i 
 
 
 
 
 
 
 
Assessing dose-response of antibiotics and monitoring degradation of RNA aptamer 
biosensor on microfluidic devices 
 
by 
 
Jing Dai 
 
 
 
 
A dissertation submitted to the Graduate Faculty of 
Auburn University 
in partial fulfillment of the  
requirements for the Degree of 
Doctor of Philosophy 
 
Auburn, Alabama 
December 14, 2013 
 
 
 
Keywords: Microfluidics, Antibiotics, Dose-response,  
RNA aptamer biosensor, Degradation characterization   
 
 
Copyright 2013 by Jing Dai 
 
 
Approved by 
 
Jong Wook Hong, Chair, Associate Professor, Materials Research and Education Center, 
Mechanical Engineering 
Jeffery W. Fergus, Professor, Materials Research and Education Center, Mechanical Engineering 
Zhongyang Cheng, Professor, Materials Research and Education Center, Mechanical 
Engineering 
Sang-Jin Suh, Associate Professor, Biological Sciences 
 
 
 
ii 
 
 
 
 
 
 
Abstract 
 
In the last century, the technologies developed to miniaturize transistors and manufacture 
microprocessors have enabled the miniaturization and integration of tools in biology, chemistry, 
biotechnology and medical fields. These tools have low reagent consumption, display high levels 
of integration, parallelization and automation, can carry out fast reactions, and are portable. The 
miniaturized and integrated microfluidic platforms are capable of integrating multiple analysis 
steps: sample preparation, reaction, and detection onto a single chip, termed as ?Lab-on-chip? or 
?TAS (micro total analysis system). This technology has altered and influenced the way various 
questions are addressed in biology, chemistry, and biotechnology.  
In this dissertation, we investigated the concentration- and time-dependent response of 
cell (bacteria) or molecule (RNA aptamer-based biosensor) to reagents (antibiotics or degrading 
agents) by using microfluidic systems. We obtained useful information for evaluating the 
attenuated inhibitory effect of antibiotics by bacteria?s resistance and differentiating degrading 
agents through monitoring the degrading profiles of RNA biosensor. Two microfluidic systems 
were used in this study. The first microfluidic system has multiplex reactors, and the second 
microfluidic system integrates a concentration gradient generator, reagent mixing and reaction 
sections. Both systems enable simultaneous, parallel and independent reactions, requiring 
nanoliter amount of reagents, unlike the conventional test tube or microtiter method. Our 
microfluidic tools are potential alternative to replace conventional batch culture methods.  In 
addition, they have a great potential for screening drug molecules, determining drug?s potency as 
iii 
 
well as replacing conventional monitoring methods in transcriptomics. So, these devices are 
highly potential to benefit the process of lead identification and optimization during drug 
discovery as well as promote the transcriptomic researches. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
iv 
 
 
 
 
 
 
Acknowledgement 
 
?To strive, to seek, to find, and not to yield.? 
- Ulysses by Afred, Lord Tennyson, 1842 
 
 
 
At first, I would like to sincerely thank my advisor Dr. Jong Wook Hong for his guidance 
and support. I am very grateful for Dr. Hong?s help to open the door to an amazing research area 
for me. With Dr. Hong?s mentoring during my Ph.D study, I developed both critical and logical 
thinking which are essential for academic success.   
I am also grateful to my committee members, Dr. Jeffery W. Fergus, Dr. Zhongyang 
Cheng, and Dr. Sang-Jin Suh for their valuable suggestions and expertise. I want to thank Dr. 
Jacek Wower for being an external reader. 
I want to thank Dr. Jae Young Yun, Dr. Sachin Jambovane, and Dr. Morgan Hamon for 
their valuable suggestions and generous help during my research, specially thank Dr. Hamon for 
providing valuable suggestions during preparation of this dissertation. I want to thank Ms. Hye 
Young Sim, Mr. Hoon Suk Rho, Ms. Kirn Cramer, Ms. Ting Chen, Ms. Lauren Bradley, and Mr. 
Austin Adamson for their collegiality, friendship and help. 
I want to thank Dr. Lingzhao Kong, a great friend and faithful brother, Dr. Min Shen, Dr. 
Jiawei Zhang, Dr. Qing Dai, Dr. Xiaoyun Yang, Dr. Dan Liu, Dr. Lin Zhang, Dr. Yingjia Liu, 
Ms. ZhiZhi Shen, Mr. Honglong Wang, Ms. Jing Zou, Ms. Wei Wang, Dr. Yu Zhao, Mr. Yang 
Xu, Dr. Shiyu Wang for their support and help. 
v 
 
I also want to express thanks to my friends and relatives: Mr. Le Mu, Ms. Lei Lei, Ms. 
Qingjun Miao, Mr. Yunfei Yan, Mr. Wenxin Shi, Mr. Jiapei He, Ms. Miao Li, Mr. Xinjun He, 
Ms. Ping Shi, Mr. Mingang Shi, and Ms. Shudi Shi.    
Finally, I have to thank my parents, Qin Shi and Yongming Dai, the most important and 
beloved persons in my life. They devoted their unconditional love to me. Their love motivated 
me an endless strength to overcome the hardest times in my life.    
This research was supported by U.S. National Institute of Health, USDA Nanotechnology 
Program, and National Science Foundation. We also acknowledge 21st Century Frontier R&D 
Program of Microbial Genomics and Applications Center, Republic of Korea. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
vi 
 
 
 
 
 
 
Table of Contents 
 
Abstract ................................................................................................................................... ii 
Acknowledgments ................................................................................................................... iv 
List of Tables ........................................................................................................................ viii 
List of Figures ......................................................................................................................... ix 
List of Abbreviations............................................................................................................. xiii 
Chapter 1 INTRODUCTION   ................................................................................................. 1 
1.1 Introduction ............................................................................................................ 1 
1.2 Background and Motivation  ................................................................................... 1 
1.3 Objective and organization of dissertation  ............................................................ 18 
Chapter 2 LITERATURE REVIEW  ...................................................................................... 21 
2.1 Introduction .......................................................................................................... 21 
2.2 Conventional methods to assess activity of antibiotics........................................... 21 
2.3 Microfluidic methods to assess activity of antibiotics ............................................ 23 
2.4 Conventional methods for monitoring RNA degradation ....................................... 27 
2.5 Microfluidic methods for monitoring RNA degradation ........................................ 28 
Chapter 3 CHARTING MICROBIAL PHENOTYPES IN MULTIPLEX NANOLITER BATCH   
                 BIOREACTORS  .................................................................................................. 29 
 
3.1 Introduction .......................................................................................................... 29 
3.2 Objective .............................................................................................................. 31 
3.3 Materials and Methods .......................................................................................... 32 
vii 
 
3.4 Result and Discussion ........................................................................................... 40 
3.5 Conclusion ............................................................................................................ 50 
Chapter 4 DETERMINATION OF EC50 OF BACTERICIDAL ANTIBIOTICS AT TIME- 
                COURSE ON A MICROFLUIDIC CHIP ............................................................... 51 
 
4.1 Introduction .......................................................................................................... 51 
4.2 Objective .............................................................................................................. 53 
4.3 Materials and Methods .......................................................................................... 53 
4.4 Result and Discussion ........................................................................................... 56 
4.5 Conclusion ............................................................................................................ 66 
Chapter 5 MONITORING OF RNA DEGRADING AGENTS WITH A NOVEL APTAMER- 
                BASED BIOSENSOR ............................................................................................ 67 
 
5.1 Introduction .......................................................................................................... 67 
5.2 Objective .............................................................................................................. 68 
5.3 Materials and Methods .......................................................................................... 69 
5.4 Result and Discussion ........................................................................................... 72 
5.5 Conclusion ............................................................................................................ 78 
Chapter 6 SUMMARY AND FUTURE PERSPECTIVES ..................................................... 79 
6.1 Summary .............................................................................................................. 79 
6.2 Future perspectives ............................................................................................... 80 
Reference   ............................................................................................................................. 82 
Appendix A   .........................................................................................................................101 
Appendix B   .........................................................................................................................107 
Appendix C ...........................................................................................................................112 
 
 
viii 
 
 
 
 
 
 
List of Tables 
 
Table 1.1 Mechanism of action of antibiotic ............................................................................ 4 
Table 1.2 Sequence of valve operation to create a sequential fluid motion in a micromixer  ... 17 
Table 4.1 EC50 (?g/ml) of gentamicin and ciprofloxacin ........................................................ 63 
 
 
 
 
 
 
 
 
 
 
 
 
ix 
 
 
 
 
 
 
List of Figures 
 
Figure 1.1 Scanning electron micrograph of Escherichia coli. ................................................. 1 
 
Figure 1.2 A typical growth curve of bacteria in a batch culture.  ............................................ 2 
 
Figure 1.3 Drug discovery and development process.  ............................................................. 6 
 
Figure 1.4 A typical dose-response curve.  .............................................................................. 7 
 
Figure 1.5 A single-stranded RNA.  ......................................................................................... 8 
 
Figure 1.6 The sequence of a RNA aptamer.  ........................................................................... 9 
 
Figure 1.7 Basis of microfluidics. (a) Turbulent and laminar flow. (b) Wetting on hydrophilic 
surface and hydrophobic surface.  .......................................................................................... 13 
 
Figure 1.8 Structure of polydimethylsiloxane (PDMS). ......................................................... 14 
 
Figure 1.9 Process flow of multiplayer soft lithography. ........................................................ 15 
 
Figure 1.10 A two-layer PDMS valve. An elastomeric membrane is formed where a control 
channels lies orthogonal to and below fluidic channel. When pressurizing the control channel, 
the membrane will deflect upward, thus cutting off the flow in fluidic channel. ...................... 16 
 
Figure 1.11 (a) A peristaltic pump using three valves in a series. (b) A circular micromixer using 
a peristaltic pump. .................................................................................................................. 17 
 
Figure 2.1 A disk diffusion test with an isolate of Escherichia coli from a urine culture. The 
diameters of all zones of inhibition are measured. .................................................................. 22 
 
Figure 2.2 A broth microdilution susceptibility plate containing 96 wells. ............................. 23 
 
Figure 2.3 The MicroScan WalkAway system. ...................................................................... 23 
 
Figure 2.4 Design and picture of the microfluidic device. (a) Schematic representation of the 
functional circuit used for long-term bacterial colony monitoring and antibiotic testing. (b) 
Enlarged functional microstructure of the microfluidic device corresponding to the dotted-line 
x 
 
square in (a). (c) Optical image of the actual device. A one cent coin was employed to show the 
size of the device. (d) SEM image of the functional microstructure. ....................................... 24 
Figure 2.5 The microfluidic device for assessing antibiotic susceptibility of biofilms. (a) 
Schematic diagram and optical image of the microfluidic device. The microfluidic device 
consists of gradient generator and main detection microchannel. (b) A 3-D plot of the 
fluorescence image obtained from 0 mm to 8 mm position in the detection microchannel. (c) 
Profiles of cross section at each representative position. A flow rate of 0.1 ml min-1 is used in the 
experiment. ............................................................................................................................ 25 
 
Figure 2.6 Schematic drawing illustrates the formation of droplets containing bacteria, viability 
indicator, and an antibiotic at varying concentrations. (a) A capillary cartridge loaded with drug 
trails and gas space plug. (b) A drug trial cartridge is connected to a microfluidic channel, and 
then drug trials are flowed to merge with viability indicator and bacterial solution. Thus, droplets 
containing various drug trials are generated at T-junction. ...................................................... 26 
 
Figure 2.7 A schematic drawing demonstrates the operation of the microfluidic system. (a) Four 
T-junctions generate packets of microdroplets and merged them to create microdroplets with a 
defined composition. The sequence of microdroplets formed in the device was incubated off-chip. 
After incubation, the microdroplets were loaded into a microchannel and the intensity of 
bacterial viability indicator was detected. (b) A linear relationship of volumes of droplets and 
valve opening time (?open) is obtained for fluids with different viscosities (1, 2, 3, 30, 100 mPas) 
in two capillaries (5m and 10m in length). .............................................................................. 27 
 
Figure 2.8 Degraded and intact total RNA were run beside RNA markers on a 1.5% denaturing 
agarose gel. The 18S and 28S ribosomal RNA bands are clearly visible in the intact RNA sample 
while no bands are visible in degraded RNA. ......................................................................... 28 
 
Figure 3.1 Flow chart of chip fabrication using multiplayer soft lithography.......................... 34 
 
Figure 3.2 Experimental setup for chip control and imaging. ................................................. 38 
Figure 3.3 Operation of the device. (a) Schematic drawing of the entrapment of cells into three 
replicate cultivation reactors (I). The middle control channel C2 is closed to load cell suspension 
(shown in yellow) and culture media (shown in light blue) into the reactors (II). Cells and the 
medium are mixed by opening C1 and closing both C1 and C3 (III). The cell cultures are 
sequestered from the fluid channels and the cells start to grow (IV). Simultaneous triplicate 
experiments for testing a culture condition can be performed in a single run. Note that control 
channels and flow channels are at different layers (see Figure A1 in Appendix A for details) (b) 
Time-lapse micrographs showing the mixing of green and red dyes in M9 medium at 37 ?C. The 
process of diffusion was initiated by opening C2, and was almost completed in 4 min. Scale bar, 
200 ?m. (c) A graph representing the time profiles of mixing efficiency of 2 dyes dissolved in 
M9 minimal medium, LB complex medium, and 5 % PEG solution at 22 ?C or 37 ?C. Error bar 
denotes the standard error of the mean from three replicate reactors. ...................................... 41 
Figure 3.4 Loading of a uniform number of particles into the reactors. (a) Fluorescent beads (2.0 
?m) are loaded into each of the 12 reactors by opening C3 control channel, followed by the 
closing of C3 channel, which retains the beads. Arrows denote the direction of bead suspension 
xi 
 
flow. Scale bar, 200 ?m. (b) Number of beads loaded into each of the reactors. Error bar denotes 
the standard error of the mean from nine separate experiments. .............................................. 43 
 
Figure 3.5 Effect of humidity controls on volume changes of the medium inside the reactors. 
The graph represents the volume changes of the initial medium over a period of time (24 h 
duration) based on the presence and/or absence of humidity control and/or anti-evaporation 
channels: in the absence of humidified incubator (denoted with a black line), in the presence of 
operation of anti-evaporation channels (green line), in the presence of a humidified incubator 
(red line), and in the presence of both humidified incubator and the operation of anti-evaporation 
channels (blue line). Error bar denotes the standard error of the mean, obtained from three 
replicate reactors. ................................................................................................................... 45 
 
Figure 3.6 Growth of E. coli cells during the batch culture in a nanoliter reactor. (a) Micrographs 
showing time-lapse cell growth on the LB complex medium in a nanoliter reactor. Scale bar, 10 
?m. (b) A graph showing the comparison of cell growth on LB medium in the nanoliter reactor 
(represented by ?) with that in a 14-mL test tube containing 4 mL of LB medium (?). The cell 
numbers in the reactors were counted every 2 h after inoculation and were normalized to the 
initial number of cells. The cell density in a tube culture was measured in OD600. Values in Y 
axes are in log scale. Error bar denotes the standard error of the mean from 3 replicate reactors or 
test tubes. (c) A graph showing the diauxic growth on M9 minimal medium with glucose (0.04 % 
wt/vol) and lactose (0.2 %) in a nanoliter reactor. ................................................................... 47 
 
Figure 3.7 Microbiological assay for antibiotics using a nanoliter reactor and a test tube. The 
graph shows the comparison of antibiotic effects of gentamicin on the growth of P. aeruginosa 
harboring the EGFP plasmid, following 24 h of culture in nanoliter reactors (represented using 
grey bars, on the left side), with that in the 14-mL test tubes (represented using hatched bars, on 
the right side). The X axis denotes concentration of gentamicin in log scale, and the Y axis 
denotes the fluorescence intensity (F.I.) normalized to the initial intensity of the cell culture, 
immediately following inoculation. The error bar represents the standard error of the mean from 
three replicate reactors or test tubes. ....................................................................................... 49 
 
Figure 4.1 Microfluidic chip with 14 processors and operation of one processor. (a) Design of 
microfluidic chip with 14 processors where bacterial cells grow with 14 concentrations of 
antibiotics. (b) Operation of one processor. Bacterial cells are introduced into half of reactors, 
then, antibiotics and dilution buffer are introduced into metering channels. After introducing 
metered antibiotics to reactors, cells and antibiotics are mixed by three mixing valves. .......... 58 
 
Figure 4.2 Introduction and growth of PT5-EGFP cells in reactors. (a) Cells are loaded into 14 
reactors through ?cell in? inlet, and cells are trapped in reactors by closing surrounding valves. (b) 
Number of cells in each reactor after introduction. Data are represented as means ? SD of three 
independent experiments. There was no significant difference in number of cells in 14 reactors. 
(c) The fluorescence intensity of cells after 24 h cultivation. Data are represented as means ? SD 
of three independent experiments. There was no significant difference in cell growth across 14 
reactors. ................................................................................................................................. 60 
 
xii 
 
Figure 4.3 Growth inhibition profiles of two bactericidal antibiotics and EC50 values for on chip 
and test tube cultures. (a) Gentamicin, and (b) Ciprofloxacin. Bacterial cells were treated with 14 
concentrations of antibiotics. Cell growth was normalized by the fluorescence intensity. EC50 
values in (c) and (d) were obtained from on chip and test tube cultures. Data are presented as 
means ? SD of three independent experiments. Asterisks in (a) and (b) indicate statistical 
significance compared to test tube culture, p<0.05. ................................................................. 63 
 
Figure 5.1 Design and working principle of fluorescent RNA aptamer biosensor. .................. 72 
 
Figure 5.2 Concentration gradient formation of degrading agents on a microfluidic chip. (a) 
Microfluidic chip with concentration gradient generators, (b) Step-by-step process of 
concentration gradation formation. ................................................................................................. 75 
 
Figure 5.3 Characterization of the concentration- and time-dependent degradation of the 
biosensor. Degradation profile (top) and scanned images (bottom) of the biosensor by different 
concentration of (a,b) lead acetate, (c,d) RNase T1, and (e,f) RNase A. Data are presented as 
means ? SD of three independent experiments... .................................................................... 77  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
xiii 
 
 
 
 
 
 
List of Abbreviations 
 
RNA  Ribonucleic acid 
RNase  Ribonuclease 
DNA  Deoxyribonucleic acid 
PDMS  Polydimethlysiloxane 
SAV  Surface area to volume ratio 
EPS  Extracellular polymeric substance 
CFU  Colony forming unit 
IC50  Half maximal inhibitory concentration 
DFHBI 3,5-difluoro-4-hydroxybenzylidene imidazoline 
 
 
 
 
1 
 
Chapter 1 
INTRODUCTION 
1 .1 Introduction 
In this chapter, I first review the basis of bacteria, bacterial growth phenotype, pathogenic 
bacteria, antibiotics, dose-response analysis in drug discovery and development, RNA and 
RNase, and aptamer. The fundamental concepts of microfluidics, fabrication method, and key 
components are described. Finally, the objectives and outline of this dissertation are presented. 
1.2 Background and Motivation 
1.2.1 Bacteria 
Bacteria have been living on our planet for about 3.8 billions of years [1]. They are 
prokaryotic microorganisms with micrometer sizes and have a wide range of shapes including 
spheres, rods and spirals [2]. Escherichia coli, for example, is a rod-shaped bacterium (Figure 
1.1). Bacteria are highly diverse and widely spread on Earth including inside of human bodies. 
As of 2011, it is estimated that there here are >10 million species [3]  and approximately 5?1030 
bacteria living on Earth [4].  
 
Figure 1.1 Scanning electron micrograph of Escherichia coli [5]. 
2 
 
Bacterial growth phenotypes 
Phenotypes are observable characteristics of an organism. They are shaped by gene 
expression and modified by environments or by both [6]. Bacterial phenotypes include any 
bacterial cell property, such as, shape, color of colonies, formation of biofilms or spores, 
mechanisms of cell-to-cell interactions, as well as growth patterns [7]. Bacterial growth 
phenotypes define whether or how fast bacteria grow under particular conditions. Typical 
bacterial growth in batch cultures under laboratory conditions experiences four different phases: 
lag phase, log phase, stationary phase, and death phase, as shown in Figure 1.2. Growth 
phenotypes in batch cultures can be easily observed and quantified without an expensive and 
sophisticated technology. In addition, growth curves can provide information about nutrient 
utilization profiles and growth kinetics required for genetic analysis. This information is also 
used to measure the impact of environmental and genetic perturbations [7, 8]. 
 
