Besides the selected concerns noted above which are severe problems within the
US. (e.g., MRSA) or considered a global health emergency (such as for WR—XDR TB),
there is also a broader range of diseases that may benefit from the development and
availability of such devices. According to Peter Hotez, a leading expert in neglected
tropical diseases (NTDs), globally the total number of lives and disability adjusted
lifeyears lost from HIV/AIDS, NTDS, Malaria, and Tuberculosis exceeds 220 million (5).
Based on the potential for massive social and economic impacts within and outside of
third world countries, rural areas, and resource-poor regions, there is a growing interest
for low cost POC devices (2, 4, 7, 9, 10, 16). Estimates for the global POC diagnostics
market in 2005 was $32.2 billion, which was predicted to grow to $45.6 billion by 2010,
with infectious disease testing contributing an estimated 37% of this market. With the
US and Europe holding 90% and the rest of the world accounting for the remaining 10%,
the availability of a low-cost point of care device could substantially increase markets in

the developing world and low-resource settings.

1.2 COMPARATIVE EVALUATION OF SELECTED REAL TIME PCR DEVICES

Of the many genetic detection approaches, real time PCR (RT-PCR), first
described in 1983, offers the highest level of sensitivity and quantification. Despite the
advantages of this technique, there are still less than a dozen PCR diagnostic tests (17)
that are approved by the Food and Drug Administration. This is due to the high cost of
RT-PCR instruments, significant training to run the instrument, and need for storage of
reagents. As listed in Table 1.1, a number of options are available to carry out this assay.
This is not an exhaustive list and certain instruments that are still under development
have been left out due to lack of information and also because they are known mostly
through publications and have not yet arrived in the market.

Gene-based assays using RT-PCR generally take about 3-4 hours to complete and
FDA has approved some of these assays for clinical use (e.g., for methicillin resistant
Staphylococcus aureus). However, the cost of real-time PCR machines (tens of thousands
of dollars; Table 1.1) generally requires that the specimens be sent to a centralized
laboratory. This delays diagnosis causing the prescription of broad-Spectrum antibiotics
or delaying treatment, increasing potential for the emergence of a new more resistant
strain and fatality. When portable Options are available (albeit at very high prices), the
validation of genes is done on a very limited set of organisms (e.g., anthrax for bio-Seeq
and Bird Flu for Palm-PCR of Neuzil’s group). For other pathogens of interest, users are
expected to develop the assay on their own (perhaps in collaboration with the equipment
suppliers). The high cost and the need for further development virtually make it
impossible for these options to serve as truly low cost point-of-care options for most

other applications.

Since all devices use amplification based strategies, parameters such as sensitivity
are similar for all RT-PCR machines; however, only 2 Of the portable devices listed in
Table 1.1 are quantitative. The Hand-Held offers the same level of throughput as the
other portable devices in terms of targets that can be assayed simultaneously. Potential
cross contamination problems that can be associated with dispensing multiple samples in
a small space make multiplexing with respect to samples less attractive. However, since
both the equipment and the chip are inexpensive, and the time of assay is short, analyzing
single samples is not a significant disadvantage. It is the gap in the availability of low
cost genetic analysis point of care system (from here onward referred to as gene analyzer)
that the current research aims to fill.

Table 1.1: Comparison of parameters for representative RT-PCR machines

 

 

 

 

 

 

 

 

 

 

 

Smiths
PCR CMP "“1"” Detection Idaho PCR-Chip Idaho BloTrOVO
Hand-Held microPCR (BMW RAZOR Bench mp R.A.P.I.D. NTCyelor
Vllldauon Focus MRSA Blrd Flu Anthrax Open Open Open Open
Base Unit $1,000 Unknown $31,000 $35,000 $25,000 $55,000 3100.000
Consumable Price (U) $10 Unknown $38 $18 $100 $400
Tlmo 5 mln 1 mln 20-30 min 30 min 30 min 30 min 3.5 hr
I of Samples 1 1 a 12 56 10 48
t of Targets -5-10 ~6—10 1-4 12 50 10 18-56
Sample volume 1000 nl 1000 nl 10 ”I 5.20 "I 5.50 ”I 5-20 pl 33 nl
Quantitation Yes Yes No No Yes Yes Yes
Sensitivity 10 copies 10 copies 5 copies 5 copies 5 copies 5 copies 5 copies
POMDlllty (Weight) < 1 lb 5 lb 6.5 lb 9 lb 30 lb 50 lb 100 lb
G081t l A ~‘n
‘ ‘ , <— , l
, i i

 

 

1.3 SELECTION OF PARAMETERS IN DEVELOPING THE GENE ANALYZER

Numerous prototype microfluidic chips and analyzers have been described in
literature that may be capable of carrying out PCR in hand-held or portable devices (1, 3,
6, 8, 13-15). Despite indications that development is near completion, few systems are
commercially available that are hand-held, multiplex, and cost-efficient. This is due to
several reasons including: i) the sensitivity required for RT-PCR amplification requires
bulky, expensive, complex, and power consuming systems for excitation (mercury lamps
or lasers with a number of optical filters) and detection of light (Photomultiplier tubes
and high cost charged coupled devices), ii) lack of expertise in embedding devices within
the settings of a single university laboratory, and iii) complexity and/or lack of portability
associated with real time detection schemes for multiple assay amplification chips. In
essence, innovation, adaptation, and cost reduction are still needed before hand-held
microbial diagnostics technologies can impact health care in a point-of-care scenario with
resource-limited settings (1 7).

As evident, a number of choices must be made and depending upon those choices,
the cost, portability and power of the device will be determined (Figure 1.1). Often, the
application may dictate that multiplexing with respect to gene (e. g., for MRSA) or
mutation (e.g., for MDR/XDR TB) must be incorporated in order for the device to be
useful for that application. Based on the molecular strategies adopted, essential
characteristics, and the application-defmed requirements (level of specificity and
throughput), three separate products are being developed in a mutually synergistic
manner, each suitable for a specific application. Table 1.2 (at the end of this chapter) lists
the target value for many of the parameters that are necessary for a device of this nature.

TB and MDR/XDR TB are taken as examples of two separate applications, one requiring

the detection of a limited set of genes (similar to gene analyzer) and the other requiring

the detection of a large number of mutations.

1)

2)

3)

Photodiode-based gene analyzer: Gene analyzer allows selection of LAMP or
RT-PCR molecular amplification with an optical setup consisting of an array of
sixteen individually controlled blue light emitting diodes (LEDs), and a single
photodiode to read the sixteen reaction wells one by one using individual optical
fibers. Figure 1.1 illustrates the approximate values chosen for the gene analyzer.
Only the gene analyzer is the topic of this research.

CCD-based LAMP Array: This product focuses on the measurement of
approximately 1,000 LAMP reactions in real time (with options surface based
hybridization to increase throughput) using a CCD camera with exposure control.
Signals are extracted as SNR using custom software being developed as part of
the project. Temperature control, fabrication technique of disposable chips, and
microcontroller are based on the developments made for the gene analyzer.

Single Photon Detection-based Rotating Disk Array: This product uses a
rotating disk instead of an optical fiber and LED harness system to move a
circular chip underneath a fixed single photon detector set up. This allows rapid
monitoring of the amplification product in multiple wells (the number of wells
will depend upon the diameter of the chip but it could potentially be much larger
in 1003 to 10008. Temperature control, fabrication technique of disposable chips,

and microcontroller are similar to that used in the gene analyzer.

.L
O
O

1.000 100 1000 50 100 100 100 100 High Excellent
a _ a _ _ A
T lib I-h

L

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—
0
--- mu :- -' a I-l'
s s 3 lb 8
0 0 1 0.5 1 50 1 1 Low Poor
ll-l Ill 0 O U) ll- 2
Q 2 w E o E {’3 I3 0 o
n: I: E g -| 9 ur a Lu T:
0. 0. (D m IL 0 E (0 <
.— >. < — I! < I—
z < 0: Lu ,_ In I'-
D 3 o 3, 5
< D- D
0

Figure 1.1. Depiction of the options for various parameters related to an analytical device
illustrating various levels of equipment price, assay price, speed, portability, limit of
detection, specificity, number of targets, number of samples, ease of use, and
quantitation. Approximate values of these parameters are shown here for gene analyzer.
Similar choices have been made for other instruments where speed, and the number of
targets and number of samples are also maximized (e.g., single photon detector based
rotating disk analyzer and CCD based LAMP microarray instrument).

1.4 OVERVIEW OF THE PROJECT

Ideally an instrument must be developed first and the assay validation follows.
However, to save time and due to the sheer volume of the assays to be validated, a
parallel approach was adopted meaning the assays were validated on existing real time
PCR instruments and hybridization systems with the assumption that when the hand-held
gene analyzer is developed, these assays will be ported and if needed optimized on the
portable system. As illustrated in Figure 1.2, the overall project was highly team

oriented, which is generally not encountered in academic setting where students work on

his/her topic with limited interaction to other students. However, this project required a

large number of interdependent tasks, ranging from assay design to microfluidic chip
development and validation to selection of integrated circuits and printed circuit board
design and its evaluation (Figure 1.2). Seven of eight team members worked at Michigan
State University and one member worked at the University of Michigan.
MOI ECLLAR DI.—\GN(:)STI(,‘S
F’IIPCP. 3:, “mm-.7

Reagent Stability LAMP

\( 0‘ v
”‘3‘ “Al Emil-:2 Ihrx Iriirr‘ut

Eurbo 3 2 mg and. \ir-M.

ELECTRONICSCONI'ROL

:SBDAO

 

 

Figure 1.2. Major considerations for the devices are divided into four categories. These
aspects have been subdivided into developmental tasks. Color codes refer to persons that
have taken the lead on a task or share the task with other (light blue represents everyone,
Robert Stedtfeld in black, Dieter Tourlousse in green, Tifimy Stedtfeld in dark blue,
Farhan Ahmed in red, Gregoire Seyrig in purple, Deric Patton in gray, Aaron Thompson
in orange, and Onnop Srivannavit in brown).

1.5 ORGANIZATION OF THIS THESIS
The thesis is divided into two main sections: Part 1: Validation of amplification

and hybridization-based molecular approaches, and Part II: Development and validation

of the hand-held real-time gene analyzer. Part I consist of two chapters that are already
published (out of 15 publications). Chapter 2 focuses on the validation of primers using
an existing high throughput real time PCR machine (Biotrove Inc). Primers Similar to
these will be used in the gene analyzer after it is fully developed. Chapter 3 evaluates the
limits of hybridization-based approach for detection of closely related fi'agments. Lessons
learned in this chapter will be usefirl in developing hybridization-based assays for

mutation detection. A summary of these chapters is provided below.

Chapter 2 Development and experimental validation of a predictive threshold cycle (Ct)
qun for quantification of virulence and marker genes using high-throughput
nanoliter PCR on OpenArraysm: Development of quantitative PCR (QPCR) assays
typically requires extensive screening within and across a given species to ensure specific
detection and lucid identification among various pathogenic and nonpathogenic strains
and to generate standard curves. To minimize screening requirements, multiple virulence
and marker genes (V MGS) were targeted simultaneously to enhance reliability, and a
predictive threshold cycle (Ct) equation was developed to calculate starting copies based
on experimental Ct. The empirical equation was developed with SYBR Green detection
in nanoliter QPCR chambers (OpenArrayTM) and tested with 220 previously unvalidated
primer pairs targeting 200 VMGS fi'om 30 pathogens. A high correlation (R2 = 0.816) was
observed between predicted and experimental Ct based on the organism's genome size,
GC content, amplicon length, and stability of primer 3' end. The performance of the
predictive Ct equation was tested using 36 validation samples consisting of pathogenic

organisms spiked into genomic DNA extracted from three environmental waters. In

10

addition, primer success rate was dependent on GC content of the target organisms and
primer sequences. Targeting multiple assays per organism and using the predictive Ct
equation is expected to reduce the extent of validation necessary when developing QPCR

arrays for a large number of pathogens or other targets.

Chapter 3 Influence of Dangling Ends and Surface-Proximal Tails of Targets on
Probe- Target Duplex Formation in 16S rRNA Gene-Based Diagnostic Arrays:
Dangling ends and surface-proximal tails of gene targets influence probe-target duplex
formation and affect the signal intensity of probes on diagnostic micrOarrays. This
phenomenon was evaluated using an oligonucleotide microarray containing 18-mer
probes corresponding to the 168 rRNA genes of 10 waterborne pathogens and a number
of synthetic and PCR-amplified gene targets. Signal intensities for . Klenow/random
primer-labeled l6S rRNA gene targets were dissimilar fi'om those for 45-mer synthetic
targets for nearly 73% of the probes tested. Klenow/random primer-labeled targets
resulted in an interaction with a complex mixture of 168 rRNA genes (used as the
background) 3.7 times higher than the interaction of 45-mer targets with the same
mixture. A 7-base-long dangling end sequence with perfect homology to another Single-
stranded background DNA sequence was sufficient to produce a cross-hybridization
signal that was as strong as the signal obtained by the probe-target duplex itself. Gibbs
free energy between the target and a well-defined background was found to be a better
indicator of hybridization signal intensity than the sequence or length of the dangling end
alone. The dangling end (Gibbs free energy of —7.6 kcal/mol) was found to be

significantly more prone to target-background interaction than the surface-proximal tail

ll

(Gibbs fiee energy of —64.5 kcal/mol). This study underlines the need for careful target
preparation and evaluation of signal intensities for diagnostic arrays using 168 rRNA and
other gene targets due to the potential for target interaction with a complex background.
Part II consist of three chapters (Chapters 4-6). Chapters 4 and 5 focus on the
development of gene analyzer and Chapter 6 is a summary of this thesis and perspectives
toward further development and commercial use of the gene analyzer. A summary of

Chapters 4 and 5 is provided below.

Chapter 4 Development of Embedded Control Systems for Gene Analyzer: The
selection and control of components toward the development of a hand-held device for
real time detection of loop mediated amplification is described in detail. Initial control of
components was obtained using National Instruments LabVIEW with USB compatible
data acquisition accessories and several integrated circuits (ICs). A rapid response
thermocouple, PWM driver IC, and PID feedback-control was used for precise
temperature control of a Kapton heater with a standard deviation from setpoint of 0.28
°C. The optical setup for capturing real time amplification signals in a multicharnber well
consisted of sixteen light emitting diodes (LEDs), optical fibers sending fluorescent light
from each well to a Single photodiode, which was used to acquire change in signal
emission. Development of optics control included individual control of all 16 LEDs with
a 4 to 16 line decoder IC, generating accurate and constant current for all LEDs with a
LED driver IC, synchronized timing of LEDs with the photodiode (PD), and reducing
and conditioning noise acquired from the PD. Once a control strategy was established

and used to test components, a mechanism was selected and tested for embedded control.

12

This included the design of a custom PCB, replacing USB-DAQs and PC with an ARM
microcontroller, designing a cartridge to package the device, and devising a circuit to

communicate with an iPod Touch.

Chapter 5 Validation of Optical System for Gene Analyzer: The development of a hand-
held real-time gene analyzer to detect amplification products from loop mediated
isothermal amplification (LAMP) is described. The high yield of amplification product
and isothermal characteristics of Bst polymerase allow the use of simple and low cost
components to drive the reaction and detect the resulting product. The device consist of:
i) a Kapton resistive heater to isothermally heat a chip with sixteen wells, ii) an array of
sixteen blue light emitting diodes (LEDs), and iii) a single photodiode to read the sixteen
wells one by one using individual optical fibers. The system is simple because it
eliminates costly lenses, CCD imaging, or moving parts. The nricrofluidic chips consist
of sixteen 3-mm long capillaries embossed in 100-um thin Zeonor film using a modified
heat embossing technique. A positive mold with shoulders fabricated using
stereolithography and a counter tool consisting of acrylonitrite butadiene styrene (ABS)
having low glass transition temperature was used to fabricate the disposable chips. The
shoulders allowed a larger process window and reproducibility. The molded ABS tool
served as both an alignment guide between the chips and optical components and
optically isolated the reaction wells. Real time LAMP experiments performed with
Staphylococcus aureus methicillin resistance gene (mecA) using calcein as a fluorescent

dye yielded results comparable to a state-of-the-art real-time nucleic acid amplification

13

instrument. This simple, small, and low-cost system is expected to play a key role in

propagating the use nucleic acid assays in resource-limited settings.

1.6 OBJECTIVES OF THIS RESEARCH

The goal of this research was to develop a low cost and rugged quantitative

nucleic acid analyzer, offering a high level of specificity. The Specific objectives were:

1. Validate a large set of real time PCR primers to be employed in any on-chip

device.

2. Evaluate the extent of differentiation between two similar sequences by on-chip

hybridization.

3. Develop a portable gene analyzer capable of quantitatively detecting gene

signatures in DNA matrix. The boundary conditions for this device were:

Component cost: less than $1000 in component cost

Assay cost: less than $5 per sample

Weight: less than 1 pound

Size: a hand-held system similar in size to an iPod Touch

As part of the gene analyzer system, the following tasks were identified:

Development of microfluidic chips.

Evaluation and selection of temperature and optical components for
heating and detecting amplification.

Control of temperature and optical components with LabVIEW.
Embedded control with ARM Microcontroller.

Housing design

14

Interfacing device with an iPOd.

Validation of all aspects.

I have taken the lead role and/or shared the following tasks interdependently:

Target selection, database development, design, and validation of RT-
PCR and microarray based assays.

Development of microfluidic chip including design and fabrication of
molds, optimization of hot embossing process for reproducible
fabrication of chips, enclosing and sealing chips.

Selection of heaters, temperature sensors, and development of
feedback-control.

Devising mechanism to control optical components.

Selection and evaluation of ARM microcontroller for embedded
control of components.

Design of PCB to reduce size of circuit.

Selection of ICs for regulating voltage to various components within
packaged device.

Validation and characterization of gene analyzer.

15

Table 1.2 Comparison of user needs specified in the Customer Requirement Document-
Primary Care Level (provided by Foundation for New Innovative Diagnostics or FIND)
to the proposed capabilities of various devices in this project.

 

 

 

 

 

 

Desired Minimum Routine TB Assay M-ITRT—XDR—
(Gene Analyzer) TB Assay
(SPD and
CCD based
Am!)
A. Workflow
1. Sample
1.]. Sample Sputum, skin, Sputum, skin, Sputum Sputum
type breath, urine breath, mine,
(at least one blood, sennn
of these (at least one of
sample types) these sample
W5)
1.2. Sample None Simple I to 10 None, sample processing None, sample
processing two-step integrated in the processing
procedure with microfluidic chips/device integrated in
less than 15 the
min hands-on microfluidic
time chips/device
2. Time-to- Less than 10 Less than 2 Less than 20 min, including Less than 30 I
result min, including hrs, including sample processing and min, including
sample sample readout sample
processing processing and processing and
and readout readout readout

 

 

3.

Instrumentation

4. Additional

equipment

5. Bio-safety

 

Table 1.2 continued

None

None

No need for
biosafety
cabinet,
Disposal of
clinical
material
without
further

treatment

B. Diagnostic performance

1. Sensitivity
1.]. Diagnostic

sensitivity

2. Specificity

 

Either:

a) rule-in-test:

>75%
b) rule-out-

test: >95%

 

 

None, or
small,
maintenance-
free device
Less than 2
devices with
small footprint
No need for
biosafety

cabinet

Either:

a) rule-in-test:
>30%

b) rule-out-

test: >85%

17

 

 

Portable and battery-

powered device

None

No need for biosafety
cabinet due to fully closed
and self-contained assays in

microfluidic chips

Sensitivity: 90% or higher

 

 

Portable and
battery-
powered
device

None

No need for
biosafety
cabinet due to
fully closed
and self-
contained
assays in
microfluidic

chips

Sensitivity:

90% or higher

 

 

 

Table 1.2 continued

2.]. Diagnostic Either:
specificity a) rule—in-test:
>97%
b) rule-out-
test: >50%
3. Effect of No change in
HIV co— performance
infection
C. Product design

]. Stability/storage requirements

1.1.Kit

stability

 

36 months at
35°C and 70%
humidity —
including
transport
stress (48 hrs

at 50°C)

 

 

Either:

a) rule-in-test:
>97%

b) rule-out-
test: >50%
50% fall in
sensitivity in
HIV infected

patients

9 months at
30°C and 3
months at
35°C including
transport stress
(48 hrs at

50°C)

18

 

 

Specificity: higher than 95%

Similar to that of other

nucleic acid based assays

At least 12 months at 30°C

and 3 months at 35°C

 

 

Specificity:

close to l00%

Similar to that
of other

nucleic acid

based assays

At least 12
months at
30°C and 3
months at

35°C

 

Table 1.2 continued

1.2.

Reconstituted

reagents

stability

2. Calibrators

3. Controls

 

 

Ready-to-use
units, no

reconstitution

Fixed cut-off

Full-process
positive

control

 

1 week at

35°C

Fixed cut-Off

No control

19

 

Ready-to-use and single-use
microfluidic chips with
stabilized reagents, no need

to store reconstituted

reagents

LAMP: time-to—positivity

Internal cellular control
(full-process positive

control)

 

Ready-to-use
and single-use
microfluidic
chips with
stabilized
reagents, no
need to store
reconstituted
reagents
LAMP: time-
tO-positivity
Microarray:
signal-to-noise
ratio

Internal
cellular control
(full-process
positive

control)

 

 

4.
Determinations
per
‘reconstituted

reagent unit’

5. Results
captlu'ing and

documentation

6. Instrument

design

 

Table 1.2 continued

Ready-to-use
units, no

reconstitution

Yes

No
instrumentatio
n or battery
operated.
Device
platform can
also be used
for other
infectious

agents

 

Opened
packaged
should last for

at 2 weeks

115/220 V AC

Operates at
35°C

20

 

Ready-to-use and single-use
microfluidic chips with
stabilized reagents, single

determination/chip

Yes, CCD imaging and

image/data analysis

Battery-operated. Can be
used for other infectious
agents by using microfluidic
chips with different primer

sets

 

Ready-to-use
and single-use
microfluidic
chips with
stabilized
reagents,
single
determination/
chip

Yes, CCD
imaging and
image/data
analysis
Battery-
operated. Can
be used for
other
infectious
agents by
using
microfluidic
chips with
different
primer sets and

probes

 

Table 1.2 continued .

7. Training and
education

requirements

 

 

 

Less than 1
hour, high
school
doplaome or

equivalent

D. Additional features

 

 

Less than 2
days, nurse

level skill set

 

 

Less than 1 hour due to full
automation, high school

doplaome or equivalent

Routine TB: Quantitative

Equipment cost: $1,000

$0.25 per assay for six

genes

 

 

Less than 2
days, nurse
level skill set,
most skills in
data

interpretation

- Routine TB:
Quantitative
- MDR/XDR
TB:
Plus/Minus
Equipment
cost: $1,000

<35 per assay

 

21

 

1.7 REFERENCES

10.

Bbattacbaryya, A., and C. M. Klappericb. 2007. Design and testing of a
disposable microfluidic chemiluminescent immunoassay for disease
biomarkers in human serum samples. Biomed. Microdevices. 9:245-251.

Herr, A. E., A. V. Hatch, D. J. Throckrnorton, H. M. Tran, J. S. Brennan,
W. V. Giannobile, and A. K. Singh. 2007 . Microfluidic irnmunoassays as
rapid saliva-based clinical diagnostics. Proc. Natl. Acad. Sci. U.S.A.
104:5268-73.

Hofmann, 0., X. Wang, A. Cornwell, S. Beecher, A. Raja, D. D. Bradley,
A. J. Demello, and J. C. Demello. 2006. Monolithically integrated dye-doped

PDMS long-pass filters for disposable on-chip fluorescence detection. Lab
Chip 6:981-987.

Holland, C. A., and F. L. Kiecble. 2005. Point-of-care molecular diagnostic
systems — past, present and future. Curr. Opin. Microbiol. 8.

Hotez, P. 2009. Devastating global impact of neglected tropical diseases.
Microbe 4:366-372.

Huen, W., A. Chow, C. McCloskey, Y. Tong, L. Hu, Q. You, C. P. Kelly,
H. Kong, Y. W. Tang, and W. Tang. 2008. Application of isothermal
helicase—dependent amplification with a disposable detectio device in a simple
sensitive stool test for toxigenic Clostridium dzflicile. J. Molec. Diag. 10:452-

458.

Inganas, M., H. Derand, G. Eckersten, A.-K. Ekstrand, G. Honerud, G.
Jeeson, T. Thorsen, T. Soderman, and P. Andersson. 2005. Integrated
microfluidic compact disc device with potential use in both centralized and
point-of—care laboratory settings Clin. Chem. 51:1985—1989.

Liao, C. S., G. B. Lee, H. S. Liu, T. M. Hsieh, and C. H. Luo. 2005.
Miniature RT—PCR system for diagnosis of RNA-based viruses Nucleic Acids
Res. 33:e156.

Linder, V. 2007. Microfluidics at the crossroad with point-of-care
diagnostics. Analyst 132:1186 - 1192.

Malamud, D., H. Bau, S. Niedbala, and P. Corstjens. 2005. Point detection
of pathogens in oral samples. Adv. Dent. Res. 18:12-16.

22

11.

12.

13.

14.

15.

16.

17.

Mokdad, A. H., J. S. Marks, D. F. Stroup, and J. L. Gerberding. 2000.
Actual causes of death in the United States. JAMA 291:1238-1245.

Panel, T. T. D. E. E. 2008. Evaluation of diagnostic tests for infectous
diseases: general principles.

Pipper, J., Y. Zhang, P. Neuzil, and T. M. Hsieb. 2008. Clockwork PCR
including sample preparation. Angew. Chem. Int. Ed. Engl. 47:3900—3904.

Smith, M. C., G. Steimle, S. lvanov, M. Holly, and D. P. Fries. 2007. An
integrated portable hand-held analyser for real-time isothermal nucleic acid
amplification. Anal. Chim. Acta 598:286-294.

Wang, J., Z. Chen, P. L. Corstjens, M. G. Mauk, and H. H. Bau. 2006. A
disposable microfluidic cassette for DNA amplification and detection. Lab
Chip 6:46-53.

Yager, P., T. Edwards, E. Fu, K. Helton, K. Nelson, M. R. Tam, and B. H.
WeigL 2006. Microfluidic diagnostic technologies for global public health.
Nature 442:412-418.

Yager, P., J. Gonzalo, G. J. Domingo, and J. Gerdes. 2008. Point-of-Care
diagnostics for global health. Annu. Rev. Biomed. Eng. 10:107—44.

23

CHAPTER Two

Stedtfeld, R. D., S. W. Bausbke, D. M. T ourloussc, S. M. Mller, T. M.
Stedtfeld, E. Gulari, J. M. Tiedje, and S. A. Hasbsham. 2008. Development
and experimental validation of a predictive threshold cycle equation for
quantification of virulence and marker genes by high-throughput nanoliter-

volume PCR on the OpenArray platform. Appl. Environ. Microbiol. 74:3831-8.

24

DEVELOPMENT AND EXPERIMENTAL VALIDATION OF A PREDICTIVE THRESHOLD
CYCLE (CT) EQUATION FOR QUANTIFICATION 0F VIRULENCE AND MARKER GENES
USING HIGH-THROUGHPUT NANOLITER PCR ON OPENARRAYSTM

2.1 ABSTRACT

Development of quantitative PCR (QPCR) assays typically requires extensive
screening within and across a given species to ensure specific detection and lucid
identification among various pathogenic and nonpathogenic strains and to generate
standard curves. To minimize screening requirements, multiple virulence and marker
genes (V MGs) were targeted simultaneously to enhance reliability, and a predictive
threshold cycle (Ct) equation was developed to calculate starting copies based on
experimental Ct. The empirical equation was developed with SYBR Green detection
in nanoliter QPCR chambers (OpenArrayTM) and tested with 220 previously
unvalidated primer pairs targeting 200 VMGS from 30 pathogens. A high correlation
(R2 = 0.816) was observed between predicted and experimental Ct based on the
organism’s genome size, GC content, amplicon length, and stability of primer 3 ’ end.
The performance of the predictive Ct equation was tested using 36 validation samples
consisting of pathogenic organisms spiked into genomic DNA extracted from three
environmental waters. In addition, primer success rate was dependent on GC content
of the target organisms and primer sequences. Targeting multiple assays per
organism and using the predictive Ct equation is expected to reduce the extent of
validation necessary when developing QPCR arrays for a large number of pathogens

or other targets.

25

2.2 INTRODUCTION

Detection and identification of multiple pathogens requires simultaneously
targeting a large number of virulence and marker genes (V M63) (14, 31, 35, 43, 50,
54). This is because: i) use of multiple markers for the same pathogen enhances
specificity and reliability of the assay, ii) often the allelic variability requires multiple
primers for the same gene, iii) some of the markers may be less specific (i.e., they
may also be found in nonpathogens, and iv) the VMGS for a given pathogen may be
unevenly distributed among various strains. Use of a varied number of VMGS to
differentiate clinical isolates has been described for many pathogenic organisms
including: Staphylococcus aureus (3, 49), Aeromonas spp. (8), Campylobacter jejuni
(39), Pseudomonas aeruginosa (13), Vibrio spp. (50), Escherichia coli, and Shigella
flexneri (5, 53).

When many pathogens must be screened in parallel with the ability to
characterize the uneven distribution of the associated VMGS and their allelic
variability, optimization is necessary for high throughput tools with high sensitivity
and specificity, e.g., quantitative PCR (QPCR) (6, 14, 19, 27, 33, 45, 51, 52). This
optimization may be cumbersome because it requires generation of multiple standard
curves with the caveat that all primer sets must perform under the same amplification
conditions. Approaches that increase reliability of primer design or avoid the use of
standard curves all together should be extremely useful in developing such parallel

assays. Development of such approaches will undoubtedly depend upon the sequence

26

characteristics (GC content, Tm, amplicons length, etc.) of primers, amplicons,
genome size, amplification conditions, and the matrix in which the target is present.
However, the influence of these factors on the performance of primer sets, especially
on the threshold cycle number (Ct) extensively used in QPCR to predict the
abundance of pathogens has not been explored fully.