Figure 1.2 A typical growth curve of bacteria in a batch culture [9]. 
Pathogenic bacteria 
Most bacteria are harmless or beneficial; only a small fraction of them are capable of 
causing disease in plants, animals and humans. Notable pathogenic bacteria, such as 
3 
 
Streptococcus spp., Pseudomonas spp., Mycobacteria spp., Escherichia coli, and Salmonella 
servovars, can cause pneumonia (by Streptococcus and Pseudomonas), tuberculosis (by 
Mycobacteria), and foodborne illnesses (by Escherichia coli and Salmonella). Infectious diseases 
caused by pathogenic bacteria have played a significant role in shaping human history and 
development. For example, Yersinia pestis, a bacterium carried by rat flea, has been responsible 
for the Black Death that decimated almost 50 % of the European population during the Middle 
Ages [10].    
 
1.2.2 Antibiotics 
In 1877, Pasteur and Robert Koch observed that an airborne bacillus could inhibit the 
growth of Bacillus anthracis. This phenomenon, later called antibiosis by Pasteur?s pupil Paul 
Vuillemin, is a cornerstone in the discovery of antibacterial agents. The most celebrated 
antibiosis effect led to the discovery of penicillin by Alexander Fleming in 1929 who observed 
that the growth of staphylococci colonies was inhibited by the mold Penicillium notatum [11]. In 
1939, the success of purification and stabilization of penicillin enabled its therapeutic potential. 
In the later stages of World War II, the mass-produced penicillin saved the lives of millions of 
wounded soldiers who would otherwise have succumbed to bacterial infections. In the following 
decades, enormous efforts were devoted to the discovery of new antibacterial agents, which 
renamed antibiotics by the American microbiologist Selman Waksman in 1942. Since then, 
thousands of antibiotic molecules were screened and isolated. Nowadays, over 100 different 
antibiotics are commercially available. Antibiotics are classified as bactericidal if they kill 
bacteria or bacteriostatic if they prevent bacterial growth [12]. In addition, most antibiotics used 
to treat bacterial infections can be classified according to their principal mechanism of action 
4 
 
[13]. Five major classes of mechanism of action are categorized: (1) interference with cell wall 
synthesis, (2) inhibition of protein synthesis, (3) interference with nucleic acid synthesis, (4) 
disruption of bacterial membrane structure, and (5) inhibition of metabolic pathway as 
summarized in the Table 1.1.  
Table 1.1 Mechanism of action of antibiotics (from reference [14]) 
  Mechanism of action  Antibiotics 
1  Interference with cell wall synthesis  ?-Lactam, Glycopeptide 
2  Inhibition of protein synthesis  Macrolide, Chloramphenicol, Clindamycin, Quinupristim-dalfopristin, Linezolid, Aminoglycoside, Tetracycline, Mupirocin 
3  Interference with nucleic acid synthesis  Fluoroquinonlne, Rifampin 
4  Disruption of bacterial membrane structure  Polymyxin, Daptomycin 
5  Inhibition of metabolic pathway  Sulfonamide, Folic acid analogue 
 
1.2.3 Bacterial resistance to antibiotics 
The inappropriate and excessive use of antibiotics has led to the emergence of pathogenic 
bacteria that are resistant to currently available antibiotics [14-17]. Bacteria achieve active 
resistance to antibiotics through three major mechanisms: (1) efflux of antibiotics from cells, (2) 
enzymatic degradation and modification of antibiotics, and (3) target site alteration [13]. First, 
bacteria have efflux pumps that extrude the antibiotic from cells before the drug reaches its target 
sites and exerts its effect. Second, bacteria may acquire genes encoding enzymes that are able to 
destroy the antibacterial agent before it can have an effect. Third, bacteria may acquire several 
genes to reprogram biosynthetic pathways which produce altered bacterial cell walls that no 
longer contain the binding site for antibacterial agents or even through mutation that limits the 
access of antibacterial agents to the intracellular target sites. Some bacteria species have passive 
resistance to antibiotics. For example, Gram-negative bacteria have outer membrane to serve as a 
5 
 
significant barrier to penetration of antibiotics, restricting the rate of penetration of molecules 
[13].         
 
1.2.4 Dose-response analysis in drug discovery and development  
Generating a new drug is an expensive and time-consuming process. On average, 10~15 
years and $ 800 million to 1 billion are required. This process includes thousands of failures: for 
every 5,000~10,000 compounds that enter into the research and development pipeline, ultimately 
only one receives governmental approval [18].   
The process of generating new drugs consists of two main stages: drug discovery and 
drug development, as shown in the Figure 1.3. The drug discovery stage includes target 
selection, lead identification and optimization, and preclinical studies, while the drug 
development stage includes clinical trials, manufacturing and product management. The first step 
in drug discovery is to identify a drug target that can interact with a drug candidate. Once 
successful compounds (hits) are identified, the molecules, now called ?lead?, will be optimized. 
The lead will be tested in progressively more complex systems including cells and model 
animals. Only a few candidates out of thousands can enter into the drug development stage. In 
the process of lead optimization, establishing a dose-response relationship is of critical important 
step. It involves a so-called secondary screen. In the secondary screen, a range of drug 
concentration prepared by serial dilution is tested to assess the dose dependence of the assay?s 
readout. It is essential to quantitatively characterize the inhibitory potency of potential drug 
candidates to determine the best candidate [19].   
6 
 
 
Figure 1.3 Drug discovery and development process [20]. 
In a medical definition, the dose-response relationship describes the pattern of 
physiological response to varied dosage (as of a drug or radiation) after certain exposure time. 
Response to dose follows a sigmodial curve increasing rapidly over a relative small change in 
dose, and eventually reaching a plateau level. Figure 1.4 shows a typical dose-response curve. 
EC50, half maximal effective concentration referring to the concentration of a compound where 
50% of its maximal effect is observed, can be determined from a dose-response curve. It is 
commonly used as a measure of agonist drug?s potency [21]. EC50 value can be determined 
through curve?fitting of the four-parameter nonlinear-logistic-regression model based on 
obtained inhibition data. The four-parameter model is shown below: 
null = nullnullnullnull + nullnullnullnullnullnullnullnullnullnullnullnullnull(nullnullnull(nullnullnullnull)null[null])null                                                                               (1.1) 
7 
 
where I represents inhibiting potency (%), IMax and IMin is the maximum and minimum inhibiting 
potency, [I] represents concentration of inhibitor, and h is hill slope. 
 
Figure 1.4 A typical dose-response curve. 
 
1.2.5 RNA and RNase 
Ribonucleic acid (RNA) is a polymeric molecule made up of one or more kinds of 
nucleotides. A strand of RNA is a chain with a ribonucleotide at each chain link. Each 
ribonucleotide is made up of a base (adenine (A), cytosine (C), guanine (G), and uracil (U)), a 
ribose sugar, and a phosphate [22]. Figure 1.5 shows a diagram of a single-stranded RNA. RNA 
plays a central role in the information transfer from DNA to proteins, known as the "Central 
Dogma" of molecular biology [23]. There are 4 main classes of RNA species: (1) Messenger 
RNA (mRNA), (2) Transfer RNA (tRNA), (3) Ribosomal RNA (rRNA), and (4) Regulatory 
RNA. 
8 
 
 
Figure 1.5 A single-stranded RNA. 
 
Ribonucleases (RNases) are small and compact proteins that cleave phosphodiester bonds 
that link ribonucleotides. They play important roles in catalyzing degradation of RNA during 
nucleic acid metabolism. RNases are found in both prokaryotic and eukaryotic cells. They can be 
divided into 2 classes: (1) endoribonucleases and (2) exoribonucleases [22]. RNase 
contamination in laboratory selections will compromise results of RNA-based experiments [24, 
25]. Therefore, RNase contamination is of great concern for researchers at both academic 
research laboratories and biotechnology corporations.  
 
 
9 
 
1.2.6 Aptamer  
Aptamers are short synthetic oligonucleic acid or peptide molecules that display high 
selective affinity to specific targets such as small molecules, proteins, nucleic acids, metal ions, 
viruses and cells [26]. Figure 1.6 shows the secondary structure of a RNA aptamer. Besides 
comparable target binding affinity/selectivity to antibodies, aptamers also offer advantages such 
as easier engineering and synthesis, lower batch-to-batch variability, better thermal stability, 
smaller size, lower immunogenicity, and more versatile chemistry among others [27]. Aptamers 
have become increasingly important molecular tools for diagnostics and therapeutics [28-30]. 
Aptamer-based biosensors have been applied for accurate and rapid detection of a diverse set 
target proteins [31], small molecules [32], metal ions [33], and many others. 
 
Figure 1.6 The sequence of a RNA aptamer [34]. 
 
1.2.7 Microfluidics 
Microfluidics is a interdisciplinary field of physics, chemistry, biotechnology, 
engineering and microtechnology. Microfluidics is using a small platform consisting of channel 
systems with dimensions of 10~100 micrometers to precisely control and manipulate small (10-9 
to 10-18 liters) amounts of fluids that are dominated by surface tension and laminar effects [35]. 
10 
 
Microfluidics offers many advantages that make it a great potential for chemistry and 
biotechnology [20, 36]. The micrometer-scale channels facilitate the handling nanoliter to 
femtoliter volumes and significantly reduce sample and reagent consumption. Moreover, 
microfluidics offers many significant improvements over traditional methods with respect to fast 
speed of analysis, precise temporal- and spatial- control of environmental conditions, capability 
of manipulating and detecting single cell and single molecule at a high resolution and sensitivity, 
and a simultaneous detection and capability to handle parallel and high-throughput analyses [37-
40]. However, microfluidics is still in its infancy. Therefore, a great amount of work is required 
to fully demonstrate its application in fields other than academic researches [35]. 
Microfluidic systems originated from the development of microanalytical methods, 
biodefense, molecular biology, and microelectronics [35]. The integration of multiple analysis 
steps (sample preparation, reaction, and detection) onto a microfluidic system suitable for the 
advanced applications in the field of biology, biochemistry, chemistry, and pharmaceutical 
science [41]. Highly developed microfluidic systems are used for manipulating stem cell cultures 
[42, 43], monitoring bacterial growth at different conditions [44-46], detection of pathogenic 
bacteria [47], separation and manipulation of single cells [48] or single molecules [49], 
extraction and amplification of DNA [50, 51], digital PCR [52], protein crystallization [53, 54], 
determination of enzyme kinetic parameters [55], dose-response analysis [56, 57], and high-
throughput drug screening [58, 59].  
The earliest microfluidic systems were made of silicon and glass. Because silicon is 
opaque to visible and ultraviolet light, these systems were not optimal when using conventional 
optical detection methods. Neither glass nor silicon has the property of gas permeability required 
to grow live mammalian cells. In addition, it is difficult to fabricate microfluidic components 
11 
 
such as pumps and valves in rigid materials. Nowadays, a polymer-polydimethysiloxane (PDMS) 
has been widely used to fabricate microfluidic devices with its superior characteristics, such as 
optical transparence, softness, and biocompatibility [35]. The unique properties of PDMS make 
multilayer soft lithography technique possible to fabricate microfluidic components such as 
pneumatic valves, peristaltic pumps, and mixers [60]. 
 
1.2.8 Basic concepts of microfluidics 
Surface area to volume ratio 
One characteristic of microfluidic device is the high surface area to volume ratio (SAV). 
SAV increases several orders of magnitude when dimension decreases from macroscale to 
microscale. Large SAV can make surface forces (such as surface tension) the dominant forces to 
drive flow by capillary effects, enhance heat and mass transfer, and create a large free surface for 
macromolecule absorption [61]. However, this unique feature elevates the rate of evaporation 
and presents a challenge for maintaining cell culture [62].  
 
Laminar flow 
Fluid flow is categorized into two flow regimes: laminar and turbulent (Figure 1.7a). 
Laminar flow is a smooth and constant fluid flow where the motion of the particles of fluid is 
very orderly with all particles moving in straight lines parallel to the pipe walls. Turbulent flow 
is the fluid flow which undergoes irregular fluctuations. The value of Reynolds number (Re) is 
used to determine the type of fluid flow. Reynolds number is the ratio of inertial forces to 
viscous forces. Thus, at low Reynolds number (Re<2300), laminar flow dominates. In 
microfluidic systems, Re is typically smaller than 100 and flow is considered to be laminar [63].  
12 
 
Mixing 
 Mixing is a result of molecular diffusion. At macroscale, mixing is generally achieved by 
turbulent flow which segregates fluid into small domains, causing an increase in contact surface 
and a decrease in diffusion path. In microfluidic systems, the absence of turbulent flow makes 
diffusion difficult to occur, leading to slow mixing. To overcome this problem, two groups of 
micromixers, passive and active mixers, have been described. Passive micromixers contain no 
moving parts and require no energy input except for the pressure that drives the fluid flow at a 
constant rate. Mixing in passive micromixers relies mainly on chaotic advection effects realized 
by manipulating the laminar flow within the microchannels or by enhancing the molecular 
diffusion through increasing the contact area and contact time between the different mixing 
reagents. The examples of passive micromixers are T- or Y- shaped micromixers and chaotic 
advection micromixers [63, 64]. Active micromixers use external energy source to stir or agitate 
the fluid flow. Active mixers use many techniques including acoustic/ultrasonic, 
dielectrophoretic and electrokinetic techniques to enhance mixing performance [63, 64].  
 
Wetting 
Wetting is an interfacial phenomenon describing how a liquid maintains contact with a 
solid surface. It occurs at three interfaces: solid/gas, liquid/solid and liquid/gas. Between each of 
these interfaces, there is an associated surface energy ?sg, ?sl, and ?lg, respectively. Young?s 
equation: nullnullnull = nullnullnull +nullnullnullnullnullnullnull, relates these three surface energy and contact angle ? which 
determines hydrophilicity (0?<?<90?) or hydrophobicity (90???<180?) of the surface (Figure 
1.7b). Wetting effect is crucial at microscale as the surface forces become dominant over body 
forces. It plays an important role in digital and droplet microfluidics [65].  
13 
 
 
Figure 1.7 Basis of microfluidics. (a) Turbulent and laminar flow. (b) Wetting on hydrophilic surface and 
hydrophobic surface [66]. 
 
1.2.9 Fabrication of microfluidic systems 
In order to fulfill the needs of different disciplines, a microfluidic system should employ 
a series of functional components for introducing reagents, moving fluids inside channels, 
mixing fluids, and integrating various other devices. The development of fabricating prototype 
device by soft lithography [67] and the development of fabricating pneumatic valves and mixers 
[60] on the basis of multilayer soft lithography techniques, have enabled the design of 
complicated devices and opened new areas of applications. 
 
Soft lithography 
Soft lithography was first developed by Whitesides and colleagues [68] as a cheaper and 
more biocompatible alternative to photolithography. This method refers to a collection of 
techniques for creating microstructures and nanostructures based on printing, moulding and 
embossing. This technique is termed ?soft lithography? since it is based on using a patterned 
14 
 
elastomeric polymer as a mask, stamp or mould, to pattern ?soft materials? (polymers, gels and 
organic monolayers). Soft lithography techniques use polydimethylsiloxane (PDMS) (Figure 1.8) 
as the major component. PDMS is a silicon rubber (Momentive Performance Materials RTV 615) 
consisting of two components: A and B. Part ?A? contains polydimethlysiloxane bearing vinyl 
groups and a platinum catalyst; Part ?B? contains a cross-linker with silicon hydride (Si-H) 
groups to form a covalent bond with vinyl groups of  part ?A? after applying external thermal 
energy. The excellent properties of PDMS make it become a widely used material for the 
application of soft lithography in microbiology and biology; it is durable, biocompatible, 
permeable to gases and only moderately permeable to water, optically transparent, flexible, 
unreactive, and suitable for chemical surface treatment [69]. The prepolymer of PDMS is 
commercially available, inexpensive, and easy to prepare.    
 
Figure 1.8 Structure of polydimethylsiloxane (PDMS). 
 
Multilayer soft lithography 
Multilayer soft lithography was developed by Quake?s group, this technique uses soft 
lithography to fabricate patterned elastomer layers that are then bonded together [60]. Figure 1.9 
shows the process flow of multiplayer soft lithography. The two component silicone rubber 
(PDMS) is used to fabricate patterned layers. The composition of each layer is designed to 
contain an excess of specific components (bottom layer has an excess of A, while the upper layer 
has an excess of B), this composition difference makes the reactive molecule remain at the 
15 
 
interface between the layers to form the irreversible bonding. The simplicity of producing 
multilayers makes it possible to fabricate multiple layers of fluidics, a difficult task using 
conventional microfabrication method. Interlayer adhesion failure and thermal stress problems 
can be avoided because layers are monolithic (all layers are made of same material) [60]. This 
technique has become one of the most popular in microfluidics for applications in biotechnology. 
 
Figure 1.9 Process flow of multiplayer soft lithography [60]. 
 
Pneumatic valve 
Valve is the basic unit for fluid-handling and controlling the movement of cells and 
particles in microfluidic channels. Quake?s group developed a monolithic pneumatic valves by 
using multilayer soft lithography [60]. The monolithic valves are based on two PDMS layers 
produced by replica molding from two masters and sealing the layers together as shown in 
Figure 1.10. Two separate layers are produced: one for the control channels and one for the flow 
16 
 
channels. Typically, the channels are 100 ?m wide by 10 ?m high to make active area of the 
valve 100 ?m by 100 ?m. These two layers are bonded together to form a thin layer of PDMS 
membrane (10 ?m~30 ?m thick) where the control channel and flow channel intersect 
orthogonally. When pressure (usually nitrogen coming from an external source such as 
pressurized gas tank) is applied to the control channel, the membrane deflects upward to close 
flow channel. The flow channel must have a rounded profile to enable the valve to close 
completely.  
 
Figure 1.10 A two-layer PDMS valve. An elastomeric membrane is formed where a control channels lies 
orthogonal to and below fluidic channel. When pressurizing the control channel, the membrane will deflect upward, 
thus cutting off the flow in fluidic channel. 
 
 
 
17 
 
Peristaltic pump 
  A peristaltic pump contains an array of valves actuated in a sequence (Figure 1.11a). 
Three valves are placed in a series and the peristaltic pumping occurs when three valves are 
actuated in a 6-sequence pattern shown in Table 1.2. In this case, the control channels are filled 
with air and the valve actuation frequency is limited by the maximum frequency of the off-chip 
solenoid control pumps. Because of the sequential motion of the valves, a fluid flow in the flow 
channel can be generated. When this peristaltic pump operates in a loop, as shown in the Figure 
1.11b, a fluid can be circulated within the loop to achieve the efficient mixing of reagents by 
rapidly stretching the interface between fluids to reduce the diffusion distance for mixing [70].  
Table1.2 Sequence of valve operation to create a sequential fluid motion in a micromixer 
  Sequence 1 Sequence 2 Sequence 3 Sequence 4 Sequence 5 Sequence 6 
Valve 1        
Valve 2        
Valve 3        
                                                                                                                        : Valve open     : Valve closed 
 
Figure 1.11 (a) A peristaltic pump using three valves in a series. (b) A circular micromixer using a peristaltic pump 
[60].  
 
 
 
 
18 
 
1.3 Objectives and organization of dissertation 
1.3.1 Objectives 
Studying responses of cells (bacteria) or molecule (RNA aptamer-based biosensor) to 
reagents (antibiotics or degrading agents) in a concentration- and time-dependent manner can 
provide useful information for pharmaceutical drug discovery and development of biological 
detection systems.  
The rapidly growing resistance of pathogenic bacteria to antibiotics resulted in a demand 
for new antibiotics. The discovery of new antibiotics requires understanding of bacteria?s 
responses to antibiotics, for example, dose-response relation of bacterial growth to antibiotics 
and bacteria?s resistance to antibiotics. Our microfluidic systems can overcome the limitations 
when using conventional methods and other microfluidic systems.     
The rapid and effective detection of RNase contamination is important in RNA related 
experiments and analyses. The application of RNA aptamer-based biosensors for detecting 
RNase contamination requires the understanding of its concentration- and time-dependent 
degradation behavior by the degrading agents, such as, RNases and metal ions. The development 
of our RNA aptamer-based biosensor can contribute to stimulating transcriptomic researches to 
easily detect RNase contaminations in lab environment. The integration of our biosensor and our 
microfluidic systems can serve as a detection device for rapid detection of RNase contaminations. 
The objectives of this dissertation are as follows: 
1) Investigate the dose-response of bacterial growth to bactericidal antibiotics in a time-
course manner. Through monitoring EC50 value over time, the attenuated inhibitory effect 
as a result of bacteria?s resistance to antibiotics during their growth can be evaluated.      
19 
 
2)  Monitor the concentration- and time-dependent degradation of a novel RNA aptamer-
based biosensor. By characterizing degradation of RNA biosensor, its degradation 
profiles can be obtained to differentiate degrading agents. 
To achieve these goals, two microfluidic platforms were used:   
1) A multiplex nanoliter reactor array chip was developed to validate the response of cells 
(bacteria) or molecule (RNA aptamer-based biosensor) to the specifically defined conditions (e.g. 
growth nutrients/antibiotics or degrading agents) in microenvironments. This device has several 
unique features including simple architecture that requires no complicated active mixing 
components, ease to operate without requiring continuous flow of reagents, and simple 
evaporation suppression component. The detailed work is presented in Chapters 3 and 5. 
2) An integrated microfluidic chip was used to investigate the response of cells (bacteria) or 
molecule (RNA aptamer-based biosensor) to different concentrations of reagents (e.g. 
bactericidal antibiotics or degrading agents) in a time-dependent manner. These devices is 
capable of simultaneously generating a wide range of concentrations without requiring 
continuous flow and growing bacteria in a zero-flow environment. The detailed work is 
presented in Chapters 4 and 5.  
 
1.3.2 Organization 
This dissertation is composed of six chapters. They are organized in the following order: 
Chapter 1 introduces fundamental concepts of pathogenic bacteria, antibiotics, bacterial 
phenotypes, dose-response analysis, RNA and RNase, and aptamer. In addition, the fundamental 
concepts of microfluidics, fabrication method, and key components are described. Finally, the 
objectives of this dissertation are presented. 
20 
 
Chapter 2 presents the literature survey of conventional methods and microfluidic devices 
reported for accessing inhibitory effects of antibiotics on bacterial growth and monitoring RNA 
degradation. 
Chapter 3 describes monitoring of the bacterial response to culture conditions such as 
growth nutrients and antibiotics on a multiplex nanoliter reactor array chip. Both the diaxuic 
growth and inhibited growth of bacteria were observed in microenvironments. 
Chapter 4 investigates the response of bacteria to various concentrations of bactericidal 
antibiotics at a time-course manner on an integrated microfluidic chip. The dose-response 
relation of bacterial growth to bactericidal antibiotics was obtained and bacteria?s resistance to 
antibiotics was discussed.  
Chapter 5 reports the monitoring of concentration- and time-dependent degradation of a 
novel RNA aptamer-based biosensor on microfluidic chips. The detection of biosensor in 
micoenvironments was validated on the multiplex nanoliter reactor array chip. The 
concentration- and time-dependent degradation profiles of biosensor by degrading agents were 
characterized on the integrated microfluidic chip.     
Chapter 6 summarizes the key research achievements, and provides the perspectives for 
their future applications. 
 
 
 
 
 
21 
 
Chapter 2 
LITERATURE REVIEW 
2.1 Introduction 
This chapter provides a brief review of methods currently used for assessing inhibitory 
effects of antibiotics on bacterial growth and for monitoring RNA degradation. The working 
principle as well as advantages and limitations of these methods are discussed. 
 
2.2 Conventional methods to assess activity of antibiotics 
The activity of antibiotics depends on their ability to inhibit bacterial growth. Several 
techniques have been described to quantitatively determine the activity of antibiotics. One 
classical method involves agar diffusion assays in which filter paper disks soaked with an 
antibiotic are placed on the surface of an agar plate that has been inoculated with the test 
bacterium. During the incubation, the antibiotic diffuses to create a concentration gradient that 
produces a zone of bacterial growth inhibition [13] (Figure 2.1). The disadvantages of the agar 
diffusion test are the lack of mechanization or automation of the test. Another classical method is 
the macrobroth or tube dilution test in which a serially diluted solutions of an antibiotic is added 
to the liquid bacteria culture. After incubation, the bacterial growth is measured by means of 
turbidity or by determining the viable cell numbers. Because bacterial cultures absorb some of 
the light, the higher the cell concentration is, the higher the turbidity [71] (Figure 2.2). In the 
early 1970s, automated systems of macrobroth dilution test were developed for the assay of 
bacterial antibiotic susceptibility [72]. The major disadvantages of the macrodilution method are 
the tedious tasks when prepared manually, possibility of errors in preparation of the antibiotic 
solutions, and the relatively large amount of reagents and space required for each test. The 
22 
 
miniaturization of broth dilution test by using small, disposable, plastic ?microdilution? plates 
has made this test popular and practical. A standard plate contains 96 wells, each well 
accommodating a volume of 100 microliter, allowing 12 antibiotics to be tested in a range of 8 
two-fold dilutions [46]. Microdilution is prepared using dispensing instruments such as a pizeo 
dispenser [73] or a focused acoustic dispenser [74], that are able to dispense small aliquots of 
diluted antibiotics into individual wells of the plate. Hundreds of identical plates can be prepared 
from the same master set of solutions in a relatively short period of time. Several automated 
systems with microdilution method have been developed to rapidly generate results of 
susceptibility test [75]. The MicroScan WalkAway? plus system (Siemens Healthcare 
Diagnostics, Figure 2.3) is a large self-contained incubator/reader device that can incubate and 
analyze 40~96 plates at a time. This system uses standard size microdilution plates that are 
hydrated and inoculated in the instrument. The instrument incubates plates and periodically 
examines them with either a photometer or a fluorometer to measure growth [76]. However, 
uncontrolled evaporation of dispensed liquid remains challenging for these systems.  
 
Figure 2.1 A disk diffusion test with an isolate of Escherichia coli from a urine culture. The diameters of all zones 
of inhibition are measured [76]. 
23 
 
 
Figure 2.2 A broth microdilution susceptibility plate containing 96 wells [76]. 
 
Figure 2.3 The MicroScan WalkAway? plus system [77]. 
 
2.3 Microfluidic systems to assess activity of antibiotics 
Microchannel based microfluidic systems have been used for assessing the inhibitory 
effect of antibiotics on bacteria [78-80]. Sun et al [78] described a nanoliter-scale reactor array 
used for bacterial growth under antibiotic treatments as shown in Figure 2.4. In this study, fresh 
nutrients or antibiotics are continuously supplied via diffusion to bacterial culture reactors. 
However, the continuous flow of reagent that is supplied outside chip is more likely to cause 
cross-contamination. Other microfluidic systems have been developed to investigate antibiotic 
24 
 
susceptibility of bacterial biofilms [79, 80]. In these systems, rather than preparing antibiotic 
concentrations outside chip, antibiotic concentration gradient is generated by networks of 
microchannels [81] within the microfluidic chip, as shown in Figure 2.5. However, these flow-
based microfluidic devices require a continuous flow to generate and maintain constant 
concentration gradient, and additional efforts are needed to calibrate concentration gradient 
profiles. 
 
Figure 2.4 Design and picture of the microfluidic device. (a) Schematic representation of the functional circuit used 
for long-term bacterial colony monitoring and antibiotic testing. (b) Enlarged functional microstructure of the 
microfluidic device corresponding to the dotted-line square in (a). (c) Optical image of the actual device. A one cent 
coin was employed to show the size of the device. (d) SEM image of the functional microstructure [78]. 
25 
 
 
Figure 2.5 The microfluidic device for assessing antibiotic susceptibility of biofilms. (a) Schematic diagram and 
optical image of the microfluidic device. The microfluidic device consists of gradient generator and main detection 
microchannel. (b) A 3-D plot of the fluorescence image obtained from 0 mm to 8 mm position in the detection 
microchannel. (c) Profiles of cross section at each representative position. A flow rate of 0.1 ml min-1 is used in the 
experiment [79]. 
 
Microfluidic platforms that use a constant inflow of immiscible liquids into junctions, 
(e.g. flow-focusing [82, 83] or T-junction[84, 85]) can produce monodispersed droplets. Droplet-
based microfluidic systems have been developed for antibiotic susceptibility screening of 
bacteria [39, 86]. Boedicker et al. [39] utilized reagent-loaded cartridges and a T-junction flow to 
generate droplets containing bacteria and various concentrations of antibiotics. In Figure 2.6, 
various concentration of drug trails and spacer plugs were pre-stored in a capillary (reagent 
26 
 
loaded cartridge). Then, drug trails and spacer plugs were flowed to microfluidic channels where 
they were merged with bacteria in a droplet. However, the preparation of a cartridge requires 
elaborate steps [87] such as dispensing nanoliter of reagents and aspirating reagents into a 
capillary. In addition, a gas segment must be applied to separate two reagents to prevent cross-
contamination. Churski et al. [86] employed four microfluidic T-junctions to create 
microdroplets for antibiotic screening shown in Figure 2.7. Droplets containing bacteria, 
antibiotics, and culture medium are finally merged and incubated. By varying the droplet volume 
of antibiotic and culture media, a concentration range of antibiotics is generated. The volume of 
droplet is controlled by the opening time of external solenoid valves [88] in which longer 
opening time leads to more dispensed volume. Consequently, volumes vary from several 
hundred nanoliters to several microliters. However, the typical volume of droplets generated by 
other types of valves is from several picoliters up to several nanoliters [89-91].  
 
 
Figure 2.6 Schematic drawing illustrates the formation of droplets containing bacteria, viability indicator, and an 
antibiotic at varying concentrations. (a) A capillary cartridge loaded with drug trails and gas space plug. (b) A drug 
trial cartridge is connected to a microfluidic channel, and then drug trials are flowed to merge with viability 
indicator and bacterial solution. Thus, droplets containing various drug trials are generated at T-junction [39].    
27 
 
 
Figure 2.7 A schematic drawing demonstrates the operation of the microfluidic system. (a) Four T-junctions 
generate packets of microdroplets and merged them to create microdroplets with a defined composition. The 
sequence of microdroplets formed in the device was incubated off-chip. After incubation, the microdroplets were 
loaded into a microchannel and the intensity of bacterial viability indicator was detected. (b) A linear relationship of 
volumes of droplets and valve opening time (?open) is obtained for fluids with different viscosities (1, 2, 3, 30, 100 
mPas) in two capillaries (5 m and 10 m in length) [86].  
 
2.4 Conventional methods for monitoring RNA degradation  
The most common method for monitoring RNA degradation is to incubate an RNA 
substrate with a degrading agent and then check for degradation of the RNA by ethidium 
bromide stained agarose gel electrophoresis. Figure 2.8 shows the results of gel-electrophoresis.  
This test has very low sensitivity and requires a subjective judgment [92]. Other more sensitive 
tests require a radio-actively RNA substrate or a fluorescence polarization instrument. Both tests 
are time-consuming, costly, and require large amount of reagents [93]. 
28 
 
 
Figure 2.8 Degraded and intact total RNA were run beside RNA markers on a 1.5 % denaturing agarose gel. The 
18S and 28S ribosomal RNA bands are clearly visible in the intact RNA sample while no bands are visible in 
degraded RNA [94]. 
2.5 Microfluidic methods for monitoring RNA degradation 
To date, the application of microfluidic devices as a platform for monitoring RNA 
degradation has not been reported yet.  
 
 
 
 
 
 
 
 
29 
 
Chapter 3 
CHARTING MICROBIAL PHENOTYPES IN MULTIPLEX NANOLITER BATCH 
BIOREACTORS 
3.1 Introduction 
One of the most fundamental challenges in biology is to understand the relationship 
between a DNA sequence (genotype) and the engendering observable cellular trait (phenotype) 
of an individual cell [95]. Recent advances in high-throughput DNA sequencing technology and 
molecular measurement techniques allow ?omics?-scale analysis of genetic information flow 
from genome to transcriptome to proteome to metabolome of a microorganism [96-99]. 
Traditionally, phenotypes have been referred to as characteristics related to cell growth, shape, 
color of the colonies, formation of biofilms or spores, cell-to-cell interactions, as well as growth 
patterns [7]. Growth phenotypes are the expression of genotypes and reflect microbial 
physiology and metabolism. For example, the growth curve experiments, which are mostly 
performed in batch cultures, generate primary information such as nutrient utilization profiles 
and growth kinetics data required for genetic analysis. This information is also used in systems 
biology to measure the impact of environmental and genetic perturbations on complex cellular or 
organismal networks [8, 96, 100, 101]. Consequently, a great deal of effort has been recently 
directed toward the development of novel high-throughput quantitative techniques for microbial 
growth assessment [102-105].  
Conventional batch culture techniques for studying cell physiology and developing 
bioproducts remain unchanged for many decades and are still in use in many academic 
laboratories and by the biotechnology industry. They require a large number of test tubes, 
30 
 
Erlenmeyer flasks, or agar plates, consume large amounts of media and apply laborious manual 
procedures for optical density measurements or microscopic observations. To overcome these 
challenges, a commercial system has been developed to test microbial strains on 96-well plates, 
which require an expensive instrumentation. Indeed, it is not ideal for automation and 
miniaturization [7, 96, 106]. Consequently, efforts have been directed toward the application of 
microtiter plates to scale-down (submilliliter) bioreactors for batch and fed-batch cultivations, 
and to improve the control over culture parameters such as temperature, dissolved oxygen, pH, 
and medium composition [102, 107-109]. 
Recent developments in microfluidics provide highly multiplex and quantitative 
measurements of individual cells, which replace the conventional liquid culture methods with 
economical and versatile miniaturized devices [36, 78, 110-114]. The application of soft 
lithography technology, using elastomers such as polydimethylsiloxane (PDMS) in microfluidics, 
has enabled fabrication of micro- or nanostructures such as valves, pumps, chambers, and 
channels in microfluidic cell culture systems [36, 60, 111, 115]. The microfluidics-based 
bioreactors offer several advantages over flask cultures or multi-well plates, including precise 
control of growth conditions and manipulation of nano- or picoliter volumes of culture media, 
capacity to integrate real-time growth and image measurements, ease of automation, ability to 
perform a large number of growth experiments under different conditions at the same time, and 
cultivation at the single-cell level. 
Most of the microfluidic cell culture systems reported to date operate in semicontinuous 
[44] or continuous culture modes [45, 78, 109, 114, 116], the so-called microchemostat, where a 
continuous exchange of fresh medium with the same component occurs through a multitude of 
growth chambers. However, practically, the functionality of any proposed microchemostat 
31 
 
differs in cells cultured in a batch process. In addition to the presence of a high risk of 
contamination due to intermittent cleaning of the reactor surfaces, the microchemostat devices 
often utilize a single growth medium, which is insufficient for testing multiple growth 
environments. Besides, in most of the laboratories, a large number of cell culture processes are 
operated in batch mode as it generates important growth kinetics data required for genetic-
phenotypic analysis, and systems modeling purposes. This necessitates the development of a 
multiplex microfluidic culture system that is able to simulate macro-cultures, which have been 
previously performed in flasks, tubes, or microtiter plates.  
Despite its importance in genetic-phenotypic analysis and systems biology, a batch 
culture process on the nanoliter scale has rarely been realized. This is primarily because PDMS is 
permeable to water vapor and the culture medium inside a PDMS nanoliter reactor easily 
evaporates, leading to complete drying over time [111, 117]. Also, implementation of on-chip 
mixing requires more operational steps, further complicating the system. Recently, a microfluidic 
batch chip was developed to have 120 culture reactors (50 nL each) in which cell suspension was 
circulated and the whole chip was put in a water bath for temperature and humidity control [113]. 
However, the active mixing by circulation of the cell suspension mimicked the previous 
microchemostats, which makes the platform difficult to test multiple culture conditions without 
changing the reactor design.  
3.2 Objective 
 In the present study, we devised a multiplex microfluidic batch culture chip that enables 
microbial growth in 24 sets of discrete bioreactors (1 nL each), while simultaneously allowing 
the assessment of eight different culture conditions in parallel. The flow channels and culture 
reactors were designed to maintain a uniform distribution of cells in each reactor and to conduct 
32 
 
experiments in triplicates in a single run. The evaporation of the cell culture medium present in 
each batch reactor was overcome by placing the chip in a humidified incubator and incorporating 
anti-evaporation channels around the reactor within the chip. Further, we demonstrated the utility 
of our system in characterizing microbial growth phenotypes by monitoring the growth of 
Escherichia coli with different carbon sources, obtained by both chip and tube cultures. We also 
demonstrated that the system could be successfully integrated with a fluorescence detection 
device for real-time observation of cell growth via online monitoring of fluorescent reporters of 
the opportunistic pathogen Pseudomonas aeruginosa incubated under different antibiotic 
concentrations. 
 
3.3 Materials and methods  
Chip fabrication 
The platform employed in the present study was fabricated by standard multilayer soft 
lithography [55, 60, 89]. The mask was designed using AutoCAD software (AutoDesk Inc., San 
Rafael, CA, USA), which was printed on a transparent film at 20,000 dpi (CAD/Art Services, 
Inc., Bandon, OR, USA). Molds for the two layers, fluidic and control layers, were prepared by a 
photolithographic technique using a positive photoresistor (AZ P4620; AZ electronic materials, 
Branchburg, NJ, USA) onto a 4-inch silicon wafer. This was followed by UV exposure and 
development. To facilitate reliable opening and closing of the valves, the photoresistor for fluidic 
layer was rounded by heating the mold at 130 ?C for 2 min. We fabricated the top thick fluidic 
layer of the chip by pouring uncured PDMS (RTV615; Momentive Performance Materials, 
Waterford, NY, USA; elastomer/crosslinker=10:1) onto the fluidic layer mold to achieve a 
thickness of 5 mm. Then, the bottom control layer of the chip was fabricated by spin-coating 
33 
 
uncured PDMS (elastomer/crosslinker=20:1) onto the control layer mold at 3,000 rpm for 1 min. 
Subsequently, the fluidic layer and control layer were cured at 80 ?C for 1 h and 45 min, 
respectively. Thereafter, the fluidic layer was peeled off from the mold, and holes for inlet and 
outlet ports to flow channels through the thick layer were created using a 19-gauge punch 
(Technical Innovations, Inc., Brazoria, TX, USA). The fluidic and control layers were aligned 
and bonded by baking at 80 ?C for 80 min, followed by punching holes for inlets to the control 
channels. The resulting PDMS chip was placed on a pre-cleaned glass slide (Fisher Scientific, 
Pittsburgh, PA, USA) and cured by incubating in an oven at 80 ?C for 18 h. The chip fabrication 
process is schematically illustrated in Figure 3.1. 
34 
 
 
Figure 3.1 Flow chart of chip fabrication using multiplayer soft lithography. 
 
 
35 
 
Loading of fluorescent beads or bacterial cells in nanoliter reactors  
Fluorescent beads (diameter: 2.0 ?m; concentration, 2.84?108 particles/mL) or bacterial 
cells were loaded through ?cell in? channels by applying 5 psi loading pressure. By opening C1 
and C3 control channel (Figure 3.3a), beads or cells were loaded into 12 nanoliter reactors. After 
maintaining the loading pressure for 1 min, the C1 and C3 channels were closed to retain beads 
or cells in the nanoliter reactors. 
 
Bacterial strains 
E. coli K12 (ATCC 25404) and its lactose non-fermenting (lac-) mutant (AB3568, F-, 
?(cod-lacI)6, rpsL60(strR)) [118] were purchased from the American Type Culture Collection 
(Manassas, VA, USA) and The Coli Genetic Stock Center (New Haven, CT, USA), respectively. 
P. aeruginosa PT5 was a gift from Dr. Thilo K?hler from University of Geneva, Switzerland. 
The pE21-EGFP vector having a GFP mutant gene was donated by Dr. Moonil Kim from 
Tuskegee University, USA. The EGFP, which is a red-shift mutation of the wild type GFP, has 
an excitation wavelength of 488 nm and an emission wavelength of 507 nm. pE21-EGFP was 
transformed into PT5 cells to construct the PT5-EGFP strain, using the standard transformation 
protocol [75]. 
 
Cultivation of E. coli in tubes and nanoliter reactors 
A seed culture was prepared by growing cells in a 14-mL test tube containing 4 mL of 
LB medium in a shaking incubator with 160 rpm at 37 ?C overnight (1.0~1.4 at OD600). For cell 
growth in a tube, the seed culture was diluted with a fresh LB medium to obtain an OD600 of 0.01. 
A volume of 4 mL of the cell inoculum was aerobically grown in a 14-mL tube at 37 ?C with 
36 
 
shaking at 160 rpm. As for lac- mutant, the LB medium was supplemented with 40 ?g/mL 
streptomycin (Sigma-Aldrich, St. Louis, MO, USA). For cell growth in a nanoliter reactor with 
LB medium, the seed culture was diluted with a fresh LB medium to obtain an OD600 of 0.1, and 
was subsequently introduced into a reactor. For cell growth in a reactor with M9 medium, cells 
from the seed culture were harvested by centrifugation at 7,711g for 1 min, washed 3 times with 
M9 medium, and then suspended in M9 medium to obtain an OD600 of 0.1. Thereafter, they were 
introduced into each of the nanoliter reactors containing M9 medium with different carbon 
source(s): no carbon source, glucose (0.4 % wt/vol), lactose (0.4 %), and lactose (0.4 %) plus 
glucose (0.08 %). Finally, the chip was incubated at 37 ?C in a humidity control system 
developed in our laboratory. 
 