This study used the nanoliter volume BioTrove OpenArrayTM platform (34)
to examine its capacity for highly parallel diagnostics of human pathogens, and to
systematically examine the influence of target and primer sequence characteristics on
specificity, sensitivity and Ct value. Results were used to establish a predictive Ct
equation to estimate starting copy without the use of standard curves. The study was
performed with SYBR Green I and used approximately 220 primer sets targeting 200
VMGS for 30 human pathogens. Performance of the predictive Ct equation is
examined and the success rate of previously un-validated pathogen targeted primer
sets is also presented. These results have significance in developing high throughput
and reliable screening tools for large numbers of pathogens without extensive

validation.

2.3 MATERIALS AND METHODS

2.3.1 DNA targets. Multiple organisms were used to develop and validate the
predictive Ct equation. Genomic DNA (gDNA) from 21 bacterial pathogens (Table
2.1) was used including: Closlridium perfiingens, Enterococcus faecalis, Listeria
monocytogenes, Legionella pneumophila, Pseudomonas aeruginosa, Salmonella

enterica, Vibrio parahaemolyticus, Yersinia enterocolitica type strains, which were

27

obtained from the American Type Culture Collection (ATCC; Manassas, VA) and
grown as per the protocol provided. For Helicobacter pylori, Campylobacter jejuni,
Cryptosporidium parvum, Giardia intestinalis, Staphylococcus aureus, Vibrio
cholerae, Mycobacterium genetalium, Haemophilus influenzae, Leptospira
interrogans, Bacillus cereus, and Bordetella pertussis, gDNA was obtained directly
from ATCC. Genomic DNA for Escherichia coli was kindly provided by Dr.
Thomas Whittam at Michigan State University. DNA from pure cultures was

extracted using Promega’s Wizard DNA Extraction Kit (Promega, Madison, WI).

2.3.2 Complex background gDNA. Genomic DNA from pure cultures was
spiked into complex background gDNA extracted fiom environmental waters. River
water was collected from the Red Cedar River and activated sludge and tertiary
effluent from a wastewater treatment plant in East Lansing (MI). Samples were
filtered through 0.45 pm nitrocellulose filters (Millipore, Billerica, MA) immediately
after collection. Genomic DNA was extracted from the filters as instructed with the
MegaPrep UltraCIeanTM Soil DNA Kit (Mo Bio Laboratories, Carlsbad, CA).
Genomic DNA was purified fiirther through ethanol precipitation. DNA
quantification and purity was examined with the NanoDrop® ND-1000
spectrophotometer (NanoDrop Technologies, Wilmington, DE). Prior to QPCR
experiments, background DNA was tested for inhibition by PCR amplification with a

universal primer set targeting the 16S rRNA gene.

28

2.3.3 Design of PCR primers to establish the predictive Ct equation. This
set of primers was designed to target 20 human pathogens, and was used to examine
relationships between primer sequence characteristics and experimental Ct. Prior to
primer design, consensus sequences were generated by aligning sequences for each of
96 VMGS using Kodon (Applied Maths, Austin, TX). The consensus sequence was
used for primer design with Primer Express (Applied Biosystems, Foster City, CA).
A majority of designed amplicons had a maximum length of 150 bases and primers
had a theoretical melting temperature of 59 °C. Some genes required longer
amplicons (<250 bases) for generating acceptable primer sets. From the list of
primers provided by Primer Express, optimal primers were selected using a script that
automatically highlighted unspecific primers. The script used the NCBI BLAST (2)
to check specificity against the GenBank database. Specificity was based on the
extent of 3’ end perfect matches to non-targeted bacterial sequences. Sequences were
selected manually based on results of BLAST output. When available, primers
described in the literature for successful QPCR were also used. Overall, 110 primers
pairs were designed and extracted from the literature. Including primers previously
described in the literature, 3,687 VMG sequences were targeted with this primer set

(determined by BLAST analysis with GeneBank, May 2006).

2.3.4 Design of PCR primers to validate correlations used in development
of predictive Ct equation. To validate the predictive Ct equation and further
examine the success rate of primers, 111 new primers were designed. The new group

of primers were designed with the same criteria as primers designed to establish the

29

predictive CT equation. In addition, primers were filtered further to select sequences
with lowest possible percentage of GC bases in the 3’ end of the primer. One non-
specific primer targeting C. parvam was removed fi'om the set due to false positive
observations. Thus, a total of 220 primers sets (109 used to establish the predictive
Ct equantion, and 111 used to validate the predictive Ct equation) targeting 200
VMGS for 30 pathogenic bacteria were tested (Table 2.1, the new VMGS targeted

with this primer set are in bold).

2.3.5 PCR on BioTrove OpenArrayTM plates. Primer sets were tested
simultaneously on the BioTrove OpenArrayTM plates. Primers were synthesized by
Sigma-Aldrich (St. Louis, MO) and pre-loaded (128 nM for primers designed to
establish the predictive Ct equation and 400 nM for primers designed to validate the
predictive Ct equation) into BioTrove OpenArrayTM plates (Wobum, MA)(34, 47). Two
to four subarrays (each with 64 wells for 56 separate assays and eight loading
controls) were used for each PCR sample. PCR mixtures (5 pl for each sample array)
consisted of l x LightCycler Fast Start DNA Master SYBR Green I Kit (Roche
Applied Sciences, Indianapolis, IN), 1.6 x SYBR Green I, 0.5% Glycerol, 0.2%
Pluronic F-68, 1 mg per ml BSA (New England Biolaboratories, Beverly, MA), 2.5
mM MgC12, 8% formarnide, and a DNA mixture. Afier initial enzyme activation at 95
°C for 10 min, 36 cycles of the following program was used for amplification:

denaturation at 95°C for 10 s, annealing at 53 °C for 10 s, elongation at 72 °C for 10 S.

30

2.3.6 Design of sample mixtures for development of predictive Ct
equation. Samples were mixed to evaluate the dependence of sequence
characteristics of primers, amplicons, genome size, amplification conditions, and the
matrix in which the target is present on specificity and sensitivity. To evaluate
specificity without a complex background, gDNA from 14 pathogenic organisms
(pure cultures; ATCC number listed in Table 2.1) was tested individually (6 ng in
total sample or 20 pg per reaction well). To develop standard curves and further
evaluate specificity and sensitivity within a complex background, gDNA from 14
pathogens was mixed and spiked at various concentrations into gDNA from waste
water tertiary effluent and river water samples (20 pg, 2 pg, 200 fg, 20 fg, 2 fg of each
of 14 pathogens mixed together spiked into 66.6 pg of background gDNA per
reaction well). The mixture of varying concentrations was also examined without a
background to serve as a control. Complex background samples were also tested
without spiking with gDNA from pure cultures. All of these samples were tested on
the set of PCR primers designed to establish the predictive Ct equation. These
samples were examined further to evaluate influence of sample inhibition and

variation between OpenArrayTM plates. All samples were tested in triplicate.

2.3.7 Design of validation sample mixtures for validation of predictive Ct
equation. Validation samples were used to fmther evaluate the characteristics of
template and primer sequences on primer success rate and to evaluate the predictive
Ct equation. For the validation samples, gDNA from 21 organisms (ATCC number

listed in Table 2.1, organism used solely with these mixtures are in bold) were spiked

31

at varying concentrations, for an absolute abundance of approximately 10, 100, and
1,000 genomic copies per reaction well, into gDNA extracted fi‘om either river water,
tertiary effluent, or activated sludge (0.99 ng per reaction well). In total, 36

validation samples were prepared. All samples were tested in triplicate.

2.3.8 Data analysis. For all analysis, data was filtered to differentiate true
and false positive and negative signals. Amplification was considered positive if the
threshold cycle was less than 26 for all three replicates and experimental melting
temperature was consistent. The influence of primer 3 ’ end GC content on specificity
was analyzed using the primer (either forward or reverse) with the highest GC
content. Primers were grouped based on the number of GC bases within last five
bases on the 3’ end. For developing the predictive Ct equation, only the primer sets
displaying true positive amplification were considered .

The GC content of primer and target organisms influence on success rate of
novel primers was examined based on average sum of successes (taken between 36
sample mixtures for validating the predictive Ct equation) for all assays targeting an
organism. Assays were considered successful if it displayed a true positive or true
negative signal, and an organism was deemed present if two or more assays (targeting
one organism) displayed amplification with a threshold cycle less than 26 for all three
replicates. The average GC content of all primers used for a targeted organism was
used for this figure. The 3D plot was generated with a loess smoother and 1.0
sampling proportion. This smoothing was performed to identify characteristics of the

population.

32

To compare with the predictive Ct equation, standard curves were generated
using an average slope and intercept fi'om all three replicates in all three backgrounds
(control, gDNA spiked in river and tertiary gDNA) from sample mixtures designed to
develop a predictive Ct equation. PCR efficiency was examined to determine
influence of sample background on quantitative values, and was calculated from the

slope of the standard curves with the following equation:
PCR Efliciency = (—1 +10-1’S’0P9)

2.4 RESULTS AND DISCUSSION

2.4.1 Sensitivity and Specificity. The influence of primer and template
sequence characteristics on specificity and sensitivity was examined using the primer
sets and sample mixtures designed to develop the predictive Ct equation. The
percentage of assays displaying false positive/negative signals for primer sets tested
individually and within a complex background did not change for target
concentrations above a range of 1 to 10 copies (Figure 2.1, top panel). Within the 1
to 10 copy range (i.e., with 20 fg of target gDNA present per reaction well),
approximately half of targeted assays displayed positive amplification. Depending on
the organisms, this corresponds to 1.5 to 11 genomic copies per reaction well. A
majority of the assays displaying true positive amplification at this concentration
were for the following organisms: L. monocytogenes (6.3 copies), S. enterica (3.7
copies), V. parahaemolyticus (3.6 copies), H. pylori (11.1 copies), C. jejuni (11.6
copies), and V. cholerae (4.6 copies). Assays targeting all other organisms within this
concentration had a detection limit higher than 10 copies per reaction well, implying

those organisms were not detectable at this concentration. In addition, a high

33

majority of the targeted assays not displaying amplification at higher concentrations
(in Figure 2.1, top panel) were for primer sets targeting the same organisms. For
example, with 2,000 fg (100 to 1,000 copies) of target gDNA per reaction spiked into 66.6 pg
per reaction, the following organisms displayed false negative signals in at least two of the
three replicates: P. aeruginosa (three out of five primer sets), S. aureus (1 out of 4 primer

sets), G. intestinalis (4 out of 5 primer sets). This accounts for all assays (8 of 69)
Showing false negative amplification at 100 to 1,000 copies per reaction well. A
closer look indicated that the GC content was either very low (37% for S. aureus) or
very high (67% for P. aeruginosa). G. intestinalis had the largest genome of all the
organisms targeted (12 Mbp).

Previous PCR studies on organisms with high GC content have found similar
influences and suggest it is due to stabilizing secondary structure influencing
polymerase extension or primer-template annealing, causing false negative
observations (16, 24, 37). Solutions to minimize the effect include using a higher
denaturation and annealing temperature, or using additives such as glycerol, betaine,
orrnamide, or dimethyl sulfoxide. The false negative signal observed with S. aureus
was likely due to the low GC content of the organism requiring the length of the
primers to be increased to maintain suitable stability. The reverse primer for the S.
aureus assay displaying a false negative signal was longer (32 bp) than any other
primer tested in this experiment. Internal complements within longer primers may
reduce annealing to target sequences (40).

The false negative signal observed with G. intestinalis may be due to lower
relative abundance caused by having a larger genome, and a potential reduction in

availability of target. The influence of genome Size on amplification potential has

34

been described previously (1 l, 15), and may be due to a decrease in relative
abundance of template over nontarget DNA. Garner pr0posed that in addition to
decreased relative abundance, there is an increased chance of non-specific annealing
of primers to nontarget regions, diminishing annealing of primers to the target strand
(15). Optimizing the PCR cycle conditions or concentration of reagents may alleviate
false negative signals (41); however, changing these parameters may influence the
specificity and sensitivity of targeted assays that behaved well.

Since primers were designed to have the same theoretical melting temperature
(Tm) the terminal 3’ end of primers was also examined. The GC content, Tm, and
binding energy within the terminal 7, 5, 3 and 2 bases on the terminal 3’ end of
primers was considered for all false positive signals. Correlations were used to
determine that both the Tm of the last 7 bases and the GC content of the last 5 bases
had the highest influence on false positive signal (considering targets alone and
spiked into a background). To demonstrate, primer sets were grouped based on the
GC content within the last 5 bases of the terminal 3’ end (Figure 2.1, bottom panel).
For this figure, primer sets were grouped based on the GC content within the terminal
3’ end. Two out of 7 primer sets (28%) with 5 GC within the last 5 hp at the 3’ end
of the primer displayed false positive amplification. The percentage of false positive
signals decreased with the amount of GC in primers 3’ end. Influence of the 3’ end of
a primer on the specificity of amplification has been described previously (32). As a
result, many primer design software’s now analyze the 3’ end of potential primers,

while Primer3 emphasizes stability of 5 base segments of the terminal 3’ end (38).

35

It should be noted that results obtained with the OpenArraym nanoliter
volume reactions are comparable to conventional microliter QPCR. Low volume PCR
has been optimized in nanoliter volume reactions by adjusting surface chemistry (34),
ramping rates and decreased annealing temperatures (47), and adjusting PCR Master
mix composition with extra SYBR Green 1, BSA and forrnarnide added to the
standard PCR mixture. Cross platform comparisons between PCR performed in the
microplate format (10 and 20 ul volume reactions in the 7900HT) and with the
OpenArray have Shown high similarity in PCR efficiency, and detection limit
(Ortenberg, E. and Roberts D., unpublished data; 7). The comparison also
demonstrated a high correlation (between the two platforms) of Specific gene
regulation patterns between experimental (diseased heart) and control (normal adult
heart) tissues. The PCR efficiency observed with the experiments described in this
study (described below) also demonstrates success of PCR primers with the

OpenArray environment.

2.4.2 Development of predictive Ct equation. An empirical equation was
developed using the sequence Specific results observed with the primers designed to
develop the predictive Ct equation. Multiple parameters were considered for the
predictive Ct equation including size of genome, GC content of genome, and primer binding
energy. Since all primers were designed with the same theoretical melting temperature (Tm)
(for Simultaneous amplification of all primer sets on the OpenArray plate), the terminal 3’
end of primers was also considered. This included the binding energy, position of G and C
bases, Tm, and GC content for the terminal 7, S, 3 and 2 bases on the primers 3’ end.

Correlations were used to determine which parameters had a greater influence on the Ct

36

value. Inclusion of parameters other than those chosen either had no affect or decreased the
correlation between the predictive and experimental Ct. In addition, using the general linear
model requires that all variables are independent, thus only the top independent parameters
were used. For example, parameters such as binding energy are not entirely independent of
the Tm of the terminal 7 bases, as a primer with a high binding energy may have a higher Tm

on terminal 3’ end. Five parameters were identified that influenced the Ct of each
primer set. These were: i) genome size of the targeted organism, ii) target organism
concentration, iii) GC content of the targeted organism, iv) amplicon length, and v)
and theoretical melting temperature of the last 7 bases on the primer 3’ end. The
correlation of these variables on the threshold cycle was in the order listed (i.e.
genome size of the targeted organism had the highest correlation, and Tm of last 7
bases on primer 3’ end had the lowest).

A multiple parameter linear regression curve was used to place a weighted

influence on each of these parameters and the following equation was developed:

 

(Ct—25.43)+(7.755xGC)+(0.01275xa1)+(0.0084638xTmlast7)+(0.254293biase)]

Y = 10[ —3.6841

where;
GC = GC content (e.g. an organism with 30%; GC = 0.3)
a1 = amplicon length (bases long)
Ct = experimental threshold cycle
Y = starting copies

Bbase = Size of genome (Mbp)

37

Tmlast 7» = melting temperature calculated with 7 terminal bases on primer 3 ’
end; lowest temperature (°C) of the two primers with base stacking
calculation (42).

A high correlation (R2= 0.816) between threshold cycle predicted with the
predictive Ct equation and the experimental threshold cycle was observed (Figure
2.2). Since the number of a gene copies will vary based on the targeted organisms and
gene, the predictive Ct equation solely predicts starting copies per reaction. The
accuracy of the equation was tested and compared with standard curves using
validation sample mixtures with the primer set designed to validate the empirical Ct
equation.

It is likely that constants in the predictive Ct equation will change based on the
chosen system, reaction conditions, and reagents fi‘om different vendors. Therefore, some
validation is necessary for setting up novel highly parallel assays. BioTrove suggests using
the same reaction conditions for all SYBR based diagnostics on Openarl'ayTM plates.
Therefore, users of BioTrove for diagnostics of microbial communities can readily apply the
predictive Ct equation. Development of a predictive Ct equation for a new assay will consist
of validating with organisms from targeted genera (with varying target characteristics), and
establishing a multiple parameter linear regression curve to place a weighted influence on
parameters. To normalize for the stability of the reagent (fresh one vs a less fresh one), lot
number between reagents, sample loading, and inhibition within various samples, inter assay
normalization with an internal control can be used with the predictive Ct equation. Internal

controls can consist of DNA strands absent in the diagnostic target sample (1, 9, 30), and/or a

universal and stable unregulated endogenous standard (28, 29, 36).

38

2.4.3. Validation of an organism’s sequence characteristics on primer
success rate. Validation sample mixtures were tested with primers designed to
validate the predictive Ct equation to determine the success rate of novel primers.
The influence of primers and targeted organism GC content on the success of primers
was observed (Figure 2.3). Results Show that the sequence of organisms with
extreme GC content (low and high) had lower success rates. A study by Housley et
al., (18) observed similar results concerning influence of GC content and success of
primers, describing a success rate of 56.9% for primers designed to target an
amplified region with GC content greater than 50%, and 74.2% success for primers
designed to target an amplified region with GC content less than 50%. Organisms
targeted with high GC content tend to frequently give weak signals in amplification
(due to secondary structure and template-template annealing), and primers with high
GC content will amplify non-targeted regions (due to high stability). In addition,
organism with a low GC content in the genome will have a higher rate of false
negative signals. The number of assays required for confidently calling presence or
absence of an organisms is dependant on the success rate of designed primers.
Therefore, targets with extreme GC content will require more assays for presence and
absence calls.

These results suggest that validation requirements can be reduced if multiple
primer sets are assayed simultaneously, overcoming issues in specificity and
sensitivity observed with organisms and genes with extreme GC content. Typically,
amplification reactions can be optimized to overcome issues with false positive or

negative signals. However, large-scale PCR diagnostics such as the platform

39

described here require broad conditions to satisfy suitable reactions for the majority
of targeted assays. Using redundant probes to increase confidence in presence and
absence calls has been described for other high-throughput platforms such as
microarrays (31, 54). It Should be noted that these observations may not be observed
using methods that provide. greater differentiation between true positive and false
positive amplification; such as TaqMan QPCR, dissociation analysis, or ligation
based. However, cost and time to complete diagnostics significantly increased with

these approaches, especially for high-throughput diagnostics.

2.4.4. Validation of predictive Ct equation. The predictive Ct equation was
examined and compared with standard curves (obtained from dilution experiments
performed with primers designed to develop the predictive Ct equation), by predicting
the starting copy number in the 36 validation samples tested with the primers
designed to validate the predictive Ct equation. A distribution of predicted values
shows a high distribution of predicted starting copies around 20, 100, and 1,000,
which is close to the actual starting copy numbers of 10, 100, and 1,000 spiked into
the validation samples (Figure 2.4). Comparing predicted starting copies using the
Wilcoxon Signed-Rank Test showed that the standard curve tended to predict higher
starting copies than using the predictive Ct equation. Using the predictive Ct
equation made differentiating between assays spiked at 10 and 100 starting copies
unclear.

The large distribution observed with both quantitative measures may be due to

differences in reaction constitute lot number (between experiments performed to

40

establish and validate the predictive Ct equation), manually mixing of samples, and
the chosen calibration parameters. The influence of reagent formulation and
calibration parameters (e. g. replication, serial x-fold dilution) on precision of standard
curves have been documented previously (12, 25). Inaccuracies in quantifying gDNA
using UV absorbance may have also occurred. The correlation of predictive starting
copies calculated using the standard curve and predictive Ct ecluation was R = 0.87.
This suggests that the error between predicted and actual starting copies may have
been due to inaccuracies in mixing validation samples and/or variations caused by
reagents and not the predictive methods themselves.

Inhibitions within various environmental samples may have also influenced
quantification. The distribution of PCR efficiencies shifted when targets were spiked
into gDNA from different environmental waters (Figure 2.5, left panel), as observed
by testing primers and samples designed to validate the predictive Ct equation.
Tertiary effluent background shifted the distribution of PCR efliciency above one and
the river water shifted the distribution below one. An analysis of variance showed
significant inequality involving the mean PCR efficiencies between the control and 2
backgrounds. Approximately 83% of targeted assays with 1,000 to 10,000 starting
copies had a Ct standard deviation equal to or less than 0.35, 80% with 100 to 1,000
starting copies, 33% with 10 to 100 starting copies, and 15% with 1 to 10 copies. A
Ct standard deviation of 0.35 corresponds to a coefficient of variation of 25% for
estimated starting copies (assuming a standard curve slope of -3.3), which has been
described in previous reports as a cutoff for acceptable precision in QPCR diagnostics

(4, 20, 44). A correlation matrix was used to examine if primer characteristics

41

influenced assays with high standard deviations (data not shown). Analysis showed
that size of genome had the highest weighted influence on standard deviation of the
Ct, followed by number of starting copies and GC content of the target genome.
Other studies have observed an influence of GC on primer success. A study by
Vanichanon et al. (48), observed reduced repeatability with primers sets with high GC
content. Thus, primers with GC content closer to 50% will be ideal for maintaining
high reproducibility (10, 21, 23). This fluctuation in efficiency could be due to the
designed assays, enzyme instability, and sample dependant inhibitions (22).

Random micro-scale defects of the interior polymer surface coating of
OpenArrayTM plates, as suggested by Morrison et al. (34), may have also contributed
to differences between predicted and actual starting copies. Inna-assay
reproducibility was examined with three replicates performed on three separate
OpenArrayTM plates ran simultaneously with primers and samples designed to
establish the predicted Ct equation. Results illustrate that 93% of the assays
displayed a coefficient of variation equal to or less than 25% between OpenArrayTM
plates for 1,000 to 10,000 starting copies, 93% for 100 to 1,000 starting copies, 70%
for 10 to 100 starting copies, and 33% for l to 10 starting copies (Figure 2.5, right
panel). Results suggest greater deviation was observed between sample backgrounds
than between plates, and the accumulation of both aspects may have influenced the
distribution between predicted and actual starting copies.

Future diagnostics will include both an exogenous internal positive control
and a universal assay targeting the 16S rRN A gene to allow normalization for the

potential influence of sample inhibitions, plate variation, lot constitutes, and assays on

42

threshold cycle (17, 26, 52). Cumulatively, the large distributions observed with
predicting starting copies with standard curves along with the cumbersome
requirement of validation make using the predictive Ct equation an attractive
potential alternative. One other alternative strategy to quantify starting copies
without generation of standard curves has been described using competitive PCR with
a fluorescence quenching (46). This alternative has the potential to quantify DNA
within presence of inhibitions as tested with high concentrations of gDNA fiom pure
cultures Spiked into Hurnic acid, however, the accuracy and precision of
quantification and influence of inhibitors was not described with samples containing

less than 1,000 starting copies.

2.5 CONCLUSION

The influence of sequence characteristics of primers, amplicons, target genome size,
amplification conditions, and the matrix in which the target is present on the Ct was
explored. A predictive Ct equation was experimentally established by examining
PCR-based assays targeting multiple VMGS, and may be an efficient alternative to
generating standard curves. These results are valuable when developing high
throughput and reliable screening tools for a large number of pathogens without

extensive validation.

2.6 ACKNOWLEDGMENTS

We are grateful for the help fiom Elen Ortenberg with a proof of principal study

with BioTrove.

43

This work was supported by the National Institutes of Health (Grant R01
RR018625-01), Environmental Protection Agency (RD83162801-0 &
RD83301001), and 21St Century Michigan Economic Development Corporation

(GR-476 PO 085P3000517).

44

Table 2.1 Pathogens and VMGS targeted with OpenArrayTM plates and organisms

used for validation.

mixtures prepared to validate the predictive Ct equation.

The VMGS in bold were only tested in validation sample

 

Genus

Aeromonas

Bacillus

Bordetella

Burkholderia

C ampylobacter

Clostridium

C oryhebacterium
Caxtella

Enterococcus

Escherichia

Francisella

Haemophilus

Species

hydrophila

cereus, anthracis

pertussis

mallet, pseudomallei

jejuni, coll

perfringens.
dtflicile, botulinum

diphtheria
burnetii

faecaltls, faecium

call
(including Shigella)

tularensis

influenzae

Marker
Genes

Targeted

alt. ech. tapA.

arcV, ascT
atxA, capA,

lef, cerA, clo,
IlyIII, Iner,

msA, plcR
cyaC, dnt,
ptx, tcfA
pilA. pilD.
SLI6310MP,
BPSSl407

hipO, racR,
maPA. gym.
cdtC, cdtB,
cth, ciaB,
cfrA, tonB

6pc, pic. pfo,
Cpbsl. etx,

bontA, In33,

ha70, cth,
cdtB

dtxr

dotA, icmP,
ich, mip
cylA, ace,

cylLS, esp, gelE

uidA, stxl, stx2,
eae, papG, lucC,

elle, sfaA,
ipaH, InxiH
acpA, clpB,
mglA
lic2A, tbpl,
tbp2

45

Designed Primer Sets

Primer
sets

5

9

10

ll

from
Literature

0

0

Organisms
tested

(ATCC)

10987

BAA-5890

700819

12916

19433

BAA-460D

51907D

Table 2.1 continued

Helicobacter

Klebsiella

Legionella

Leptospira

Listeria

Mycobacterium

Illycoplasma

Proteus

Pseudomonas

Salmonella

S erratic

Staphylococcus

pylori

pneumoniae

pneumophila

interrogans

monocytogenes

MAC
paratuberculosis,
MT B tuberculosis,
leprae
genetalium

mirabilzls

aeruginosa

marcescens

aureus

cagA. cagE, 10
ureA, virD4,

virB9, virBl l,

urel, ureB,

flaB

nuc, magA, 5
kng, kci, kca

lepB, ich, 9
IBM. mip,

dotA, icmB,

icmR, IspH,

lssD

hlyB, lipA, 6
chpl, lip21,

omle, chpK,

etpK

actA, hlyA, 8
inlA, mp1,

pch, lisA,

pIcA. tap.

prfA

erp, glnAl, 6
mmpA,

mmpB, 18900,

mmpC, pch

bww3, Pl, 4
P200, P30

lpp, uca, 4
zapA, zapD

exoS, iasA, ll
perv,

pilDchA,

pepD. popB.

Ias, etA, prfA,

pilD

fimA, flicC, 15
fljB, invA,

MA, MD,

invE, prgH,

sipB, sipC,

spar. tyv

flhDC, hasD, 4
lipA, safA

alphahly, seA, lO
seC, tsstl,

IukE, lukF,

lytS, nuc, seA,

tst

46

700392

33152

43642

15313

33530D

10145

13311, 19585

700699

Table 2.1 continued

Strephocococcus

Vibrio

Yersim'a

Cryptosporidium

Giardia

Total: 30

pyogenes

cholerae, mimicus,
parahaemolyticus,
vulm'ficus

enterocolitica,
pestis,
pseudotuberculosis

parvam, homim's

intestinalis, lamblia

ssa, mtsA,
ska, speA,
spec
ace, ompU,
tdh. tth. tlh.
toxR, ctxA,
cth, mshA,
tcpA, zot, trhl,
VPL Whas WP,
tagA, trbl
CFPs YOPD,
yscD, ystsA,
bipA. yadA.
YSPB, YSCN,
ail
gp40, hsp70,
COWP, cp23,
cgd7
B-giardin,
VSPH71,
VSP4177,
VSP4173, vsp,
VSPH7

200

47

14

(200

39315
43996

55075

PRA-67

30888 A-l

2.7 FIGURES

Figure 2.1 Impact of gDNA fiorn various environmental water samples and
primer stability on specificity and sensitivity. Top panel: the percent of
targeted and non-targeted primer assays displaying amplification at various
dilutions of organisms spiked in gDNA from environmental samples. Errors
bars represent standard deviation between replicates performed on 3 plates.
Bottom panel: the sum of GC bases on the terminal 3’ end (various size
circles) versus the percentage of primer sets displaying false positive
amplification when targets are spiked into background gDNA and not spiked
into background gDNA.

48

H
O
O

OO
O

  
 

  

+ Targeted (river bkg)
-0... Non-targeted (river bkg)
+ Targeted (effluent bkg)
-------A Non-targeted (effluent bkg)
+ Targeted (no bkg)

o Non-targeted (no bkg)

O\
O

.b
O

N
O

 

"a v. fi

0.1t01.0 1.0t010 lOtolO2 102t0103 103t0104

Assays with amplified signal (%)

Genomic copies per reaction

DJ
£11

GC in last 5 bases of primer 3’ end

N
LII

l—I
LII

U!

 

7

l l I 1

5 15 25 35
False positive assays: no bkg (%)

False positive assays: spiked in bkg (%)

49

28-

26i ”-1
24" W
22J m
2320‘ l‘.“
at

161 “In "a!

14‘

124

 

 

 

10 l I l l I l I l 1

10 12 14 16 18 20 22 24 26 28
Experimental Ct

Figure 2.2 Experimental versus predicted Ct using predictive Ct equation, results
obtained with primer designed to establish the predictive Ct equation. The equation is
derived using amplicon length, starting genomic copies, number of base pairs in
target organism’s genome, GC content of target organism’s genome, and theoretical
melting temperature (Tm) of last 7 bases of primers 3’ end. Errors bars represent
standard deviation of experimental Ct between replicates on three plates.

50

 

F that. .e. a. v... J'E. new; L. at», 3133,35; "Imfltfi‘ “Fm
HEHIIII‘I|“““\‘
_ I, : ““‘I‘l‘l‘hhh‘lms
wfimmerfix .. ..

Success rate %

 

 

 

  
   

a 0.40
’a 0.45
0
*0 0.50
o 030
. 0. .
0% 0.60 055 0.50 0.45 040 35
(‘3 0.65 0.60 .
91' Genome GC Content

Figure 2.3 Success rate (%) of all assays targeting an organism. Success rate is
defined as the sum of all true positive and negative assays divided by the sum of all
assays targeting a single organism. The success rate was taken as an average of all 35
validation samples mixtures prepared to validate the predictive Ct equation.

51

30..

25-

N

O
s
l
l

u—
Us
4

O
a

 

Number of predicted assays

l

r --

e
v V i

0 10 20 30 40 50

M
s_

 

 

0

+ Standard curve
—0— Predictive Ct equation

 

 

 

 

 

30 a 30 -
25 d 25 "
20 .
)1-
15 -
t)
10 q
5 a
0 ’ .s 1 I 7
100 200 300 400 1500 2500 3500
Starting copies

Figure 2.4 Distribution of predicted starting copies numbers using predictive Ct
equation and standard curves for validation samples and primers designed to
validated the predictive Ct equation. Note: x-axis of the 3 figures have different
scales. Template of all organisms were spiked at either 10, 100, or 1,000 genomic
copies per reaction well (indicated by dotted lines). Errors bars represent standard
deviation between three replicates performed on the same OpenArrayTM plates.

52

t—-‘
A

 

 

 

 

O
1.3- .
I
E 1.2. . .
.2 1.1
a
% 1.0.
8‘3
on 0.9-
O
0.8 o E
0.7 6\ “60‘ $3.6“
00° _ 6 «195
saw“
100 ..... up“
I”
I .‘
80 I,’ g
A t’ j
3360 / j
(I) .
5‘ l ..'
m I I
<40 'l .-
1’ ll _10:t010‘3‘
10 to 10
l .. ..............
20 if ...... 10to 102
______ ,z. - ------- 1.0t010
0.1 1 10

Ct standard deviation between plates

Figure 2.5 Factors potentially influencing quantitative inaccuracies as observed with
primers and samples designed to establish the predictive Ct equation. Top panel: box
plot showing distribution of PCR efliciency for target organisms Spiked into gDNA
fi'om complex environmental waters and a control (no background). Bottom panel:
the cumulative frequency distributions of the standard deviation of Ct determined
between three separate OpenArrayTM plates with various transcript concentrations.

53

2.8 REFERENCES

10.

Abdulmawjood, A., S. Roth, and M. Bultc. 2002. Two methods for construction
of internal amplification controls for the detection of Escherichia coli 0157 by
polymerase chain reaction. Mol. Cell. Probes 16:335-339.

Altschul, S. F., T. L. Madden, A. A. Schafl'er, J. Zhang, Z. Zhang, W. Miller,
and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of
protein database search programs. Nucleic Acids Res. 25:3389—402.

Becker, K., A. W. Friedrich, G. Peters, and C. von Eifl'. 2004. Systematic
survey on the prevalence of genes coding for staphylococcal enterotoxins SEIM,
SEIO, and SEIN. Mol. Nutr. Food Res. 48:488-495.

Behets, J., P. Declerck, Y. Delaedt, L. Verelst, and F. Ollevier. 2007. A duplex
real-time PCR assay for the quantitative detection of Naegleria fowleri in water
samples. Water Res. 41:118-126.

Blanco, J., M. Blanco, J. E. Blanco, A. Mora, E. A. Gonzalez, M. I.
chardez, M. P. Alonso, A. Coir-a, A. Rodriguez, J. Rey, J. A. Alonso, and
M. A. Usera. 2003. Verotoxin-producing Escherichia coli in Spain: Prevalence,
serotypes, and virulence genes of 0157 : H7 and non-0157 VTEC in nmiinants,
raw beef products, and humans. Exp. Biol. Med. 228:345-351.

Bonants, P., M. Szemes, R. van Doom, and C. Schoen. 2007. Quantitative
multiplex detection of plant pathogens using FRI-lock probes and universal, ultra-
high-throughput real-time PCR on OpenAnays (TM). BMC Genomics 97:812-
812.

Brenan, C., and T. Morrison. 2005. High throughput, nanoliter quantitative
PCR. Drug Discovery Today: Technologies 2:247-253.

Chacon, M. R., M. J. Figueras, G. Castro-Escarpulli, L. Soler, and J. Guarro.
2003. Distribution of virulence genes in clinical and environmental isolates of
Aeromonas spp. Antonie Van Leeuwenhoek 84:269-278.

Cubero, J., J. van der Wolf, J. van Beckhoven, and M. M. Lopez. 2002. An
internal control for the diagnosis of crown gall by PCR. J Microbiol Methods.
51:387-392.

Diefi'enbach, C. W., T. M. J. Lowe, and G. S. Dveksler. 1993. General
Concepts for PCR Primer Design. Per-Methods and Applications 3:S30—S37.

54

11.

12.

l3.

14.

15.

16.

17.

18.

19.

20.

21.

Farrelly, V., F. A. Rainey, and E. Stackebrandt. 1995. Effect of genome size
and rrn gene copy number on PCR amplification of 16S ribosomal-RNA genes
fi'om a mixture of gacterial species. Appl. Environ. Microbiol. 61:2798-2801.

Farriol, M., and X. Orta. 2005. Influence of reagent formulation on mRNA
quantification by RT-PCR using imported external standard curves. Acta
Biochirn. Pol. 52:845-848.

Feltman, H., G. Schulert, S. Khan, M. Jain, L. Peterson, and A. R. Hauser.
2001 . Prevalence of type 1H secretion genes in clinical and environmental isolates
of Pseudomonas aeruginosa. Microbiology 147:2659-2669.

Fukushima, H., Y. Tsunomori, and R. Seki. 2003. Duplex real-time SYBR
green PCR assays for detection of 17 species of food- or waterborne pathogens in
stools. J. Clin. Microbiol. 41:5134-5146.

Garner, T. W. J. 2002. Genome size and microsatellites: the effect of nuclear
size on amplification potential. Genome 45:212-215.

Grassi, M., E. Volpe, V. Colizzi, and F. Mariani. 2006. An improved, real-time
PCR assay for the detection of GC-rich and low abundance templates of
Mycobacterium tuberculosis. J. Microbiol. Methods 64:406-410.

Hartman, L. J., S. R. Coyne, and D. A. Norwood. 2005. Development of a
novel internal positive control for Taqman based assays. Mol. Cell. Probes.
19:51-59.

Housley, D. J. E., Z. A. Zalewski, S. E. Beckett, and P. J. Venta. 2006. Design
factors that influence PCR amplification success of cross-Species primers among
1147 mammalian primer pairs. BMC Genomics 7.

Hsu, C. F., T. Y. Tsai, and T. M. Pan. 2005. Use of the duplex TaqMan PCR
system for detection of Shiga-like toxin-producing Escherichia coli 0157. J. Clin.
Microbiol. 43:2668-2673.

Huang, H. Y., and T. M. Pan. 2004. Detection of genetically modified maize
MON810 and NK603 by multiplex and real-time polymerase chain reaction
methods. J. Agric. Food Chem. 52:3264-3268.

Innis, M. A., and D. H. Gelfand. 1990. Optimization of PCRs, p. 3-12. In M. A.
Innis, D. H. Gelfand, J. J. Sninksy, and T. J. White (ed.), PCR protocols: A guide
to methods and application. Academic Press, Inc., New York.

55

22.

23.
24.
25.
26.
27.
28.
29.
30.

31.

Kains, P. 2000. The PCR plateau phase-towards an understanding of its
limitations. Biochim. Biophys. Acta 1494:23-7.

Kim, H. S., and O. Smithies. 1988. Recombinant fragment assay for gene
targetting based on the polymerase chain-reaction. Nucleic Acids Res. 16:8887-
8903.

Kureishi, A., and L. E. Bryan. 1992. Pre-boiling high GC content, mixed
primers with 3' complementation allows the successful PCR amplification of
Pseudomonas aeruginosa DNA. Nucleic Acids Res. 20:1155-1155.

Lai, K. K. Y., L. Cook, E. M. Krantz, L. Corey, and K. R. Jerome. 2005.
Calibration curves for real-time PCR. Clin. Chem. 51:1132-1136.

Layton, A., L. McKay, D. Williams, V. Garrett, R. Gentry, and G. Sayler.
2006. Development of Bacteroides 16S rRNA gene TaqMan-based real-time PCR
assays for estimation of total, human, and bovine fecal pollution in water. Appl.
Environ. Microbiol. 72:4214-4224.

Lebuhn, M., M. Effenberger, A. Gronauer, P. A. Wilderer, and S. Wuertz.
2003. Using quantitative real-time PCR to determine the hygienic status of cattle
manure. Water Sci. Technol. 48:97-103.

Lee, C., S. Lee, S. G. Shin, and S. Hwang. 2008. Real-time PCR determination
of rRNA gene copy number: absolute and relative quantification assays with
Escherichia coli. Appl. Microbiol. Biotechnol. 78:371-376.

Livak, K. J., and T. D. Schmittgen. 2001. Analysis of relative gene expression
data using real-time quantitative PCR and the 2(T)(-Delta Delta C) method.
Methods 25:402-408.

Malorny, B., J. Hoorfar, C. Bunge, and R. Helmuth. 2003. Multicenter
validation of the analytical accuracy of Salmonella PCR: towards an international
standard. Appl. Environ. Microbiol. 69:290-296.

Miller, S. M., D. M. Tourlousse, R. D. Stedtfeld, S. W. Bausbke, A. B.
Herzog, L. M. Wick, J.-M. Rouillard, E. Gulari, J. M. Tiedje, and S. A.
Hasbsham. 2007. An in situ synthesized virulence and marker genes (V MG)
biochip for detection of waterborne pathogens. Appl. Environ. Microbiol. In
Review.

56

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

Miura, F., C. Uematsu, Y. Sakaki, and T. Ito. 2005. A novel strategy to design
highly specific PCR primers based on the stability and uniqueness of 3 '—end
subsequences. Bioinforrnatics 21:4363-43 70.

Molenkamp, R., A. van der Ham, J. Schinkel, and M. Beld. 2007.
Simultaneous detection of five different DNA targets by real-time Taqman PCR
using the Roche LightCycler480: Application in viral molecular diagnostics. J.
Virol. Methods 141:205-211.

Morrison, T., J. Hurley, J. Garcia, K. Yoder, A. Katz, D. Roberts, J. Cho, T.
Kanigan, S. E. Ilyin, D. Horowitz, J. M. Dixon, and C. J. H. Brenan. 2006.
Nanoliter high throughput quantitative PCR. Nucleic Acids Res. 34.

Pannucci, J., H. Cai, P. E. Pardington, E. Williams, R. T. Okinaka, C. R.
Kuske, and R. B. Cary. 2004. Virulence signatures: microarray-based
approaches to discovery and analysis. Biosens Bioelectron 20:706-718.

Pfafl'l, M. W. 2001. A new mathematical model for relative quantification in real-
time RT-PCR. Nucleic Acids Res. 29:e45.

Raychaudhuri, A., and P. A. Tipton. 2004. Protocol for amplification of GC—
rich sequences fi'om Pseudomonas aeruginosa. Biotechniques 37 :752, 754, 756.

Rozen, S., and H. Skaletsky. 2000. Primer3 on the W for general users and
for biologist programmers. Methods Mol. Biol. 132:365-86.

Rozynek, E., K. Dzierzanowska-Fangrat, P. Jozwiak, J. Popowski, D.
Korsak, and D. Dzierzanowska. 2005. Prevalence of potential virulence markers

in Polish Campylobacter jej uni and Campylobacter coli isolates obtained from
hospitalized children and fiom chicken carcasses. J. Med. Microbiol. 54:615-619.

Rychlik, W., and R. E. Rhoads. 1989. A computer-program for choosing
optimal oligonucleotides for filter hybridization, sequencing and invitro
amplification of DNA. Nucleic Acids Res. 17:8543-8551.

Sahdev, S., S. Saini, P. Tiwari, S. Saxena, and K. S. Saini. 2007. Amplification
of GC-rich genes by following a combination strategy of primer design, enhancers
and modified PCR cycle conditions. Mol. Cell. Probes 21:303-307.

SantaLucia, J. 1998. A unified view of polymer, dumbbell, and oligonucleotide
DNA nearest-neighbor thermodynamics. Proc. Nat. Acad. Sci. U S A 95: 1460-
1465.

57

43.

44.

45.

46.

47.

48.

49.

50.

51.

Sergeev, N., M. Distler, S. Courtney, S. F. Al-Khaldi, D. Volokhov, V.
Chizhikov, and A. Rasooly. 2004. Multipathogen oligonucleotide microarray for
environmental and biodefense applications. Biosens Bioelectron 20:684-698.

Shindo, Y., H. Kuribara, T. Matsuoka, S. Futo, C. Sawada, J. Shono, H.
Akiyama, Y. Goda, M. Toyoda, A. Hino, T. Asano, M. Hiramoto, A. Iwaya,
S. I. Jeong, N. Kajiyama, H. Kato, H. Katsumoto, Y. M. Kim, H. S. Kwak, M.
Ogawa, Y. Onozuka, K. Takubo, H. Yamakawa, F. Yamazaki, A. Yoshida,
and T. Yoshimura. 2002. Validation of real-time PCR analyses for line-specific
quantitation of genetically modified maize and soybean using new reference
molecules. J. AOAC Int. 85:1119-1126.

Skanseng, B., M. Kaldhusdal, and K. Rudi. 2006. Comparison of chicken gut
colonisation by the pathogens Campylobacter jejuni and Clostridium perfiingens
by real-time quantitative PCR. Mol. Cell. Probes 20:269-279.

Tani, H., T. Kanagawa, S. Kurata, T. Teramura, K. NakamuraO, S.
Tsuneda, and N. Noda. 2007. Quantitative method for specific nucleic acid

sequences using competitive polymerase chain reaction with an alternately
binding probe. Anal. Chem. 79:974-979.

van Doorn, R., M. Szemes, P. Bonants, G. A. Kowalchuk, J. F. Salles, E.
Ortenberg, and C. Schoen. 2007. Quantitative multiplex detection of plant

pathogens using a novel ligation probe-based system coupled with universal,
high-throughput real-time PCR on OpenArraysTM. BMC Genomics 8:276.

Vanichanon, A., N. K. Blake, J. M. Martin, and L. E. Talbert. 2000.
Properties of sequence-tagged-site primer sets influencing repeatability. Genome
43:47-52.

von Eifl', C., A. W. Friedrich, G. Peters, and K. Becker. 2004. Prevalence of
genes encoding for members of the staphylococcal leukotoxin family among

clinical isolates of Staphylococcus aureus. Diagn. Microbiol. Infect. Dis. 49:157-
162.

Vora, G. J., C. E. Meador, M. M. Bird, C. A. Bopp, J. D. Andreadis, and D.
A. Stenger. 2005. Microarray-based detection of genetic heterogeneity,
antimicrobial resistance, and the viable but nonculturable state in human
pathogenic Vibrio spp. Proc. Natl. Acad. Sci. U S A 102:19109-19114.

Wang, L. X., Y. Li, and A. Mustapha. 2007. Rapid and simultaneous
quantitation of Escherichia coli 0157 : H7, Salmonella, and Shigella in grolmd
beef by multiplex real-time PCR and immunomagnetic separation. J. Food Prot.
70:1366-1372.

58

52.

53.

54.

Welti, M., K. Jaton, M. Altwegg, R. Sahli, A. Wenger, and J. Bille. 2003.
Development of a multiplex real-time quantitative PCR assay to detect Chlamydia
pneumoniae, Legionella pneumophila and Mycoplasma pneumoniae in respiratory
tract secretions. Diagn. Microbiol. Infect. Dis. 45:85-95.

Wick, L. M., W. H. Qi, D. W. Lacher, and T. S. Whittam. 2005. Evolution of
genomic content in the stepwise emergence of Escherichia coli 0157 : H7. J.
Bacteriol. 187:1783-1791.

Wilson, W. J., C. L. Strout, T. Z. DeSantis, J. L. Stilwell, A. V. Carrano, and
G. L. Andersen. 2002. Sequence-specific identification of 18 pathogenic
microorganisms using microarray technology. Mol. Cell. Probes 16:119-127.

59

CHAPTER THREE

Stedtfeld, R.D., Wick, L.M., Bausbke, S.W., Tourlousse, D.M., Herzog, A.B., Xia,
Y., Rouillard, J.M., Klappenbach, J.A., Cole, J.R., Gulari, E., Tiedje, J.M.,
Hashsham, SA 2007 . Influence of Dangling End and Surface Proximal Tail of Targets
on Probe-Target Duplex Formation in 16S rRNA Gene-Based Diagnostic Arrays. Appl

Environ Microbiol. 73:380-9.

60

INFLUENCE OF DANGLING END AND SURFACE PROXIMAL TAIL 0F TARGETS ON PROBE-

TARGET DUPLEX FORMATION IN 168 RRNA GENE-BASED DIAGNOSTIC ARRAYS

3.1 ABSTRACT

Dangling ends and surface proximal tails of gene targets influence the probe-
target duplex formation and affect the signal intensity of probes on diagnostic
microarrays. This phenomenon was evaluated using an oligonucleotide microarray
containing 18-mer probes corresponding to the 168 rRNA gene of 10 waterborne
pathogens and a number of synthetic and polymerase chain reaction amplified gene
targets. Signal intensities for Klenow/random primer labeled 16S rRNA gene targets vs.
45-mer synthetic targets were dissimilar for nearly 73% of the probes tested.
Klenow/random primer labeled targets resulted in 3.7 times higher interaction with a
complex mixture of 168 rRNA genes (used as background) compared to the interaction
of 45-mer targets with the same mixture. An 7 bases long dangling end sequence with
perfect homology to another single stranded background DNA sequence was sufficient to
produce a cross hybridization signal that was as strong as the Signal obtained by the
probe-target duplex itself. Gibbs free energy between the target and a well defined
background was found to be a better indicator of hybridization signal intensity than the
sequence or length of the dangling end alone. The dangling end was found to be
Significantly more prone to target-background interaction (Gibbs fi'ee energy of -7.6
kcal/mol) than the surface proximal tail (Gibbs free energy of -64.5 kcal/mol). This study

underlines the need for careful target preparation and evaluation of signal intensities for

61

diagnostic arrays using 163 rRNA and other gene targets due to the potential for target
interaction with a complex background.
3.2 INTRODUCTION

There are many factors that influence the hybridization Signal intensity between
long strands of DNA (referred to as targets) and oligonucleotide probes in microarray-
based diagnostics and gene expression studies (14). Some of these factors include
characteristics of fluorescent molecule used to label the target, labeling efficiency, length
of probe, target secondary structure, and hybridization conditions (3, 4, 6, 9, 15, 22, 28,
33-35, 37). However, factors such as interaction of dangling end and surface proximal
tail of targets with background DNA has not yet been systematically characterized. In
this study, the term dangling end refers to the sequence of target that extends beyond the
distal end of the probe and surface proximal tail refers to the sequence of target that
extends beyond the anchored end of the probe.

Target preparation and resulting length, sequence, double versus single stranded
composition, position of fluorescent label, and secondary structure of the dangling end
and surface proximal tail of targets influence the stability of duplex formation and
resulting signal intensity (25, 31, 44). The targets may also interact with other non-target
DNA sequences present in the background, previously referred to as “hitchhiking” (23).
This is especially true for 168 rRNA gene-based diagnostic microarrays where dangling
ends and surface proximal tails will generally represent conserved regions and may have
greater potential to interact with the background DNA or other targets. Because
background will be sample specific, comparative evaluation of two samples on the same

microarray using target mixtures that interact with each other may pose problems. Low

62

reproducibility and false positive/negative Signals with 16S rRNA gene based arrays is
partly accredited to bias caused by characteristics of target molecules (8, l7), and target-
background interaction may be one of the main reasons for this phenomenon. Poor
statistical relationships between experimental and predicted signal intensities (from Gibbs
free energy calculations) suggest that thermodynamic parameters between target and
probe duplexes are not fully understood (26). Thus, an evaluation of the influence of
dangling end and surface proximal tail on signal intensity and target-background
interaction is needed in order to interpret the signal intensities on 16S rRNA gene based
diagnostic and microbial community analysis arrays. The interaction among probe (P),
target (T), and background (B) can be represented by:

Pn <:> Tn c> Bn
Where, n represents the length of the corresponding DNA sequence. A given background
DNA sequence may also interact with the probe sequence (Pn <=> Bn). But it was not
considered significant in this study because signal due to background DNA when
hybridized alone was absent for the targeted probes.

The objective of this study was to systematically examine the impact of
dangling end and surface proximal tail of targets on hybridization Signal intensity and
evaluate target-backgromd interaction by: i) comparing products of unknown length
generated by random primer labeling with targets of known length, and ii) examining
interactions between well defined targets and backgrounds. This study underlines the
need for careful target preparation and evaluation of signal intensities for diagnostic
arrays using 16S rRNA and other gene targets due to the potential for target interaction

with a complex background.

63

3.3 MATERIALS AND METHODS

Influence of background on dangling end and surface proximal tail was evaluated
using an oligonucleotide microarray containing l8-mer probes corresponding to the 16S
rRNA gene of 10 waterborne pathogens and a number of synthetic and polymerase chain
reaction amplified gene targets. Using a set of 45-mer synthetic targets and
Klenow/random primer labeled targets generated by PCR amplification (assumed to be
approximately 200 to 1,500 bp long), the signal intensities were compared for a total of
95 probes targeting the 168 rRNA genes of 10 waterborne pathogens. Both types of
targets were then spiked individually into a complex mixture of 168 RNA genes to study
the influence of the type of background DNA that will be present as a result of target
amplification using universal primers and random primer labeling. Using a Cy3 end-
labeled 59-mer sequence as a model background DNA sequence designed to complement
the dangling end of 5 different targets of variable length (and the associated Gibbs free
energy), it was demonstrated that 8 to 12 bases of contiguous homology with the
dangling end is sufficient to cause cross hybridization. A second 59-mer Muence served
as a model background designed to complement the Slu'face proximal tail of 11 different
targets of various sequence similarity and Gibbs fiee energy. Sequences used for the
model backgrounds were from the dangling ends of non-targeted 16S rRN A genes for
which probes were also present on the array. Synthetic sequences used for the well

defined model background complemented the dangling end of non-target probes.

64

3.3.1 Preparation of target DNA. Four different types of targets were prepared:
i) mixture of Klenow/random primer labeled (with Cy5) targets prepared from 16S rRNA
genes of 10 pathogens (Tk-m), ii) mixtlue of Cy3 end-labeled 45-mer synthetic targets
corresponding to 95 probes for the 10 pathogens (T45-m), iii) one individual Cy3 end-
labeled synthetic target of length 102 bases (Td) to examine dangling end length, 4
additional Cy3 end-labeled synthetic target of length 47, 60, 80, and 106 bases (Tn,
where n is the length of the target) to study the effect of increasing the length of dangling
end, and iv) 11 individual Cy3 end-labeled synthetic targets of lengths 89 to 102 bases
(Ts) were used to study the effect of sequence similarity in the surface proximal tail of
targets. The sequence of dangling end and surface proximal tail for all the synthetic
targets (T45-m, Td, Tn, and Ts) matched the 16S rRNA gene sequence of the
corresponding pathogens. The approach to prepare these targets is summarized below.

Klenow/random primer labeled target mixture of 16S rRNA genes of pathogens
(Ht-m). Genomic DNA fi'om 10 bacterial pathogens (Table l) was used as the source of
16S rRNA genes. E. coli, L. pneumophila, P. aeruginosa, S. enterica, S. typhimurium, Y.
enterocolitica type strains were obtained fi'om the American Type Culture Collection
(ATCC; Manassas, VA) and grown as per the protocol provided. For H. pylori and C.
perfiingens, only genomic DNA was obtained from the ATCC. E. feacalis and C. jejuni
were kindly provided by Drs. Joan B. Rose and Vincent Young, respectively; both at
Michigan State University. DNA from pure cultures and the environmental sample was
extracted using Promega Wizard DNA Extraction Kit (Promega, Madison, WI).

The 16S rRNA gene was amplified from the respective genomic DNA using 27F

(5’-AGAG'ITI‘GATCMTGGCTCAG—3’) and 1525R (5’-AAGGAGGTGWTCCARCC-

65

3’) primer pairs (18) and Platinum Taq DNA Polymerase (Invitrogen; Carlsbad, CA).
Thirty cycles of the following temperature program was used for amplification:
denaturation at 94 °C for 30 s, annealing at 52 °C for 45 s, and elongation at 72 0C for 90
S. All PCR amplicons were cleaned using a Qiagen PCR clean up kit (Qiagen Inc.,
Valencia, CA).

The amplified 16S rRNA gene (250 ng) fiom each pathogen was individually
labeled with Cy5 using the Bioprime DNA Labeling Kit (Invitrogen, San Diego, CA).
Briefly, the protocol included a 90-min incubation of the amplicon with Klenow
polymerase and 5:1 amino-allyl-dUTPsz'TP (Ambion, Austin, TX) followed by Cy5
labeling. All amino-allyl-dUTP-labeled products were cleaned using a Qiagen PCR clean
up kit with modified phosphate wash buffer (5 mM K2HPO4, pH 8.0, 80% EtOH) and
phosphate elution buffer (4 mM K2HPOq, pH 8.5). Cyanine dye was attached by
incubating 3 to 5 pg of amino-allyl-dUTP-labeled DNA for 1 h in 50:50 mixture of 0.1 M
sodium carbonate buffer (pH 9.3) and N-hydroxysuccinimide ester Cy (dye (prepared in
fresh dimethyl-sulfoxide). DNA product from dye coupling was cleaned using a Qiagen
PCR clean up kit. A mixture of Cy5 labeled 16S rRNA gene was then prepared by
mixing 10 pmol (157 to 357 ng DNA) of labeled product from each of the 10 pathogens.
Klenow/random primer labeled targets were estimated to be between 200 to 1,500 bp (as
assessed by gel electrophoresis). All Klenow-labeled targets are expected to have
multiple Cy5 labels. Specific activity of Cy dye was calculated by dividing measured
pmol of nucleotides with measured pmol of Cy3/Cy5 dye for each target.

Mixture of 45-mer synthetic targets (T 45-m). A mixture of synthetic target was

designed and synthesized to complement a set of 95 probes (1 8-mer) in the middle region

66

of the 45-mer targets (Figure 3.1a). Each 45-mer synthetic target had a 14 nucleotides
surface proximal tail at the 5’ terminus and 13 nucleotides dangling end at the 3’
terminus. The sequences of both the overhanging ends matched the 16S rRNA gene
sequence of the corresponding pathogen. The single stranded mixture of sequences was
synthesized at the University of Michigan using an in situ oligonucleotide synthesis
technology (11, 12), end labeled with Cy3 at the 5’ terminus, and harvested fi'orn the
solid substrate.

Synthetic target with dangling ends of variable lengths a" d). A 102-mer synthetic
target was designed to hybridize to 9 different probes on the array (Td). After
hybridization, Td was expected to have a dangling end of 8 to 16 bases at the 3’ terminus
andasurfaceproximaltailof68to76basesatthe5’ terminus,dependingupontheprobe
to which it was hybridized (Figure 3.2a). Td was synthesized and mono end-labeled with
Cy3 at the 5’ terminus by IDT (Coralville, IA). A 59-mer defined background DNA (Bd,
described below) was designed to complement the dangling end sequence from position
86 to 98 with 4 to 12 base stretches of continuous homology. The sequence similarity
betweenthesurfaceproximal tail odeandtheBdwaslessthan25%.

The dangling end interaction due to length of overhang was studied further using
4 synthetic targets of varying length T47, T60, T80, and T106 (Tn, where n refers to the
length of the synthetic target, e.g., T60 is 60 bases long). Each Tn hybridized to 10
probes that had been designed from the same 16S rRNA sequence from the position 242
to 272 (starting at the 3’ end). Each Tn was synthesized and mono end-labeled with Cy3
at the 5’ terminus by IDT (Coralville, IA), and resulted in a 3’ dangling end of 21, 34, 54,

and 80 bases for the probe designed at start position 242, and 8, 21, 41, and 67 bases for

67

the probe designed at start position 254. At the 5’ terminus, all 4 synthetic targets had a
surface proximal tail of 8 to 21 bases depending on the probe it was hybridized to. Bd
complemented the dangling end of Tn by 0, 13, 33, and 59 bases.

synthetic targets (Ts) with 10 to 100% sequence similarin at the surface proximal
tail. The effect of sequence Similarity of the surface proximal tail of target to a defined
background DNA sequence was studied using 11 different targets (T s) of similar length
(89 to 102 bases). Each Ts was synthesized and mono end-labeled with Cy3 at the 5’
terminus by IDT (Coralville, IA). All the targets had 18 bases complementary to 52
probes on the microarray (Figure 3.3a); a dangling end of l to 16 bases and surface
proximal tail of 67 to 76 bases depending on the probe and target. Ts were designed to
have 10 to 100% sequence similarity to a single background sequence (BS, described

below).