Cultivation of P. aeruginosa harboring pEGFP in tubes and nanoliter reactors 
A seed culture was prepared by growing cells in a 14-mL test tube containing 4 mL of 
M9 medium supplemented with Casamino Acids (0.5 % wt/vol), MgSO4 (1.0 mM), glucose 
(11.1 mM), and 100 ?g/mL of ampicillin overnight (1.7~1.8 at OD600). Subsequently, the cells 
were harvested by centrifugation at 1,600g for 10 min, washed once with 2?M9 medium, and 
then suspended in 2?M9 medium to obtain an OD600 of 0.2. Various concentrations of 
gentamicin (Sigma-Aldrich, St. Louis, MO, USA) (0, 0.2, 0.4, 0.8, 1.6, 3.2, 6.4, and 12.8 mg/L) 
were prepared by diluting 100 ?g/mL of gentamicin stock solution in water. A volume of 2 mL 
gentamicin solution was added to 2 mL of the cell inoculum in a 14-mL tube, and the cells were 
allowed to grow aerobically at 37 ?C with shaking at 160 rpm. For cell growth in a nanoliter 
reactor, the cell inoculum along with a different concentration of gentamicin was introduced into 
37 
 
the reactor. Finally, the chip was incubated at 37 ?C in a humidity control system developed in 
our laboratory. 
 
Chip control and operation 
Figure 3.2 shows the schematics of experimental setup for chip control and imaging. A 
pneumatic control was used to operate the chip. A pressure of 15 psi was applied to control 
valves for closing control channels, and 5 psi was applied to introduce reagent to fluidic 
channels. The pneumatic control setup consists of eight-channel manifolds (Fluidigm, South San 
Francisco, CA, USA) controlled by a BOB3 control board (Fluidigm, South San Francisco, CA, 
USA). A digital I/O card (National Instruments, Austin, TX, USA) was mounted into the 
computer to control the switching of each channel of manifolds through the BOB3 control board. 
A LabVIEW (National Instruments, Austin, TX, USA) program developed by our group was 
used for automatic control of each control valve.  
 
Chip image acquisition and analysis 
The microfluidic chip was imaged using an inverted optical microscope (Axiovert 40 
CFL; Carl Zeiss, Munich, Germany) equipped with a CCD camera (Moticam 1000; Motic, Inc., 
Richmond, BC, Canada). Bright-field images were captured at 40? magnifications every 2 h and 
digitized using Motic Images Plus 2.0 software (Motic, Inc., Richmond, BC, Canada). For cell 
culture in nanoliter reactors, the fluorescence image was obtained using a modified 
ArrayWoRx? biochip scanner (Applied Precision, Issaquah, WA, USA) with 480?15 nm of 
excitation and 530?20 nm of emission. All captured images were in 16-bit grayscale; the 
resolution was 7,800 pixels per inch (PPI), and the pixel size was 3.25 ?m. The whole image (5.0 
38 
 
mm?3.8 mm) of 24 reactors was acquired by accumulating sequential image (1.25 mm?3.8 mm) 
four times (2.75 s/scan). ArrayWoRx? 2.5 software suits (Applied Precision, Issaquah, WA) 
built into the workstation automatically converted the images to an integrated image. The total 
scanning time of the integrated image (5.0 mm?3.8 mm) was approximately 11 s and exposure 
time was 0.01 s. The fluorescence images were digitized using ImageJ software (NIH, Bethesda, 
MD, USA).  
For tube culture of PT5-EGFP, the fluorescence intensity was obtained using Hitachi F-
7000 fluorescence spectrophotometer (Pleasanton, CA, USA) with 488?10 nm of excitation and 
507?10 nm of emission. The fluorescence intensity of 150 ?L of PT5-EGFP cells in a 1,000-?L 
cuvette was measured. 
Figure 3.2 Experimental setup for chip control and imaging. 
39 
 
Mixing efficiency of the reagents inside the nanoliter reactors 
We captured the mixing phenomena by using an inverted microscope (Axiovert 40 CFL; 
Carl Zeiss, Munich, Germany) equipped with a CCD camera (Moticam 1000, Motic, Inc., 
Richmond, BC, Canada). Time series analyzer of ImageJ software was employed for digitizing 
the captured images and obtaining the gray value of each reactor. When mixing was complete, 
the value remained constant. To quantify the mixing efficiency based on the average gray value, 
we used the following equation: 
null(%) = nullnullnullnullnull
nullnullnullnull
 ?100 %                 (3.1) 
where I is the mixing efficiency; and Gi, G, and Gf are the average gray values of initial time, 
given time, and final time, respectively. 
 
Statistical analyses 
The ANOVA test was used to determine the statistical significance of distribution of 
beads loaded per bioreactor. Normality and constant variance assumption were checked by using 
Anderson-Darling and Levene?s tests, respectively (MINITAB 14, Minitab Inc., State College, 
PA, USA). The bootstrapping method in the R statistical package was used to assess the 
significance of the correlation between cell densities and log-transformed antibiotic 
concentrations. 
 
 
 
 
40 
 
3.4 Results and Discussion 
Design and operation of the multiplex microfluidic nanoliter device 
The PDMS device features a 20-?m thin control layer that is overlaid with a 5-mm thick 
fluidic layer, which is placed on a glass slide (Figure A1 in Appendix A). The fluidic layer has 
two parallel rows of 12 culture reactors, each with a volume of 1 nL (Figure A1b in Appendix 
A). Each row is connected to a single-cell suspension channel and four different medium 
channels, where each of them correspond to three culture reactors. This design enables 
simultaneous triplicate experiments for each of the eight culture environments in a single run. 
Cell suspension and fresh medium were loaded into three reactors present as a single unit, 
by opening two control channels located at the ends of the reactors (C1 and C3 in Figure 3.3a) 
and closing a control channel at the middle of the reactors (C2). Once they were filled into each 
half of the reactor, the modes of the channels were switched to open C2 and to close C1 and C3, 
causing diffusive mixing of the cells and the medium. The diffusion efficiency was determined 
by monitoring the mixing of green and red dyes (ESCO Foods, Inc., San Francisco, CA, USA) 
dissolved in M9 medium, Luria?Bertani (LB) complex medium, or 5 % polyethylene glycol 
(PEG) solution (Figure 3.3b,c; Video A1 in Appendix A). When the temperature was increased 
from 22 to 37 ?C, we observed an approximately 33-55 % increase in the mixing efficiency due 
to increased energy for diffusion. The difference among mixing efficiency for all tested reagents 
is due to the different viscosity of reagents (viscosity: M9<LB<5 % PEG). Because the 
temperature-dependent viscosity is different for three tested reagents [119], all tested regents 
showed different level of increased mixing efficiency at 37 ?C. Three mixing efficiency values 
are around 102 % for LB at 37 ?C, we attributed these values to the fluctuations of light source 
intensity during video capture. All the tested media showed complete mixing in approximately 9 
41 
 
min, which is negligible when compared to the time taken for the generation of microbial cells in 
growth curve experiments (for example, at least 12 h for the fast-growing E. coli).  
 
Figure 3.3 Operation of the device. (a) Schematic drawing of the entrapment of cells into three replicate cultivation 
reactors (I). The middle control channel C2 is closed to load cell suspension (shown in yellow) and culture media 
(shown in light blue) into the reactors (II). Cells and the medium are mixed by opening C1 and closing both C1 and 
42 
 
C3 (III). The cell cultures are sequestered from the fluid channels and the cells start to grow (IV). Simultaneous 
triplicate experiments for testing a culture condition can be performed in a single run. Note that control channels and 
flow channels are at different layers (see Figure A1 in Appendix A for details) (b) Time-lapse micrographs showing 
the mixing of green and red dyes in M9 medium at 37 ?C. The process of diffusion was initiated by opening C2, and 
was almost completed in 4 min. Scale bar, 200 ?m. (c) A graph representing the time profiles of mixing efficiency 
of 2 dyes dissolved in M9 minimal medium, LB complex medium, and 5 % PEG solution at 22 ?C or 37 ?C. Error 
bar denotes the standard error of the mean from three replicate reactors. 
Uniform distribution of particles loaded per bioreactor  
The flow of cell-suspension channel perpendicular to the reactors caused the fluid to be 
pumped tangentially along the entrance of each of the reactors (Figure 3.4a). Distribution of 
particles confined in the reactors was analyzed using 2.0-?m fluorescent polystyrene beads 
(Polysciences, Inc., Warrington, PA, USA). An applied pressure of 5 psi to the suspension flow 
forced a portion of the fluid into the reactors. No build-up of the beads was observed at the 
entrances of the reactors, and hence, closing of the entrances led to the retention of the particles 
in the reactors, which thus resulted in the uniform distribution of beads trapped in the reactors 
(29?2 beads per reactor, P-value=0.713) (Figure 3.4b). Thus, loading of a nearly constant 
number of cells into each of the reactors allows direct and unbiased comparison of growth 
phenotypes from the reactors. 
43 
 
 
Figure 3.4 Loading of a uniform number of particles into the reactors. (a) Fluorescent beads (2.0 ?m) are loaded 
into each of the 12 reactors by opening C3 control channel, followed by the closing of C3 channel, which retains the 
beads. Arrows denote the direction of bead suspension flow. Scale bar, 200 ?m. (b) Number of beads loaded into 
each of the reactors. Error bar denotes the standard error of the mean from nine separate experiments. 
Evaporation-free batch cultivation 
A major challenge in employing microfluidic PDMS devices for cell cultivation at the 
nanoliter scale is that PDMS, being permeable to water vapor, allowing a rapid evaporation of 
the culture medium inside the microfluidic chambers [111, 117]. Evaporation is a matter of great 
concern in a batch cultivation, where the initially supplied liquid medium is not sufficient to 
support cell growth in a long-term culture. In contrast, this issue can be avoided by the use of 
fed-batch [44] or continuous [45, 109, 116] cultures through continuous renewal of the medium. 
In this study, we resolved this problem by placing the whole device in a humidified incubator 
and incorporating water-filled anti-evaporation channels around the nanoliter reactors within the 
device (Figure A1b in Appendix A). Water-filled channels installed around or above reactors 
44 
 
can suppress evaporation by letting water to saturate surrounding PDMS [120, 121]. The use of 
water jackets for long term cultivation could raise a concern of different salt concentrations as 
the culture chambers. However, as long as the medium in each chamber is retained to be constant 
during the culture, the variation in the medium concentration should be negligible.  
To determine the effects of anti-evaporation channels and humidified incubator on the 
degree of evaporation of the media inside the reactors, we measured the amount of green dye that 
was retained in the reactors at 37 ?C after 24 h (Figure 3.5; Figure A2 in Appendix A). Notably, 
during the initial stage, when the anti-evaporation channels and the humidified incubator were 
not operated, evaporation of the medium was observed, which resulted in 7 % of volume loss in 
4 h to ~26 % in 12 h. We observed the shrinkage of reactors and a dark green oval in the middle 
of reactors. It is likely that the rapid and elevated water diffusion occurred through a thin 
membrane (~10 ?m thick) between control channel (C2) and the middle of the reactors, causing 
the concentrated dye molecules (Figure A2 in Appendix A). Operating anti-evaporation 
channels under a pressure of 15 psi retarded evaporation for 8 h. The intensive drying was also 
observed in the middle of the reactors, possibly due to water diffusion from reactors to control 
channel (C2). We assumed that the upward deflection of the thin membrane separated the reactor 
into two parts (Figure A2 in Appendix A). Furthermore, it was noted that placing the device in a 
humidity-saturated incubator led to approximately 2 % volume loss in 24 h. That loss seems to 
be negligible and to be caused by the diffusion through thin membrane between a reactor 
chamber and a control channel. However, even relatively small shifts in the osmolality of culture 
media by evaporation through PDMS can drastically affect sensitive mammalian cell behavior 
[122].  In addition, when the system is applied to slowly growing cells such as myxobacteria 
(e.g., Sorangium cellulosum with doubling time of 16 h) [123], the advserse effect caused by the 
45 
 
volume change can be profound. Strikingly, by operating anti-evaporation channels along with 
use of the humidified incubator, virtually 100 % of the initial volume in the reactors was 
maintained throughout incubation. Therefore, our finding highlights the importance of 
conditioning microfluidic devices in humidified environment in preventing evaporation and 
further demonstrates that a continuous supply of water molecules from the anti-evaporation 
channels contributes to the saturation of PDMS backbone-pores around the reactors. 
 
Figure 3.5 Effect of humidity controls on volume changes of the medium inside the reactors. The graph represents 
the volume changes of the initial medium over a period of time (24 h duration) based on the presence and/or absence 
of humidity control and/or anti-evaporation channels: in the absence of humidified incubator (denoted with a black 
line), in the presence of operation of anti-evaporation channels (green line), in the presence of a humidified 
incubator (red line), and in the presence of both humidified incubator and the operation of anti-evaporation channels 
(blue line). Error bar denotes the standard error of the mean, obtained from three replicate reactors. 
 
46 
 
Application of the device to measure bacterial growth phenotypes 
As denoted above, we succeeded in developing a novel microfluidic batch culture system 
that enables uniform distribution of cells in each bioreactor and aids evaporation-free long term 
cultivation, which have been the major challenge in the development of PDMS-based 
microculture devices for batch cultivation. A cultivation-associated physiological environment 
can be easily changed when the reactor volume changes from macro- to microscale [111, 124]. 
Therefore, we investigated whether the designed system would allow sensitive and accurate 
measurements of bacterial growth phenotypes that are equivalent to the traditional methods, by 
performing growth curve experiments with E. coli and microbiological assays of antibiotics 
against the opportunistic pathogen P. aeruginosa. 
 
Growth curve experiments 
We demonstrated the utility of the developed device to replace the traditional methods by 
performing batch cultivation using one of the most commonly used laboratory strains of E. coli 
K-12. We obtained micrographs of the cells grown in the LB medium every 2 h after inoculation 
and counted the number of cells (Figure 3.6a). The growth curves were compared with the 
results obtained from the batch cultures of E. coli grown in 14-mL tubes, each containing 4 mL 
of LB medium (Figure 3.6b). The growth profiles derived from both methods reflected the 
typical batch growth curve of E. coli, and were highly similar to each other as shown in Figure 
3.6a. When grown in M9 minimal medium with glucose (0.04 % wt/vol) and lactose (0.2 %) as 
carbon sources, nanoliter reactors reproduced the diauxic growth curve [125] that is 
characterized by two exponential phases separated by a lag period during which cells produce 
enzymes required for metabolizing the less-preferred carbon source (Figure 3.6c). E. coli prefer 
metabolize glucose over lactose because glucose transportation can block of inducer of the lac 
47 
 
operon which is responsible for transport and metabolism of lactose [126]. The short lag period 
may be a result of lower glucose concentration in mixed carbon source as it takes short time for 
enzymes responsible for metabolizing lactose to build up after glucose is exhausted. Furthermore, 
in order to determine the ability of the device in accurately characterizing mutant phenotypes, we 
assessed the growth patterns of a known mutant with different carbon sources (Figure A3 in 
Appendix A). Notably, lactose nonfermenting (lac-) mutant of E. coli K12 [118] failed to grow 
on lactose as a sole carbon source, and did not show the diauxic growth pattern when grown on 
glucose and lactose as it lacked the lactose-metabolizing capability. 
 
Figure 3.6 Growth of E. coli cells during the batch culture in a nanoliter reactor. (a) Micrographs showing time-
lapse cell growth on the LB complex medium in a nanoliter reactor. Scale bar, 10 ?m. (b) A graph showing the 
comparison of cell growth on LB medium in the nanoliter reactor (represented by ?) with that in a 14-mL test tube 
containing 4 mL of LB medium (?). The cell numbers in the reactors were counted every 2 h after inoculation and 
were normalized to the initial number of cells. The cell density in a tube culture was measured in OD600. Values in Y 
axes are in log scale. Error bar denotes the standard error of the mean from 3 replicate reactors or test tubes. (c) A 
48 
 
graph showing the diauxic growth on M9 minimal medium with glucose (0.04% wt/vol) and lactose (0.2%) in a 
nanoliter reactor. 
 
Microbiological assay for antibiotics 
We validated the capability of the device for determining the concentration-dependent 
antibacterial activity of gentamicin against the opportunistic pathogen P. aeruginosa, which is 
responsible for nosocomial infections such as severe ventilator-associated pneumonia and 
bacteremia [127]. For real-time measurement of the cell growth, we employed a fluorescence-
based scanner to capture and quantified fluorescence intensity of the growing cells. For this 
purpose, we employed P. aeruginosa that constitutively expressed plasmid-encoded enhanced 
green fluorescent protein (pEGFP) . Fluorescence intensity and the cell number in the bioreactors 
were measured by employing a modified biochip scanner, following visual inspection using an 
optical microscope. We observed that there was a significant linear correlation (R2=0.986) 
between the fluorescence signal and the cell density (Figure A4 in Appendix A). Strikingly, in 
such a small culture volume (1 nL), cells were diffusively mixed and reached reasonably 
homogeneous growth (Figure A5 in Appendix A; see Figure 3.6a for E. coli cultivation). Most 
of microbioreactor systems reported to date use active mixing components for microbial culture 
to ensure homogenous growth conditions [44, 113, 128, 129], which requires delicate and 
complicated steps for the fabrication and operation of such devices. In this regard, homogenous 
conditions achieved by diffusive mixing in our system can contribute to the development of 
high-throughput microfluidic devices for phenotypic analysis. 
Further, we measured and compared the concentration-dependent inhibitory effects of 
gentamicin on P. aeruginosa harboring the EGFP plasmid in the cultures obtained from both 
49 
 
nanoliter reactors and test tubes (Figure 3.7; Figure A5 in Appendix A). In the case of test tube 
cultures, at 24 h following the inoculation, the cell density was found to be inversely correlated 
significantly with log-transformed antibiotic concentration (r=-0.98, P-value?0), which is in 
accordance with previous observations [130, 131]. Remarkably, similar results were obtained 
using cultures from nanoliter reactors (r=-0.87, P-value<0.007), indicating that our system could 
be employed for microbiological assay of antibiotics. In addition, significant inhibitory effects 
were observed for the concentration above 1.6 mg/L in microfluidic cultures. We assumed that 
bacteria were rapidly killed by high concentration of gentamicin due to the small number of cells 
introduced at the beginning. The slight different inhibitory effects in the microfluidic cultures 
from those in the 14-mL test tube cultures might be a result of diffusion-dependent cell behavior 
in microenvironments [124, 132, 133].   
 
Figure 3.7 Microbiological assay for antibiotics using a nanoliter reactor and a test tube. The graph shows the 
comparison of antibiotic effects of gentamicin on the growth of P. aeruginosa harboring the EGFP plasmid, 
following 24 h of culture in nanoliter reactors (represented using grey bars, on the left side), with that in the 14-mL 
test tubes (represented using hatched bars, on the right side). The X axis denotes concentration of gentamicin in log 
scale, and the Y axis denotes the fluorescence intensity (F.I.) normalized to the initial intensity of the cell culture, 
50 
 
immediately following inoculation. The error bar represents the standard error of the mean from three replicate 
reactors or test tubes 
3.5 Conclusion 
We demonstrated the utility of the device in charting phenotypic changes of E. coli by 
assessing the growth patterns with different carbon sources and also of P. aeruginosa cells in 
response to growth inhibitors with various concentrations of antibiotics through fluorescence-
based growth measurement. Our system was designed to conduct experiments in triplicate in a 
single run and ensure uniform distribution of cells in nanoliter reactors at the inoculation stage, 
which enabled direct comparison of the results from eight different culture environments. 
Importantly, virtually complete prevention of evaporation of the cell culture media was achieved 
by placing the device in a humidified incubator and installing water-filled anti-evaporation 
channels around the nanoliter reactors within the device. The present device has a simple 
architecture, ease of operation, and ability to precisely control multiple culture conditions while 
the batch culture format simulates a conventional microbial assay. This could further be 
exploited for investigating host-microbe interactions, and applied to a variety of cell types, not 
limited only to bacteria. In addition, by increasing the number of reactors and adding a 
concentration gradient generator at the inlets of the reactors, the device could be used to screen a 
multitude of metabolites, peptides, and proteins, thereby significantly reducing time and 
expenditure involved in drug-screening processes. Collectively, our study confirmed that high-
throughput growth phenotyping using multiplex nanoliter batch reactors, in combination with 
real-time growth and image measurements, will be able to meet the increasing demand for 
genome-wide studies, single cell analyses and drug discovery. 
 
51 
 
Chapter 4 
DETERMINATION OF EC50 OF BACTERICIDAL ANTIBIOTICS AT TIME-COURSE 
ON A MICROFLUIDIC CHIP 
4.1 Introduction 
In antimicrobial drug discovery and clinical practice, the knowledge of antimicrobial 
pharmacodynamics (PD) is useful for developing new antimicrobial drugs and establishing 
guidelines for therapy of infections. PD focuses on the relationship between drug concentrations 
and their antimicrobial effects [134]. The dose-response relationship is used to determine EC50, 
the antibiotic concentration that induces 50 % maximal inhibitory effect. EC50 is an important 
parameter to quantitatively characterize the inhibitory potency of antibiotics. Generally, the 
dose-response data are obtained from end-point readout of inhibitory effect after certain period 
of antibiotic treatment [65], thereby, a EC50 value is determined. Two classes of antibiotics 
(bactericidal and bacteriostatic antibiotics) are primarily used in antimicrobial therapy. 
Bactericidal antibiotics kill bacteria while bacteriostatic antibiotics prevent the growth of 
bacteria [12]. For most infections, bactericidal and bacteriostatic antibiotics function to prevent 
the spread of bacteria and permit the body?s defense mechanisms to eradicate infectious agents. 
However, bactericidal antibiotics are preferred for severe or difficult-to-treat bacterial infections 
(e.g. endocarditis or meningitis) [81] and may be required when the immune defenses are not 
effective or in chronic conditions [12].        
Eliminating bacteria by antibiotics is a complicated process involving transportation of 
the drug to appropriate sites and its interactions with targets [63]. The transportation of 
bactericidal antibiotics, such as aminoglycosides and quinolones, requires the penetration 
through cell wall to reach the target sites [43]. Metabolic activities of bacteria provide the 
52 
 
essential energy for uptake of antibiotics. For example, the aerobic catabolism of carbon source 
by Pseudomonas aeruginosa provides energy for aminoglycoside transportation and interaction 
with targets [135]. Bacteria resist antibiotics by employing three mechanisms: efflux of 
antibiotics from cells, enzymatic degradation and/or modification of antibiotics, and target site 
alteration [64]. The efflux pump is an energy-dependent mechanism by which the cell can 
become resistant to antibiotics, including aminoglycosides and quinolones, by preventing drug 
entry into the cytosol. It is an important property of P. aeruginosa and many other bacteria that 
have broad resistance to for antibiotics [43]. Because efflux systems typically require energy 
expenditure by the cell, metabolic activity of the bacterial cell is important for its effectiveness. 
Thus, determining EC50 by conventional end-point analysis in the dose-response data is 
insufficient to demonstrate the attenuation effect caused by restricting antibiotic uptake. In 
contrast, determining EC50 values from time-course readouts allows correlating the inhibitory 
effect with bacterial growth and resistance. 
Currently, effect of antibiotics is determined by the inhibition of bacterial growth in test 
tubes, microtiter plates and on solid medium using disk diffusion method [61]. Although 
effective, these traditional methods require a large volume of reagents. This is an important 
consideration when the availability of antibiotics to be tested is limited. The sophisticated 
concentration gradient preparation through dilution also can cause high possibility of errors. The 
automated high-throughput screening (HTS) method has been used for measuring dose-response 
of antibiotic molecules [136]. However, this method is also limited due uncontrolled evaporation 
of dispensed liquid at nanoliter or picoliter scales [137]. Microfluidic methods that can precisely 
manipulate small volume of reagent and generate concentration gradient are reported to be 
effective in study of antibiotic susceptibility of bacterial biofilms [79, 80] and effects of drug 
53 
 
concentration on bacterial chemotaxis [138-140]. However, these microfluidic devices require 
continuous flow to generate and maintain concentration gradient. Thus, cross-contamination of 
reagents is more likely to happen.  
 
4.2 Objetive 
At present, the investigation of time-course EC50 values from dose-responses of 
antibiotics on a microfluidic platform has not been reported yet. In this study, we determined 
EC50 for two bactericidal antibiotics on a microfluidic system device we previously reported [56]. 
Our data clearly demonstrate the feasibility of this microfluidic device for maintaining bacterial 
growth and assessing dose-response of bactericidal antibiotics (gentamicin and ciprofloxacin) 
against growth of P. aeruginosa. In addition, we demonstrate that the EC50 values acquired from 
the microfluidic device are comparable to those obtained from conventional test tube cultures. 
Furthermore, we demonstrate our microfluidic device for detecting attenuation of bactericidal 
activity of antibiotics as a function of bacterial growth and metabolism. Thus, our microfluidic 
device is a viable alternative to the conventional methods for effective determination of EC50 
while offering advantages including generation of a wide concentration range and conservation 
of precious reagents. 
 
4.3 Materials and methods 
Chip design and concentration generation  
The microfluidic device used in this work was previously reported by our laboratroy [56]. 
We fabricated microfluidic device by using standard multilayer soft lithography [60], the 
microfluidic device is composed of two polydimethylsiloxane (PDMS) layers. This microfluidic 
54 
 
chip has 14 parallel processors for generating 14 concentrations and bacterial cell culture. 
(Figure B1 in Appendix B) shows the step-by-step operation of generating 14 concentrations. 
Antibiotic solution and dilution buffer are introduced to metering channels. By varying the 
length ratio of metering channels, 14 different antibiotic concentrations ranging from 0 to 1000? 
are generated. Then, diluted antibiotics are mixed with bacterial cells by accelerated diffusion 
through fluid circulation [70].    
 
Antibiotics and bacteria 
Gentamicin was purchased from Sigma-Aldrich (St. Louis, MO, USA). Ciprofloxacin 
was purchased from Santa Cruz Biotechnology (Dallas, TX, USA). Stock solutions of antibiotics 
were prepared for use. P. aeruginosa (PT5-EGFP) was constructed by transforming a pE21-
EGFP vector containing a GFP mutant gene into a wild type P.aeruginosa PT5 by following the 
standard transformation protocol [75]. The EGFP, which is a red-shift mutation of the wild GFP, 
has an excitation wavelength of 488 nm and emission wavelength of 507 nm.  
 
Bacterial cell culture and suspension preparation for on chip and test tube culture 
An overnight cell culture was prepared by growing a single colony in a 45-mL test tube 
containing 19 mL of M9 medium supplemented with Casamino Acids (0.5 % wt/vol), MgSO4 
(1.0 mM), glucose (11.1 mM), and 100 ?g/mL of ampicillin (1.3~1.5 at OD600). Subsequently, 
the cells were harvested by centrifugation at 1,600? g for 10 min, washed once with 2?M9 
medium, and then suspended in 2?M9 medium to obtain a cell density of 3~4?108 CFU/mL for 
subsequent experiments. Cell density was determined using a normal plate count method.  14 
concentrations of antibiotics were prepared by diluting stock solution in water. For test tube 
55 
 
culture, 1.5 mL of antibiotic solutions was added to 1.5 mL of cells in a 14-mL tube, and they 
were grown aerobically at 37 ?C with shaking at 160 rpm. For on chip culture, cells were loaded 
into 14 reactors through ?cell in? channels by applying 5 psi pressure. After maintaining the 
loading pressure for 1min, separation valves around reactors were closed to trap cells in reactors. 
Then, cells are mixed with various concentrations of antibiotics with three sequential peristaltic 
pumps for 10 mins. Three sequential valves were filled with water and kept pressurized at 2 psi 
throughout experiment. The chip was placed at 37 ?C in a humidity control system developed in 
our laboratory [46]. 
 
Measuring number of cells in reactors 
The number of cells was visually counted under an inverted optical microscope (Axiovert 
40 CFL; Carl Zeiss, Munich, Germany) at 40? magnifications. 
 
Fluorescent signal acquisition and analysis 
For test tube culture, the fluorescence intensity was obtained using Hitachi F-7000 
fluorescence spectrophotometer (Pleasanton, CA, USA) with 488?10 nm of excitation and 
507?10 nm of emission. The fluorescence intensity of 150 ?L of PT5-EGFP cells in a 1000-?L 
cuvette was measured. For on chip culture, the fluorescence image was obtained using a 
modified ArrayWoRx? biochip scanner (Applied Precision, Issaquah, WA, USA) with 480?15 
nm of excitation and 530?20 nm of emission. All captured image were 16-bit grayscale, the 
resolution was 7,800 pixels per inch (PPI), and the pixel size was 3.25 ?m. The whole image 
(34.6 mm?4.2 mm) of 14 reactors was obtained by accumulating a sequential image (1.47 
mm?4.2 mm) twenty-six times (2.85 s/scan). ArrayWoRx? 2.5 software suites (Applied 
56 
 
Precision, Issaquah, WA, USA) built into the workstation automatically converted the images to 
an integrated image. The total scanning time of the integrated image (34.6 mm?4.2 mm) was 
approximately 74 s and exposure time was 0.01 s. The fluorescence images were digitized using 
ImageJ software (NIH, Bethesda, MD, USA).   
 
Growth quantification, data fitting and statistical analysis 
The cell growth was normalized by using the following equation:  
Normalized growth = null.null.nullnull.null.nullnull null nullnullnullnullnullnullnullnull nullnull (null.null.nullnull.null.nullnull null null) nullnullnull nullnull nullnullnullnullnullnullnullnullnullnullnullnullnullnull 
The normalized growth points were fitted by using a four parameter nonlinear-logistic-
regression model: 
 null = nullnullnullnull + nullnullnullnullnullnullnullnullnullnullnullnullnull(nullnullnull(nullnullnullnull)null[null])null , where G is normalized growth, GMax is maximum normalized 
growth, GMin is minimum normalized growth, [c] is antibiotic concentration, and h is hill slope. 
Then, EC50 values were determined. 
ANOVA test was used to determine the statistical significance of distribution and growth of cells 
loaded in reactors. Student?s T-test was used to determinate the statistical significance of 
inhibition data between the on chip and test tube culture method. Normality and constant 
variance assumption were checked by using Anderson-Darling and Levene?s tests, respectively 
(MINITAB 14, Minitab Inc., State College, PA, USA). 
 
4.4 Results and Discussion 
Operation of microfluidic device to grow bacteria with antibiotics 
A microfluidic device with 14 parallel processors [56] was used to grow bacteria with 
antibiotics in a zero-flow enviornment. (Figure 4.1a) Each processor consists of an 8-nanoliter 
57 
 
rectangle shaped reactor for cell growth and metering channels where antibiotics are diluted. 
Figure 4.1b shows the steps for introducing antibiotics to cell cultures to determine their 
inhibitory effect on bacterial growth in reactors. First, bacterial culture is introduced through 
?Cell in? inlet and cells are trapped in the first half of the reactor. Antibiotic solution and dilution 
buffer are then introduced into the metering channels through ?Antibiotics? and ?Buffer? inlets. 
After metering, both antibiotic solution and dilution buffer are pushed into the second half of 
reactor, separated from the cell solution compartment by separation valves. Each reactor is then 
isolated from the rest of the device and the separation valves is opened. Three mixing valves are 
activated sequentially to generate a fluidic flow in each reactor and mix the reagents.  
58 
 
 
Figure 4.1 The microfluidic chip with 14 processors and operation of one processor. (a) Design of microfluidic chip 
with 14 processors where bacterial cells grow with 14 concentrations of antibiotics. (b) Operation of one processor. 
Bacterial cells are introduced into first half of reactors, then, antibiotics and dilution buffer are introduced into 
metering channels. After introducing metered antibiotics to second half of reactors, cells and antibiotics are mixed 
by three mixing valves.      
 
 
 
 
59 
 
Cell introduction and growth in reactors 
A wild type P. aeruginosa strain that constitutively expresses plasmid-encoded enhanced 
green fluorescent protein (pEGFP) was used for real-time monitoring of cell growth [46]. A 
significant linear correlation between fluorescence intensity and cell density was observed 
(R2=0.992) (Figure B2 in Appendix B). Bacterial cell culture was pumped into half of reactors 
along the cell entrance channel which connects 14 reactors (Figure 4.2a). Cells were forced to 
enter the reactors under a 5 psi driven pressure and trapped by closing separation valves which 
resulted in the uniform distribution of cells in reactors (184?23 per reactor, p-vale=0.787) 
(Figure 4.2b). Thus, a nearly constant number of cells in reactors allowed unbiased comparison 
of cell growth. The trapped cells were mixed with water from the second half of reactor. Then, 
the growth of cells cultured in 1? culture medium was monitored. We further validated the 
uniformity of culture conditions within this microfluidic device by comparing the cell growth in 
14 reactors. By comparing the fluorescence intensity of cells in the reactors after 24 h of growth 
(Figure 4.2c), we observed no significant difference across 14 reactors (p-value=1.000). 
Therefore, this microfluidic system can maintain consistent growth of bacteria in each reactor to 
enable multiple antibiotic testing.  
 
 
60 
 
 
Figure 4.2 Introduction and growth of PT5-EGFP cells in reactors. (a) Cells are loaded into 14 reactors through 
?cell in? inlet, and cells are trapped in reactors by closing surrounding valves. (b) Number of cells in each reactor 
after introduction. Data are represented as means ? SD of three independent experiments. There was no significant 
difference in number of cells in 14 reactors. (c) The fluorescence intensity of cells after 24 h cultivation. Data are 
represented as means ? SD of three independent experiments. There was no significant difference in cell growth 
across 14 reactors.    
 
Oxygen supply and consumption in reactors 
Oxygen plays an important role for organisms that rely on aerobic respiration including P. 
aeruginosa [9]. Some investigators [141, 142] used oxygenator systems incorporated in the 
perfusion circuits to provide a sufficient supply of oxygen for tissue and cell culture. The high 
surface-to-volume (S/V) ratio characteristic of microfluidic devices makes it an effective 
approach to provide oxygenation inside a microfluidic cell culture system [143]. PDMS-based 
microfluidic devices have been used for cell culture [74, 82, 87] because PDMS is a polymer 
with high oxygen diffusivity [84]. In the PDMS reactors on this microflidic device, the 
61 
 
maximum amount of oxygen flux Fmax by diffusion through PDMS can be estimated by the 
following equation [82, 85]: 
nullnullnullnull ? nullnullnullnullnull null?null?nullnull                                                  (4.1) 
in which DPDMS denotes the diffusivity of oxygen in PDMS, DPDMS= 4.1?10-9 m2/s [84]; ?null 
denotes the difference of oxygen concentration across the PDMS layer, the oxygen concentration 
in atmosphere is 2.0 mol/m3 and used to estimate the maximum oxygen difference [82]; ?null 
denotes the thickness of PDMS layer, the PDMS layer for this microfludic device is 5 mm. 
Because the height of a reactor (10 ?m) is approximately two order of magnitude smaller 
compared to the length and the width, the oxygen supply is dominated by the top surface. For 
one reactor, the total surface area is 9.37?10-7 m2, and the maximum oxygen flux can be 
estimated to be 1.53?10-13 mol/s. The oxygen consumption rate of P. aeruginosa has been 
reported to be 5.97?10-19 mol/cell/s when cultured in LB medium at 32 ?C with agitation at 150 
rpm [83]. Based on the correlation of cell density and fluorescence intensity from our published 
work [46], the bacteria concentration in a 8 nL reactor after 24 h culture under the experimental 
condition (Figure 4.2c) is estimated to be 0.98?109 CFU/mL with the oxygen consumption rate 
of 4.68?10-15 mol/s. Given these conditions, the maximum oxygen supply flux in our system is 
almost 33 times higher than the oxygen consumption rate. Therefore, this microfluidic device 
can provide adequate amount of oxygen for sustaining the aerobic growth of bacteria. 
 
Glucose supply and consumption in reactors 
Carbon compounds such as glucose are major sources of cellular carbon and energy [9]. 
The culture medium contains 11.1 mM glucose and, therefore, the total amount of glucose 
supplied in a reactor is 8.88?10-11 mol. It has been reported that the maximum glucose uptake 
62 
 
rate of P. aeruginosa is 1.2?10-9 mol/min/mg dry cell at 30 ?C [144]. Estimating the mass of a 
dry bacterial cell to be 1.0?10-9 mg [9] and bacteria concentration in a reactor to be 0.98?109 
CFU/mL after 24 h of growth, the maximum amount of glucose consumption during 24 h culture 
is 1.35?10-11 mol. The amount of supplied glucose is almost 7 times higher than the amount of 
glucose consumption. So, a sufficient amount of glucose is supplied for maintaining bacterial 
growth for at least 24 h under experimental conditions. 
 
Dose-responses of P. aeruginosa to bactericidal antibiotics on chip and in test tubes 
Two bactericidal antibiotics, gentamicin and ciprofloxacin, were used to investigate the 
inhibitory effect on growth of P. aeruginosa. P. aeruginosa is a common opportunistic human 
pathogen that causes various acute and chronic infections [51]. Gentamicin is an aminoglycoside 
antibiotic that inhibits protein synthesis by interrupting the 30S ribosomal subunit [145]. 
Ciprofloxacin is a highly active fluoroquinolone antibiotic that interferes with the DNA gyrase 
and Topoisomerase IV that are involved in uncoiling of the supercoiled DNA during replication 
and decatenation of the replicated circular DNA [145] .  
We first monitored the growth of P. aeruginosa under the treatment of gentamicin or 
ciprofloxacin on chip and in test tubes (Figure B3 in Appendix B). Then, we assessed the 
inhibitory effects at three different growth times. We benchmarked the normalized growth of 
bacteria on chip with those of the test tube grown cultures (Figure 4.3a and b). The inhibitory 
effects of antibiotics on chip were similar to those obtained in test tubes for both antibiotics. 
Based on the growth inhibition profiles, we determined the EC50 value for each antibiotic (Table 
4.1). Standard antibiotic susceptibility test requires a certain culture time (16~24 h), therefore, 
the inhibitory effect of antibiotics are generally evaluated within this time frame [65]. In this 
63 
 
study, we evaluated the inhibitory effects of antibiotics at 16, 20, and 24 h, respectively. The 
average of EC50 values obtained on chip is close to that obtained in test tubes for both gentamicin 
and ciprofloxacin. Therefore, this microfluidic system was validated to be accurate for assessing 
inhibitory effects of antibiotics on bacterial growth. The EC50 values obtained in this study were 
of the same order of magnitude as IC50 values reported previously for P. aeruginosa: 1.69 ?g/ml 
for gentamicin [146], 0.06 ?g/ml for ciprofloxacin [54]. 
 
Figure 4.3 Growth inhibition profiles of two bactericidal antibiotics. (a) Gentamicin, and (b) Ciprofloxacin. Bacterial cells were treated with 
14 concentrations of antibiotics in a wide concentration range. Cell growth was normalized by the fluorescence intensity. Data are presented as 
means ? SD of three independent experiments. Asterisks indicate statistical significance compared to test tube culture, p<0.05.  
 
Table 4.1 EC50 (?g/ml) of gentamicin and ciprofloxacin  
Time (h)  Gentamicin  Ciprofloxacin  On chip Test tube  On chip Test tube 
16  1.88 ? 0.38 2.35 ? 0.33  0.083 ? 0.003 0.067 ? 0.012 
20  2.39 ? 0.41 2.69 ? 0.29  0.088 ? 0.001 0.066 ? 0.015 
24  2.97 ? 0.68 2.81 ? 0.30  0.095 ? 0.001 0.079 ? 0.025 
 
64 
 
Bacterial resistance to antibiotics 
Because transportation of gentamicin through inner bacterial membrane to reach its target 
site is an oxygen-requiring process [78] and the interaction with target occurs in an energy-
dependent step [147], we calculated the supply and consumption of oxygen and glucose during 
growth of P. aeruginosa in reactors as well as in test tubes (data not shown).  Since the supply of 
oxygen and glucose is adequate, we assumed that the effect of limitation of oxygen and glucose 
on reducing gentamicin uptake was minimal in our system. In contrast to gentamicin, because the 
uptake of ciprofloxacin is an energy-independent process of passive diffusion [148], we do not 
believe the ciprofloxacin uptake was affected  in our device. 
    In our study, we observed an increase among average EC50 values of gentamicin for 
both on chip culture and test tube culture (Table 1). For ciprofloxacin, we observed an increase 
among average EC50 values only on chip. A possible reason for the difference in bacterial 
counteracting to these two antibiotics may be due to the effect of the exopolysaccharide alginate 
on the drugs. Alginate forms a layer of anionic polysaccharide which surrounds cells to serve as 
a barrier to restrict uptake of aminoglycoside [43]. The binding of gentamicin to alginate may 
reduce the rate of gentamicin diffusion as the antibiotic-binding sites act as sinks to reduce free 
concentration of gentamicin [149]. Therefore, the accumulation of alginate layers produced 
during P. aeruginosa growth could be a likely cause to reduce the uptake of gentamicin, leading 
to increased average EC50 values. 
Efflux systems play a key role in bacterial resistance to antibiotics. It is well known that 
P. aeruginosa uses four efflux systems to extrude antibiotics to attenuate inhibitory effect of 
antibiotics [150] . The efflux pump MexA-MexB-OprM of wild type P. aeruginosa is expressed 
during bacterial growth where it contributes to the intrinsic resistance to quinolones [91]. It has 
65 
 
been reported that the expression of MexA-MexB-OprM efflux gene is regulated by growth 
phase of P. aeruginosa [47]. The expression of MexA-MexB-OprM parallels to cell growth and 
may be a reflection of increased metabolic activity of cells at log phase. For on chip culture, we 
observed that the growth of bacteria was at late-log phase within 16~24 h (Figure B3b in 
Appendix B). Therefore, we speculated the inhibitory effect of ciprofloxacin may have been 
attenuated as a result of the continuous expression of MexA-MexB-OprM efflux genes at late-
log phase. The increase of average EC50 values for on chip culture (Table 4.1 and Figure 4.3d) 
validated our assumption. However, for test tube culture, the growth of bacteria was at late-log to 
early stationary phase (Figure B3b in Appendix B) during which metabolic rate was decreased, 
and no increase of average EC50 values from 16~20 h was observed (Table 4.1 and Figure 4.3d).  
Ciprofloxacin is considered to be a concentration-dependent bactericidal antibiotic [151]. 
However, above 0.5 ?g/ml of ciprofloxacin, both on chip and test tube culture showed reduced 
growth inhibition (Figure 4.3b). We attributed this phenomenon to the survival of antibiotic-
tolerant persister cells induced by ciprofloxacin [152-154].    
Features of our microfluidic device 
This microfluidic chip integrates concentration generation components (metering 
channels) which facilitate the preparation of a wide range of antibiotic concentrations and avoid 
pipetting errors during antibiotic dilution. The bacterial growth in our microfluidic device 
mimics batch culture mode where bacteria grow in test tubes or microtiters in most laboratories 
[76]. The zero-flow culture environment of this device prevents cross-contamination of ragent.  
Therefore, the unique features of this microfluidic method make it a viable alternative to 
conventional methods to study dose-responses of antibiotics against bacteria. 
 