3.3.2 Preparation of background DNA. Two types of background DNA were
used in this study: i) a mixture of 16S rRNA genes obtained by PCR fiom an
environmental sample referred to as complex background DNA (B165), and ii) two
defined background DNA of known sequences (both were 59-mers, referred to as Bd and
BS) with varying potential of interaction with dangling end of Td and surface proximal
tail of Ts. For Td and TS, synthetic sequences complemented the dangling end of non-
target probes. These are described below.

Complex background DNA (316s). A mixture of 168 rRNA genes amplified fiom
genomic DNA of activated sludge from the wastewater treatment plant in East Lansing,

MI) served as the source of complex background DNA. The genomic DNA was obtained

68

using the protocol described above except that a purification step was also performed
using Wizard DNA clean-up kit (Promega, Madison, W1). A similar PCR protocol was
used as in preparation of Tk-m, but only 25 cycles were used and annealing temperature
was set to 50 °C for 45 s. The extent of sequence similarity of the complex background
DNA to target sequences was unknown but expected to be substantial (because of
conserved regions of 16S rRNA genes) and a function of the length of background and
target DNA.

Synthetic background DNA of defined length Two 59-mer synthetic defined
background sequences (Bd and BS) were designed to complement the dangling end of Td
and surface proximal tails of Ts (11 different T, sequences were used). The extent of
sequence similarity for defined backgrormd DNA was carefirlly controlled to either result
in continuous homology of variable length at the dangling end of Td or varying sequence
similarityatthe surface proximaltailofTs. For Td, the sequence odehad4to 12bases
of continuous homology fiom position 86 to 98 considering the 5’ end as 0 (Table 1 and
Figure 3.2a). The Gibbs free energy between Td and Bd varied between -3.2 to -14.9
kcal/mol for 4 to 12 base stretches of continuous sequence homology at the dangling end.
Gibbs free energy between Bd and P18 (complementing the Td) was between -4.3 and -
4.6 kcal/mole.

When compared with Tn (T47, T60, T80, and T106), the sequence of Bd had 0,
13, 33, and 59 bases matching within the dangling end oan, and a Gibbs free energy of-
2.3 to -74.9 kcal/mol. The Gibbs free energy between Bd and P18 (complementing the

Tn) was between -3.2 and -5.3 kcal/mol.

69

The second defined background DNA sequence, BS, was designed to complement
the surface proximal tail of 11 different Ts (as shown in Table 1 and depicted in Figure
3.33) with various degrees of sequence similarity. The surface proximal tail of Ts had an
8 to 16 base overhang (duplex dependent) with 0% sequence homology to BS followed by
a stretch of 59 bases with variable regions of contiguous perfect matches and sequence
similarity with BS. This resulted in a sequence similarity of 0 to 90% on the whole
surface proximal tail and 10-100% sequence similarity considering only the 59 bases on
the 5’ endofTs. ThedanglingendofTshad 1 to 16basesdependingontheprobeitwas
hybridized. For all interactions between BS and Ts, the Gibbs free energy was always less
negative than -5.7 kcal/mol. The Gibbs fi'ee energy between BS and P18 (complementing

the Ts) was between -3.8 and -8.1 kcal/mol.

3.3.3 Probe design and microarray synthesis. Probes were designed using a
Perl script developed in house that screened for all 18-mers of 168 rRNA genes that had
at least two mismatches to every other sequence in the RDP-II (http://rdp.cme.msu.edu/).
Screened probes for a given pathogen were ranked based on melting temperature (Tm)
and GC content; following the parameters incorporated into OligoArray 2.0 probe design
software (32). The final array included a set 146 probes (P18) targeting 21 different 16S
rRNA gene sequences and mismatch probes; identical to the perfect match sequence
except for l and 2 incorrect bases in the middle of the oligomer to measure the degree of
cross hybridimtion. Microarrays containing the above set of probes were synthesized by

Xeotron Corporation, Houston, TX (now part of Invitrogen, Carlsbad, CA) using a

70

proprietary in situ synthesis technology developed by the University of Michigan (11,

12).

3.3.4 Experimental approach. A). Comparison of 45-mer synthetic target
mixture vs. Klenow/random primer labeled target mixture. Two-sarnple comparative
hybridization, similar to that used in gene expression studies (5), was performed to
evaluate the differences and similarities between synthetic targets and Klenow/random
primer labeled targets. A similar experiment was conducted to study the impact of
complex background 16S rRNA gene mixture on target signal intensity. A total of three
comparative hybridizations were performed in triplicate (shown by double-arrows in
Figure 3.1a). These were: i) T45-m vs. Tk-m - to compare the hybridization behavior of
two types of target mixtures, ii) T45-m vs. B16s - to evaluate the impact of background
DNA on 45-mer synthetic target mixture and iii) Tk-m vs. B165 - to evaluate the impact
of background DNA on Klenow/random primer labeled target mixture The complex
background mixture B163, Bd, and Bs was also hybridized alone to ensure that target
signals for the probes to be tested did not cross hybridize with various backgrounds. For
hybridization (i), 20 pmol of T45-m (1,069 ng DNA) was mixed with 100 pmol (2,481 ng
DNA) of Tk-m. For hybridization (ii), 20 pmol (1,069 ng of DNA) of Cy3-end-labeled
T45-m was mixed with 200 pmol (6,550 ng DNA) of Cy5-labeled B16s. For
hybridization (iii), 100 pmol of Cy5-labeled Tk-m was mixed with 200 pmol (6,550 ng
DNA) of Cy3-labeled B163. The last mixture was obtained by combining 10 pmol of

labeled 16S rRNA gene hour each of the 10 pathogens.

71

B. Evaluation of length and sequence of dangling end of synthetic targets. The
effect of dangling end length was studied by varying the length of continuous sequence
homology between Td and Bd fiom 4 to 12 bases as shown in Figm'e 3.2a. The dangling
end had the sequence 5’-(86 bases)-CGACTI‘GCATGTGTTG—3’, wherein bold type
nucleotides complement the sequence of Bd. In order to generate the increasing dangling
end overhang of Td fiom 8 to 16 bases, nine separate probes with increasing offsets
relative to the target. The sequence of background DNA (Bd) included the
complementary sequence for the 12 bases of dangling end to result in target-background
interaction, which was expected to increase with an increase in dangling end length. The
hybridization included 67 pmol of Cy3 labeled Td (3,557 ng) and 37 pmol of Cy5 labeled
Bd (1,540 ng).

The dangling end interaction due to Gibbs fiee energy was examined firrther using
four synthetic targets with 100% sequence similarity and varying length (T 47, T60, T80,
and T106) along with a 59-mer Bd that was designed to complement the dangling end of
Tn by 0, 13, 33, and 59 bp (having Gibb’s fiee energy of -8.8, -14.8, -43.8, -74.9
kcal/mol, respectively). The Ed for this examination was the same as before; however
the Tn targeted 10 different probes than Td. The hybridization included 52 to 84 pmol of
Cy3 labeled Tn (1522 to 4277 ng) and 37 pmol of Cy5 labeled Bd (1,540 ng).

C. Evaluation of sequence similarity in surface proximal tail of a synthetic target.
The dependence of sequence similarity at the surface proximal tail of a synthetic target
with background DNA was studied by synthesizing 11 different synthetic targets of
similar length Ts (Figure 3.3a) and a new synthetic background DNA sequence (BS). The

Gibbs free energy of Ts sequences varied between -9.2 and -76.8 kcal/mol with sequence

72

Similarity varying between 10 to 100% (Figure 3.3a). Each Ts was end-labeled with Cy3
at the 5’ end and complemented 3 to 8 probes. The hybridization included a total of 160

pmol of Cy3 labeled Ts (7,895 ng) and 30 pmol of Cy5 labeled Bs(1,115 ng).

3.3.5 Hybridization and scanning. Target and background mixtures were
prepared in 100 pl of hybridization solution containing 35% formamide, 0.4% Triton X-
100, and 6X SSPE. SSPE buffers were made from a stock of 18X SSPE, which is 2.7 M

NaCl, 130 mM Na2P04, 18 mM NazEDTA (adjusted to 6.6 pH with HCl). The

hybridization solution was heated at 95 °C for 3 min, cooled on ice for 1 min, and passed
through 0.22 pm filter. All hybridizations were carried out in triplicate for 16 to 18 hr at
20 °C using 3 M-2 microfluidic station (Xeotron Corporation). A flow rate of 500 pl per
min was used for recirculation of hybridization solution through the microfluidic array
during hybridization (42). After overnight hybridization at 20 0C, the microarray was
washed using wash buffer 2 (6X SSPE, 0.2% Triton X-lOO), wash buffer 4 (1X SSPE,
0.2% Triton X-100), and wash buffer 2 with no Triton X-100 in series for 2.2 minutes
each (500 pl per min, 20 °C). A non-equilibrium thermal dissociation approach adapted
to the Xeotron platform was used in all hybridization experiments (3, 20, '39, 42). The
protocol was based on earlier studies utilizing a dissociation curve approach for
diagnostic arrays (8, 20, 39, 42). The microarrays were washed with a high stringency
wash buffer (10 mM Na2HP04, 5 mM EDTA, pH 6.6; flow rate 500 pl per min) for 2.2
min at increasing temperatures from 20 to 70 °C at 2 °C intervals. Experimental
conditions have been examined previously for optimal specificity (42). Signal intensities

were quantified after each wash using a GenePix 4000B non-confocal laser scanner

73

(Axon Instruments, Inc, Foster City, CA) at a photomultiplier tube voltage setting of 650
V for Cy5 (635 nm laser) and 600 V for Cy3 (532 nm laser). For complex background
DNA (8165), this setting was 750 V (for Cy5) and 500 V (for Cy3). With this setting,
photo-bleaching caused an average decrease in signal intensity of 0.85% for Cy5-labeled

targets and 0.54% for Cy3-labeled targets between each of the 20 scans at 20 oC.

3.3.6 Data analysis. Data was analyzed using an XL script that imported raw
GenePix dissociation curve data between 20 and 70 °C into Excel and generated a
sigmoid curve for each microarray feature. Signal to noise ratio (S/N) at a given
temperature was calculated from three replicate hybridization experiments. Noise was
defined as the average signal of twenty empty spots (containing linker chemistry and no
probes). The average signal of 20 randomly selected non-target probes (80.3 i 10.6
arbitrary units with 635 nm, 148.3 1 16.7 a.u. with 532 nm laser) was similar to average
signal of 20 empty spots (83.6 i 6.4 an. with 635 nm, 155.9 3: 10.6 a.u. with 532 nm
laser) when microarray was hybridized with complex background. Compared to no
background, the average noise increased slightly when a complex background DNA
(B165) was hybridized (1.19-fold for Cy3 and 1.54-fold for Cy5). For all experiments, an
average S/N greater than 3.0 was considered positive to balance the maximum percentage
of true positive calls of targeted probes and true negative calls of non-targeted probes.
Figure 3.1c is presented as S/N for the area under the dissociation curve to include
variation in both dissociation and signal intensity. For statistical compalison of probe
signal intensity when hybridized with Klenow/random primer fragmented and 45-mer

targets, a two-tailed inference about differences in population mean for independent

74

samples was performed with 95% confidence interval. The free energy of hybridization
was computed using the two-state hybridization server (www.bioinfo.rpi.edul) developed
by Dimitrov and Zuker (7) with 43 °C instead of the actual 20 °C. This is because the
hybridization buffer contained 35% formamide, which is expected to destabilize duplexes

equivalent to increasing the temperature by 21 to 25 °C (2, 39).

3.4 RESULTS AND DISCUSSION

3.4.1 Klenow/random primer labeled target mixture vs. end-labeled 45-mer
synthetic target mixture. The hybridization behavior of Klenow/random primer
labeled target mixture compared to the 45-mer synthesized target mixture, plotted as S/N
of T45-m and Tk-m is shown in Figure 3.1b. Tested probes in Figure 3.1b were sorted
based on the difference in S/N between T45~m and Tk-m. Considering a S/N greater than
3 as positive signal, 87 out of 95 tested probes yielded positive Signal for T45-m while 91
probes gave positive Signal for Tk-m. Approximately 5.4% of non-targeted probes gave
false positive signal with T45-m, and 10.9% gave false positive signal with Tk-m. A
higher amount of non-specific hybridization with T'k-m compared to Single stranded
DNA targets has been reported previously (10). Most of the probes lmving Similar S/N
withboththetargetsgenerallyhadaS/Noflessthan10(probes21to46). TheS/Nratio
for 73 out of the 95 probes differed Significantly between the two target mixtures (95%
confident interval considering mean difference). There was no systematic pattern when
signal intensities of all probes for a single 16S rRNA gene were compared for the two
targets. Often the difference in S/N was as large as 33-fold (e.g., for probe numbers 80 to

95).

75

While a difference in signal intensity with various target mixtures is expected,
higher signal with single stranded targets compared to Klenow/random primer labeled
targets has not been reported previously. More specifically, probes 1 through 30 had
much higher S/N with T45-m tlmn with Tk-m, while the opposite was observed for
probes 60 to 95. For 32 out of 95 probes, the S/N was Significantly higher with T45-m
than with Tk-m (P 5 0.05). Dissimilar results were observed in a study by Franke-
Whittle et al. (10), in which S/N was higher for a majority of duplexes targeted with
Klenow fragmented targets compared to single stranded targets. Higher S/N with
Klenow fragmented targets is expected because longer dangling ends in T'k-m provide
opportunity for attachment of many more Cy5 dye molecules compared to only one Cy3
dye in T45-m. The average specific activity for T'k-m was 72 :t 20 nucleotides per Cy5
dye (calculated using arithmetic average of specific activity for each of 10 targets that
were labeled separately). This implies that the average number of Cy5 dye was between 2
and 30 (assuming a target length between 200 and 1,500).

It is evident from Figure 3.1b that normalization of S/N for the higher number
of dyes in Tk-m or dye bias would not result in a profile that overlaps completely with the
S/N profile of T45-m. This implies that factors in addition to differences in label
abundance and dye bias are responsible for the differences in S/N. These factors include:
i) secondary structure of longer dangling ends of Tk-m making certain regions
inaccessible for hybridization with probes (4, 19, 28), ii) Tk-m and T45-m displacement
from probes due to complementing strands of double stranded Tk-m (25) known as a
zipper effect (30), iii) bias in Klenow-labeling due to higher GC content (38), iv) probe

sequence variability and linear probes influences on thermodynamic stability (31), v)

76

physical quenching of Cy dyes (6), and vi) conserved regions on overhang of hybridized
Tk-m interacting with complementing strands from 1 or more of the 10 organisms also
targeted in the mixture (23) creating chains of targets. Interactions occurring in solution
between 163 rRN A gene targets fiom various organisms may influence the spot Signal
intensity by competing with the targets bound to the probe.

Further experiments were performed to examine and quantify the extent of target-
background interaction on the two dissimilar targets by spiking both Tk-m and T45-m in
background DNA. When the two target mixtures were spiked (separately) into a complex
mixture of non-target 16S rRNA genes (B 16s), the S/N of Tk-m changed significantly
compared to the S/N of T45-m (Figure 3.1c). The X-axis in Figure 3.1c represents the
factor by which the S/N ratio of the target mixture (T45-m or Tk-m) is influenced by the
background, B16s, and Y-axis is frequency of probes with a given influence factor. For a
given probe, influence factor was computed as S/N in absence of Bl6s divided by the
S/N in the presence of B163. For targets not impacted by the background DNA, this
influence factor should be close to unity (implying no efl‘ect) and the spread of the bell-
shaped curve Should be minimal. For T45-m, the influence factor was close to 1 while
for Tk-m, it was approximately 4. Similarly, the Spread of T45-m influence curve was
much smaller compared to that obtained for Tk-m. A simple comparison of the areas
under each influence factor curve indicated that the interaction of T'k-m with the
background was 3.7 times higher compared to T45-m (obtained by comparing the areas
under the curve of T'k-m and T45-m which was 69.5 and 19 arbitrary units, respectively).
A Wilcoxon signed-rank test was used to statistically conclude that distribution of

influence factor was greater for Tk-m than T45-m with 95% confidence interval.

77

Therefore, background significantly influences signal intensity of Tk-m and not T45-m.
Since the same probes were targeted with both target mixtures, the influence of
background on signal intensity is dependent on the characteristics of the target molecules,
and independent on the spatial region on the array. The B16s was also hybridized alone
to ensure that targeted probes did not produce any signal with the background 168 rRNA
gene mixture.

The influence of background on signal intensity of double stranded Tk-m may
be the result of target interactions influenced by complementary and non-target strands in
background DNA. In one study, greater hybridization efficiency was observed with
single-stranded targets compared to double-stranded targets (1 3). This is due to
competition between complementary strands and probes with double-stranded DNA.
Related studies have reported increased competition between the complementing strand
and probe as the length ofthe dangling end increased, while increasing the length ofthe
surface proximal tail did not influence hybridization (25). Interactions between the
conserved regions of target dangling ends and sequences from a second target have also
been observed (23). These interactions were minimized by using single stranded RNA
instead of double stranded DNA. If this observation can be extended to DNA sequences,
the use of asymmetric PCR for target preparation may be advantageous in minimizing
target-background interaction.

Target-background interaction on the dangling ends of the target may also
influence Signal intensity by affecting the secondary structure. As studied by Chandler et
al. (4), signal intensity was substantially increased as target secondary structure was

relieved using a second probe in solution that annealed to target dangling end.

78

Interaction between the target dangling end and a background DNA sequence may have a
similar effect. Peplies et al. (24) observed similar results for a majority of tested probes.
However, the signal of some probes decreased with a second probe in solution,
suggesting that the opening of a selected binding site may lead to a reorganization of one
or more secondary structures in other target regions.

There are additional motives for using shorter targets with diagnostic arrays.
Shorter targets reduce intermolecular structures that occur more fiequently with single
stranded targets compared to stiffer double stranded products (36). Lane et al. (15) used
amplicons of various lengths and concluded that shorter fragments reduced false
negatives due to high intermolecular secondary structure in longer targets. F luorophore
position on a target DNA sequence can also influence signal intensity (44). Signal
intensity decreases as the proximity of the fluorescent molecule from a probe target
duplex increases. Thus, target preparation strategies should include amplifying shorter
targets with specific asymmetric PCR, or fiagrnent longer amplicons to produce a range
of desired length.

Wilson et al. (43) suggested using fiagrnents under 100 bp compared to full-
length amplicons to obtain a stronger hybridization signal. A number of fragmentation
strategies have been described for obtaining a Size range between 35 and 200 bases with
pre- and post-labeled targets and asymmetric PCR amplified products (1, 21, 27, 41).
Subsequent studies comparing cost, efficiency, and influence on Signal intensity for some
of these methods are currently underway in our laboratory. Target-target and target-
background interactions are expected to be less for targets other than 168 rRNA genes,

especially if the targets have less conserved regions.

79

3.4.2 Synthetic targets of increasing dangling end length. The effect of
dangling end length was evaluated using a target with an increasing stretch of continuous
homology. A Cy3 end-labeled synthetic 102-met target (Td) was designed to hybridize
to 9 different probes on the array. After hybridization, Td was expected to have a
dangling end of 8 to 16 bases (Figlne 3.2a). A 59-mer synthetic background DNA
sequence (Bd) end-labeled with Cy5 was designed to complement the dangling end
sequence from position 86 to 98 with 4 to 12 base stretches of continuous homology.
The signal to noise ratio odeand Td at 26 °C is Shown for each ofthe 9 separate probes
(Figure 3.2b). In order to generate the increasing dangling end overhang on the Td,
probes were offset by one nucleotide relative to the target. Considering a S/N greater
than 3 as positive signal, Bd hybridized to the dangling end of Td only when the dangling
end stretch of continuous homology was 7 bases or longer. Signal from Bd was not
observed when hybridized alone (i.e., without Td), and mismatch probes did not display
signal for Bd or Td; directly implicating that Ed did not cross hybridize with the probe.
AS evident by the lack of Signal for probes with target dangling end of 6 bases or less,
interaction of surface proximal end of Td with Bd was insignificant. The S/N of Bd for
11 and 12 baseoverhangswashigherthanthe S/N givenby Td. Thismaybedueto dye
bias between Cy5 and Cy3 and higher specific activity of Cy5-labeled Bd. The 59-mer
Bd may have resulted in a greater percentage of dye labeled product compared to the
102-mer Td, as purity of oligonucleotide synthesis decreases with increasing length.

The dangling end interaction due to length of overhang was reexamined using

four synthetic targets of varying lengths (T47, T60, T80, and T106 together represented

80

by Tn) along with the same 59-mer Bd that was designed to complement the dangling
end of Tn by 0, 13, 33, and 59 bp. Dangling end interaction occurred between T60, T80,
T106, and Bd but not between T47 and Bd (data not Shown). The zero sequence
homology between T47 and Bd was the obvious reason for this lack of interaction.

Gibbs free energy calculations incorporate influences such as length, sequence
similarity, secondary structures and hairpin loops caused by near perfect match sequences
with insertions or deletions (16). It was used to examine interaction between the Td, Tn
and Bd. For Td, an overhang with 4 sequential perfect match bases had a Gibbs free
energy of -3.2 kcal/mol, while an overhang with 12 sequential perfect match bases had a
Gibbs fiee energy of -14.9 kcal/mol. The S/N of Bd became greater than 3 when the
stretch of continuous bases on the dangling end was 7 or higher and Gibbs free energy
was -7.6 kcal/mol or more negative. With the experimental conditions used in this study,
target-background interaction on dangling end can occur with Gibbs free energy that is
more negative than -7.6 kcal/mol. However, due to phenomenon such as difference in
intermolecular structure and competitive influence of complementary strands (28), the
interaction observed with this fiee energy may not be extensible to double-stranded DNA
targets. For calculating the Gibbs free energy between Bd and Td, only the 3’ end of Td
was considered to circumvent the influence of probes on dangling end interaction.
Neglecting the presence of the probe, the Gibbs fiee energy of the duplex formed
between Td and Bd was -15.4 kcal/mol. Because the 5’ end of Bd preferentially
hybridizes to the 3’ dangling end of Td, the interaction occurs solely on the dangling end

rather than the surface proximal end of Td or the probe.

81

The S/N of Bd increased as Gibbs fi'ee energy became more negative and length
of continuous homology on Td dangling end increased. Gibbs free energy between Bd
and T47, T60, T80, and T106 was -2.3, -14.2, -43.8, -74.9 kcal/mol, respectively.
Previous studies have also reported successful hybridization to 9-mer probes, with a
Gibbs free energy of -6 to -10 kcal/mol producing weak signals and -8 to —12 kcal/mol
producing strong Signals (29). It should be noted that binding fiee energy has been
shown to be a function of the surface material, surface charge density, length of linker

molecule, and other experimental conditions (40).

3.4.3 Synthetic targets with surface proximal tail sequence complementary to
background DNA. The effect of sequence similarity of the surface proximal tail of
target to sequence of defined background DNA was studied using 11 different targets
(TS) of similar length (89 to 102 bases). All the targets had 18 bases complementary to
52 probes on the microarray (Figure 3.3a), a dangling end of l to 16 bases and surface
proximal tail of 67 to 76 bases depending on the probe and target. Targets were designed
to have 10 to 100% sequence similarity with Bs. Probes were not offset (as with
dangling end experiments); however 11 targets were used resulting in a Gibbs free energy
between -9.2 to -76.8 kcal/mol for Ts<:> BS duplex (Figure 3.3b). The Gibbs free energy
for P184: Ts varied between -17.5 and -24.1 kcal/mol. Continuous stretches of 38 and
45 perfect matches at the surface proximal tail of TS (with free energy of -64.5 to -65.9
kcal/mole) displayed target-background interaction (Figure 3.3c), while continuous

stretches of 19 or less perfectly matching bases (with free energy of 42.] kcal/mol or

82

more positive) did not result in signals attributable to BS. Targeted probes did not display
signal when Bs was hybridized alone (i.e. without Ts).

These results suggest that target-background interaction at the surface proximal
tail is different compared to a Similar interaction at the dangling end. While a Gibbs flee
energy of -7.6 kcal/mol (stretch of 7 continuous perfect matches) was sufficient for
dangling end interaction, 3 Gibbs flee energy more positive than -64.5 kcal/mol
(including a stretch of 19 continuous perfect matches) on surface proximal tail did not
display interaction. Further examination of the Ts with a Gibbs flee energy of -36.8
kcal/mol when hybridized with BS (19 base stretch of continuous homology) showed no
secondary structure on surface proximal tail. Thus external influences that are not
incorporated into Gibbs flee energy calculations may be causing greater destabilization
on surface proximal end. A study examining the influence of target overhang length on
hybridization efficiency showed that the dangling end of a target interacted with its
complementary strand more than the surface proximal tail (25). It was suggested that this
is because the dangling end is exposed to the liquid phase providing better access to
complementary strands in solution. On the other hand, the surface proximal tail is closer
to the surface, thus limiting availability to other strands in solution and reducing
association with complementary strands. It is obvious that the difference in target-
background interaction on surface proximal tail and dangling end requires further

investigation.

83

3.5 CONCLUSIONS

The length and sequence of surface proximal tail and dangling end of targets and
resulting target-background interaction influences signal intensity and decreases
Specificity of oligonucleotide probe-target hybridizations. The dangling end was found to
be significantly more prone to target-background interaction compared to surface
proximal tail. Gibbs free energy between the target and background was found to be a
better indicator of hybridization Signal intensity than the sequence or length of the
dangling end alone. This study underlines the need for carefirl target preparation
(fragmentation and asymmetric PCR) and evaluation of Signal intensities for diagnostic
arrays using 16S rRNA and other gene targets due to the potential for target interaction

with a complex background.

3.6 ACKNOWLEDGMENTS

We are grateful for the help fl'om Jimmy Johnson and Weihong Qi with scripts for
sequence analysis, and Alok Dhawan for critically reading this manuscript.

This work was supported by the National Institutes of Health (Grant R01
RR018625-01) and Michigan Economic Development Corporation (GR-476 PO

O85P3000517).

84

Table 3.1 Identity of pathogens, number of probes and range of Gibbs flee energy for the

 

 

set of probes.
Gibbs Free
Pathogen Name 1:13:30. No. of Probes 5:15;? range
fical/mol)
Burkholderia pseudomallei 1026b 2 -l9.6
Campylobacter jej uni 700819 1 -21 .1
Clostridium perfringens 13124 5 -2l .3 to -26.7
Enterococcusfaecalis 19433; B/66 15 -17.3 to -24.1
Escherichia coli 11775; M25588 15 -20.8 to -24.9
Helicobacter pylori 43504; 49396T 16 -l 7.9 to -20.6
Legionella pneumophila 33215 12 -18.1 to -21
10145; TERI
Pseudomonas aeruginosa 2014; 33 -18.6 to -23.
DSM50071
Salmonella enterica 1331 l; 76; 12-1 18 -20 to -23.4
Salmonella lyphimurium 13314 5 -18.2 to -19.2
Salmonella enleritidi.s' SE22 3 -22.1 to -22.8
Vibrio parahaemolyticus 13; 17 11 -l 7.4 to -23.4
Yersinia enterocolitica 9610 10 -15.9 to -20.2

 

85

Table 3.2 Characteristics of various DNA sequences used to study the effect of dangling

end and surface proximal tail on signal intensity.

 

A

45-mer mixture and probe sequence (in bold) for one of

95 targets

TTACGACTT CACCCCAATC GCTGACCCT
ACTGCGGCCG CTGCCTC

l8-mer probe

AGTAGGGTC AGCGATTGG

B

102-met target (Td) and probe sequence 1-9 (in bold)

GGACGTTAT CCCCCACTAC CAGGCAGATT
CCTAGGCATT ACTCACCCGT

CCGCCGCTGA ATCCAAGGCA AGCTCCCTT C

ATAAGCTCGA CTTGCATGTG TTG

59-mer background (Bd)

AGAGTTTGA TCCTGGCTCA GAGTGAACGC

TGGCGGCGTG CCTAATACAT GCAAGTCGAA

86

Free energy of
duplex
(kcal/mol)

-15.9 to -26.7
between T45-m
and P18

-3.2 to -14.9
between dangling
end of Td and Bd

-4.3 to -5.4 between
Ed and P18

-18.6 to -21.4
between Td
and P18

Table 3.2 Continued
C

Ts and probe sequence (in bold) for one of 11 targets

CCCGI I ICC GGACGTATCC CCCACTACCA
GGCAGATTC CTAGGCATTA

CTCACGTCCG CCGCTGAANC CAGGAGCAAC
GTCCGTCATC CGCTCGACT

59-mer background (BS)

TCAGCGGCG GACGGGTGAG TAATGCCTAG

GAATCTGCCT GGTAGTGGGG GATAACGTCC

-9.2 to -76.8
between surface
proximal tail of Ts
and BS

-3.6 to -8.1 between
BS and P18

-17.4 to -24.9
between Ts
and P18

 

87

3.7 Figures

Figure 3.1 Klenow/random primer labeled target mixture vs. end-labeled 45-
mer synthetic target mixture. a) Experimental strategy. b) Signal to noise ratio
of synthetic targets T45-m (A) and Klenow/random primer labeled targets
Tk-m (I) at 26 °C. The average Signal for three replicate experiments is
plotted and the error bars are standard deviation. c) Ratio of signal from
hybridizations with target alone over targets spiked in background DNA for
T45-m (A) and Tk-m (I). The X-axis represents the factor by which the S/N
ratio of the target mixture (T45-m or Tk-m) is influenced by the background,
B163, and Y-axis is frequency of probes with a given influence factor. For a
given probe, influence factor was computed as S/N ratio in absence of Bl6s
divided by the S/N ratio in the presence of 8165. Signal in Figure 3.1c is S/N
of area under the dissociation curve to include variation in both initial Signal
intensity and dissociation.

88

  

Fl”

 

Klenow-labeled 16S rRNA gene
target mixture Of200-1500 bp, Tk-m :

o. “

I'o . :::
"Vi e <—) :
/'\—- .....IIIII.I \llu-III-IIII" .
5’ \ i

Cy5
45-mer synthetic target Complex backgroundC16S rRNA gene
mixture, T45-m mixture of 200-1500 bp, 3163

I.
.II .

   
   

  

 

 

! 14 bases. ....lIIII-nllllllI'..'.-.-I.. i.‘.iifi:¥§;f*
'“ 5’ 13 b '
Linker 18-mer probe ases Cy5
(b) 1000 «-

é

T45-m labeled with Cy3-W1

W if a. iiiiiii iijfifllfi

iii

  
 

S/N ratio at 26 °C

Tk-m labeled with Cy5 g l i iiéi

 

 

0 20 40 00 80 95

Probe number — sorted by difference in Signal intensity between Tk-m and T45-m

 

 

89

Figure 3.1 continued

 

m
K
T

 

 

  

 

 

 

I

m

e4.

T ‘- IIIIIIIII'

A Ila-natal...- :4
IQIIIIIIIIIIII IIIII

P h b b p
1 q q q a
5 0 5 0 5
2 2 1 1

08333.. no .3532

Influence Factor

90

Figure 3.2 Synthetic targets of increasing dangling end length. a)
Experimental strategy. b) The signal to noise ratio of Bd and Td at 26 °C (Y-
axis) is shown for each of 9 separate probes offset by one nucleotide relative
to the target, in order to generate the increasing dangling end overhang of Td
(X-axis). Clear bars represent S/N of Td and dark bars are S/N of Bd. Signal
of Bd is observed when target dangling end overhang has a stretch of
continuous homology greater than 7 bases. The average signal for three
replicate experiments is plotted and the error bars are standard deviation.

91

4 to 12 bases long dangling end (5’-(86
bases)—CGACTTGCATGTGTTG—3 ’)
of the 102-mer synthetic target to
interact with BD

(3)

102-mer individual synthetic target, TD

3".

9
‘2’ <’O> Cy3

 

e CNS
': 68to76 bases long ‘
.' surface proximal tail 3(—
'......IIII. 3’ I."
.IIIIIIIIIIIIIIIH'HMMHI no .3

   
 

 

 

 

 

5’:, ............. .
Linker 18-me' Pmbe ' 59—mer single
. .............. stranded DNA
3” used as defined
background. 80
Gibbs Free T overhan
Energy with BD D g
kcal/mol
-32 LEGTTG'?
-4_3 QAIGTGTI'G-3'
-6.9 GCATGTGTTG-3’
-7.6 LQCATQIGTTG-C”
-8.3 TTGCATGTGTTG-3’
-9.8 CTTGCATGTGTTG—3’
-1 1.1 ACTTGCATGTGTTG-3’
-12.0 GACTTGCATGTGTTG-3’
-149 CGACTTGCATGTGTI'G-3'

 

92

Figure 3.2 continued

SIN ratio at 26 °C

400!

350'

300-

2501

2001

150'-

1001

(b)

l- TD signal

”L + i +

 

 

 

 

 

FiT

 

 

 

 

 

l‘__

 

 

l I
U U V

4 5 6 7

Number of bases in the dangline end of Td with perfect match to Bd

+-_

 

BD signal

 

10

[Gibbs Free Energy (kcal/mol)]

93

0......“

 

11
[-3.2] [-4.3] [-6.9] [-7.6] [-8.3] [-9.8] [-11.1] [-12] [-14.9]

 

12

 

.r

Figure 3.3 Synthetic targets with surface proximal tail sequence complementary
to background DNA. a) Experimental strategy. b) Mismatches, insertions, and
deletions influencing Gibbs flee energy between the Bs and 11 different Ts,
resulting in a Gibbs flee energy for Ts a BS duplex formation that ranged
between -9.2 to -76.8 kcal/mol. c) S/N at 26 °C of TS and Bs (Y-axis) versus
Gibbs flee energy (kcal/mole) between Ts and Bs (X-axis). Clear bars represent
S/N of Ts and dark bars are S/N of BS. Only 7 out of 11 results are displayed
because the BS did not interact with other Ts, The average signal for three
replicate experiments is plotted and the error bars are standard deviation.

94

I " '3- ----------- 89-102-mer synthetic targets, Ts

an... (a)

I
. ........ . ,
1.3;», 3 ............ 59—mer single stranded DNA used as
1":5' defined background, BS
:9 5: 2:4
'mwifiiflifltzffinfirflnyuuv” '\ Cy5
O

   

Linker 18-mer probe

TS Number Gibbs free energy with BS
kcal/mol

- 76.8
- 64.9
- 64.5
- 42.1
- 38.9
- 36.8
- 28.5
- 13.5
- 12.5
- 11.1
- 9.2

AOCDCDVODUI-th—x

A—‘L

95

Figure 3.3 continued

.m- Coe- mmV-Iiul-fiumu-u'efllh-ugl-Efiiulnlun- 29.7 :V- .m
.m- 265- 03 #33887-EuuueubeFF-uu—Euu-Eufiuglfiue.-quub¢—ul- Coe- mHV- .m
.m- noos- HeV- unau-uueuu--kuuuu<uhu<kpeunu<uuukknu <uuu<u-<-u--uuuu

-.m

 

.m- Coe- HNV- usS-uuuuu--5u-u<ub<t<u-u<-uuthu- <uuu<u—<-u--uuuuuh<tiu- Cos- 3V-
.m- Tos- mmV- nab-Ugo:-Guuu<u..u<t<u—u<—uut.—u- <uuu<u—<-u--uuuuuk<te-ms-.m
.m- Cos- NMV -<ubuuuuuu--5 a... ub<t<uuu<5ut<u- -<uuu<uu<5<-8882-538 C27 8-
.m- nnms- oeV -<u»uuuuuuu--kuuuu<uku<hh<uuu<huupk<u- <uuu<uu<ku<uunuuuupuhhuunan-.m
.m- mews- va -<ukuuuuuuu--huuuu<uhu<hk<uuu<huukk<u- <uuu<uu<hu<-uuuuuk<hkuu<ue-.m

.mlkU<UuquUU

<UUUUFU<UF<<PUUUF<UU<<FU FUUUFUUF<UP OUUUO<F<<UUPUU1.M mm

§

09.-
mdw-
mem-
adm-
Pd?-
m. we.
m. Vo-
w. mu-
m...

96

Figure 3.3 continued

1000 -—

S/N ratio at 26 00

d
O
I
l

moniiiiii iii}

Ts SW (C) BS S/N

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[-28.5] [-36.8] [-38.9] [-42.1] [-64.5] [-64.9] [-76.8]
Gibbs Free Energy (kcallmol) between T8 and Bs

97

3.8 REFERENCES

10.

11.

Bekal, S., R. Brousseau, L. Masson, G. Prefontaine, J. Fairbrother, and J.
Harel. 2003. Rapid identification Of Escherichia coli pathotypes by virulence
gene detection with DNA microarrays J. Clin. Microbiol. 41:2113—2125.

Blake, R. D., and S. G. Delcourt. 1996. Thermodynamic effects of formarnide
on DNA stability. Nucleic Acids Res. 24:2095-2103.

Bodrossy, L., N. Stralis-Pavese, J. C. Murrell, S. Radajewski, A. Weilharter,
and A. Sessitsclr. 2003. Development and validation Of a diagnostic microbial
microarray for methanotrophs. Environ. Microbiol. 5:566—5 82.

Chandler, D. P., G. J. Newton, J. A. Small, and D. S. Daly. 2003. Sequence
versus structlue for the direct detection of 168 rRNA on planar oligonucleotide
microarrays. Appl. Environ. Microbiol. 69:2950-2958.

Churchill, G. A. 2002. Fundamentals of experimental design for cDNA
microarrays. Nat. Genet. 32:490-495.

Cox, W. G., M. P. Beaudet, J. Y. Agnew, and J. L. Ruth. 2004. Possible
sources of dye-related signal correlation bias in two-color DNA microarray
assays. Anal. Biochem. 331:243-254.

Dimitrov, R., and M. Zuker. 2004. Prediction of hybridization and melting for
double-stranded nucleic acids. Biophys. J. 87 :215-216.

El Fantroussi, S., H. Urakawa, A. E. Bernhard, J. J. Kelly, P. A. Noble, H.
Smidt, G. M. Yershov, and D. A. Stahl. 2003. Direct profiling Of environmental
microbial populations by thermal dissociation analysis of native rRNAS
hybridized to oligonucleotide microarrays. Appl. Environ. Microbiol. 69:23 77-
2382.

Franke-Whittle, I. H., S. H. Klammer, and H. Insam. 2005. Design and
application of an oligonucleotide microarray for the investigation Of compost
microbial communities. J. Microbiol. Methods 62:37-56.

Franke-Whittle, I. H., S. H. Klammer, S. Mayrhofer, and H. Insam. 2006.
Compaison of different labeling methods for the production of labeled target
DNA for microarray hybridization. J. Microbiol. Methods 65:117-126.

Gao, X. L., E. Gulari, and X. C. Zhou. 2004. In situ synthesis of
oligonucleotide microarrays. Biopolymers 73:579-596.

98

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

Gao, X. L., E. LeProust, H. Zhang, 0. Srivannavit, E. Gulari, P. L. Yu, C.
Nishiguchi, Q. Xiang, and X. C. Zhou. 2001. A flexible light-directed DNA

chip synthesis gated by deprotection using solution photogenerated acids. Nucleic
Acids Res. 29:4744-4750.

Guo, Z., R. A. Guilfoyle, A. J. Thiel, R. Wang, and L. M. Smith. 1994. Direct
fluorescence analysis of genetic polymorphisms by hybridization with
oligonucleotide arrays on glass supports. Nucleic Acids Res. 1 1:5456—5465. .

Hashsham, S. A., L. M. Wick, J. M. Rouillard, E. Gulari, and J. M. Tiedje.
2004. Potential of DNA microarrays for developing parallel detection tools

(PDTS) for microorganisms relevant to biodefense and related research needs.
Biosens. Bioelectron. 20:668-683.

Kane, M. D., T. A. Jatkoe, C. R. Stumpf, J. Lu, J. D. Thomas, and S. J.
Madore. 2000. Assessment of the sensitivity and specificity of oligonucleotide
(50mer) microarrays. Nucleic Acids Res. 28:4552-4557.

Koehler, R. T., and N. Peyret. 2005. Thermodynamic properties Of DNA
sequences: characteristic values for the human genome. Bioinformatics 21:3333-
3339.

Koizumi, Y., J. J. Kelly, T. Nakagawa, H. Urakawa, S. El-Fantroussi, S. A]-
Muzaini, M. Fukui, Y. Uruslrigawa, and D. A. Stahl. 2002. Parallel
characterization of anaerobic toluene- and ethylbenzene-degrading microbial
consortia by PCR-denaturing gradient gel electrophoresis, RNA-DNA membrane
hybridization, and DNA microarray technology. Appl. Environ. Microbiol.
68:3215-3225.

Lane, D. J. 1991 . 16S/23S rRNA sequencing, p. 115—148. In E. Stackebrandt and
M. Goodfellow (ed.), Nucleic Acid Techniques in Bacterial Systematics. John
Wiley, New York.

Lane, 8., J. Evermann, F. Loge, and D. R. Call. 2004. Amplicon secondary
structure prevents target hybridization to oligonucleotide microarrays. Biosens.
Bioelectron. 20:728-735.

Li, E. S. Y., J. K. K. Ng, J. H. Wu, and W. T. Liu. 2004. Evaluating single-
base-pair discriminating capability of planar oligonucleotide microchips using a
non-equilibrium dissociation approach. Environ. Microbiol. 6:1197-1202.

Longo, M. C., M. S. Berninger, and J. L. Hartley. 1990. Use of uracil DNA
glycosylase to control carry-over contamination in polymerase chain reactions.
Gene 93:125-128.

99

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

Loy, A., C. Schulz, S. Lucker, A. Schopfer-Wendels, K. Stoecker, C. Baranyi,
A. Lehner, and M. Wagner. 2005. 168 rRNA gene—based oligonucleotide

microarray for environmental monitoring of the betaproteobacterial order
“Rhodocyclales”. Appl. Environ. Microbiol. 71:1373-1386.

Palmer, C., E. M. Bik, M. B. Eisen, P. B. Eckburg, T. R. Sana, P. K. Wolber,
D. A. Relman, and P. 0. Brown. 2006. Rapid quantitative profiling of complex
microbial populations. Nucleic Acids Res. 34:e5.

Peplies, J., F. O. Glockner, and R. Amann. 2003. Optimization strategies for
DNA microarray-based detection of bacteria with 16S rRNA-targeting
oligonucleotide probes. Appl. Environ. Microbiol. 69: 1397-1407.

Peytavi, R., L.-y. Tang, F. R. Raymond, K. Boissinot, L. Bissonnette, M.
Boissinot, F. J. Picard, A. Huletsky, M. Ouellette, and M. G. Bergeron. 2005.
Correlation between nricroarray DNA hybridization efficiency and the position of
short capture probe on the target nucleic acid. BioTechniques 39:89-96.

Pozhitkov, A., P. A. Noble, T. Domazet-Loso, A. W. Nolte, R. Sonnenberg, P.
Staehler, M. Beier, and D. Tantz. 2006. Tests of rRNA hybridization to

nricroarrays suggest that hybridization characteristics of oligonucleotide probes
for species discrimination cannot be predicted. Nucleic Acids Res. 34:e66.

Ramakrishnan, R., D. Dorris, A. Lublinsky, A. Nguyen, M. Domanus, A.
Prokhorova, L. Gieser, E. Touna, R. Lockner, M. Tata, X. Zhu, M.
Patterson, R. Shippy, S. T.J., and A. Mazumder. 2002. An assessment of
Motorola CodeLinkTM microarray performance for gene expression profiling
applications. Nucleic Acids Res. 30:e30.

Ratushna, V. G., J. W. Weller, and C. J. Gibas. 2005. Secondary structure in
the target as a confounding factor in synthetic oligomer microarray design. BMC
Genomics 6:31.

Reyes-Lopez, M. A., A. Mendez-Tenorio, R. Maldonado-Rodriguez, M. J.
Doktycz, J. T. Fleming, and K. L. Beattie. 2003. Fingerprinting of prokaryotic
16S rRN A genes using oligodeoxyribonucleotide microarrays and virtual
hybridization. Nucleic Acids Res. 31:779-789.

Reynaldo, L. P., A. V. Vologodskii, B. P. Neri, and V. I. Lyamichev. 2000.
The kinetics of oligonucleotide replacements. J. Mol. Biol. 297 :51 1-520.

Riccelli, P. V., K. E. Mandell, and A. S. Benight. 2002. Melting studies of
dangling-ended DNA hairpins: effects of end length, loop sequence and
biotinylation of loop bases. Nucleic Acids Res. 30:4088-4093.

100

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

Rouillard, J.-M., M. Zuker, and E. Gulari. 2003. OligoArray 2.0: design of
oligonucleotide probes for DNA microarrays using a thermodynamic approach.
Nucleic Acids Res. 31:3057-3062.

Sengupta, S., K. Onodera, A. Lai, and U. Melcher. 2003. Molecular detection
and identification of influenza viruses by Oligonucleotide microarray
hybridization. J. Clin. Microbiol. 41:4542-4550.

Small, J., D. R. Call, F. J. Brockman, T. M. Straub, and D. P. Chandler.
2001. Direct detection of 16S rRNA in soil extracts by using Oligonucleotide
microarrays. Appl. Environ. Microbiol. 67:4708-4716.

Southern, E., K. Mir, and M. Shchepinov. 1999. Molecular interactions on
microarrays. Nat. Genet. 21:5-9.

Southern, E. M., S. C. Case-Green, J. K. Elder, M. Johnson, K. U. Mir, L.
Wang, and J. C. Williams. 1994. Arrays of complementary oligonucleotides for
analysing the hybridisation behaviour of nucleic acids. Nucleic Acids Res.
22:1368-1373.

Taroncher-Oldenburg, G., E. M. Griner, C. A. Francis, and B. B. Ward.
2003. Oligonucleotide microarray for the study of functional gene diversity in the
nitrogen cycle in the environment. Appl. Environ. Microbiol. 69:1159-1171.

Tiquia, S. M., L. Wu, S. C. Chong, S. Passovets, D. Xu, Y. Xu, and J. Zhou.
2004. Evaluation of 50-mer Oligonucleotide arrays for detecting microbial
populations in environmental samples. BioTechniques 36:664—70, 672, 674-5.

Urakawa, H., P. A. Noble, S. El Fantroussi, J. J. Kelly, and D. A. Stahl. 2002.
Single—base-pair discrimination of terminal mismatches by using Oligonucleotide
microarrays and neural network analyses. Appl. Environ. Microbiol. 68:235-244.

Vainrub, A., and B. M. Pettitt. 2003. Surface electrostatic effects in
oligonucleotide microanays: Control and Optimization of binding
thermodynamics. Biopolymers 68:265-270.

Wang, H., J. Li, H. Liu, Q. Liu, Q. Mei, Y. Wang, J. Zhu, N. He, and Z. Lu.
2002. Label-flee hybridization detection of a single nucleotide mismatch by
immobilization of molecular beacons on an agarose film. Nucleic Acids Res.
30:e61.

Wick, L. M., J. M. Rouillard, T. S. Whittam, E. Gulari, J. M. Tiedje, and S.
A. Hasbsham. 2006. On-chip non-equilibrium dissociation curve and dissociation

101

 

43.

44.

rate constant as methods to assess Specificity of Oligonucleotide probes. Nucleic
Acids Res. 34:e26.

Wilson, W. J., C. L. Strout, T. Z. DeSantis, J. L. Stilwell, A. V. Carrano, and
G. L. Andersen. 2002. Sequence-specific identification of 18 pathogenic
microorganisms using microarray technology. MOI. Cell Probes. 16:119-127.

Zhang, L., T. Hurek, and B. Reinhold-Hunk. 2005. Position of the fluorescent

label is a crucial factor determining signal intensity in microarray hybridizations.
Nucleic Acids Res. 33:e166.

102

CHAPTER FOUR

DEVELOPMENT OF EMBEDDED CONTROL SYSTEMS FOR GENE ANALYZER
4.1 INTRODUCTION

Development of embedded control systems and housing, generally assumed to be
mature technologies, may pose a major challenge during development of a device in a
university setting. Even though the technology is mature, the service is only available
through expensive powerhouses requiring 6 to 24 months worth of consulting fee for a
team of engineers amounting to several hundred thousand dollars. At universities, the
skill may exist in the appropriate disciplines but not in a Single laboratory. Thus the
development of basic controls, internal structure, and housing becomes a problem in
itself. Without these components, the device doesn’t exist and work related to validation
is not possible. Considering the above factors, a decision was made to develop the
embedded system for the gene analyzer using basic skills in microcontroller
programming, data communication, software programming, and CAD to design the
housing. This chapter describes selection and control of components for the gene
analyzer. Compared to the structure of previous chapters that were written as journal
manuscripts, the structure Of this chapter differs for two reasons: i) the content Of this
chapter is more suitable for a book chapter, and ii) there was a need to document the
embedded system programming more thoroughly for future students.
Control and electronic related aspects in development of the gene analyzer consisted

of the following tasks:

1. Temperature control using LabVIEW, USB-DAQS, and breadboard.

2. Detection system control using LabVIEW, USB-DAQS and breadboard.

103

3. Design Of custom PCB to accommodate ARM microcontroller, ICS, and LED
drivers.

4. Replacement Of USB-DAQS and PC with custom ICs and ARM microcontroller.

5. Development of circuit to establish communication between ARM
microcontroller and an iPod Touch, and

6. Design of housing.

 

 

 

 

 

 

 

 

 

Breadboard Stage PCB Stage Integrated Device
1. Heater Control 3, Develop PCB 5. Communication with iPod
2. Detectron Control 4. Detach from DAQ/PC 5. Design of Housing

Figure 4.1 Schematic depicting the various stages during the development Of gene
analyzer.

4.2. TEMPERATURE CONTROL USING LABVIEW AND BREADBOARD

Initial control of components and data acquisition was obtained using National
Instruments LabVIEW (Laboratory Virtual Instrumentation and Engineering Workbench)
with USB compatible data acquisition accessories (U SB-DAQ) as analog and digital
inputs/outputs to and flom a personal computer (PC). The USB-DAQ boards along with
LabVIEW graphical interface provided a flexible means Of control with minimal
programming experience. Several integrated circuits (IC) were also used to aid in critical
aspects of control (Figure 2). The small, single structured assembly of ICs made them

ideal for deployment into a hand-held based platform.

104

led

ach'

may

requ

deri‘

00111

then

anal

belo

 

Figu

4.2.1

temp,

aCCur;

In general, a control strategy had to provide both accurate and precise temperature
to drive the amplification reaction and synchronize control of a detector scheme to
acquire amplification signal measurements in real-time. A means of feedback control
may be required to drive the amplification reaction with precise temperature
requirements. Use of pulse-width—modulation (PWM, 3, 8), or proportional—integral—
derivative control (PID, 1, 5, 7) is well known for accurate and precise temperature
control. The combined use Of both PWM and PID together on a system with very small
thermal mass has resulted in one Of the most rapid cycling rates (4). For the gene

analyzer, developed as part of this study, a coupled PID/PWM was used as described

 

 

 

 

 

 

 

 

 

 

 

 

 

below.
0.1 mA, 10V
Detection l E I
Control [ Photodiode l 1 LED Driver I0 I
4 to 16 line decoder DC
. ‘i IC |
PC With LabVlew contro|}_ NI USB D AQ power
—| PWM IC l SUPPIV
Heater NI therrncouple DAQ Heater With rapld
Control thermocouple
0.7 amps. 28 V
Max Duty Cycle

Figure 4.2 Schematic of temperature and detection system control for the gene analyzer.
4.2.1 HARDWARE FOR COUPLED PID/PWM IMPLEMENTATION

Critical aspects of temperature control included using a rapid response
temperature sensor, a heater that allowed rapid temperature cycling, and a strategy for

accurate/precise feedback control. Signal flom the temperature sensor had to be

105

amplified and conditioned for accurate control. Final hardware components for the
device consisted of a rapid response thermocouple, a pulse width modulator IC (Texas
Instruments DRV102T, $8.40), 2 USB DAQS flom National Instruments (N1), the heater
itself (2.54 x 7.62 cm Kapton heater, 40 Ohm, ElectrO-Flex Heat, Inc., $40), and a
LabVIEW VI developed in-house (Figure 3).

Rapid response temperature measurements were Obtained using a type T
thermocouple with a 0.004” sheath diameter (SSRTC-TT-T-40-36, Omega, $15.00) with
a response time of approximately 0.1 sec and a specified standard limit of error less than
1 °C. An extension cable was used to increase the size of the thermocouple wires for
easier handling and preventing damage that could influence the temperature measurement
(TECT10-13, omega, $15.00). Initial evaluation Of temperature sensors included surface
mountable thermistors, however, these devices did not have the response time required
for accurate temperature control causing drastic overshoot flom the set-point before the
senor responded to the correct temperatrn'e. Voltage output produced by the thermocouple
was amplified and conditioned using the National Instruments USB-DAQ (9211, $530).
The 9211 DAQ automatically converted the thermocouple output into temperature, which
was read directly by the LabVIEW V1 for feedback control. A LabVIEW PID subVI
(graphical interface with a specific function inside of a graphical interface) provided ari
analog output (flom multifunction NI USB-DAQ 6009, $280) based on the deviation
from the set-point temperature (called using a signal generator VI described below).
Large temperature deviations below the set point produced an output less than 0.5 V,
small deviation below the set point produced a larger voltage (0.56 to 3.9 V), and

deviation above the set point produced an output of 4 V. This output had to be

106

 

programmed into the DAQ assistant (described below) that was used for the analog

output.

PWM Driver (DRV102) Multifunction USB-DAQ _

Thermocouple USB-DAQ

 

 

Thermal Switch
~ 1 l
. V DC power supply
Oscrllatorl I 28 V :
PWM \ Kapton heater with rarllid
' £3 +— thermocouple

 

 

 

Cycle 0 .
I Adjust 4 :Analog Output 0. l to 4 V. 8 mA

________ _(._________.(_

Figure 4.3 Detailed schematic Of circuit used for feedback control of the heater on the
bench-top device including description of how components were connected to USB DAQ
and PWM IC. Dotted lines represent signal and solid lines represent power.

eel]

i

A pulse width modulation (PWM) driver was used for highly precise temperature
control. The analog output flom the DAQ was connected to the duty cycle pin of the
PWM driver. When a larger duty cycle is required, a relay connecting the power supply
(V 5 pin) with the heater (Out pin) inside the PWM IC is closed for a larger flaction of the
time, thus the average power delivered to the heater is increased. This allows the amount
of power provided to the heater over time to be changed without using a potentiometer.
One advantage of this strategy is that it reduces the total amount of power delivered to a
load without losses normally incurred with sources limited by resistive means. All other
pins on the PWM driver were lefi unconnected, except the ground. The selected PWM
driver is ideal for smaller heaters and can supply a large range of voltages (8 to 60 V), but

is not rated for a current above 2.7 A (the selected Kapton heater used a maximum

107

current of 0.45 A with 28 V). Therefore, larger heaters and devices that are driven with
more power such as thermoelectric Peltiers will require a different means Of PWM
control. Some important notes for using the PWM IC include using a l to 10 k resistor
connected in series between the analog output and duty cycle pin to limit swing to ground
to be no less than 0.1V. In addition, a grounded 0.1 IIF capacitor was placed in parallel
between the power supply and IC to reduce noise to other sensors within the circuit. The
PWM IC has a thermal shutdown switch to protect itself when close to overheating that
will reset when temperature drops below 125 °C. The driver also has a pin that can be
connected to a LED to signal when thermal shutdown has occurred. To date, thermal
shutdown has never occurred when controlling a doped silicon heater or the Kapton

heater.

4.2.2 SOFTWARE FOR COUPLED PID/PWM IMPLEMENTATION

The LabVIEW VI consisted Of five main mechanisms, a PID SubVI, a signal
generator to call the temperature profile, the analog inputs to read the thermocouples,
analog output for feedback control of the PWM driver, and a program to write
temperature data to a file (Figure 4). The Simulate Arbitrary Signal SubVI was used to
generate a user specified temperature profile over time. The user can change the
temperature profile by double clicking on the Arbitrary Signal SubVI, clicking on define
signal, and manually entering time and temperature set-points in a tabular fashion. The
signal generator was wired to the set-point Of the PID SubVI. Analog inputs were
obtained using a series of DAme SubVIs to obtain Signal from the USB-DAQ 9211 and

to report errors. Input flom the DAme SubVI was wired to the process variable pin of

108

the PID SubVI. The output flom the PID was wired to a DAQ assistant SubVI, which
can be used for both analog and digital inputs/outputs. The DAQ assistant makes
connection to the multifirnction USB-DAQ simple, leading the user through the process
of selecting if Signal will be acquired or generated, what is to be measured (e.g. voltage,
temperature, current, resistance), and suggests pin connection to the USB-DAQ. This is
the first tool that should be used by an inexperienced LabVIEW user for accepting and
generating signals flom the PC via the multifunction USB-DAQ. The V1 shown in
Figure 3 also has the input for an IC thermocouple amplifier and conditioner, as a low
cost alternative to the USB DAQ 9211 (described in section 4.4). The VI reads a voltage
using the DAQ assistant, and a linear equation is used to convert the voltage to
temperature. The conversion equation was provided within the specification of the IC
conditioner. Finally, the temperature flom all thermocouples is recorded along with a
time stamp into a tab-delimited file that can be named by the user by double clicking on

the Write to Measurement File SubVI.

109

(Jr‘s-urn
"mun-nupt-

0 -
Alum-V‘s rd" ('-
I'T

,. -
Ph m .111 Inn-d l Firm :1 kit—Inn] ‘
m __ fflflwzh— "" A _._ I]: «r mu.- . my; 2
' i‘ .w—J—i # 4
Al‘- 1 "j A117». 17.3 MAI-(mun- 1
I l. .
ml-DIQ In Two E Thu“ “9" T‘?’ {1 l
fl . . m. Zorn Had-
I It. Vle 4 It Value:
1 '1 lion—I It hand ‘

 

1 M's-nun: x. \l'tlodule‘

El: _
Mm\ duo dept-3

 

nu
\lnIIn—sz-ul-gill !
"fl

 

 

,3 1.," 'r

  

 

 

 

Thermocouple

 

mL————J

 

 

 

1 input from 921 1

 

 

 

 

 

 

 

C‘allulg telllpelanue Win" loop Tm" rum
profile m—

i
“217.4

bI-rrlrn -

,__.. a
~ ' Manna-nu 2
[no J—I',_
‘ Annual I . L1;
- m , I .. 4m:
1, q , “L ‘ E ‘
run-«ml Ila-funnel
°
II
'mn.
lumen-“an an
13‘1le NS” " "m.

 

 

 

 

. Mean-Hue
an. l

Measurements to
file

Analog output (to

PWM driver)

 

 

PID SubVI /

Figure 4.4 The temperature control VI developed to drive the amplification reaction,
acquires analog input measurements fl'om the thermocouple, and generates analog output
with PID based feedback control

The flont panel for the temperature control V1 is user fliendly in that only a few
parameters need to be selected when starting the program (lower left corner, Figure 5).
First, the user must set the optimal PID control parameters. The user must then select the
physical channels for each of two thermocouples, and type of thermocouples (default
selection shown if Figure 4). Two thermocouples can be used for dual measurements of
either the difference in temperature across the surface of the amplification chip, or to
examine the temperature change between the top and bottom of the amplification chip.