66 
 
4.5 Conclusion 
Because metabolic activities and efflux pumps of bacteria can affect the uptake of 
antibiotics, EC50 value may change during bacterial growth. In this study, we investigated dose-
responses of bactericidal antibiotics against growth of P. aeruginosa and obtained EC50 values. 
Our results are the first efforts to demonstrate the efficacy of using a microfluidic device in the 
time-course analysis of EC50. Our analysis provided a method to evaluate the inhibitory effect 
associated with the uptake of gentamicin and ciprofloxacin. The unique features of this 
microfluidic device make it a viable alternative to conventional methods to study dose-responses 
of antibiotics against bacteria. We envision this microfluidic tool can be further applied to study 
the concentration effect of different molecules such as ions, nutrients, or antibiotic degrading 
enzymes as well as the combinatorial effect of antibiotics on multidrug resistant pathogenic 
bacteria. 
 
 
 
 
 
 
 
 
 
67 
 
Chapter 5 
MONITORING OF RNA DEGRADING AGENTS WITH A NOVEL APTAMER-BASED 
BIOSENSOR 
5.1 Introduction 
The transcriptome is the complete set of RNAs in a cell [155], reflecting genes expressed 
in a specific cell at a specific time. The study of the transcriptome is critical for interpreting the 
functional elements of the genome, revealing the molecular constituents of cells [156] and tissues 
[157], and understanding development of organisms [158] and molecular origins of diseases 
[159-162]. The accuracy of gene expression profiling is influenced by the quantity and quality of 
starting transcripts. RNA purity and integrity are critical for the successful realization of RNA-
based analyses [163, 164] as starting with low quality transcripts may strongly compromise the 
results of labor-intensive, time-consuming and very expensive experiments [24, 25] (quantitative 
RT-PCR, micro-arrays, northern blot analysis, RNA mapping, cDNA library construction and 
any applications where the integrity of RNA should be checked). 
The most important factor obstructing the analysis of gene expression in human cells and 
tissues is the rapid degradation of cellular RNAs [165]. After extraction of a tissue from its 
normal environment, the relative rates of RNA metabolism change, resulting in changes in the 
relative proportions of various RNA species within a tissue, and an overall reduction of RNA 
concentrations, due to degradation by endogenous or exogenous RNases [165, 166]. To reduce 
the rate RNA catabolism after tissue collection, several approaches have been used. Although 
RNA degradation from endogenous RNases can be prevented (for review, see Srinivasan et al. 
(2002) [167]), contamination of samples with exogenous ribonucleases (RNases) remains an 
important cause of RNA loss. Careful handling during the experimentation can limit RNA 
68 
 
degradation [168] but simple mistakes from inexperienced researcher can easily ruin controlled 
samples [169]. When RNase contamination is suspected, detection of the contamination source is 
primordial. However, common electrophoresis-based methods have a very low sensitivity and 
depend on a subjective judgment [92]. Other tests for RNase require a radio-actively RNA 
substrate or a fluorescence polarization instrument [93]. Although those assays are very 
sensitive, they are time-consuming, lab-intensive, and expensive. Consequently, when RNase 
contamination is suspected in one of the reagents, the simplest solution is to throw away all the 
solutions and re-start the experiment with fresh reagents. 
Aptamers are nucleic acids or peptides that exhibit selective affinity for a target molecule, 
often higher than antibodies [26], and that have little cross-reactivity to other homologous 
compounds [26, 170]. They are thus currently considered to be attractive detection and 
diagnostic tools in various domains from health to food science [26, 170-173] because they are 
inexpensive to generate and produce, easy to modify and easy to combine with other molecules 
to fit testing/diagnostic purposes [26], including fluorescent dyes [174, 175] that can be activated 
or inactivated in the presence of target molecules [176]. 
 
5.2 Objetive 
In this work, we report a novel RNA aptamer-based biosensor for the detection of RNA 
degrading agents. Using microfluidic systems, we demonstrated the capability of our biosensor 
to detect different degrading agents and, for the first time, we observed the degradation profiles 
of the biosensor as a function of time and concentration of degrading agents, in a single 
experiment. With the advantages of rapid characterization and small reagent consumption, 
69 
 
microfluidic systems are superior to conventional methods to fully characterize degrading effects 
of enzymes and metal ions. 
 
5.3 Materials and methods 
Chip design and fabrication 
In this work, we used a microfluidic device, which was previously reported by our group 
[56]. This microfluidic chip has 14 processors. Each processor consists of an 8-nanoliter 
rectangle shaped reactor and a metering channel. We fabricated a polydimethylsiloxane (PDMS) 
based microfluidic device by using standard multilayer soft lithography [60].  
Materials 
The DFHBI-binding aptamer used in this study is composed of two RNA strands, 
denoted as strands A and B. Both strands were synthesized by transcription of duplex synthetic 
DNA templates with T7 RNA polymerase. DNA template for the strand A was formed by 
annealing deoxyoligonucleotides:  
AATTCCTGCAGTAATACGACTCACTATAGACGCGACTGAATGAAATGGTGAAGGAC
GGGTCCAGGTGCGGCTGC and 
GCAGCCGCACCTGGACCCGTCCTTCACCATTTCATTCAGTCGCGTCTATAGTGAGTC
GTATTACTGCAGGAAT. DNA template for strand B was formed by annealing 
deoxyoligonucleotides: 
AATTCCTGCAGTAATACGACTCACTATAGCAGCTGCACCTTGTTGAGTAGAGTGTGA
GCTCCGTAACTAGTCGCGTC and 
GACGCGACTAGTTACGGAGCTCACACTCTACTCAACAAGGTGCAGCTGCTATAGTGA
GTCGTATTACTGCAGGAATT. These deoxyoligonucleotides were synthesized by Eurofins 
70 
 
MWG Operon and purified by preparative 20 % polyacrylamide gel electrophoresis. The purified 
deoxyoligonucleotides were stored in 10 mM Tris-HCl, pH=7.5 at -20 ?C. Templates were 
prepared by heating the two complementary DNA strands to 70 ?C for 3 minutes and then 
cooling the DNA on ice. RNA strands A and B were produced by incubating 0.2~1.0 ?M 
template with T7 RNA polymerase (0.1 mg/mL) together with 4 mM ATP, 4 mM GTP, 4 mM 
CTP, 4 mM UTP in 40 mM Tris-HCl (pH=8.0), 10 mM MgCl2, 1 mM spermidine, 5 mM 
dithiothreitol and 0.01 % Triton X-100 for 1 hour at 37 ?C. Transcripts were purified by 
electrophoresis in a 1.5-mm thick 20 % polyacrylamide sequencing gel, located by UV 
shadowing, eluted by crushing in 0.25 mM ammonium acetate/10 mM Tris-HCl (pH=8)/1 mM 
EDTA, and concentrated by DEAE?cellulose chromatography and ethanol precipitation as 
described by Cazenave and Uhlenbeck [177]. RNA was dissolved in sterile water and stored at -
20 ?C. The DFHBI-binding aptamer was assembled by heating equimolar aliquots of strands A 
and B in 40 mM HEPES-KOH (pH=7.5) buffer containing 125 mM KCl and  5 mM MgCl2 for 3 
min at 65 ?C and then incubating them for 1 hour at 37 ?C. The DFHBI-binding aptamer was 
stored at -20 ?C. To construct RNA aptamer biosensor, DFHBI was bound to its two-stranded 
aptamer by combining equal volumes of 1.2 ?M  RNA and 1.2 ?M DFHBI, both were 
resuspended in 40 mM HEPES-KOH (pH=7.5) buffer containing 125 mM KCl and  5 mM 
MgCl2  and incubating them at 37 ?C for 15 min. 
Lead acetate (Pb(Ac)2) is from Sigma-Aldrich (St. Louis, MO, USA), RNase T1 is from 
Ambion (Grand Island, NY, USA), and RNase A is from Worthington Biochemical Corp., 
(Lakewood, NJ, USA). RNase A was diluted in 30 mM Tris-HCl, pH=7.5 buffer, RNase T1 and 
lead acetate were diluted in RNase-free water. RNA aptamer biosensor was suspended in 40 mM 
HEPES-KOH, 125 mM KCl, 5 mM MgCl2, 5 % DMSO, pH=7.5 buffer. All agents are prepared 
71 
 
at room temperature. The final concentrations generated in reactors were: 0, 0.05, 0.125, 0.25, 
0.375, 0.5, 1.25, 2.5, 3.75, 5, 12.5, 25, 37.5, 50 mM for lead acetate; 0, 0. 5, 1.25, 2.5, 3.75, 5, 
12.5, 25, 37.5, 50, 125, 250, 375, 500 U/?L for RNase T1; 0, 0.05, 0.125, 0.25, 0.375, 0.5, 1.25, 
2.5, 3.75, 5, 12.5, 25, 37.5, 50 U/?L for RNase A. 
Image acquisition and analysis 
Fluorescence image was obtained using a modified ArrayWoRx? biochip scanner 
(Applied Precision, Issaquah, WA, USA) with 480?15 nm of excitation and 530?20 nm of 
emission. All captured image were 16-bit grayscale, the resolution was 7,800 pixels per inch 
(PPI), and the pixel size was 3.25 ?m. The whole image (5.0 mm?3.8 mm) of multiplex reactor 
array chip was acquired by accumulation sequential image (1.25 mm?3.8 mm) four times (2.75 
s/scan). The whole image (34.7 mm?4.5 mm) of integrated chip was obtained by accumulating a 
sequential image (1.47 mm?4.5 mm) twenty-seven times (2.89 s/scan). ArrayWoRx? 2.5 
software suits (Applied Precision, Issaquah, WA, USA) built into the workstation automatically 
converted the images to an integrated image. The total scanning time of the integrated image (5.0 
mm?3.8 mm) and (34.7 mm?4.5 mm) was approximately 11 s and 78 s, respectively, exposure 
time was 0.01 s. The fluorescence images were digitized using ImageJ software (NIH, Bethesda, 
MD, USA).   
To characterize RNA biosensor degradation, a reference image was captured before 
mixing reagents. Right after mixing, time-lapse images of reactors are obtained by 12 
consecutive image scans. Total incubation time is 60 min. Reagents are circulated within reactors 
before each scan. The intensity of 12 consecutive images is compared by intensity of the 
reference image.  
 
72 
 
5.4 Results and Discussion 
Design and working principle of fluorescent RNA aptamer biosensor 
The fluorescent RNA aptamer biosensor (S4) is constructed with a RNA aptamer and 
non-fluorescent 3,5-difluoro-4-hydroxybenzylidene imidazoline (DFHBI) (Figure 5.1a). When 
bounded to the RNA aptamer, DFHBI becomes fluorescent under excitation. The two-stranded 
design increases sensitivity of this biosensor. In the presence of a RNA degrading agent, RNA 
aptamer is cleaved and DFHBI is released. As DFHBI is a conditional fluorophore, its free form 
is not fluorescent (Figure 5.1b).        
 
Figure 5.1 Structure and working principle of biosensor. (a) Sequence and structure of biosensor. (b) Working 
principle of biosensor. 
 
73 
 
Validation of biosensor in a microfluidic chip  
We first demonstrated that our biosensor can detect RNA degrading agents in a 
microfluidic chip that was described in chapter 1 (Figure C1 in Appendix C) [46]. The device is 
composed of two units of 12 reactors. The 12 reactors are divided into 4 sub-units of 3 reactors 
each, allowing testing 8 different conditions in triplicate in a single experiment. Each reactor is 
divided into two halves by a central valve. When the valve is closed, two different reagents can 
be introduced in each half-reactor. When the valve is open the two reagents are mixed by 
diffusion. In the first unit, we introduced RNase-free water, the RNA aptamer without dye, the 
dye without aptamer, and the RNA biosensor in triplicate in the 4 sub-units. In the second unit, 
the central valve was closed and the biosensor was introduced in the first half of the reactors. The 
second half was then filled with RNase-free water, RNase T1 (3 U/?L), RNase A (0.5 U/?L), or 
lead acetate (50 mM). As a control, we introduced RNase-free water, the RNA aptamer, DFHBI, 
and the biosensor S4, alone, in triplicate, in the 4 sub-units of the second unit. At t=0, the valve 
was open and fluorescence intensity was measured at fixed time with a modified biochip scanner. 
When incubated alone or with RNase-free water, fluorescence remained constant in time 
showing the stability of the RNA aptamer-DFHBI complex and the possibility to detect nanoliter 
volumes of RNA aptamer biosensors in a microfluidic environment. A significant decrease of 
fluorescence level was observed when the RNA biosensor was incubated with Pb(Ac)2 and 
RNase A, suggesting that a single cleavage in the two-stranded RNA aptamer is sufficient to 
extinguish fluorescence. However, we observed that RNase T1 did not affect the fluorescence at 
3 U/?L concentration, probably due to the low number of unpaired guanosine residues, which 
are specifically recognized by RNase T1, in the RNA aptamer [178] (Figure C1c in Appendix C). 
 
 
74 
 
Characterization of concentration- and time- dependent degradation of biosensor  
To determine the minimum detectable concentration of an RNA degrading agent and 
quantify the concentration- and time-dependant degradation process, we used a microfluidic chip 
that we previously developed for wide range concentration gradient formation [56] (Figure 
5.2a). The device has three gradient generators (GGs) with 4 parallel processors each. Every 
processor contains a metering section and a reactor section, separated by a pneumatic valve. 
Each reactor integrates a valve-based peristaltic pump for mixing of the reagents in the reactor. A 
step-by-step process of the chip operation is shown in Figure 5.2b. Briefly, a log scale 
concentration gradient of RNA degrading reagents (Pb(Ac)2, RNase T1, or RNase A) was 
generated in the GGs (from 0 to 100 mM, 0 to 1000 U/?L, and 0 to 100 U/?L, respectively) and 
introduced into the first half of the reactors. Fluorescent RNA biosensor was introduced into the 
second half of the reactor. Consequently, the concentration of the degrading agent was different 
for each reactor while the concentration of the RNA biosensor was identical along the chip. Both 
reagents were 2-fold diluted after mixing inside reactors. The fluorescent intensity inside each 
reactor was measured with a modified biochip scanner at specific times. A curve of the 
normalized fluorescence intensity was then plotted for each time point. As a preliminary 
experiment, we generated a gradient of our RNA aptamer biosensor to determine the minimum 
fluorescence intensity, and thus the minimum amount of RNA aptamer biosensor, that could be 
monitor with our system. As shown in Figure C2 in Appendix C, we could detect as low as 0.84 
?M of RNA aptamer biosensor. 
75 
 
 
Figure 5.2 Concentration gradient formation of degrading agents on a microfluidic chip. (a) Microfluidic chip with 
concentration gradient generators, (b) Step-by-step process of concentration gradient formation.    
 
As expected, we observed that the degrading process is a concentration- and time-
dependent process (Figure 5.3). For each of the three degrading reagents, we observed a 
decrease of the fluorescence intensity when the concentration of the degrading agent reached a 
threshold concentration, 0.375 mM, 37.5 U/?L and 0.375 U/?L for Pb(Ac)2, RNase T1, and 
RNase A, respectively. Below the threshold concentrations, the presence degrading agents did 
not influence the fluorescence intensity, while at high concentration, RNA biosensor was entirely 
degraded in the first time after mixing (Figure 5.3). After 60 min incubation, we were able to 
detect 0.375 mM of Pb(Ac)2, 37.5 U/?L of RNase T1 and 0.375 U/?l of RNase A. For Pb(Ac)2, 
we observed that at concentrations above 3.75 mM, the fluorescence intensity tended to increase 
76 
 
with the concentration. Close observation of the reactors under microscope showed aggregated 
particles inside the reactors (data not shown). We assumed that DFHBI recombined with random 
RNA oligomers aggregated by the high concentration of divalent ions (here Pb2+) [179] and 
recovered its fluorescence. 
The differences observed in the degradation pattern of both enzymes are due to the 
nucleotide composition of the RNA biosensor which has less cleavage target sites for RNase T1 
than RNase A. RNase A is a ubiquitous endoribonuclease that cleaves single-stranded RNA at 
3?-ends of unpaired cytosine and uracil residues [180]. In contrast, RNase T1 is an endonuclease 
that specifically cleaves RNA at 3?-ends of unpaired guanine residues [178]. Because 
degradation kinetics is different for RNase A and RNase T1, our detection system can support 
which of these two enzymes contaminates analyzed samples. In addition, the integration of 
microfluidic chip and portable detection system can provide a sensing platform for rapid and 
cost-effective RNase detection using this RNA biosensor. 
77 
 
 
78 
 
Figure 5.3 Characterization of the concentration- and time-dependent degradation of the biosensor. Degradation 
profile (top) and scanned images (bottom) of the biosensor by different concentration of (a,b) lead acetate, (c,d) 
RNase T1, and (e,f) RNase A. Data are presented as means ? SD of three independent experiments. 
 
5.5 Conclusion 
The integrity and purity of RNA is essential for the successful RNA-related analyses 
while RNase contamination can affect the accuracy of these results. To date, aptamer based 
biosensors have rarely been developed for detection of RNase contamination. In this paper, a 
novel fluorescent RNA aptamer-based biosensor was reported for the first time. We 
demonstrated the possibility to detect RNA degrading agents at very low concentration. The 
concentration- and time-dependent degrading profiles were rapidly obtained using only nanoliter 
volume of reagents. Our biosensor could detect degrading agent within minutes, while traditional 
gel electrophoresis-based methods require at least several hours [181]. We believe that our 
system will stimulate transcriptomic research world-wide, as laboratories lacking experience in 
RNA research will be able to easily monitor RNase contaminations in solutions and lab 
equipment. In addition, we demonstrated that microfluidic systems can be used to give 
alternative and complementary information for the development of molecular tools and, when 
integrated with a portable detection system, can provide a platform where rapid detection of 
RNases can be realized with this novel biosensor. 
 