However, only one of the thermocouples is used for feedback control. TO determine

110

which thermocouple is connected to the PID controller, the user can start the VI, watch
the graph on the lower right comer, and look for a temperature response by placing the
thermocouple between their fingers. The thermocouple used for temperature control
should be placed between the heater and the chip. The lower right graph also shows the
set-point temperature and the voltage output from the PID used to increase duty cycle of
the PWM IC (Figure 5). The top right graph also shows the set point, but over a longer
period of time, and the top left graph shows the temperature of the second thermocouple
(not controlling the PID). The axis of all graphs can be changed to reduce or increase
temperature and time scales. This can become useful when manual tuning the PID to a
single temperature. For example, when timing the PID for maintaining temperature to
drive loop mediated amplification; the scale of the y-axis on the lower right graph can be
reduced to 60 to 66 °C. The V1 can be stopped using the button on the fiont panel or the

stop button on the toolbar.

111

Measured Temperature thermocouple 2 Setpoint 3321-3'

_ ram and.
9‘ :1

9|)
85
m
'5
U
05
an
55
I)

   

8

WM! ”:4“
£0£4J"" . Q fins-Jar L"
Joan Jtoou P
Ila—mue'C) Max—vhflpgo fluntx!)
;t-Toa&““*‘“" rgumm' ‘ , .__
.“H-UU n
mar 95.—earn» .. ' 1
JT {)1 ,nnn‘o

    

 

fl

.,, .

,W WWW.-. , r.
TOIN‘EIC) — 7L".w ..

“ml-(inc) — 3"

came-M! 37””
Figure 4.5 Front panel of temperature control VI. User changeable parameters including
constants for PID control are in the lower lefi comer (default parameters are shown). The
lower right graph shows the temperature measured by the thermocouple used for
feedback control (yellow), set-point (red), and the output voltage from the PID (green).
The upper right graph shows the set points over time, and the upper lefi graphs show the
temperature measured with the second thermocouple

4.2.3 PERFORMANCE OF COUPLED PID/PWM IMPLEMENTATION
Results of using PWM along with PID control showed that both control methods
were important for precise control of the heater and to reduce overshoot and oscillation
from the set points. Using a l x 1 x 0.05 cm piece of doped silicon as a test heater, the
combined control of PWM and PID reduced overshoot and standard deviation of the set-
point from 0.80 °C with PID alone to 0.28 °C with PWM and PID (measured during 72
oC set-point for 8 seconds, Figure 6 lefi panel). Compared to the sole use of PID (to

open/close a relay between the heater and power supply), added precision using PWM

112

with PID permitted high temperature set points to be reached more rapidly without
overshoot. For future refinement of the device for quantitative PCR, this setup has the
potential for running a 30 cycle real-time PCR assays in approximately 12 minutes using
1 s denaturing, l s annealing, and 8 3 extension (Figure 6 right panel shows 8 cycles in 3
min). This makes a heating rate of approximately 20 °C/s and a cooling rate of 3.5 °C/s.
A 0.1 A 12 volt fan (EC5010M12C, Evercool) was turned on for approximately 6 5
during cooling fiom 95 to 55 0C using a relay switch that was controlled with a timer

through LabVIEW. Compared to the silicon doped heater, the Kapton heater had 20

 

times the surface area and used nearly the same amount of power for temperature change,

0.5 A times 28 V for Kapton heater and 1.13 amp times 10 V with the doped silicon

 

 

 

 

 

 

 

 

 

heater.
100 * — PWM and PID 100
\ PID alone ""

90 1 A 90
E ‘ a
£801 2 so
*' B
E "'\.V_Vv E
2.70‘ g. 70‘
E a

60 . I I" 60 t

50 - . so . . . c . .

 

 

0 5 10 15 20 25 30 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (s) . .
Time (mm)

Figure 4.6 Left panel demonstrates a thermocycle temperature profile using PID/PWM
feedback control and PID control without PWM (tested using doped silicon heater). The

right panel shows 8 cycles performed in approximately 3 minutes using a 0.3 mm thick
Kapton heater.

PID tuning was essential for reducing deviation from set-points and limiting
overshoot during rapid temperature changes. For obtaining the rapid cycling

demonstrated in figure 4.6, the PID parameters were manually tuned as follows. The Kp

113

parameter was the first to be tuned while the K1 and KD were kept constant at zero. A
higher Kp parameter increased the amount of change in the output for a given change in
the error. Thus, a Kp that was too high caused the system to become unstable, producing
large fluctuations in temperature from the set-point. In contrast, a small Kp resulted in a
small output response to a large input error, making feedback less responsive and even
prohibiting the heater from reaching the set-point temperature. For our setup, the Kp term
heavily contributed to the majority of feedback control compared to the K1 and KD
constants. Next, the K; constant was adjusted by finding the optimal constant between
reaching steady-state more rapidly (higher K;) and reducing overshoot fi'om the set-point
(lower K1). The K9 is typically used to reduce overshoot, however, no observable change
was measured, and therefore a low KD constant was used. The PID constants used for the

measurements in Figure 4.5 were a Kp of 1, a K. of 0.05, and a KD of 0.005.

4.3 DETECTION SYSTEM CONTROL USING LABVIEW AND BREADBOARD

An optical setup for capturing real time amplification Signals in a multichamber
assay will require synchronized control between the light source and the detection
system. Multiple solutions exist for this purpose and have been in use in larger more
expensive instruments. Key requirements dictated by the small footprint of gene analyzer
were: i) the total detection system must not be larger than 20 cm3 and the total cost must
not exceed $200. This was achieved by the development of an optical system consisting
of: i) 16 LEDS coupled to an IC that allowed multiplexing, ii) individual LED drivers to
provide accurate and constant current to all LEDS, iii) synchronized control of LEDS with

a photodiode (PD) used to acquire change in signal emission, and iv) an IC to reduce the

114

conditioning noise acquired from the PD. Details of the detection system are described

below.

4.3.1 HARDWARE FOR DETECTION SYSTEM CONTROL

Critical aspects in controlling the detection setup included synchronized timing of
components, controlling multiple LEDS with a limited number of digital outputs,
obtaining a constant and programmable current to the LEDS, reducing noise from the
photodiode, and using a strategy to prevent the photodiode from overloading. Control of
the optical setup consisted of 16 light emitting diodes (LEDS) for exciting a dye in the
amplification reaction, a photodiode (PD) for measuring emission, three separate ICs to
aid in control, the same multifunction NI USB DAQ (6009) used for the temperature
control, and a LabVIEW VI developed in-house. The following describes the hardware
in more detail and connection between components, ICS, and the multifunction USB-
DAQ (Figure 4.7).
The nunnber of controllable LEDS can be limited to the number digital outputs on a
processor (U SB-DAQ has 12). Therefore, a 4—tO-16 line decoder/demultiplexer
(CD74HCT4514, Texas Instruments, $2.25) was used to individually control all 16 LEDS
using only 5 digital outputs from the USB-DAQ. For communicating with the decoder, a
high logic signal (3.3 V) was provided at all times to the LE pin. The on/off logic
configuration of 4 other digital outputs from the USB-DAQ was used for gingival control
of the LEDS. The small amount of current provided via the digital lines from the USB-
DAQ was sufficient to power the decoder without connecting a power supply to the Vcc

pin. The LE pin provides the possibility of cascading between multiple decoders using

115

only 1 additional line per decoder. To summarize, the first decoder requires five digital
outputs and each new set of 16 LEDS only requires one additional digital output. In total,
a maximum set of 128 addressable LEDS could be used with the 12 digital outputs on the
USB-DAQ board.

Continuous current had to be delivered to the LEDS for reproducible and
quantifiable measurements of the amplification reaction. This was obtained using a 50
mA stand-alone linear LED driver IC (STmicroelectronics, STLAOl, $0.95). A high
logic output of 3.3 V from the decoder (pins YO to Y15 of decoder connected to the EN
pin of each driver) instructed drivers to turn on 1 of the 16 LEDS. In addition to
providing constant current, the LED driver could be programmed to supply current based
on the requirements of the user without the need for an external sensor. A power supply
provided 5 V to the LED drivers, which were programmed (using a 4.6 kilo Ohm resistor)
to provide 26 mA with :1: 5% accuracy to the 30 mA rated LEDS. A higher resistance
resistor (connected to the PROG pin) courld be used to reduce power to the LEDS and a

lower resistance resistor could be used to increase the deliverable current up to 50 mA.

A critical aspect in using the photodiode for accurate, reproducible, and
quantifiable detection of emitted light was reducing noise and preventing the PD from
overloading. Overloading occurred when too much light was detected, causing the
photodiode to generate inaccurate measurements, which could only be reset by briefly
removing power. This problem was resolved by powering down the photodiode after
every LED measurement cycle using a soft relay (Potter and Brumfield, T77SID3-05,
$1.80). A 5 V signal from the analog output of the USB-DAQ closed the relay switch,

connecting the power supply to the photodiode for measurements during the LED

116

cycling. After each of 16 LEDS had been on for a user-defined amount of time and all
PD measurements had been acquired, the relay would open, powering down the
photodiode. Noise was also reduced using a grounded 0.1 uF capacitor in parallel with
the V+ and V- pins (connected to a 10 volt power supply) as instructed in the photodiode
specifications manual. The output pin of the photodiode was connected to the analog
input on the USB-DAQ, which was also connected in parallel to a grounded 0.1 uF
capacitor for additional noise reduction. Output reading fiom the photodiode and the
circuit ground was connected to two analog inputs on the USB-DAQ for a Obtaining a

differential measurement (also implemented to reduce noise).

Through-hole configurations (DIP) were selected for all ICS to develop the
prototype on a standard solderless breadboard (K and H products, SD-24, Figure 4.12A).
The LED drivers were only available as a surface mountable configuration (DFN);
therefore, they were mounted to DFN to DIP boards (E6-0091, ePBoard Design) and
soldered using services provided by ePBoard Design The total cost for the board and
mounting was $12.15 per LED driver. It should be noted that all ICS were ordered from
DigiKey. Ordering directly fiom the manufacturer typically required large volume, while

single piece orders could usually be placed through DigiKey.

117

________________ Multifunction USB~DAQ " _( 17.50; Iain]

 

 

 

 

+ Digital Outputs ' "Q 0.05 to 5 volts.
I High/low signal I
I 10 v 0.1 amp '

I Relay I
| I
I Decoder + I
I CD74HCT45|4 a , fl” - - —'
l a. . ,, LT .
l a IPhotodiode
I t:
I
I- g LED r-ra
I o
l— — a
|_ _
l
'— —
I
+ t
vcc LED
L - 5 v 0.2 amp Driver
STLA01
. PROG - -|
- - >- - "': EN '
High/low
signal E

 

Figure 4.7 Detailed schematic of circuit used for the optical detection on the bench-top
device including description of how components were connected to the multifunction
USB DAQ and the ICS. Dotted lines represent signal and solid lines represent power.

4.3.2 SOFTWARE FOR DETECTION SYSTEM CONTROL

Precise timing of the device was obtained using a LabVIEW VI developed in-
house. The VI for control of optical components consisted Of six major mechanisms
including a Boolean array for generating digital outputs to control LEDS, a DAQ assistant
generating an analog output for controlling the relay between the PD and power supply, a
DAQ assistant for analog input measurements from the photodiode, a VI to average all
PD samples, flat sequence structure and loops for precise timing between LED and PD

measurement acquisition, and a program to write PD data to a file (Figure 4.9). A

118

Boolean array (described in more detail in software section below) allowed user defined
control of how many and which LEDS to utilize for a given experiment. The array of
true/false calls was converted into an unsigned 32-bit integer and wired to a while loop to
define how many LEDS to excite and PD readings to take per cycle. Inside and outside of
the while loop was a DAQ assistant subVI that generated a signal to open and close the
relay used between the photodiode and power supply. A flat sequence structure (looks
like film reel) inside the while loop was used for synchronized control between the
individual LEDS for a user specified amount of time, and reading the photodiode after the
LED has been on for a user specified amount of time. The initial frame in the flat
sequence loop generated digital outputs (to the decoder) from the unsigned integers. The
latter of the two flat sequence frames acquired signals fiom the photodiode with a rate of
sampling and number of samples defined by the user. The rate of sampling and number
of samples can be changed by double clicking this DAQ assistant SubVI in the block
diagram. An amplitude and level measurement subVI was used to generate a single data
point by averaging all readings taken from the PD after each measurement. The average
data point was time stamped and saved into a tab-delimited file named by the user. The
data file can be directly imported into excel for analysis of real time amplification curves.
An additional noise reduction strategy was implemented within the optical detection VI
by changing the maximum and minimum anticipated voltage from 0 to 2.5 V (changed in
DAQ assistant SubVI). The number of cycles can be selected by the user by entering a
constant that is wired to a “for loop” that encompasses the entire block diagram.
The fiont panel of the optics control VI was designed to allow the user to

change the following: 1. Number and position of LED readings using the Boolean

119

 

array, 2. Number of cycles, 3. Time for which the LEDS are on, 4. The amount of
time to wait between measurement cycles, and 5. The time to allow the LED to warm

prior to acquiring PD samples (Figure 4.8).

 

    

    

 

 

 

 

azaam II E
00000 C,
000.02% 3: Eg
ooooosgg 3' 22
00.0.2 F E: V”
oooooagg.§s E
00000— e: =
0.00028 >=
. oooooazg g 'E
‘— ooooowga $3
@3333: “9* Eco - ~ ~ ' or
OOOOOOw 3 ”WM” *5
OOOOOOv g
.00...“ :
g...‘..“ a
OOOOOO~ a
>OOOOOO° A 3
5:19.939; g g
:%E. g g
r] g g
E E é a
a 5 E “E
a 2: O _
a $8 “*8 QL
iii gs £13 gig: is
N a e us

 

Figure 4.8 Front panel of optical component control VI. User changeable parameters
include time LED is on, number of cycles, time to wait between cycles, time to wait for
LED to warm before taking PD measurement, and the Boolean array for controlling
individual LEDS. The chart shows the average PD measurement and the last
measurement of multiple samples taken with the PD. Constants shown in the screen shot
were used for experiments described in Chapter 5.

120

. The LED array. This array is used to communicate with LED drivers to turn them
on/off. Each column represents the sequential calling of the individual LEDS (0
refers to LEDl, 15 refers to LED 16), and each row represents the digital output from
the DAQ board to decoder addresses (A0 to A3 in the table below). A green pattern
represents a high signal to be sent from the DAQ to the decoder and a black pattern
represents a low signal. The first four rows are used to call the LEDS. The decoder
accomplishes demultiplexing of LEDS (channel 0 to 15) using select inputs (rows 14)
as addresses (A0 to A3). The fifth row is always on high to communicate with this
decoder. If a second set of 16 LEDS are to be used (i.e. 32 LEDS instead of 16), the
fifth row and a sixth row could be used as a chip select so that multiple decoders can
be cascaded. The sixth row would be highlighted for LEDS 17 to 32 (while fifih row
is off), and the fifth row would be highlighted to command LEDS 1 to 16 (while sixth
row is off). For using less than 16 LEDS, the chosen column can be deleted by right
clicking on the array.

. The number of cycles. This parameter is user-defined. For real time PCR, this
number would be adjusted to 40 to 45 cycles. The number of cycle for LAMP will be
dependant on how many cycles will be obtained within a 30-minute period. This
number will be dependant on the number of channels used, and the amount of time
specified for switching on each LED. For the default numbers provided (Figure 4.8),

approximately 50 cycles can be used for a 30 min reaction.

121

 

 

 

  

 

Measurements to
file

 

Average PD
readings

 

 

 

 

 

DAQ assistant
measuring PD

 

 

 

 

DAme digital
output to LED

 

 

 

 

DAQ assistant

/ for soft relay

 

 

 

 

 

 

Boolean Array
and mechanisms
for cycling LEDS

 

 

 

 

 

 

 

Figure 4.9 Block diagram of optical control VI developed to detect the amplification

reaction, generate digital outputs for individual LED excitation using a Boolean array,

and acquire PD measurements using DAQ assistant for analog input. A flat sequence
structure loop was used to synchronize the LEDS with PD data acquisition.

122

3. The amount of time to turn on the LEDS (ms). This constant allows the user to
determine how long each LED should be on. This includes time to allow LED to
warm (control 5) and time to stay on for the PD reading. A default value of 2000 ms
was used for characterization experiments described in chapter 5.

4. The time to wait before running next cycle. This time will be dependant on how
many channels will be read per cycle, how long the LEDS will be on, and how much
time is necessary between taking readings. For example, if 6 LEDS are measured and
are on for 2 seconds each, and the user prefers 20 s between each measurement, a
wait time of 8000 ms will be entered here.

5. Amount of time to wait before LED is measured. This control allows the user to
determine how long to wait before measuning the PD to ensure that LEDS had reached

a steady and high level of brightness.

4.3.3 PERFORMANCE OF THE DETECTION SYSTEM

The amount of time required for measurement of each well is critical in affecting
the number of cycles (data points) that can be obtained to plot the real time amplification
curve. The parameters used for experiments described in Chapter 5 are displayed in
Figure 4.8. This included turning each LED on for 2 s, and acquiring a PD measurement
after one second Of allowing the LED to warm up. The average of 100 sample readings
taken during a 0.2 see time interval was recorded for the amplification curves. However,
using these parameters reduced the amount of readings that could be taken during the

exponential phase Of amplification to 4 to 5 data points. Additionally, the amount Of time

123

required to read all of the reaction wells may influence CV% of Tt between reactions on

the same chip, subsequently reducing quantitative accuracy and reproducibility.

     
 

1000
100
yr,”
10
c,\"
>
U
l
-0-— no LED noise
f‘ ......o. amplified signal
0.1 1.1.0....35 --v-- non amplified signal

0.01

 

Time (s) to measure all 16 channels (PD measurment parameters)

Figure 4.10 Coefficient of variation (%) for 4 replicate readings (y-axis on log scale)
taken within the same reaction channel and a variation in the PD measurement
parameters and therefore, the amount of time required for a one cycle measuring all 16
channels (x-axis). Readings were taken from 3 different channels including one without
an LED, 3 channel with amplification, and a channel with no amplification

Further experiments were performed to examine the minimum amount of time
that could be used for reproducible Signal. Results demonstrated that a time lower than

1.5 S for acquiring reading fiom all 16 channels (0.09 8 per channel) significantly

124

decreased the ability to obtain reproducible readings (Figure 4.10). Therefore, the 2 8
cycle time per LED (used for data acquired for chapter 5) could be reduced to 0.09 s by
obtaining 500 samples at a 10 kilo-hertz sampling rate. Irreproducible Signal observed
with reduced acquisition times may have been due to limitations of LEDS, PD, USB-
DAQ, or timing of the LabVIEW VI. For example, the 10k sampling fiequency of the
PD always has a lower CV% than the 100k sampling fiequency. Regardless of the cause,
a total time of 1.5 s for measuring all 16-reaction wells should aid in the reduction of the

CV% (described in Chapter 5).

4.4. DESIGN OF CUSTOM PCB To ACCOMMODATE ARM MICROCONTROLLER, ICS,
AND LED DRIVERS

To reduce size of the circuit, a custom printed circuit board (PCB) was designed
and fabricated to replace the breadboard. Using a PCB for the circuit also allows low
cost mass production of the device (~$10 each for a 100 piece order). Issues related to
PCB layout were critical for obtaining reproducible fabrication of the PCB with low
noise, cost, and reducing accidental contact between two points in the circuit. A
schematic of the circuit was drawn in ExpressSCH for initial determination of wire
placement (Figure 4.11). An important point in using ExpressSCH, is that component
footprints do not need to be accurate, however, a critical parameter for the development
of the final PCB layout is the correct placement of wires between components.
Therefore, each component (including resistors and capacitors) requires unique names
with correctly identified pin numbers. Trace lines can be crossed as the software

identifies connections via a node symbolized by a small square, which is only placed by

125

 

the user clicking on the line or pin. Simple components such as resistors, capacitor,
ground, and ribbon connectors could be added fiom the component library. However, all
ICs selected for the hand-held device had to be drawn manually. Once the schematic was
completed, it was imported into the ExpressPCB software to aid in the correct
connections between pins. This helped the user focus on critical PCB design such as
spacing between traces instead of correct wiring. More specifically, choking on a pin of
a component caused all connections to be highlighted, thus traces could be drawn without
continuously cross-referencing the schematic.

For prototyping punposes, the first set of PCB boards was designed using Express
PCB’S free software. Two separate PCBs were designed and fabricated using
ExpressPCB. The first PCB functioned to hold all of the LEDS and consisted of a ribbon
cable, and multiple through-holes for the LED pins (Figure 4.12, left panel). The second
PCB was the circuit for the [CS and microcontroller and had a bottom and top layer of
copper consisting of traces and pads for components, and a ground and power plane
sandwiched in the middle (Figure 4.12, right panel). The 4—1ayer design was selected to
reduce traces required for grounding and power, and for increased noise immunity for the
PD and thermocouple sensors. In detail, the boards were designed with several important
parameters including the following: vias (also called feed-through holes) were used to
move traces from one layer to the other. The vias or through-holes can be connected to,
or isolated fiom, the inner planes of the 4-layer board. A size of the hole used for all vias
was 0.008". The vias become important when traces need to be crossed. An ideal layout
for a board will consist of vertical traces on one side and horizontal traces on the other,

with horizontal and vertical traces connected using the via. A width 0.010" was used for

126

most of the traces, as it is ideal for low current (less than 0.3 A) digital and analog
signals. A trace of 0.025" (suggested for less than 1.0 A) was used to power the heater.
While a minimum spacing of 0.010" was suggested between traces, a spacing of 0.050”
was used to reduce the risk of a shorts developing in the board manufacturing process.
One important design consideration is to avoid using sharp right angle turns when placing
narrow traces (0.012” or less). A suitable alternative is to use two 45-degree bends with a
short leg in between. A minimum of 0.021" space must remain between adjacent holes
andthegapbetweeneveryitem(padtopad,padtotrace,tracetotrace)was0.007" or
greater. A 26-pin connection for a mountable ribbon cable was used to allow connection

between the LED PCB and the component PCB.

l

a
d
o
a
d
o
a.
a
d

D
c:
c:
:3
In
I:
I?
c:
u:
D

.iflflfl'flpflflm.

 

Figure 4.11 Schematic of circuit layout drawn in ExpressSCH

127

 

Figure 4.12 Screenshot of PCB layouts for the LED array and for ARM
microcontroller and ICs

Many of the component le had standard sizing, therefore dimensions could be
found in the component manager. However, some ICS were not in the component
manager, requiring component specifications to be examined carefully for drafting in
ExpressPCB. Pads for the majority of surface mountable ICS were placed on the bottom
of the PCB while le requiring through-holes were placed on the top Of the PCB (Figure
4.13B and 4.13C). Heat generated by the power regulating ICS and the PWM drivers was
dissipated using copper pads with footprints suggested by specification manuals. Since
most of the ICS can be purchased in several configurations (i.e. wide or thin, surface
mountable or through-hole), it is important to ensurre that the configuration selected for
the PCB are not obsolete or require large quantity orders. For example, the decoder can

bepurchasedinathinconfigurationthatis7mmwideoraconfigurationthisis 14mm

128

wide. However, the thin configuration can only be purchased in a minimum order of 285

pieces.

 

Figure 4.13 Picture of constructed circuit used to control and test components. A)
bench-top control with circuit connected using a bread-board (USB-DAQ in the back).
First version of custom PCB with B) fi'ont-side for through-hole ICs (only 1 is attached),
and C) back-side of PCB for surface mountable ICs (ARM microcontroller and voltage
regulators are shown).

This version of the PCB was designed with USB-B connections for
communicating between the ARM device and a PC. Three identical component boards
were printed and shipped for approximately $100. The MiniBoard manufacturing option

was chosen which limits overall dimensions to 3.8 x 2.5”, but reduces costs by $70.

129

 

These boards include the green solder mask and white silkscreen layers (with trace

outlines and text to determine proper placement of ICS, Figure 4.13B and 4.13C).

4.5. REPLACEMENT OF USB-DAQS AND PC WITII CUSTOM ICS AND ARM
MICROCONTROLLER

Embedded control of the LAMP hand-held device required replacing the
LabVIEW Multifirnction USB-DAQ, thermocouple conditioning USB-DAQ, and the PC
with inexpensive alternatives suitable for a portable deployment. Microcontrollers are a
cost and space-effective solutions for control compared to the large and costly USB-
DAQ. Critical aspects in selection of a microcontroller include a simple and fast means
of generating code. Familiarity with the LabVIEW programming environment dictated I

the selection of NI LabVIEW embedded module for ARM microcontrollers.

LabVIEW has recently released a module to aid in the programming of
microcontrollers using their graphical interface (6). This module has been described as
an ideal means of embedded control for engineers that are not accustomed to C/C+-I-
prograrnnring (6). As of 2009, ARM (advanced reduced instruction set computer
machine or advanced RISC machine) processors account for approximately 90% of all
embedded 32-bit RISC processors, and are deployed in a wide range of applications such
as iPods, game consoles, calculators, GPS navigators, industrial, and medical
applications. Furthermore, approximately 98% of mobile phones sold each year use at
least one ARM processor (2). In addition, the ARM microcontroller utilizes a CPU

design with simplified instructions that can be executed quickly for obtaining a high level

130

of performance. Therefore, this processor was selected to control components and
acquire data in the hand-held device.

Processor specifications for the 2 x 2 cm microcontroller (ARM7 NXP LPC2378)
selected with the LabVIEW embedded module include the following: 58KB RAM, CPU
clock up to 72 MHz, USB 2.0 Full Speed Device Controller, lO-bit analog-to-digital
converter (ADC) with 8 channels, and a lO-bit digital-to-analog converter (DAC) with 1
channel. The module also contained an evaluation board (MBC2300), which allowed the
firnctionality of the ARM microcontroller to be tested with the various components.
Specification for the evaluation board include: serial interfaces, speaker, through-hole
leads input/output pin on the ARM for prototyping, an LCD, USB connection, Ethernet
connection, and 8 LEDS connected to digital outputs.

The LabVIEW ARM module automatically converted LabVIEW graphics based
control VIs into C/CH code and programmed the microcontroller for testing with the
evaluation board. The workflow for this consisted of opening LabVIEW 8.6 and
selecting ARM project target in the lower left portion of the “Getting Started” window.
After saving the project filename, a “Project Explorer” window was opened that allowed
the user to select digital, analog, PWM, input/outputs. After adding elemental
inputs/outputs they could be dragged fiom the “Project Explorer” window into the block
diagram and wired accordingly. Once the LabVIEW block diagram is completed, the run
button is clicked and C++ code is generated and programmed to the ARM
microcontroller via the Keil ULINK2 USB-JT AG Interface Adapter, a small hardware
adapter that connects the USB port of the PC to the JTAG port on the evaluation board

(Figure 4.14). Hand-held components were connected to the ARM controller using

131

through-holes in the prototyping area of the board. The board was powered through the
USB, which also served to transmit measurements to the PC for ensuring proper control
and measurements of the components with the ARM components.

The LabVIEW ARM module was highly efficient in terms of reducing time
required for programming the microcontroller with a small learning curve. Basic
LabVIEW VIs were written to test control of each component with the ARM module
(Figure 4.15, 4.16). Sensor currents from the PD and thermocouple were converted to a
digital value using the ARM microcontroller with an integrated analog-to-digital
converter (ADC). Output from the temperature and PD sensors were displayed in real
time on the LCD screen of the evaluation board and on LabVIEW charts (via the USB
connection to the PC). Time that was saved in using the LabVIEW module, instead of
manually writing code, became obvious when testing the components. For example, a
simplistic LabVIEW block diagram for PID based control of the heater at a single

temperature (Figure 4.15) generated 4,220 lines of code.

132

 

 

JT AG
to KEIL
Adaptor

I:1 o
n 2'.

'
mm

Prototype through-holes

Figure 4.14 Picture of the ARM evaluation board used with the LabVIEW embedded
control module. Sections Of the board used to evaluate control of hand-held components

Analog input (thermocouple)

E;

Ill

 

[I]

are highlighted in yellow.

amp e an rep y every ms
Waveform Chart

min/max output from PID ’7

    
  
 

. . PID subVI
#Tempera ure set-point ' y “do“ than '

0
‘ , .__ $9511
1 ‘7‘ .......... p“, r
-E g
PID constants m‘.--,it91’.__t-
U Analog output

 

Continue until stop button is pressed

 

Figure 4.15 Block diagram of VI constructed to test PID-based feedback control of the
heater with the LabVIEW embedded module. Approximately 4,220 lines of code were

generated for this control scheme.

133

 

 

 

 

 

 

E— u
a a
I m l
m ‘3 ww- :-
El --_ "
E."~- 3;-7- i
“z" '1': -fi 6
"‘~—.- "’ ,
“e...,_ 5

~ 1
[I -.- [I " ,‘ I ‘ i
J‘ L """"""""""""""""" M‘ i
m .m __ N“, ,,,. m,“ ___ - I , I];

 

 

 

Figure 4.16 Left panel: Testing control of LEDS with Boolean control Ofmultirple digital
output channels connected to the decoder. Right panel: Analog inputs for PD and
temperature sensor displayed on LED sereelngf evaluation board and LabVIEW charts on

For feedback control of the heater, a 10 bit (minimum output set to 0 and
maximum output set to 1024) resolution generated by the PID (subVI was converted to
an analog signal that varied between 0 and 3.3 V for adjusting the duty cycle on the
PWM IC. The evaluation board had one minor drawback in testing complete
functionality of the PWM IC. This is due to the ARM drawing power for ADC and the
DAC through the USB interface, allowing a maximum analog output voltage of 3.3 V.
The PWM IC requires up to 4 V for a 0% duty cycle. This caused the heater to run at
temperatures beyond the set-point. However, it was observed that once the temperature
set-point was reached, the PID increased the analog voltage to 3.3 V and the amount of
current drawn fiom the power supply was reduced. The maximum voltage to the ARM is
approximately 4.6 V, therefore an operational amplifier can be used to provide the ARM
DAC with enough voltage required for the PWM driver.