 
 
 
79 
 
Chapter 6 
SUMMARY AND FUTURE PERSPECTIVES 
6.1 Summary 
In this dissertation, we investigated the concentration- and time-dependent response of 
cells (bacteria) or molecule (RNA aptamer-based biosensor) to reagents (antibiotics or degrading 
agents) by using microfluidic systems. We validated the growth of bacterial cells and detection of 
the RNA aptamer-based biosensor on a multiplex nanoliter reactor array chip. With an integrated 
microfluidic chip, we further investigated the dose-responses of bacterial growth to bactericidal 
antibiotics at a time-course manner and monitored the concentration- and time-dependent 
degradation of the biosensor by RNases and metal ion. This dissertation includes following 
contributions:  
1.  We developed a multiplex nanoliter reactor array chip that is capable of cultivating 
bacteria in a batch culture mode. We validated the bacterial growth phenotypes and quantified 
the inhibitory effect of antibiotics on bacterial growth in a microfluidic environment. This 
microfluidic platform is capable of precisely controlling multiple culture conditions with 
successful evaporation suppression. It is ease of operation without requiring complicated active 
mixing components and continuous flow of reagents. Our results demonstrated this device can be 
an alternative to a conventional batch culture assay. Moreover, this device is highly applicable to 
develop a high-throughput platform for screening potential drug molecules for lead identification 
in drug discovery. 
2.  We utilized an integrated microfluidic platform which can generate a wide range of 
different concentrations and grow bacteria in a zero-flow environment to evaluate dose-response 
80 
 
of bacterial growth to bactericidal antibiotics. This device can prevent cross-contamination of 
regents and mimic conventional batch culture mode. We determined EC50, an antibiotic 
concentration which induces 50 % inhibitory effect. Our results are the first effort to demonstrate 
the efficacy of using microfluidic devices in the time-course analysis of EC50. Our time-course 
analysis of EC50 could provide a method to evaluate the attenuated inhibitory effect as a result of 
bacteria?s resistance to antibiotics during their growth.  This device is a potential alternative to 
replace conventional microtiter plate to determine drug?s potency for lead optimization in drug 
discovery.   
3.  The degradation profile of a novel RNA aptamer-based biosensor by a gradient of 
different degrading agent concentrations at specific time intervals can be rapidly obtained with 
the integrated microfluidic platform. The degradation profiles of RNA biosensors can be used for 
differentiating degrading agents. Our microfluidic approach is expected to replace costly, time-
and labor-intensive conventional methods. This biosensor can contribute to stimulating 
transcriptomic researches to monitor RNase contaminations. In addition, our microfluidic 
platform can be integrated with this biosensor to be a detection device for rapid detection of 
RNase contaminations. 
 
6.2 Future perspectives  
Since tremendous efforts have been devoted to the development of microfluidic systems 
in recent decades, microfluidic systems have brought up a new route for pharmaceutical drug 
discovery and biochemical analysis. This dissertation focused on assessing the concentration- 
and time-dependent response of cell (bacteria) or molecule (RNA aptamer-based biosensor) to 
reagents (antibiotics or degrading agents) by using microfluidic systems. These microfluidic 
81 
 
systems can be further applied to a variety of cell types and molecules, such as, yeast, 
mammalian cells, proteins, and peptides. By increasing the number of reactors of multiplex 
reactor chip, high-throughput screening of metabolites, peptides, and proteins can be achieved 
for drug discovery with reduced time and cost. With the concept of integration of multiplex 
reactor array chip and biosensors with a portable detection system, a hand-held device for rapid 
in-field detection of RNases can be developed. Furthermore, these platforms can be applied to 
investigate concentration-dependent effects of different molecules such as ions, nutrients, or 
antibiotic degrading enzymes on bacterial growth. In the absence of new pharmaceuticals, 
multidrug resistance pathogens are frequently treated with combinations of two or more 
antibiotics. Combination therapy provides a potential means to alleviate the emergence of drug 
resistance traits as microbial pathogens are less likely to simultaneously develop mutations that 
have resistance to multiple antibiotics [182].  The effects of antibiotic combinations show either 
improved (synergistic) or reduced (antagonistic) efficiency compared to each antibiotic applied 
individually. The effects of antibiotic combinations can be evaluated using our microfluidic 
systems. 
 
 
 
 
 
 
 
 
82 
 
Reference 
1. DeLong, E.F. and N.R. Pace, Environmental diversity of bacteria and archaea. Syst Biol, 
2001. 50(4): p. 470-478. 
2. Lim, D., Microbiology. third ed. 2003, Dubuque, Iowa, USA: Kendall/Hunt. 
3. Anissimov, M. How Many Species of Bacteria Are There? 2013  [cited 2013 10/30]. 
4. Biodiversity 2010: the tip of the iceberg. Nat Rev Micro, 2010. 8(6): p. 384-384. 
5. CDC, Escherichia coli Centers for Disease Control and Prevention. 
6. Hartl, D.L., Genetics: analysis of genes and genomes. 8th ed. 2012, Burlington, MA, 
USA: Jones & Bartleet learning. 
7. Bochner, B.R., Global phenotypic characterization of bacteria. FEMS Microbiol. Rev., 
2009. 33(1): p. 191-205. 
8. Karr, J.R., et al., A whole-cell computational model predicts phenotype from genotype. 
Cell, 2012. 150(2): p. 389-401. 
9. Shuler, M.L. and F. Kargi, Bioprocess engineering. 2nd ed. 2002, Upper Saddle River, 
NJ, USA: Prentice Hall. 
10. Alchon, S.A., A Pest in the Land: New World Epidemics in a Global Perspective. 2003, 
Albuquerque, N.M, USA: University of New Mexico Press. 
11. Fleming, A., On the Antibacterial Action of Cultures of a Penicillium, with Special 
Reference to their Use in the Isolation of B. influenz?. Br J Exp Pathol., 1929. 10(3): p. 
226-236. 
12. Lancini, G., F. Parenti, and G.G. Gallo, Antibiotics A Multidisciplinary Approach. 1995, 
New York, USA: Plenum Press. 
83 
 
13. Bauer, A.W., et al., Antibiotic susceptibility testing by a standardized single disk method. 
Am J Clin Pathol, 1966. 45(4): p. 493-496. 
14. Alfonso J, A., Resistance to Antibiotics: Are We in the Post-Antibiotic Era? Archives of 
Medical Research, 2005. 36(6): p. 697-705. 
15. Bax, R., N. Mullan, and J. Verhoef, The millennium bugs: the need for and development 
of new antibacterials. International Journal of Antimicrobial Agents, 2000. 16(1): p. 51-
59. 
16. Norrby, S.R., C.E. Nord, and R. Finch, Lack of development of new antimicrobial drugs: 
a potential serious threat to public health. The Lancet Infectious Diseases, 2005. 5(2): p. 
115-119. 
17. Silver, L.L. and K.A. Bostian, Discovery and development of new antibiotics: the 
problem of antibiotic resistance. Antimicrob Agents Chemother, 1993. 37(3): p. 377-83. 
18. Tonkens, R., An Overview of the Drug Development Process, in The Physician Executive. 
2005. p. 48-52. 
19. Liu, P., et al., Pharmacokinetic-pharmacodynamic modelling of antibacterial activity of 
cefpodoxime and cefixime in in vitro kinetic models. International Journal of 
Antimicrobial Agents, 2005. 25(2): p. 120-129. 
20. Kang, L., et al., Microfluidics for drug discovery and development: From target selection 
to product lifecycle management. Drug Discovery Today, 2008. 13(1): p. 1-13. 
21. Gupta, S.K., Drug Discovery and Clinical Research. 2011, New Delhi, India: Jaypee 
Brothers Medical Publishers. 
22. Clark, D.P. and N.J. Pazdernik, Molecular Biology. 2nd ed. 2013, Waltham, MA, USA: 
Academic Press. 
84 
 
23. Jan Barciszewski, B.F.C.C., RNA Biochemistry and Biotechnology. 1998, Netherlands: 
Kluwer Academic Publishers. 
24. Raeymaekers, L., Quantitative PCR: Theoretical Considerations with Practical 
Implications. Analytical Biochemistry, 1993. 214(2): p. 582-585. 
25. Imbeaud, S., et al., Towards standardization of RNA quality assessment using user-
independent classifiers of microcapillary electrophoresis traces. Nucleic Acids Research, 
2005. 33(6): p. e56-e56. 
26. Ellington, A.D. and J.W. Szostak, In vitro selection of RNA molecules that bind specific 
ligands. Nature, 1990. 346(6287): p. 818-822. 
27. Keefe, A.D., S. Pai, and A. Ellington, Aptamers as therapeutics. Nature Reviews Drug 
Discovery, 2010. 9(7): p. 537-550. 
28. Ni, X., et al., Nucleic acid aptamers: clinical applications and promising new horizons. 
Current medicinal chemistry, 2011. 18(27): p. 4206. 
29. Song, S., et al., Aptamer-based biosensors. TrAC Trends in Analytical Chemistry, 2008. 
27(2): p. 108-117. 
30. Brody, E.N. and L. Gold, Aptamers as therapeutic and diagnostic agents. Reviews in 
Molecular Biotechnology, 2000. 74(1): p. 5-13. 
31. Wang, L., et al., Fluorescent strip sensor for rapid determination of toxins. Chem. 
Commun., 2011. 47(5): p. 1574-1576. 
32. Du, Y., et al., Microfluidic Electrochemical Aptameric Assay Integrated On-Chip: A 
Potentially Convenient Sensing Platform for the Amplified and Multiplex Analysis of 
Small Molecules. Analytical Chemistry, 2011. 83(5): p. 1523-1529. 
85 
 
33. Xu, J.-P., et al., Label-free fluorescence detection of mercury(ii) and glutathione based 
on Hg2+-DNA complexes stimulating aggregation-induced emission of a 
tetraphenylethene derivative. Analyst, 2010. 135(11): p. 3002-3007. 
34. Xiao, H., T.E. Edwards, and A.R. Ferr?-D'Amar?, Structural Basis for Specific, High-
Affinity Tetracycline Binding by an In Vitro Evolved Aptamer and Artificial Riboswitch. 
Chemistry & Biology, 2008. 15(10): p. 1125-1137. 
35. Whitesides, G.M., The origins and the future of microfluidics. Nature, 2006. 442(7101): p. 
368-373. 
36. Weibel, D.B., W.R. DiLuzio, and G.M. Whitesides, Microfabrication meets microbiology. 
Nat Rev Micro, 2007. 5(3): p. 209-218. 
37. Saleh-Lakha, S. and J.T. Trevors, Perspective: Microfluidic applications in microbiology. 
Journal of Microbiological Methods, 2010. 82(1): p. 108-111. 
38. Abdelgawad, M. and A.R. Wheeler, Low-cost, rapid-prototyping of digital microfluidics 
devices. Microfluidics and nanofluidics, 2008. 4(4): p. 349-355. 
39. Boedicker, J.Q., et al., Detecting bacteria and determining their susceptibility to 
antibiotics by stochastic confinement in nanoliter droplets using plug-based microfluidics. 
Lab Chip, 2008. 8(8): p. 1265-1272. 
40. Mariella Jr, R., Sample preparation: the weak link in microfluidics-based biodetection. 
Biomedical microdevices, 2008. 10(6): p. 777-784. 
41. Auroux, P.-A., et al., Micro total analysis systems. 2. Analytical standard operations and 
applications. Analytical Chemistry, 2002. 74(12): p. 2637-2652. 
42. Kim, L., et al., Microfluidic arrays for logarithmically perfused embryonic stem cell 
culture. Lab on a Chip, 2006. 6(3): p. 394-406. 
86 
 
43. Flanagan, L.A., et al., Human neural stem cell growth and differentiation in a gradient-
generating microfluidic device. Lab on a Chip, 2005. 5(4): p. 401-406. 
44. Balagadde, F.K., et al., Long-Term Monitoring of Bacteria Undergoing Programmed 
Population Control in a Microchemostat. Science, 2005. 309(5731): p. 137-140. 
45. Groisman, A., et al., A microfluidic chemostat for experiments with bacterial and yeast 
cells. Nat. Methods, 2005. 2(9): p. 685-689. 
46. Institute, C.a.L.S., Methods for dilution antimicrobial susceptibility testing for bacteria 
that grew aerobically.Approved Standard M7-A10. 2009, Clinical and Laboratory 
Standards Institute: Wayne, PA. 
47. Mairhofer, J., K. Roppert, and P. Ertl, Microfluidic systems for pathogen sensing: a 
review. Sensors, 2009. 9(6): p. 4804-4823. 
48. Wheeler, A.R., et al., Microfluidic device for single-cell analysis. Analytical chemistry, 
2003. 75(14): p. 3581-3586. 
49. Dittrich, P.S. and A. Manz, Single-molecule fluorescence detection in microfluidic 
channels?the Holy Grail in ?TAS? Analytical and bioanalytical chemistry, 2005. 382(8): 
p. 1771-1782. 
50. Hong, J.W., et al., A nanoliter-scale nucleic acid processor with parallel architecture. 
Nat Biotech, 2004. 22(4): p. 435-439. 
51. Yang, M., et al., An integrated microfluidic device for influenza and other genetic 
analyses. Lab on a Chip, 2005. 5(10): p. 1024-1032. 
52. Ottesen, E.A., et al., Microfluidic Digital PCR Enables Multigene Analysis of Individual 
Environmental Bacteria. Science, 2006. 314(5804): p. 1464-1467. 
87 
 
53. Hansen, C.L., et al., A robust and scalable microfluidic metering method that allows 
protein crystal growth by free interface diffusion. Proceedings of the National Academy 
of Sciences, 2002. 99(26): p. 16531-16536. 
54. Zheng, B., et al., A Droplet?Based, Composite PDMS/Glass Capillary Microfluidic 
System for Evaluating Protein Crystallization Conditions by Microbatch and Vapor?
Diffusion Methods with On?Chip X?Ray Diffraction. Angewandte chemie international 
edition, 2004. 43(19): p. 2508-2511. 
55. Jambovane, S., et al., Determination of kinetic parameters, Km and kcat, with a single 
experiment on a chip. Anal. Chem., 2009. 81(9): p. 3239-3245. 
56. Yun, J.Y., et al., Log-Scale Dose Response of Inhibitors on a Chip. Analytical Chemistry, 
2011. 83(16): p. 6148-6153. 
57. Hamon, M., et al., Cell-Based Dose Responses from Open-Well Microchambers. 
Analytical Chemistry, 2013. 85(10): p. 5249-5254. 
58. Viravaidya, K., A. Sin, and M.L. Shuler, Development of a microscale cell culture 
analog to probe naphthalene toxicity. Biotechnology progress, 2008. 20(1): p. 316-323. 
59. Pregibon, D.C., M. Toner, and P.S. Doyle, Multifunctional encoded particles for high-
throughput biomolecule analysis. Science, 2007. 315(5817): p. 1393-1396. 
60. Unger, M.A., et al., Monolithic microfabricated valves and pumps by multilayer soft 
lithography. Science, 2000. 288(5463): p. 113-116. 
61. S. Pennathur, C.D.M., H. T. Soh, How to exploit the features of microfluidics technology. 
Lab on a Chip, 2008. 8(1): p. 20-22. 
88 
 
62. Heo, Y.S., et al., Characterization and Resolution of Evaporation-Mediated Osmolality 
Shifts That Constrain Microfluidic Cell Culture in Poly(dimethylsiloxane) Devices. 
Analytical Chemistry, 2006. 79(3): p. 1126-1134. 
63. Capretto, L., et al., Micromixing within microfluidic devices. Microfluidics, 2011: p. 27-
68. 
64. Lee, C.-Y., et al., Microfluidic mixing: a review. International Journal of Molecular 
Sciences, 2011. 12(5): p. 3263-3287. 
65. Berthier, J., Micro-drops and digitial microfluidics. 2nd ed. 2013, Oxford, UK: Elsevier. 
66. Nguyen, N.-T. and S.T. Wereley, Fundamentals and Applications of Microfluidics 2nd ed. 
2006, Norwood, USA: ARTECH HOUSE. 
67. McDonald, J.C., et al., Fabrication of microfluidic systems in poly(dimethylsiloxane). 
Electrophoresis, 2000. 21(1): p. 27-40. 
68. Xia, Y. and G.M. Whitesides, Soft Lithography. Angewandte Chemie International 
Edition, 1998. 37(5): p. 550-575. 
69. Whitesides, G.M., et al., Soft lithography in biology and biochemistry. Annu Rev Biomed 
Eng, 2001. 3: p. 335-373. 
70. Chou, H.-P., M.A. Unger, and S.R. Quake, A Microfabricated Rotary Pump. Biomedical 
Microdevices, 2001. 3(4): p. 323-330. 
71. Ericsson, H.M. and J.C. Sherris, Antibiotic sensitivity testing. Report of an international 
collaborative study. Acta Pathol Microbiol Scand B Microbiol Immunol, 1971. 217: p. 
Suppl 217:1+. 
89 
 
72. Isenberg, H.D., A. Reichler, and D. Wiseman, Prototype of a fully automated device for 
determination of bacterial antibiotic susceptibility in the clinical laboratory. Appl 
Microbiol, 1971. 22(6): p. 980-986. 
73. Niles, W.D. and P.J. Coassin, Piezo- and solenoid valve-based liquid dispensing for 
miniaturized assays. Assay Drug Dev Technol, 2005. 3(2): p. 189-202. 
74. Ellson, R., et al., Transfer of low nanoliter volumes between microplates using focused 
acoustics?automation considerations. Journal of the Association for Laboratory 
Automation, 2003. 8(5): p. 29-34. 
75. Patrick R. Murray, E.J.B., James H. Jorgensen, Marie Louise Landry, and Michael A. 
Pfaller Manual of Clinical Microbiology. 9th ed. Susceptibility testing instrumentation 
and computerized expert systems for data analysis and interpretation, ed. F.M. Richter SS. 
2008, Washington, DC: ASM Press. 
76. Reller, L.B., et al., Antimicrobial susceptibility testing: a review of general principles and 
contemporary practices. Clinical infectious diseases, 2009. 49(11): p. 1749-1755. 
77. Diagnostics, S.H., MicroScan WalkAway plus System. 2008: USA. 
78. Sun, P., et al., High-throughput microfluidic system for long-term bacterial colony 
monitoring and antibiotic testing in zero-flow environments. Biosensors and 
Bioelectronics, 2011. 26(5): p. 1993-1999. 
79. Kim, K.P., et al., In situ monitoring of antibiotic susceptibility of bacterial biofilms in a 
microfluidic device. Lab on a Chip, 2010. 10(23): p. 3296-3299. 
80. Kim, J., et al., A microfluidic device for high throughput bacterial biofilm studies. Lab on 
a Chip, 2012. 12(6): p. 1157-1163. 
90 
 
81. Dertinger, S.K.W., et al., Generation of Gradients Having Complex Shapes Using 
Microfluidic Networks. Analytical Chemistry, 2001. 73(6): p. 1240-1246. 
82. Anna, S.L., N. Bontoux, and H.A. Stone, Formation of dispersions using ?flow focusing? 
in microchannels. Applied physics letters, 2003. 82(3): p. 364-366. 
83. Garstecki, P., et al., Formation of monodisperse bubbles in a microfluidic flow-focusing 
device. Applied Physics Letters, 2004. 85(13): p. 2649-2651. 
84. Thorsen, T., et al., Dynamic pattern formation in a vesicle-generating microfluidic device. 
Physical review letters, 2001. 86(18): p. 4163-4166. 
85. Garstecki, P., et al., Formation of droplets and bubbles in a microfluidic T-junction?
scaling and mechanism of break-up. Lab on a Chip, 2006. 6(3): p. 437-446. 
86. Churski, K., et al., Rapid screening of antibiotic toxicity in an automated microdroplet 
system. Lab on a Chip, 2012. 12(9): p. 1629-1637. 
87. Chen, D.L. and R.F. Ismagilov, Microfluidic cartridges preloaded with nanoliter plugs of 
reagents: an alternative to 96-well plates for screening. Current Opinion in Chemical 
Biology, 2006. 10(3): p. 226-231. 
88. Churski, K., P. Korczyk, and P. Garstecki, High-throughput automated droplet 
microfluidic system for screening of reaction conditions. Lab on a Chip, 2010. 10(7): p. 
816-818. 
89. Lee, W., et al., Predictive model on micro droplet generation through mechanical cutting. 
Microfluid. Nanofluid., 2009. 7(3): p. 431-438. 
90. Zeng, S., et al., Microvalve-actuated precise control of individual droplets in microfluidic 
devices. Lab on a Chip, 2009. 9(10): p. 1340-1343. 
91 
 
91. Bo-Chih Lin, Y.-C.S., Novel on-demand droplet generation for selective fluid sample 
extraction. Biomicrofluidics, 2012. 6(2): p. 024103. 
92. Richard Curtis Bird and B.F. Smith, Genetic Library Construction and Screening: 
Advanced Techniques and Applications. 2002, Berlin, Germany: Springer. 
93. DiNitto, J.P., L. Wang, and J.C. Wu, Continuous fluorescence-based method for 
assessing dicer cleavage efficiency reveals 3'overhang nucleotide preference. 
Biotechniques, 2010. 48(4): p. 303. 
94. Invitrogen, Is Your RNA Intact? Methods to Check RNA Integrity. 
95. Bochner, B.R., New technologies to assess genotype-phenotype relationships. Nat. Rev. 
Genet., 2003. 4(4): p. 309-314. 
96. Yoon, S.H., et al., Comparative multi-omics systems analysis of Escherichia coli strains 
B and K-12. Genome Biol., 2012. 13(5): p. R37. 
97. Schneider, M.V. and S. Orchard, Omics technologies, data and bioinformatics principles. 
Methods Mol. Biol., 2011. 719: p. 3-30. 
98. Lee, S.Y., D.Y. Lee, and T.Y. Kim, Systems biotechnology for strain improvement. 
Trends Biotechnol., 2005. 23(7): p. 349-358. 
99. Buescher, J.M., et al., Global network reorganization during dynamic adaptations of 
Bacillus subtilis metabolism. Science, 2012. 335(6072): p. 1099-1103. 
100. Covert, M.W., et al., Integrating high-throughput and computational data elucidates 
bacterial networks. Nature, 2004. 429(6987): p. 92-96. 
101. Yoon, S.H., et al., Parallel evolution of transcriptome architecture during genome 
reorganization. Genome Res., 2011. 21(11): p. 1892-1904. 
92 
 
102. Duetz, W.A., Microtiter plates as mini-bioreactors: miniaturization of fermentation 
methods. Trends Microbiol., 2007. 15(10): p. 469-475. 
103. Kumar, S., C. Wittmann, and E. Heinzle, Minibioreactors. Biotechnol. Lett., 2004. 26(1): 
p. 1-10. 
104. Marques, M.P., J.M. Cabral, and P. Fernandes, High throughput in biotechnology: from 
shake-flasks to fully instrumented microfermentors. Recent Pat. Biotechnol., 2009. 3(2): p. 
124-140. 
105. Li, N., A. Tourovskaia, and A. Folch, Biology on a chip: microfabrication for studying 
the behavior of cultured cells. Crit. Rev. Biomed. Eng., 2003. 31(5-6): p. 423-488. 
106. Bochner, B.R., P. Gadzinski, and E. Panomitros, Phenotype microarrays for high-
throughput phenotypic testing and assay of gene function. Genome Res, 2001. 11(7): p. 
1246-55. 
107. Funke, M., et al., Bioprocess control in microscale: scalable fermentations in disposable 
and user-friendly microfluidic systems. Microb. Cell. Fact., 2010. 9: p. 86. 
108. Kensy, F., C. Engelbrecht, and J. Buchs, Scale-up from microtiter plate to laboratory 
fermenter: evaluation by online monitoring techniques of growth and protein expression 
in Escherichia coli and Hansenula polymorpha fermentations. Microb. Cell. Fact., 2009. 
8: p. 68. 
109. Zhang, Z., et al., Microchemostat-microbial continuous culture in a polymer-based, 
instrumented microbioreactor. Lab Chip, 2006. 6(7): p. 906-913. 
110. Dai, W., et al., A prototypic microfluidic platform generating stepwise concentration 
gradients for real-time study of cell apoptosis. Biomicrofluidics, 2010. 4(2): p. 024101. 
93 
 
111. Velve-Casquillas, G., et al., Microfluidic tools for cell biological research. Nano Today, 
2010. 5(1): p. 28-47. 
112. Yeo, L.Y., et al., Microfluidic devices for bioapplications. Small, 2011. 7(1): p. 12-48. 
113. Gan, M., et al., Massively parallel bacterial and yeast suspension culture on a chip. 
Small, 2012. 8(6): p. 863-867. 
114. G?nberger, A., et al., A disposable picolitre bioreactor for cultivation and investigation 
of industrially relevant bacteria on the single cell level. Lab Chip, 2012. 12(11): p. 2060-
2068. 
115. Vyawahare, S., A.D. Griffiths, and C.A. Merten, Miniaturization and parallelization of 
biological and chemical assays in microfluidic devices. Chem. Biol., 2010. 17(10): p. 
1052-1065. 
116. Lecault, V., et al., High-throughput analysis of single hematopoietic stem cell 
proliferation in microfluidic cell culture arrays. Nat. Methods, 2011. 8(7): p. 581-586. 
117. Young, E.W. and D.J. Beebe, Fundamentals of microfluidic cell culture in controlled 
microenvironments. Chem. Soc. Rev., 2010. 39(3): p. 1036-1048. 
118. Cook, A. and J. Lederberg, Recombination studies of lactose nonfermenting mutants of 
Escherichia coli K-12. Genetics, 1962. 47: p. 1335-1353. 
119. Seeton, C., Viscosity?temperature correlation for liquids. Tribology Letters, 2006. 22(1): 
p. 67-78. 
120. Urbanski, J.P., et al., Digital microfluidics using soft lithography. Lab Chip, 2006. 6(1): p. 
96-104. 
121. Liu, J., C. Hansen, and S.R. Quake, Solving the ?World-to-Chip? Interface Problem with 
a Microfluidic Matrix. Anal. Chem., 2003. 75(18): p. 4718-4723. 
94 
 
122. Heo, Y.S., et al., Characterization and resolution of evaporation-mediated osmolality 
shifts that constrain microfluidic cell culture in poly(dimethylsiloxane) devices. Anal. 
Chem., 2007. 79(3): p. 1126-1134. 
123. Reichenbach, H., The ecology of the myxobacteria. Environ. Microbiol., 1999. 1(1): p. 
15-21. 
124. Paguirigan, A.L. and D.J. Beebe, Microfluidics meet cell biology: bridging the gap by 
validation and application of microscale techniques for cell biological assays. BioEssays, 
2008. 30(9): p. 811-821. 
125. Monod, J., The growth of bacterial cultures. Annu. Rev. Microbiol., 1949. 3(1): p. 371-
394. 
126. Gorke, B. and J. Stulke, Carbon catabolite repression in bacteria: many ways to make 
the most out of nutrients. Nat Rev Micro, 2008. 6(8): p. 613-624. 
127. Garau, J. and L. Gomez, Pseudomonas aeruginosa pneumonia. Curr. Opin. Infect. Dis., 
2003. 16(2): p. 135-143. 
128. Szita, N., et al., Development of a multiplexed microbioreactor system for high-
throughput bioprocessing. Lab Chip, 2005. 5(8): p. 819-826. 
129. Lee, H.L.T., et al., Microbioreactor arrays with integrated mixers and fluid injectors for 
high-throughput experimentation with pH and dissolved oxygen control. Lab Chip, 2006. 
6(9): p. 1229-1235. 
130. Zuluaga, A.F., et al., Application of microbiological assay to determine pharmaceutical 
equivalence of generic intravenous antibiotics. BMC Clin. Pharmacol., 2009. 9: p. 1. 
95 
 
131. Silva, E., et al., Comparative in vitro study of the antimicrobial activities of different 
commercial antibiotic products for intravenous administration. BMC Clin. Pharmacol., 
2010. 10: p. 3. 
132. Raty, S., et al., Embryonic development in the mouse is enhanced via microchannel 
culture. Lab Chip, 2004. 4(3): p. 186-190. 
133. Yu, H., et al., Diffusion dependent cell behavior in microenvironments. Lab Chip, 2005. 
5(10): p. 1089-1095. 
134. Nayyar, G.M.L., et al., Poor-quality antimalarial drugs in southeast Asia and sub-
Saharan Africa. The Lancet Infectious Diseases, 2012. 12(6): p. 488-496. 
135. On-demand micro-encapsulation utilizing on-chip synthesis of semi-permeable alginate-
PLL capsules. Microfluidics and Nanofluidics, 2010. 
136. De La Fuente, R., et al., Small molecules with antimicrobial activity against E. coli and P. 
aeruginosa identified by high-throughput screening. British Journal of Pharmacology, 
2006. 149(5): p. 551-559. 
137. Huser, J., High-throughput screening in drug discovery. 2006, Stuttgart, Germany: 
Wiley-VCH. 
138. Jeong, H.-H., et al., Microfluidic monitoring of Pseudomonas aeruginosa chemotaxis 
under the continuous chemical gradient. Biosensors and Bioelectronics, 2010. 26(2): p. 
351-356. 
139. Englert, D.L., M.D. Manson, and A. Jayaraman, Flow-based microfluidic device for 
quantifying bacterial chemotaxis in stable, competing gradients. Appl Environ Microbiol, 
2009. 75(13): p. 4557-4564. 
96 
 
140. Kalinin, Y., et al., Responses of Escherichia coli bacteria to two opposing 
chemoattractant gradients depend on the chemoreceptor ratio. J Bacteriol, 2010. 192(7): 
p. 1796-1800. 
141. Guo, M.T., et al., Droplet microfluidics for high-throughput biological assays. Lab on a 
Chip, 2012. 12(12): p. 2146-2155. 
142. Maharbiz, M.M., et al., A microfabricated electrochemical oxygen generator for high-
density cell culture arrays. Microelectromechanical Systems, Journal of, 2003. 12(5): p. 
590-599. 
143. Jambovane, S., et al., Creation of Stepwise Concentration Gradient in Picoliter Droplets 
for Parallel Reactions of Matrix Metalloproteinase II and IX. Analytical Chemistry, 2011. 
83(9): p. 3358-3364. 
144. Eagon, R.G. and P.V. Phibbs Jr, Kinetics of Transport of Glucose, Fructose, and 
Mannitol by Pseudomonas aeruginosa. Canadian Journal of Biochemistry, 1971. 49(9): p. 
1031-1041. 
145. Reyes, D.R., et al., Micro total analysis systems. 1. Introduction, theory, and technology. 
ANALYTICAL CHEMISTRY-WASHINGTON DC-, 2002. 74(12): p. 2623-2636. 
146. Scudamore, R.A. and M. Goldner, Penetration of the outer membrane of Pseudomonas 
aeruginosa by synergistic combinations of beta-lactam and aminoglycoside antibiotics. 
Antimicrobial agents and chemotherapy, 1982. 21(6): p. 1007-1010. 
147. A microfluidic platform for on-demand formation and merging of microdroplets using 
electric control. Biomicrofluidics, 2011. 5(1): p. 011101. 
148. High-throughput sample introduction for droplet-based screening with an on-chip 
integrated sampling probe and slotted-vial array. Lab on a Chip, 2010. 10(21): p. 2864. 
97 
 
149. A gravity-actuated technique for flexible and portable microfluidic droplet manipulation. 
Microfluidics and Nanofluidics, 2010. 
150. Schweizer, H.P., Efflux as a mechanism of resistance to antimicrobials in Pseudomonas 
aeruginosa and related bacteria: unanswered questions. Genet Mol Res, 2003. 2(1): p. 
48-62. 
151. Temperature-induced droplet coalescence in microchannels. Biomicrofluidics, 2012. 6(1): 
p. 012811. 
152. Production of uniform droplets using membrane, microchannel and microfluidic 
emulsification devices. Microfluidics and Nanofluidics, 2012. 
153. Yi-Ming, T., W. Jhih-Jhe, and S. Yu-Chuan, A random-access microarray for 
programmable droplet storage, retrieval and manipulation. Journal of Micromechanics 
and Microengineering, 2012. 22(4): p. 045005. 
154. Ralf, S., et al., Droplet based microfluidics. Reports on Progress in Physics, 2012. 75(1): 
p. 016601. 
155. Bernot, A., Genome Transcriptome and Proteome Analysis. 2007, West Sussex, England: 
Wiley. 
156. Velculescu, V.E., et al., Characterization of the yeast transcriptome. Cell, 1997. 88(2): p. 
243-251. 
157. Wang, E.T., et al., Alternative isoform regulation in human tissue transcriptomes. Nature, 
2008. 456(7221): p. 470-476. 
158. Baugh, L.R., et al., Composition and dynamics of the Caenorhabditis elegans early 
embryonic transcriptome. Development, 2003. 130(5): p. 889-900. 
98 
 
159. Anisimov, S.V., E.G. Lakatta, and K.R. Boheler, Discovering altered genomic expression 
patterns in heart: transcriptome determination by serial analysis of gene expression. 
European journal of heart failure, 2001. 3(3): p. 271-281. 
160. Bakay, M., et al., A web-accessible complete transcriptome of normal human and DMD 
muscle. Neuromuscular Disorders, 2002. 12: p. S125-S141. 
161. Connolly, S.B., et al., Transcriptome profiling and the pathogenesis of diabetic 
complications. Journal of the American Society of Nephrology, 2003. 14(suppl 3): p. 
S279-S283. 
162. Meyerson, M., S. Gabriel, and G. Getz, Advances in understanding cancer genomes 
through second-generation sequencing. Nature Reviews Genetics, 2010. 11(10): p. 685-
696. 
163. Perez-Novo, C.A., et al., Impact of RNA quality on reference gene expression stability. 
Biotechniques, 2005. 39(1): p. 52-56. 
164. Fleige, S. and M.W. Pfaffl, RNA integrity and the effect on the real-time qRT-PCR 
performance. Molecular Aspects of Medicine, 2006. 27(2?3): p. 126-139. 
165. Chirgwin, J.M., et al., Isolation of biologically active ribonucleic acid from sources 
enriched in ribonuclease. Biochemistry, 1979. 18(24): p. 5294-5299. 
166. Ullrich, A., et al., Rat insulin genes: construction of plasmids containing the coding 
sequences. Science, 1977. 196(4296): p. 1313-1319. 
167. Srinivasan, M., D. Sedmak, and S. Jewell, Effect of fixatives and tissue processing on the 
content and integrity of nucleic acids. The American journal of pathology, 2002. 161(6): 
p. 1961. 
99 
 
168. Bustin, S.A. and T. Nolan, Pitfalls of quantitative real-time reverse-transcription 
polymerase chain reaction. Journal of biomolecular techniques: JBT, 2004. 15(3): p. 155-
166. 
169. Robert E. Farrell, J., RNA Methodologies: Laboratory Guide for Isolation and 
Characterization. 4th ed. 2009, London, UK: Elsevier Science. 
170. Ellington, A.D. and J.W. Szostak, Selection in vitro of single-stranded DNA molecules 
that fold into specific ligand-binding structures. Nature, 1992. 355(6363): p. 850-852. 
171. Bock, L.C., et al., Selection of single-stranded DNA molecules that bind and inhibit 
human thrombin. Nature, 1992. 355(6360): p. 564-566. 
172. Breaker, R.R., DNA enzymes. Nat Biotech, 1997. 15(5): p. 427-431. 
173. Joshi, R., et al., Selection, characterization, and application of DNA aptamers for the 
capture and detection of Salmonella enterica serovars. Molecular and cellular probes, 
2009. 23(1): p. 20-28. 
174. Holeman, L.A., et al., Isolation and characterization of fluorophore-binding RNA 
aptamers. Folding and Design, 1998. 3(6): p. 423-431. 
175. Stojanovic, M.N., P. De Prada, and D.W. Landry, Aptamer-based folding fluorescent 
sensor for cocaine. Journal of the American Chemical Society, 2001. 123(21): p. 4928-
4931. 
176. Nutiu, R. and Y. Li, Structure-switching signaling aptamers. Journal of the American 
Chemical Society, 2003. 125(16): p. 4771-4778. 
177. Cazenave, C. and O.C. Uhlenbeck, RNA template-directed RNA synthesis by T7 RNA 
polymerase. Proceedings of the National Academy of Sciences, 1994. 91(15): p. 6972-
6976. 
100 
 
178. Nicholson, A.W., Ribonucleases, Part A: Functional Roles and Mechanisms of Action, 
Volume 341 (Methods in Enzymology). 2001, San Diego, USA: Academic Press. 
179. Vlassov, A., Assay of random RNA oligomerisation in buffers with high concentrations of 
divalent metal ions. Nucleosides Nucleotides Nucleic Acids, 2004. 23(6-7): p. 999-1001. 
180. Rushizky, G.W., C.A. Knight, and H.A. Sober, Studies on the preferential specificity of 
pancreatic ribonuclease as deduced from partial digests. J Biol Chem, 1961. 236: p. 
2732-2737. 
181. Donald C. Rio, M.A.J., Gregory J. Hannon, and Timothy W. Nilsen, RNA: A Laboratory 
Manual. 2010, Cold Spring Harbor, NY, USA: CSHL Press. 
182. Chait, R., A. Craney, and R. Kishony, Antibiotic interactions that select against 
resistance. Nature, 2007. 446(7136): p. 668-671. 
 
 
 
 
 
 
 
 
 
 
 
 
 
101 
 
Appendix A 
 
Figure A1. Design of the multiplex microfluidic nanoliter chip. (a) Multi-layered structure of the device. (b) 
Photographs of the top view of the nanoliter reactor system (scale bar, 2 mm) and the enlarged version of the array 
comprising 2 ? 12 parallel nanoliter culture reactors (scale bar, 500 ?m), which are filled with cell suspension 
(denoted as yellow lines) and culture media (denoted as blue lines, M1 to M8). The anti-evaporation and control 
channels are represented as green and red lines, respectively. In a single unit, an experiment can be performed in 
triplicates. The arrows denote the direction of flow. 
  
102 
 
 
 
Figure A2. Micrographs showing time profiles (24 h duration) corresponding to the amount of green dye, which 
was retained in the reactors at 37 ?C, based on the presence and/or absence of humidity control and/or anti-
evaporation channels?absence of humidity (upper left), in the presence of operation of anti-evaporation channels 
(lower left), in the presence of a humidified incubator (upper right), and in the presence of both humidified incubator 
and the operation of anti-evaporation channels (lower right) (scale bar, 100 ?m).  
103 
 
 
 
Figure A3. A graph showing the time profile corresponding to the number of lac- E. coli mutants, which was 
normalized to the initial number of cells. Cells were loaded into the nanoliter reactors filled with M9 minimal 
medium along with different carbon sources, namely, glucose (0.2 % wt/vol) (denoted as blue circles), lactose 
(0.2 %) (red triangles), and glucose (0.04 %), lactose (0.2 %) (black squares), as well as no carbon source (green 
triangles down). Error bar denotes the standard error of the mean from three replicate reactors. 
  
104 
 
 
 
Figure A4. A graph showing the correlation between cell density (?108 CFU/mL) and fluorescence intensity (F.I., 
represented in arbitrary unit (A.U.)) of PT5-EGFP. Error bar denotes the standard error of the mean from three 
replicate reactors. 
 
 
 
  
105 
 
 
Figure A5. Concentration-dependent inhibitory effect of gentamicin on PT5-EGFP strain. (a) Time-lapse 
fluorescence images of the 3 replicate nanoliter reactors. Scale bar, 400 ?m. Graphs showing the time profiles of cell 
106 
 
density measured as fluorescence signal in (b) nanoliter reactors and (c) 14-mL test tubes. Error bar denotes the 
standard error of the mean from three replicates. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
107 
 
Appendix B 
 
 
Figure B1. Step-by-step procedure to generate 14 concentrations of antibiotics. First, antibiotics and dilution buffer 
are introduced to metering channels through ?Antibiotics? and ?Buffer? inlets; bacterial cells are introduced into half 
of reactors through ?cell in? inlet. Then, metered antibiotics are introduced to reactors to generate 14 concentrations. 
Finally, antibiotics are mixed with bacterial cells by operating three mixing valves. (Figure is modified from [56])   
 
108 
 
 
Figure B2. Correlation of cell density and fluorescence intensity. A significant linear correlation was observed. Data 
are represented as means ? SD of three independent measurements.  
 
109 
 
 
110 
 
 
111 
 
Figure B3. Time-lapse fluorescence images of 14 reactors and fluorescence intensity of PT5-EGFP under the 
inhibitory effects of two bactericidal antibiotics in reactors and test tubes. (a) Gentamicin and (b) Ciprofloxacin. 
Data are represented as means ? SD of three independent experiments. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
112 
 
Appendix C 
 
Figure C1. Time-course fluorescence intensity profiles of each reagent a microfluidic chip. (a) A multiplex 
microfluidic reactor chip. (b) Scanned images of reactors. (c) Time-course fluorescence intensity profiles of reagents 
in reactors.  S4 ( ) represents biosensor (22.4 ?M). Water+S4 ( ) represents biosensor (11.2 ?M) 
which is 2-fold diluted with water. RNase T1+S4 ( ) represents biosensor (11.2 ?M) with RNase T1 (3 
U/?L). RNase A+S4 ( ) represents biosensor (11.2 ?M) with RNase A (0.5 U/?L). Pb(Ac)2+S4 represents 
biosensor (11.2 ?M) with lead acetate (50 mM). Fluorescent molecule ( ) represents DFHBI (20 ?M) which 
113 
 
is not conjugated with RNA aptamer. RNA ( ) represents RNA aptamer (22.4 ?M). Water ( ) 
represents RNase free water. Data are represented as means ? SD of three replicate reactors. 
 
 
Figure C2. Fluorescence intensity of biosensor (S4) concentration gradient. A good linear relation between S4 
concentration (0.84 ?M~11.2 ?M) and fluorescence intensity is obtained (R2=0.994). Data are presented as means ? 
SD of three independent experiments.    
 
 
 
 
 
 
 
114 
 
 
 
 
 
115 
 
 
116 
 
 
Figure C3. Time-lapse scanned images of reactors in which biosensor was degraded by (a) lead acetate, (b) RNase 
T1, and (c) RNase A, respectively. Reaction time is from 0 min to 60 min.