Portable deployment of the device also required replacing the $530 thermocouple
conditioning USB-DAQ with an integrated circuit with similar function. The

amplifier/conditioner is critical because raw signal generated by thermocouples produces

134

small voltage differences. For example, a type K thermocouple will generate a difference
of 4 mV between 0 and 100 °C, therefore, this small change in voltage must be amplified
and conditioned for accurate and precise measurement. While multiple ICs were tested
for precise temperature acquisition with minimal noise, a monolithic chip fiorn Analog
Devices (AD595C, $20) with an instrumentation amplifier, a precalibrated accuracy of
i1°C, and thermocouple cold junction compensator offered the most accurate and precise
control. Amplified and buffered Signal from the AD595 IC provided a difference of
1,000 mV for temperatures between 0 and 100 °C. Both the NI USB-DAQ conditioner
and Analog Device 1C conditioner were tested at three separate mmpemtmes to determine
accuracy and precision. While both conditioners showed similar levels of precision, the
integrated circuit had slightly better accuracy (Figure 4.17). The IC conditioner has an
additional function in that it has a LED alarm to alert if the thermocouple has failed.

Thus, the IC is a suitable alternative to the high cost and large USB-DAQ conditioner.

135

110
100 1 96.7 i 0.06 0C

       
   

A90- NTUSB-DAQ 949+0140C

 

62.3 2: 0.15 0C

60.0 i 0.17 0C

   
  

42.5 i 0.15 0C

40.0 i 0.17 0C
' I U j
0 2 4 6 8 10
Time (min)
Figure 4.17 Temperature measured at three different set-points (40, 60, 95 °C) using a

thermocouple connected to the National Instruments conditioning USB-DAQ (gray line)
and Analog Devices 1C (black line).

 

4.6. DEVELOPMENT OF CIRCUIT TO ESTABLISH COMMUNICATION BETWEEN ARM
MICROCONTROLLER AND AN IPOD TOUCH

The hardware and software for communicating between the ARM microcontroller
and iPod Touch must also be developed and tested. A circuit for USB connection
suggested by the specifications manual for the ARM microcontroller was drawn on the
initial PCB, and will be tested. A more complex circuit may be necessary, especially if
the final device will allow Options for connecting to both the 30 pin docking device and
USB-A. A potential circuit schematic used to connect USB-A to the ARM
microcontroller may require one additional IC. Both configuration switch on a LED
when connection between the ARM and host has been established. The LabVIEW VI

must also be modified to allow commrmication between the ARM microcontroller and

136

the iPod Touch. Communication requirements include the microcontroller accepting
information fi'om user-supplied constants on the iPhone (such as time and temperature
set-points), and the ARM transferring data to the iPod Touch. SubVIS for file I/O are
supported through the LabVIEW embedded ARM module for accepting constants and
writing data to a file off of the microcontroller. A means Of programming the ARM
microcontrollers directly on the customized PCB must also be established. One potential
. strategy is to include a circuit on the next version of the customized PCB for
communication between the ARM and JT AG plug-in, which is used for programming the

ARM microcontroller on the Keil evaluation board.

4.7. Design of Housing

Critical issues in development of component housing included arrangement of
components around the amplification chip to reduce cross contamination of LED light
between wells, and reproducible alignment between disposable amplification chip and
LEDS/optical fibers. The cartridge for packaging the entire device also had to utilize
a sliding top cover for removing/adding the chip that eliminates outside light and
provides a latch for slight pressure, be astatically pleasing, allow plug-and-play
capabilities with a iPod Touch to be used for initialization, data transfer, storage and

analysis from the hand-held device.

Reproducible user alignment between the removable amplification chip and
the LEDS/optical fibers was accomplished using a microstructered piece of ABS as an

alignment guide for the reaction wells. Through-holes were drilled in the

microstructured ABS piece for LEDS and optical fibers were permanently placed

137

within the ABS apparatus. Optical fibers (sending emitted light to the PD) contacted
the sides of the shell structured amplification chip (Figure 4.18A shows 2-
dimenstional side view, Figure 4.18D shows 3-dimensional isometric view). LEDS
were placed beneath the chip to reduce the requirement of multiple wires being with
the movable top cover when disposable amplification chips are loaded/unloaded.
LEDS were also placed below the amplification chip and ABS holder to reduce
scattered light fiom LEDS. A piece (44.6 x 41.1 x 15.9 mm) was designed and
fabricated via rapid prototyping (black PC-ABS material, Stratasys, Inc.) to hold the
ABS holder, amplification chip, secure optical fibers with a 5 mm bending radius, and
mount LEDS (Figure 4.18B). A second piece (20.0 x 20.0 x 17.5 mm) was fabricated
to house the PD and emission filter and secmely hold all optical fiber ends around the
active area of the photodiode (Figure 4.18C). The black PC-ABS was chosen for
minimal light reflection, low transmission of light, .ABS like properties that with high
temperature stability Similar to polycarbonate. The modular design of the holder
allows components to be easily replaced if damaged, or changing LEDS and the filter

to detect the amplification reactions at a different wavelength.

138

 

 

  
   
 
    

Photodiode

15.9mm

 
 
  
 
 
    

  

T

Optical fiber T
Optical filter 44.6 mm\

    

ABS holder
C (alignment guide)

I 17.5 mm

Optical fibers

   

KLEDS

20.0 mm LED PCB
20.0 mm

Figure 4.18 Component layourt A) 2-dimensional view of components for the
amplification chip, B) piece fabricated in PC-ABS to house the LEDS, optical fibers
and ABS holder. C) piece fabricated to hold the PD, Optical fibers. D) Three
dimensional view of component layout for the amplification chip (PC-ABS piece is
rendered transparent to view LED and fiber through holes in the structure.
A cartridge was designed to package all components into a small portable layout.
The PCB was the largest component that had to be considered along with the foresight of
a suitable battery. Attempts to decrease the layout with more efficient placement of
traces and ICS would have been cumbersome. Therefore, the footprint required for the
PCB was used as the base for an iPhone or iPod Touch (Figure 4.19). The top of the
cartridge was designed to hold a majority of the components for detecting and heating the
amplification reaction, and the base will serve to house the PCB and a battery. An of

campus source was contracted to draft the cartridge due to the importance of designing it

with a latch—able sliding cover (for inserting/removing the disposable amplification chip)

139

that does not allow ambient light into the holder (Vowin Model Design Co. Ltd,
Schenzhen, China). Once designed, the cartridge will be fabricated out of ABS and

finished with chrome or sandblasting.

25 LED PCB
Latching cover

(————— for loading chips
55

    
 

   
   
 

‘— LED holder/stage
PD holder

30 pin dock and USB-A

120

  

 

2 Main PCB

 

 

 

59' .4 r
16.5 ‘ 75 ’

 

Figure 4.19 Schematic of the cartridge for packaging the LAMP based hand-held device
(dimensions in mm). The back of the package, which encases the PCB, is used as a base
for the iPod Touch and/or iPhone which is placed upside down and plugged into a male
30 pin docking port. The docking port connects the iPhone to the hand-held device for
heating and real-time detection of the disposable amplification chip. The chip will be
inserted/removed with a sliding self-latching door that will also serve to block all ambient
light from entering the device during testing.

4.8 CONCLUDING REMARKS

140

A control setup for driving and detection loop mediated amplification was
established and tested. Results Showed that components and control setup allow precise
and accurate temperature control for single set-point isothermal amplification and has the
potential for rapid temperature cycling for quantitative PCR. Constant current LEDS,
noise reduction strategies to the photodiode, and synchronized control strategy permitted
reproducible signal measurements to be acquired in as little as 0.09 seconds per channel.
For embedded control of components, a LabVIEW module was shown to efficiently
program an ARM microcontroller. A custom PCB was used to replace a breadboard
layout, and a cartridge has been designed for housing components and interfacing with an
iPod Touch or iPhone. This control setup serves as a basic model for a multiassay, low-
cost, miniaturized diagnostics system built with standard hardware components that are
readily available and aids in the realization of a laboratory-independent tool for rapid

detection of microbial pathogens.

141

 

4.9 References

1. Easley, C. J., J. M. Karlinsey, J. M. Bienvenue, L. A. Legendre, M. G. Roper,
S. H. Feldrnan, M. A. Hughes, E. L. Hewlett, T. J. Merkel, J. P. Ferrance,
and J. P. Landers. 2006. A fully integrated microfluidic genetic analysis system
with sample-in—answer-out capability. Proc. Natl. Acad. Sci. U S A. 103:19272-
19277.

2. Krazit, T. 2006. ARMed for the living room, CNET News.

3. Liao, C. S., G. B. Lee, H. S. Liu, T. M. Hsieh, and C. H. Luo. 2005. Miniature
RT—PCR system for diagnosis of RNA-based viruses Nucleic Acids Res. 33:e156.

4. Neuzil, P., J. Pipper, and T. M. Hsieh. 2006. Disposable real-time microPCR
device: lab-on-a-chip at a low cost. MOI. Biosyst. 2:292-298.

 

5. Pal, R., M. Yang, R. Lin, B. N. Johnson, N. Srivastava, S. Z. Razzacki, K. J.
Chomistek, D. C. Heldsinger, R. M. Haque, V. M. Ugaz, P. K. Thwar, Z.
Chen, K. Alfano, M. B. Yim, M. Krishnan, A. O. Fuller, R. G. Larson, D. T.
Burke, and M. A. Burns. 2005. An integrated nricrofluidic device for influenza
and other genetic analyses. Lab Chip 5:1024-1032.

6. R008, G. 2008, posting date. National Instruments LabVIEW targets ARM
microcontrollers. [Online.]

7. Toriello, N. M., C. N. Lin, and R. A. Mathies. 2006. Multichannel reverse
transcription-polymerase chain reaction microdevice for rapid gene expression
and biomarker analysis. Anal. Chem. 78:7997-8003.

8. Woolley, A. T., D. Hadley, P. Landre, A. J. deMelIo, R. A. Mathies, and M.
A. Northrup. 1996. Functional integration of PCR amplification and capillary
electrophoresis in amicrofabricated DNA analysis device. Anal Chem. 68:4081-
4086.

142

CHAPTER FIVE

Validation of Optical System for Gene Analyzer

5.1 INTRODUCTION

Lab on a chip based formats are well suited for point Of care screening Of
infectious diseases. Microbial agents are responsible for more than 75,000 deaths
annually in the U.S. (24). A large part of this is due to the increase in antibiotic resistance
among microbes over the past decade (18). One microbe Of particular importance is
methicillin-resistant Staphylococcus aureus (MRSA). Infections caused by this microbe
have continuously increased in the U.S. from 108,600 in 1999 to 368,600 in 2005, and
during the same period, MRSA-related deaths have increased fiom 11,000 to 19,000 (13).
Current screening practices for MRSA consist of either culture-based assays or real time
PCR (RT-PCR). Culture based assays take two to four days because Specimens must be
sent to a centralized laboratory and screened by culturing the specimen on MRSA media
specific. RT-PCR assays are faster (e.g., BD GeneOhmTM Staph SR, made by BD
Diagnostics and recently approved by FDA), taking about 2 hours. However, the cost Of
these machines (up to $30,000—$100,000) requires specimens to be sent to a centralized
laboratory, delaying diagnosis, implementing the prescription of broad-spectrum
antibiotics, and increasing potential for the emergence of a new more resistant MRSA
strain. The availability of a portable and low cost device at a point-Of-care, such as
emergency first responders, healthcare centers, or physician’s will aid in the realization of
a more rapid means of identifying a suitable prescription of the correct antibiotic,

avoiding the use of broad Spectrum antibiotics.

143

Constraints in the development of truly portable and low cost nucleic acid
analyzers are typically associated with the means of detection. The most common
schemes for real time PCR amplification are based on fluorescence detection (6, 32), with
components such as lasers and charge-coupled devices (CCDs). These apparatuses are
considered to be too complex and large for miniaturization and integration into portable
hand-held devices (14). Less costly components such as LEDS and photodiodes are
promising alternatives, however, simultaneous detection of multiple channels with single
point detectors typically requires movement or assigning independent detectors to each
channel on the device (1 1). In contrast, an array of individually controlled LEDS can be

used with optical fibers sending fluorescent signal to a single point photodiode detector.
This mechanism for detection compliments loop mediated amplification (LAMP), as it
produces more amplified product ~10 pg per 25 ul reaction (26) compared to
conventional RT-PCR which produces approximately 200 ng per 25 pl reaction. Thus, a
less sensitive, more compact, and cost efficient optical strategy can be utilized for real
time fluorescent detection (20, 36). With this in mind, a device for fluorescence real time
LAMP detection within a disposable microfluidic chip was developed (referred to as
LAMP Analyzer). The setup uses an array of sixteen LEDS as an excitation source (one
LED for each reaction well) and sixteen optical fibers to send fluorescent light fi'om each
well to a single photodiode sensor for detection. Only one LED/well/optical fiber is
turned “on” at a time and identity is determined by the associated sequence and timing of
the photodiode reading.
A critical component of the overall system is its low cost and user-scalable

disposable microfluidic chip. Typical methods for hot embossing thin wall thermal plastic

144

 

structures consists of using a pair of matching dies which require very precise alignment
of both embossing molds, and may lead to difficulties during ejection of the microchips
and mold damage when alignment is not exact. To circumvent these issues, a low glass
transition temperature thermal plastic (ABS) was used as a counter tool (12). This
method provided numerous advantages over standard hot embossing for fabricating thin-
walled structures, including the need for only a single embossing die, elimination of
alignment issues, and ease of demolding as pressed chips could be deembossed from the
counter tool instead of the mold itself. Thus, thin wall rrricrofluidic chips were
microstructured by hot embossing 100 um Zeonor film (COP) with stereolithography
molds. The process window for reproducible embossing of amplification chips was
increased using a shoulder on the perimeter of the mold. The negative piece of
microstructured ABS was utilized further to: l) aid in enclosing amplification chips under
high pressure without damaging the shelled microstructures, 2) providing an alignment
guide between optical components (optical fibers and LEDS) for replacing consumable
amplification chips in the gene analyzer, and 3) reducing potential for crosstalk of light
between adjacent reaction wells (17, 33).

Components for the gene analyzer were characterized by examining the
reproducibility among reaction wells, and influence of light crosstalk between channels.
Further validation included a comparison of quantitative capacity, sensitivity, and
reproducibility between the gene analyzer and a standard laboratory real time PCR
thermocycler. Real time LAMP characterization experiments were performed with

calcein and an assay targeting the mecA gene of MRSA.

145

 

5.2 MATERIALS AND METHODS

5.2.1 Disposable anrplg'ficaa'on chip fabrication and ABS holder: For embossing, male
molds were drawn in Cobalt, a three dimensional computer aided design software
(Ashlar-Vellum), and fabricated with 0.002 inch resolution out of Somos® NanoToolTM
using stereolithography (FineLine Prototyping). NanoToolTM pieces underwent a thermal
post-cure with UV to increase the glass transition temperature to 170 °C, and finished
with nickel-plating (SLArmor) for increased strength and ease of de-embossing. For
embossing, a 100 um piece of ZeonorFilm (ZF14-100, Zeon Chemicals L.P) was
sandwiched between the male mold and a piece of 2.2 mm thick low glass transition
temperature plastic ABS (KS-8889, K-Mac Plastics), which was used as a soft counter
mold (Figure 5.1 top panel). The embossing protocol consisted of preheating all
materials without pressure to 150 °C for 3 min, followed by adding 2,000 lbs of pressure
for 3 min using a Carver Hot Press (4386, Wabash). Subsequently, the press was cooled
to 105 °C (well below the 136 °C glass transition temperature of COP) using manual
cooling platens, while the microforming pressure was maintained. Pressure was released, .
and the formed polymer chips were deembossed from the formed ABS counter tool. The
chips were cleaned by soaking in 1% Liqui-Nox (Alconox) for 5 min, thoroughly rinsed

with distilled water for 1 min, rinsed in isoproponal, and air dried in a 65 0C oven.

146

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 |
Mold 1 Mold _ Rubber Am lification chip
ZeonorFilm ._ wJ PCR tape h!”
ABS W W
L J ABS
“wage-n. _

Figure 5.1 T op panel : Illustration of the low temperature plastic-assisted microforming
and process for enclosing reaction wells. Bottom panel: Photograph of a microshelled
lOO-um thick COP sheet containing 7 50-um deep reaction wells enclosed with a tOp
cover containing holes for adding samples and air ventilation.

To form enclosed features, the chip was bonded with 70-um thick PCR tape
(MicroAmp Optical Adhesive Film, Applied Biosystems). For introducing sample and
air ventilation, circular holes were patterned into PCR tape using a Craft ROBO
(CC33OL-20, 4). Patterned PCR tape was manually aligned onto the chips, and briefly
pressed at 500 lbs (at room temperature) by sandwiching between a piece of silicon
rubber (70 A Durometer Rating, McMaster-Carr) and the microstructured ABS. The
silicon rubber was used to evenly distribute pressure over the full substrate area between
the PCR tape and microstructured Zeonorfilm, and the ABS served to protect the shelled
structure of the Zeonorfilm and maintain channel features. Samples were loaded into the

channels, and a second piece of PCR tape was used to seal the sample ports.

5. 2.2 Component setup: Components for heating the LAMP reaction and detecting
amplification in real-time included a 2.54 x 7.62 cm Kapton heater (40 ohm, Electro—Flex
Heat, Inc., Bloomfield, CT), 16 polymer optical fibers with a 1 mm core (IF C U1000,

Industrial Fiber Optics, Tempe, AZ), a preamplified photodiode (ODA-6WB-500M, Opto

147

Diode Corp.), 16 blue 470 nm LEDS (RL3-B2030, Super Bright LEDS Inc), and a
microstructured piece of ABS. The microsturctured piece of black ABS served as an
alignment guide between the chip and the LEDS/optical fibers. For constructing the
device, the microstructured piece of ABS was fastened to a piece of black nylon. One
mm holes were drilled through the ABS directly underneath the reaction wells, and 3-mm
holes were drilled into black nylon for mounting the LEDS. Reaction wells were
individually illuminated by time staggered control of the LEDS (i.e., LEDS were turned
“on” one at a time).

Polymer optical fibers, chosen due to their low cost and high numerical aperture,
carried light from the reaction well to the PD. Optical fibers were butted against the
sidewall of the microshelled reaction wells. An apparatus was designed and fabricated
using stereolithography to receive all optical fibers with the same distance and angle from
the active area of the photodiode. Different emission filters (colored glass longpass
filters) were tested by placing in-between the optical fiber and the PD. The temperature
of the chip was measured using a type T thermocouple with a 0.004 inch sheath diameter
(5 SRTC-TI-T-40-36, Omega) with a rapid response time of approximately 0.1 sec and a
specified standard limit of error less than 1 °C. The control thermocouple was placed

between the heater and the amplification chip. A more detailed description of component

layout can be found in Chapter 4, Figure 4.18.

5. 2.3 Bench-top control: Components were interfaced to a personal computer using a
Multifunction USB Data Acquisition (USB-DAQ) card (6009, National Instruments,

Inc.). A signal generating sub-virtual instrument (SubVI) was used to call a temperature

148

profile, which was controlled by LabVIEW software using proportional integrated
derivative (PID). Voltage signals from the thermocouple were amplified and conditioned
using the LabVIEW 9211 thermocouple input module. An analog output generated from
the PID controller was used to adjust the duty cycle on a pulse width modulation (PWM)
driver (DRV102T, Texas Instruments). The PWM driver acted as a gate between a
power supply and the heater, providing fine control of power to the load without losses
that occur when using resistive means of power management. A DC power supply set to
28 V provided a current of 0.5 A to initially heat the load, and an average current of 0.1 A
was required to maintain isothermal temperatrne set-point for LAMP.

Individually controlled LEDS were turned on/off for obtaining signal with the PD
of each reaction well. A 4-to-16 line decoder/demultiplexer (CD74HCT4514, Texas
Instruments) was used to control the individual LEDS. The decoder is a high-speed
silicon-gate that allows communication with each LED using only 5 digital outputs from
the DAQ card. High logic output from the decoder signaled 1 of 16 LED drivers
(STLAOl) to provide constant current to the LEDS. The drivers were programmed to
provide 26 mA with :l: 5% accuracy. Prior to reading a signal with the PD, the LED was
turned on for 1 s. The average of 100 sample readings taken during a 0.2 s time interval
was recorded for the amplification curves. An analog switch was used to power down the
photodiode between each cycle of sampling to circumvent issues related to overloading.
The LabVIEW vi for both temperature control and optical detection are provided in

Chapter 4.

149

5.2.4 Characterization of LAMP reaca'on using gene analyzer: Genomic DNA from
methicillin-resistant S. aureus (ATCC 700699D-5) was obtained fiom the American
Type Culture Collection (ATCC) and resuspended in diethylpyrocarbonate (DEPC)-
treated and nuclease-free sterile water (Fischer Scientific, Pittsburgh, PA). The target
was amplified using a set of four LAMP primers (F 3-
AAAAAACGAGTAGATGCTCAA, R3-TGGCCAATTCCACATTGT, FIP-
TCCCAATCTAAC'ITCCACATACCATAAAACAAACTACGGTAACATTGA, BIP-
CATAGCGTCATTATTCCAGGAATGCCGGTCTAAAATITTACCACGT) along with
a pair of loop primers (LF- TTI‘AACAAAATTAAATTGAACGTI‘GCGA, and LR-
AGAAAGACCAAAGCATACATATTGAAAA). The primers were designed using
Primer Explorer version 4 (httpzllprimerexplorer.jp/e) from a consensus of 15 mecA
alleles generated with Bioedit Sequence Alignment Editor (Ibis Biosciences, Carlsbad,
CA). The specificity of the primers was ensured by checking against the GeneBank
database using NCBI BLAST, and were ordered from Integrated DNA Technologies
(IDT). LAMP reactions were performed in a total volume of 3 uL in the amplification
chips and 12 u] in the benchtop Chromo4 Real-time PCR detector (Bio-Rad
Laboratories), and consisted of 1.6 uM each of FIP and BIP primers, 0.2 uM each of F3
and B3 primers, 0.8 uM each of LF and LB primers, 0.8 M betaine (Sigma), 1.4 mM of
each dNTPs (Invitrogen Corporation, Carlsbad, CA), 20 mM Tris/HCl (pH 8.8), 10 mM
(NHa)2804, 10 mM KCl, 8 mM MgSO4, 8 mM Triton X—100, 0.64 units/uL of Bst DNA
polymerase, large fragment (New England Biolabs Inc.) and 12.5 uM calcein and 0.25
mM MnClz (half of concentration described (39)). Samples run with the Bio-Rad were

loaded in 0.2 mL PCR tubes (VWR International) and incubated at 64°C for 36 min, and

150

fluorescence was measured every 45 5. Samples in the chip were incubated at 60 °C
measured on the side of the chip furthest fiom the heater (a temperature of 70 °C was
controlled using a thermocouple placed between the heater and amplification chip), and
the PD was measured every 18 s for the 6 reaction experiment (dilution series

experiments) and every 36 s for the 16 reaction experiments.

5.3 RESULTS AND DISCUSSION

5.3.1 Microstructered amplification chip: Thin-wall chips were fabricated out of
ZeonorFilm cyclic olefin polymer (COP) due to low water absorption, superior optical
properties, inert surface, and no observable change in dimensions under high
temperatures (2, 23, 43). Thin films were molded using hot embossing, because unlike
injection molding, polymer flows a very short distance, resulting in parts with very little
stress that are well suited for optical setups (16).

Hot embossing with molds fabricated using stereolithography provided a robust
and suitable alternative compared to conventional stamping materials. Often, the
embossing mold is made of a lithographic material such as silicon (8). However, this
material is brittle resulting in a small process window for embossing microstructres with
high-aspect ratios (44). Additional advantages using stereolithography fabricated molds
include low cost and a rapid turnaround time (4 to 5 business days). Thermal expansion
of nanotool caused slight bubbling in the layer of nickel after multiple uses, however, this
did not influence hot embossing as the bubbles were pushed down during pressing. To
our knowledge, this is the first manuscript describing the use of stereolithography

fabricated molds for hot embossing microstructures (Figure 5.2).

151

O
‘ fl

' . a ' ‘. - .
, ‘ Nrckcl plfitulfitemolrmogra[flue molds
’ V A A

v A

'l'ht'r‘mrrl plastic counter—tun]

 

Figure 5.2 Picture of nickel-plated stereolithography mold pressed into a piece of low
glass transition temperature thermoplastic (ABS) used as a counter-tool for
microstructuring channels in ZeonorFilm COP. Shoulders were placed on the
perimeter of the mold to increase the embossing process window.

A shoulder was placed on the perimeter of the mold to simplify the process of
embossing (Figure 5.2). Specifically, the shoulder prevented lateral flow of the
thermoplastic and allowed a more uniform buildup of pressure for pressing reproducible
microstructures. The shoulder bordered two of forn' sides of the mold and was 100 um
higher than the tallest features. The thickness of Zeonorfilm alter pressing with molds
wasapproximately80to90umwherefeatureshadbeenpressedandwas100uminareas
without features. Pressing without a shoulder made the thickness of the film under the
features irreproducible, often resulting in the mold pressing through both layers of

ZeonorFilm and ABS. A study by Yao and coauthors (44) first described the use of

152

shoulders on molds for hot embossing microstructures. Their findings suggested that a
shoulder on the mold reduced the dependence of pressure on the embossing thickness,
and facilitated the attainment of a uniform pressure across the entire embossing area,
allowing a larger processing window.

A low glass transition temperature thermoplastic (ABS) was used for thin film
embossing of microstructures. The use of a soft support instead of a hard counter-tool
has been described previously for simplifying embossing techniques by protecting the
mold, allowing more uniform embossing pressure, and easier deembossing (12).
Compared to rubber-assisted embossing techniques described previously (27),
comparable features were only obtained when the surface area of the rubber counter mold
was ~10 times larger than the mold itself. When counter molds were only slightly larger
than the molds (in terms of spatial area), a remarkable difference in microstructured
shells was observed (Figure 5.3). The less defined and irreproducible structures observed
when using the smaller pieces of rubber may have been due to elongation of rubber
reducing pressure and ability to obtain well defined shapes. However, elongation is
reduced when the same amount of pressure is used on the larger pieces of rubber counter
tool. In addition to obtaining well defined featrnes, embossing with ABS had the added
benefit of allowing microstructured COP to be deembossed from the negative ABS
structure rather than the mold itself, preventing damage to the chip and eliminating the

need to include small draft angles in the mold (5).

153

 

 

so A‘itubbé‘r 5“ 60 A Rubber

9“ H - ‘%~w$._,"
. .,‘i"e-. “i :3 .

{I 1’" ,

'70JA.Rubber ' ‘

 

Figure 5.3 Cross sectional picture taken of 0.75 um wide microstructured channels
pressed in 100 um ZeonorFilm with rubber and low glass transition temperature thermal
plastic counter tools. All of chips (except 50 A rubber large footprint) were pressed using
counter tools with spatial area slightly larger than the molds. For the 50 A rubber large
footprint, 3 piece of rubber 10 x the spatial area of the mold was used as the counter tool.
Chips were cut with scissors, taped to the side of a pctridish and magnified 40 times
using a fluorescent microscope to highlight shelled structure of COP.

The molded ABS tool was used further to enclose channels of the chip. The ABS
served as a protective casing, allowing PCR tape to be pressed to the chip with high
pressure without damaging or deforming shelled channel microfeatures. This
circumvented the challenge of optirrrizing temperature and pressure to reduce
deformation of microchannels for pressure driven bonding. Previous methods to
overcome these issues have used costly programmable hot presses (9, 21, 28, 40),

commercial bonding systems ( 1, 3), or laminators (15, 31). However, to our knowledge,

154

 

this is the first manuscript describing the use of microstructured ABS to protect shelled
microstructures during high pressure bonding of adhesives.

A chip with 16 individual reaction channels (fluidically unconnected) was used
for the characterization experiments so that the different sample mixtures (i.e. dyes, and
concentration of gDNA) could be tested simultaneously. Microfluidic chips with a more
simplistic mechanism for loading one sample into multiple reaction chambers are also
under examination. This includes photograiting COP to reduce hydrophobicity and
increase capillary forces (22, 37). Strategies for eliminating cross contamination between
reaction wells are also currently being examined including using valve systems with
cover layers of soft PDMS (30, 35, 42), using a wax to eliminate primer dissolution and
carryover during sample loading (19), and reducing cross contamination due to primer
diffusion with increased lengths between reactions (7, 41). Alternative means of sealing
reactions such as heat sealing and capping are also currently being explored. Previous
bottlenecks for examinations of these strategies have been resolved with the

establishment of a means to reproducibly fabricate chips.

5.3.2 Characterization of the gene analyzer: Prior to running comparative
characterization experiments, the gene analyzer was tested to examine reproducibility
between reaction wells on the same chip. For this, the same amount of S. aureus gDNA
(1 x 104 genomic copies) was placed in all 16 reaction wells. While amplification was
observed in all wells (Figure 5.5 left panel), the initial background signal varied fiom
0.05 to 0.82 V and the amplification signals was 1.19 to 1.87 times greater than the initial

signal. To circumvent normalization issues caused by the variation in baseline signal, the

155

 

threshold time (Tt) was calculated as the inflection point (using the slope of 5 sequential
data points) to determine when the change in fluorescence is at its greatest (Figure 5.4,
(36). The variation in background is likely due to both a small difference in the distance
between the capillary-end and side-wall of the amplification chip and the distance of the
optical fiber fiom the active area of the photodiode. To reduce this variation, optical
fiber glue (40-0005, Industrial Fiber Optics) will be used with subsequent prototypes of
the device. In addition, multiple designs are being examined for placement of optical

fibers with equal distance around the active area of the PD.

 
     

 

 

 

 

 

 

1.8 r 6.0 0,004 r A P 0.012
r: ’55 '5 «a A
1.64 3 5, o .5 u0.010
T.“ g 5.0 0.003« 3 9' o E
. v 5
Lug 5,45 ,_ 00,. 3, 0.008
g :0 a$4.0 0.002‘ E ° 30 '6 H.006
13‘ fl :0 a >35 3 . E
E . ._ < 0 0° 6 '0.“
9 o m , 0 o o '-
1.0‘ c: . 3.0 0.0011 a .°, “1
p l 2.5 G o . M0000 0.002
oooooooooooo
0.8 . - - - - 2.0 0.000 -%J°-o
0 5 I0 15 20 0
Time (min) Time (min)

Figure 5.4 Difference in baseline signal between reaction wells and systems (left
panel) prompted calculation of the inflection point (right panel) for determination of
Tt, circumventing potential normalization bias.

Using the inflection point to determine Tt, the average Tt among 16 reaction wells
was 12 min with a standard deviation of 0.94 min, and a coefficient of variation (CV) of
7.85%. This CV is high compared to studies described by Mori and coauthors in which a
CV of 0.85% was observed using a turbidity meter for real time LAMP in 8 reaction
wells (25). The CV obtained with the gene analyzer can be decreased with further
optimization of reaction conditions in the low volume amplification chips. In addition,

the 2 3 LED pulse used to acquire signal between each reaction well may have

156

contributed to the high CV. Experiments (described in chapter 4) found that the scan rate
could be reduced to 0.09 8 per reaction well, which may serve to decrease CV in future
experiments. No correlations were observed between the Tt and background signal, Tt
and amplified signal over noise ratio, and Tt with respect to position of the reaction well
on the chip.

Prior to running comparative studies, additional experiments were performed to
ensure that the increase in signal was solely due to fluorescence and not turbidity
generated due to precipitation of manganese pyrophosphate. Positive controls with and
without calcein were tested with the hand-held components and results showed no
observable increase in signal for the wells without calcein (Figure 5.5 right panel). Since
the signal increase was caused solely by fluorescence, and there was no observable
variation in Tt attributed to well position, further characterization experiments were

conducted and compared to a standard laboratory real-time thermocycler.

 

3.0 1

2.5 ‘
,1. A 2.0 J
:3 3'
a 5'.
"" - 1.5 4
.13 a
m "’ 1.04

0.5 i

usestrtttzz:33:22323231233133313:IZ:

 

 

 

0.0 V V V V V V V V

05101520253035
Time (min) Time (min)

Figure 5.5 Left panel: LAMP in all sixteen-reaction wells spiked with 10,000 genomic
copies of DNA. Right panel: positive control with calcein (solid line) and without calcein
(dotted line) was examined to determine if turbidity influenced signal. The maximum
input voltage (used to reduce noise in PD measurements) did not provide measurements
above 2.5 V.

 

157

 

 

To examine and compare real time quantification and optical sensitivity of the
hand-held device with a standard laboratory real-time thermocycler, a dilution series of
10 to 100,000 genomic copies of S. aureus per reaction well was prepared. The Tt was _
approximately 20 s lower with the gene analyzer for 1x 105 and 1 x 104 genomic copies
and 120 3 higher with 1x103 genomic copies compared to the laboratory real-time
thermocycler (Figure 5.6 top left panel). A linear relationship with a correlation
coefficient of r2 = 0.962 was obtained with the gene analyzer (emission filter of 550 nm),
and was r2 = 0.996 with the Bio-Rad. Varying the emission filter with the gene analyzer
had little influence on the correlation coefficient r2 = 0.999 (570 nm) and r2 = 0.949 (590
nm) (Figure 5.6 top right panel). As expected, the baseline signal was significantly
reduced with the highest long pass emission filter (Figure 5.6 lower left panel), however,
very little difference between amplified and baseline signal was observed between the 3
filters (Figure 5.6 lower right panel). Therefore, the 550 nm filter may be better suited for

future experiments with the gene analyzer.

158

 

 

20'

 

 

 

 

 

 

 

 

 

 

20 v , W a
rs« . 550 l
o B‘ R d 16.
A“; '0- a E O , v Stfl
E o E, r44
:14 l-‘-‘ '
l" i 12 1 .
12 i 101 9
lo . - 1 - _ s . - . -
Ictz re+3 1e+4 le+5 lets “+2 ”+3 ”+4 ”+5 ”“6
Genomic copies Genomic copies
'5
3.0 . $93.0 ‘
D
2.5t J- g 2.5
a §
502.0 1 3 2.0 '
ELS' E 15‘
e S»
31.0t u 1.01
.e ‘8
0.5« E 0.5«
'E.
0.0 - , e ' - - 50.0 - -
550 570 S90 Bio-Rad 550 570 590 Bio-Rad

Figure 5.6 Top left panel: standard curves observed with the gene analyzer and standard
laboratory real-time thermocycler (Bio-Rad) obtained using a dilution series with the
same number of genomic copies per reaction well. Top right panel, standard curve
obtained with 3 separate emission filters on the gene analyzer. Bottom left panel displays
the baseline signal for all filters used on the gene analyzer and the Bio-Rad. Bottom right
panel: the amplified signal divided by the baseline signal for all comparisons. Error bars
represent standard deviation of three replicates.

The larger difference in Tt at lower concentration may have been due to
interactions between the chip surface and biomolecules in the LAMP solution. This is
primarily due to the increase of the surface-to-volume ratio and/or the PCR tape used to
enclose the wells (45). Both a high concentration of BSA and Pluronic F68 were added
to the amplification mixture to reduce the influence of the surface on the reaction (10,
38), however, the concentrations of this admixtures have yet to be optimized for use with

LAMP or within COP chips. The low differences in Tt observed between the gene

159

 

analyzer and laboratory system suggest that the optical setup is well suited for
quantitative detection of LAMP.

Amplification was not observed for the no template control and dilutions of 100
and 10 genomic copies per reaction in the laboratory thermocycler or gene analyzer.
Therefore, the minimum amount of template DNA showing observable amplification for
both systems was 1000 genomic copies. Prior studies have indicated that the LAMP
reaction is capable of amplifying as little as six copies of template DNA (29), and studies
examining the use of this primer set with intercalating dyes revealed a sensitivity of 100
copies (34). Further optimization of the reaction conditions such as calcein dye

concentration and reaction temperature may serve to increase the sensitivity of the assay.

160

    
   
 
     

.°
4

(B)

+ control

.9
a

water

e
in

+ control

Signal (n.u.)
6
b

t

 

 

 

0 5 10 15 20
Time (min)

 

Optical shell mrcrostructered surface
fiber COP embedded m (C) LEI) microstructered (D)
ABS holder COP—

Figure 5.7 Top Iefl panel: Computer aided draft of chip showing placement of
samples for testing cross contamination of light, and cross sectional area used for
schematics in lower panels. Top right panels; results of light scattering experiment
for signal in positive control reaction wells (solid line) and signal in adjacent wells
filled with water (dotted line). Lower left panel) cross sectional schematic of the chip
illustrating cross contamination of light using the shell microstructered COP
embedded in ABS holder versus the potential for light contamination in a surface
microstructered amplification chip (Lower right panel).

Since COP istransparentandhasarefi‘active indexhigherthanthatofair, itcan
behave as a waveguide for light, thus, experiments were also performed to investigate the
fluorescence noise due to cross-talk between adjacent channels in the multi-channel
amplification chip. The perceived mechanin for crosstalk in our setup was that
fluorescent light emission from adjacent positive reaction wells (caused bylscattering of
light fiom LEDs) may influence signal when measurements are taken for an Imamplified

reaction well. To ensure that this was not influencing signal in the gene analyzer, two

161

reaction wells were filled with water and surrounding adjacent reaction wells were filled
with positive amplification solutions (Figure 5.7 top left panel). Wells filled with water
showed no observable increase in signal suggesting that crosstalk caused by the scattering
of fluorescent emission fiom adjacent wells was not detectable with this optical setup
(Figure 5.7 top right panel). Crosstalk of light between adjacent wells has been described
for similar optical strategies using LEDS and optical fibers for detection in transparent
chips (illustrated in Figure 5.7 lower right panel). More specifically, fluorescein filled
charmels without focused LED excitation displayed 5% of signal fiom adjacent
fluorescein filled wells (2 mm apart) with a focused LED (17). The authors found the
majority of the crosstalk was due to fluorescent emission (not scattered LED light from
adjacent channels. Increasing distance between reaction channels and/or placement of a
black piece of rubber between channels reduced observable cross contamination of light.
In our setup, the ABS holder complemented the shelled chips, serving as a block between

wells, thus reducing cross contamination of light (Figure 5.7 lower left panel).

5.3.3 Refining the gene analyzer for R T-PCR. A fluorescence means of detection was
chosen (instead of turbidity) for future refinement of the gene analyzer for quantitative
PCR. This would provide the option for detection of mutation. Issues related to thermal
mass of the amplification chip have been circumvented through development of thin
walled microfluidic chips with low mass. The heater control setup has also demonstrated
accurate/precise thermocycling (Chapter 4, Figure 4.6). Therefore, two obstacles must be
overcome for quantitative PCR with the gene analyzer including increasing sensitivity

(with a potentially more complex and costly optical strategy) and overcoming issues with

162

thermal gradients between the heater, layers of the amplification chip, and the holder.
Using a thin piece of black anodized aluminum, instead of microstructered ABS, may be
a suitable means for reducing thermal gradients. The inability to obtain amplification
profiles when testing green and blue intercalating dyes with the gene analyzer (instead of
calcien) demonstrates the need for a more advanced series of excitation and emission
filters, and perhaps a more sophistical means of transferring between consumable chips
and optical fibers and the PD and optical fibers. Alternative means of increasing optical
sensitivity may include increasing the emission intensity of the reaction by supplying a
higher current to the LEDS. A study by Ren and coauthors observed a ten-fold increase
in signal intensity when current to LEDS was increased fiom 20 to 200 mA (33).

A dilution series of fluorescein was prepared to compare optical sensitivity of the
gene analyzer with the laboratory thermocycler. Results show a similar response in
signal with an increasing amount of fluorescein between the two devices, with the Bio-
Rad having a slightly increased response with higher concentration (Figure 5.9). A linear
response in signal (R2 Z 0.97) with the concentration gradient of fluorescein (0.5 to 10
uM) further demonstrates the quantitative utility of the gene analyzer (Figure 5.8). A
lower correlation (R2 = 0.95) has been described for a hand-held device with similar
optical components (36). With a response that is similar to the real-time laboratory
thermocycler, the optical performance of the gene analyzer may only require slight

modifications before it can be used for RT-PCR diagnostics.

163

 

 

 

 

 

1.0 -
0 8 . R2=0.98 0
5‘ 2..
.305 t O R ‘3-97
5t 9 "
'65 0.4 r ’ O a 0
02 ’56 a -
' A ‘ R2=0.98
0.0 . I-—-I . J—

0 2 4 6 8 10 12
Fluorescein (uMl

Figure 5.8 Dilution series of fluorescein with the gene analyzer, change in signal
with varying concentrations of fluorescein is show for three separate reaction wells.

 

'3'

g» 8

-§ 7 O Gene Analyzer 0
E O BioRad

° 6

s

G

g 5

3 4

8

a 3

.5 2

a

:8

:r l

55

.5 0

8 0 2 4 6 8 10 12
a? Fluorescein (uM)

Figure 5.9 Ratio of dilution signal over signal of wells without fluorescein (water)
tested with Bio-Rad (solid dots) and gene analyzer (open dots) shows similar
response to increasing amount of fluorescein. This data represents the average ratio
of 3 different reaction wells.

164

5.4 CONCLUSION

A device for real-time fluorescent-based detection of loop mediated amplification
(LAMP gene analyzer) was developed with a simple optical setup that is capable of
measuring 16 assays simultaneously. Strategies for hot embossing such as the use of
stereolithographic molds with shoulders and low glass transition temperature counter
tools were used to increase the process window for reproducible fabrication of
amplification chips with well-defined microstructures. The simple and low cost optical
setup consisting of individually pulsed LEDS, optical fibers for sending fluorescent
signal, and a single photodiode provided comparable real time results with a laboratory
thermocycler for calcein based LAMP detection. The small and low cost components
assembled for the device make it a promising venture in the development of a nucleic
acid analyzer that is portable, sensitive, quantitative, targets multiple assays

simultaneously, and is cost efficient.

165

5.5 REFERENCES

10.

Abgrall, P., L. N. Low, and N. T. Nguyen. 2007. Fabrication of planar
nanofluidic channels in a thermoplastic by hot-embossing and thermal bonding.
Lab Chip 7 :520—522.

Ahn, C. H., J. W. Choi, G. Beancage, J. H. Nevin, J. B. Lee, A. Puntambekar,
and J. Y. Lee. 2004. Disposable smart lab on a chip for point-of-care clinical
diagnostics. Proceedings of the IEEE 92:154-173.

Arroyo, M. T., L. J. Fernandez, M. Agirregabiria, N. Ibanez, J.
Aurrekoetxea, and F. J. Blanco. 2007. Novel all-polymer nricrofluidic devices
monolithically integrated within metallic electrodes for SDSCGE of proteins. J.
Micromech. Microeng. 17:1289-1298.

Bartholomensz, D. A., R. W. Boutté, and J. D. Andrade. 2005. Xurography:
rapid prototyping of microstructmes using a cutting plotter. J. Micromech. Syst.
14:1364-1374

Becker, H., and U. Heim. 2000. Hot embossing as a method for the fabrication
of polymer high aspect ratio structures. Sensors and actuators. A: Physical
83:130-135.

Belgrader, P., S. Young, B. Yuan, M. Primeau, L. A. Christel, F.
Pourahmadi, and M. A. Northrup. 2001. A battery-powered notebook thermal
cycler for rapid multiplex real-time PCR analysis. Anal Chem. 73:286—289.

Berthier, J., and F. Chatelain. 2005. Presented at the Multiphysics User's
Conference, Paris.

Bhattacharyya, A., and C. M. Klapperich. 2007. Design and testing of a
disposable microfluidic chemiluminescent immunoassay for disease biomarkers
in human serum samples. Biomed. Microdevices. 9:245-251.

Buch, J. S., C. Kimball, F. Rosenberger, W. E. I-Iiglrsmith Jr., D. L. DeVoe,
and C. S. Lee. 2004. DNA mutation detection in a polymer microfluidic network
using temperature gradient gel electrophoresis. Anal Chem. 76:874-881.

Cho, Y. K., J. Kim, Y. Lee, Y. A. Kim, K. Namkoong, H. Lim, K. W. Oh, S.
Kim,J.Han,C. Park,Y.E.Pak, C.S.Ki,J. R.Choi,H.K.Myeong, and C.
Ko. 2006. Clinical evaluation of micro—scale chip-based PCR system for rapid
detection of hepatitis B virus. Biosens. Bioelectron. 21:2161-2169.

166

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

Dishinger, J. F., and R. T. Kennedy. 2008. Multiplexed detection and
applications for separations on parallel microchips. Electrophoresis 29:3296—
3305.

Dreuth, H., and C. Heiden. 1999. Thermoplastic structuring of thin polymer
films. Sensors and Actuators A: Physical 78: 198-204.

Elixhauser, A., and C. Steiner. 2007. Infections with methicillin-resistant
Staphylococcus aureus (MRSA) in U.S. hospitals, 1993—2005. Healthcare Cost
and Utilization Project Statistical Brief #35, Agency for Healthcare Research and
Quality, Rockville, MD. .

Fang, T. H., N. Ramalingam, D. Xian-Dui, T. S. Ngin, Z. Xianting, A. T. Lai
Kuan, P. H. E.Y., G. Hai-Qiug, and 2009. Real-time PCR microfluidic devices
with concurrent electrochemical detection. Biosens. Bioelectron. 24:2131-2136.

Fredrickson, C. K., Z. Xia, C. Das, R. Ferguson, F. T. Tavares, and Z. H.
Fan. 2006. Effects of fabrication process parameters on the properties of cyclic
olefin copolymer microfluidic devices. J. Micromech. Syst. 15:1060-1068.

Heckele, M., and W. K. Schomburg. 2004. Review on micro molding of
thermoplastic polymers. J. Micromech. Microeng. 14:Rl-Rl4.

Irawan, R., S. Chuan Tjin, P. Yager, and D. Zhang. 2005. Cross-Talk problem
on a fluorescence multi-channel microfluidic chip system. Biomed. Microdevices.
7 :205—21 1.

Jones, R. N. 2003. Global epidemiology of antimicrobial resistance among
communityacquired and nosocomial pathogens: A five-year summary fiom the
SENTRY Antimicrobial Surveillance PrOgram (1997-2001). Semin. Respir. Crit.
Care Med. 24:121-134.

Kim, J., D. Ryan, M. G. Mauk, and H. H. Barr. 2009. A disposable, self-
contained PCR chip. Lab Chip 9:606-612.

Lee, S. Y., J. G. Huang, T. L. Chuang, J. C. Sheu, Y. K. Chuang, M. Holl, D.
R. Meldrum, C. N. Lee, and C. W. Lin. 2008. Compact optical diagnostic
device for isothermal nucleic acids amplification. Sensors and Actuators B:
Chemical 133:493-501.

Li, Y., J. S. Buch, F. Rosenberger, D. L. DeVoe, and C. S. Lee. 2004.
Integration of isoelectric focusing with parallel sodium dodecyl sulfate gel

167

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

electrophoresis for multidimensional protein separations in a plastic microfludic
network. Anal. Chem. 76:742-748.

Lin, R., and M. A. Burns. 2005. Surface-modified polyolefin microfluidic
devices for liquid handling. J. Micromech. Microeng. 15 2156—2162.

Mela, P., A. van den Berg, Y. Fintschenko, E. B. Cummings, B. A. Simmons,
and B. J. Kirby. 2005. The zeta potential of cyclo-olefin polymer microchannels
and its effects on insulative (electrodeless) dielectrophoresis particle trapping
devices. Electrophoresis 26: 1792-1799.

Mokdad, A. H., J. S. Marks, D. F. Stroup, and J. L. Gerberding. 2000. Actual
causes of death in the United States. JAMA 291:1238-1245.

Mori, Y., M. Kitao, N. Tomita, and T. Notomi. 2004. Real-time turbidirnetry of
LAMP reaction for quantifying template DNA. J. Biochem. Biophys. Methods.
59:145-57. '

Mori, Y., K. Nagamine, N. Tomita, and T. Notomi. 2001. Detection of loop—
mediated isothermal amplification reaction by turbidity derived fiom magnesium
pyrophosphate formation. Biochem. Biophys. Res. Commun. 289: 150-4.

Nagarajan, P., and D. Yao. 2009. Rubber-assisted rrricro forming of polymer
thin films. Microsyst. Techno]. 15:251-257.

Nie, Z., and Y. S. Fung. 2008. Microchip capillary electrophoresis for fiontal
analysis of free bilirubin and study of its interaction with human serum albumin.
Electrophoresis 29: 1 924-193 1 .

Notomi, T., H. Okayama, H. Masubuchi, T. Yonekawa, K. Watanabe, N.
Amino, and T. Hase. 2000. Loop-mediated isothermal amplification of DNA.
Nucleic Acids Res. 28:e63.

Oh, K. W., C. Park, K. Namkoong, J. Kim, K. S. Ock, S. Kim, Y. A. Kim, Y.
K. Cho, and C. Ko. 2005. World—to-chip microfluidic interface with built-in
valves for multichamber chip-based PCR assays. Lab Chip 5:845-850.

Olivero, D., and Z. H. Fan. 2008. Chips & tips: lamination of plastic
microfluidic devices, Lab Chip.

Ottesen, E. A., J. W. Hong, S. R. Quake, and J. R. Leadbetter. 2006.
Microfluidic digital PCR enables multigene analysis of individual environmental
bacteria. Science 314: 1464—1467.

168

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

Ren, K., O. Liang, X. Mu, G. Luo, and Y. Wang. 2009. Miniaturized high
throughput detection system for capillary array electrophoresis on chip with
integrated light emitting diode array as addressed ring-shaped light source. Lab
Chip 9:733-736.

Seyrig, G., R. D. Stedtfeld, T. M. Stedtfeld, D. M. Tourlousse, J. M. Tiedje, S.
A. Hashsham, and 2009. Accelerated real-time loop mediated isothermal
amplification with optimized fluorescent dyes and protocol for fast (less than 10
minutes) detection of nucleic acids. . For submission to Appl. Environ. Micrbiol.

Shin, Y. S., K. Cho, S. H. Lim, S. Chung, S. J. Park, C. Chung, D. C. Han,
and J. K. Chang. 2003. PDMS-based micro PCR chip with Parylene coating. J.
Micromech. Microeng. 13:768-774.

Smith, M. C., G. Steimle, S. lvanov, M. Holly, and D. P. Fries. 2007. An
integrated portable hand-held analyser for real-time isothermal nucleic acid
amplification. Anal. Chim. Acta 598:286-294.

Stachowiak, T. B., D. A. Mair, T. G. Holden, L. J. Lee, F. Svec, and M. J.
Jean. 2007. Hydrophilic sm'face modification of cyclic olefin copolymer
microfluidic chips using sequential photograiting. J. Sep. Sci. 30: 1088-1093.

Stedtfeld, R. D., S. W. Baushke, D. M. Tourlousse, S. M. Mller, T. M.
Stedtfeld, E. Gulari, J. M. Tiedje, and S. A. Hasbsham. 2008. Development
and experimental validation of a predictive threshold cycle equation for
quantification of virulence and marker genes by high-throughput nanoliter-
volume PCR on the OpenArray platform. Appl. Environ. Microbiol. 74:3831-8.

Tomita, N., Y. Mori, H. Kanda, and T. Notomi. 2008. Loop-mediated
isothermal amplification (LAMP) of gene sequences and simple visual detection
of products. Nat. Protoc. 3:877-82.

Tsao, C. W., L. Hromada, J. Lin, P. Kumar, and D. L. DeVoe. 2007. Low
temperatrne bonding of PMMA and CDC microfluidic substrates using UV/ozone
surface treatment. Lab Chip 7:499—505.

Wang, F., M. Yang, and M. Burns. 2008. Microfabricated valveless devices for
thermal bioreactions based on diffusion-limited evaporation. Lab Chip 8:88-97.

Warren, L., D. Bryder, I. L. Weissman, and S. R. Quake. 2006. Transcription
factor profiling in individual hematopoietic progenitors by digital RT—PCR. Proc.
Natl. Acad. Sci. U S A. 103:17807-17812.

169

43.

45.

Yang, Y., C. Li, J. Kameoka, K. Lee, and H. G. Craighead. 2005. A polymeric
microchip with integrated tips and in situ polymerized monolith for electrospray
mass spectrometry. Lab Chip 5:869—876.

Yao, D., and R. Kuduva-Raman-Thanumoorthy. 2008. An enlarged process
window for hot embossing. J. Micromech. Microeng. 18:7.

Zhang, C., and D. Xing. 2007. Miniaturized PCR chips for nucleic acid
amplification and analysis: latest advances and firture trends. Nucleic Acids Res.
35:4223-4237.

170

CHAPTER6

ENGINEERING SIGNIFICANCE AND FUTURE PERSPECTIVES

6.1 ENGINEERING SIGNIFICANCE

The potential for microbial diagnostics technologies to impact health care in
developing regions of the world is well recognized. However, there is a large gap in the
use of such technologies due to the unavailability of a low cost genetic screening system
that is point of care. The device developed within this project fills this gap, with the
potential for providing genetic screening at a cost that is well within reach of resource-
lirnited regions. This has been obtained through: 1) establishment of tools for efficient
design and validation of high-throughput assays, 2) development and utilization of a low
cost and user-scalable disposable microfluidic chip 3) integration of a simple and low
cost optical setup tint eliminates the complexity and size requirements associated with
real time detection of multiple assays, and 4) acquisition of expertise for embedded
control of components within a footprint that is truly portable. Overall, the gene analyzer
has a total component cost of approximately $599 (Table 6.1). The use of a portable
sample preperation device (1) to complement the gene analyzer will satisfy the the main
objective of this project for providing a device with the potential to globally impact

human health.

171

Table 6.1 Component costs for Gene Analyzer prototype with iPod Touch and

 

 

Software
Component (numberL Low volume order cost
Optical fibers (16) $10.00
Kapton heater $35.00
Temperature sensor $30.00
Photodiode $64.00
High volume (100 piece) PCB order $10.00
ARM microcontroller $10.00
PWM driver 1C $8.00
Temperature conditioner [C $19.00
4 to 16 line Decoder IC $2.00
LED drivers (16) $5.00
LEDS (16) $11.00
Optical filter $15.00
Ribbon cables, plug-in $7.00
Voltage regulators (5) $20.00
General resistor/capacitors $10.00
Hardware for ARM-iPod connection $15.00
Battery $47.00
Housing (mass production) $2.00
Total Device Cost $320.00
iPod Touch $229.00
GeneZ Software $50.00
Grand Total $599.00

 

Several novel strategies have been implemented toward the development of the
device. In detail, a predictive equation for quantification of stating copies was
experimentally established based on the sequence characteristics of primers and target
organisms as an efficient alternative to generating standard curves. Development of the
gene analyzer required the assembly of components that were not only small, and low
cost, but provided the same level of sensitivity and quantitative capacity of a standard
laboratory based RT-PCR device. The assembly of LEDS and photodiodes has been used
previously for fluorescent detection, however, the implementation of this detection
scheme with optical fibers for single point detection of multiple reactions without moving
parts, along with the integration of a precisely controlled heater for genetic assay based

amplification make the component setup and mechanism for utilization novel. To our

172

knowledge, the strategies used to reproducibly fabricate microfluidic chips including: use
of stereolithography molds along with combined use of shoulders on positive molds and
ABS counter tools to increase embossing process windows, and using microsturcutred
ABS to protect channels of microfluidic chips during enclosure are also novel
developments. One of the more critical aspects in development of Gene Analyzer is the
electronics hidden within. Thus, the acquired expertise and collection of electronic
component hardware such as ICs, LabVIEW embedded module for generating C++ code,
and the library of developed LabVIEW software provides the laboratory with a distinct
advantage in efficient development of future devices.

The advent of a prototype for the gene analyzer brings several more steps that
must be implemented prior to commercialization including: further evaluation of the
prototype using field trials, transfer of developments (gene analyzer and disposable chips)
into manufacturing, improvement of analysis software, and development of after-sale
services. Since the device will be low cost, the disposable and service based aspects will
be critical aspects of commercialization. The process of transferring a prototype into a
manufactured unit will slow availability as pilot builds, manufacturing process validation,
regulatory submissions, and manufacturing scale-up can take 12 to 36 months. Once
completed, the gene analyzer can be complemented with a sample preparation method to
allow traditional laboratory analysis to be decentralized and taken into the field. Not only
will such a device influence human health in clinical settings, but also in environmental,

agricultural, and bio—defense applications.

173

6.2 FUTURE PERSPECTIVES TOWARD FINAerrNG THE GENE ANALYZER PROTOTYPE
A few control related tasks remain before the field ready prototype of the LAMP based
gene analyzer is complete. This includes testing ICs selected for power and voltage
management and obtaining a suitable battery. Three separate ICS have been selected to
convert 12 V DC power to the following: 3.3 V 20 mA power supplies for the ARM
microcontroller (LT3014), 28 V and 700 mA max current for the heater (LT1129), and 5
V 50 mA max to the LED drivers (LT3010-5). For obtaining required voltages, all
, regulators can be programmed by replacing two resistors in the circuit. Pads for the
power regulating devices were placed on the initial design of the PCB for testing.
A second remaining task is to program the ARM microcontroller with code for
simultaneous control of all components, which may require optimization of the code
compiler to reduce memory requirements. Due to the limited amount of memory that
could be accessed using the evaluation kit of the LabVIEW embedded module (only 25%
of 512 KB memory), the original LabVIEW VIs used for bench-top control had to be
stripped to a very basic form. However, code for all tested Vls was generated using
default debugging and compiling parameters, which can be changed to require less space,
but also may run less efficiently.

Development of GeneZ (Objective C-based software) to be installed on the iPod
Touch is the third and final component. A conceptual fiarnework for this software is
under development at Michigan State University with some help from an Apple software
develOper (Figure 6.1). The software is rather simple but extremely versatile. As
illustrated in Figure 6.1, user specified firnctions include: adding any test (corresponding

to a new chip), or changing the name of the gene (the gene list window is not shown in

174

detail but it is possible to change the identity of the gene), options for an isothermal or
temperature cycling program, visualizing the results, calculating averages of identified
wells and reporting the final result of the overall test. Because users can pick/download
the list of genes for any given assay, the software is expected to be an ideal companion to
the gene analyzer. When completed, this software will be distributed through the Apple

Store with revisions provided through the same channel as necessary.

Home Screen (1) Select a Test (2) Test Settings (3)

 

  

 

         

Rcsu Its

Amplification Curves (6)

   

Test Settings continued (4 and 5)

Figure 6.1 Conceptual representation of the screens for GeneZ software to be used to a)
control the Gene Analyzer, b) visualize the amplification curves and c) report the final
results.

175

6.3 REFERENCES
l. Casper, E.T., Patterson, S.S., Bhanushali, P., Farmer, A., Smith, M., Fries,

D.P., Paul, J.H. A handheld NASBA analyzer for the field detection and
quantification of Karenia brevis. 2007. Harmful Algae. 6: 112.

176

3 1293 03063 1620

l

llllllllllll