A RAPID ASSAY TO DETECT ANTIBIOTIC RESISTANCE WITH NOVEL 3D PRINTED 

PHARMACOKINETIC/PHARMACODYNAMIC TECHNOLOGIES 

 

 

 

 

 

By 

Andrew A. Heller 

 
 
 
 
 
 
 
 
 
 
 
 
 
 

A DISSERTATION 

Submitted to 

Michigan State University 

in partial fulfillment of the requirements 

for the degree of 

 

Chemistry – Doctor of Philosophy 

 

2019 

 

 

ABSTRACT 

A RAPID ASSAY TO DETECT ANTIBIOTIC RESISTANCE WITH NOVEL 3D PRINTED 

PHARMACOKINETIC/PHARMACODYNAMIC TECHNOLOGIES 

 

By 

Andrew A. Heller 

 

The over-prescription and misuse of antibiotics has led some bacterial strains to 

become resistant to one, multiple, or all currently available antibiotics. In order to treat 

an antibiotic resistant bacterial infection, a novel antibiotic or therapeutic is required.  Up 

until the last few years, there had been a trend of fewer new antibiotics due to 

pharmaceutical companies not pursuing antibiotic development. The continuous threat 

of antibiotic resistance and the lack of antibiotic research led to the National Action 

Plan, which called for novel, rapid diagnostic tools and new therapeutics to combat 

antibiotic resistance.  The main reason for the lack of interest in antibiotic development 

is that drug development now cost a pharmaceutical greater than $2.5 billion and can 

take over 10 years.  In addition, only 10.4% of drugs that enter clinical trials eventually 

are approved.  One of the main causes of this low success rate is that the drugs do not 

have the same pharmacokinetics (PK) or pharmacodynamics (PD) as the drugs did in in 

vitro and in vivo animal models.  These differences in PK/PD can lead to safety and 

efficacy concerns in humans.   

In this dissertation, this issue is combated with new technologies for antibiotic 

resistance identification.  A rapid, static susceptibility assay was created in order 

monitor the growth of a bacterial culture by measuring the extracellular ATP/OD600, 

which in a healthy culture should increase to a maximum during early logarithmic growth 

phase and then decrease.  Adding an antibiotic to a growing culture after this 

 

 

ATP/OD600 maximum led to an increase in the ATP/OD600, while a healthy culture 

decreased leading a statistical difference (α = 0.05) in 20 – 60 minutes after adding the 

antibiotic.  This increase in the ATP/OD600 was due to the antibiotic’s ability to 

effectively kill the bacteria by lysing leading to the OD600 remaining stable and 

extracellular ATP levels to increase. This trend was not seen when an antibiotic that the 

bacteria were resistant to was added.  This procedure could also determine which 

antibiotic is killing the most bacteria in a mixed bacterial culture. 

The above procedure was adapted to be dynamic in order to expose the bacteria 

to a PK curve similar to that seen in a human so more clinically-relevant PD data could 

be measured.  This was achieved by creating a fluidic, two compartment model that was 

3D printed, which utilized porous membrane inserts that were created by novel 3D 

printed procedures to incorporate the membranes into the 3D printing structure.  The 

device was characterized using fluorescein (332.31 g/mol) due to having similar 

properties to the antibiotic, levofloxacin (361.37 g/mol).  The devices were impervious to 

leaking and were successful in replicating PK curves for an oral, intermittent intravenous 

(IV), and continuous IV administration.  Replacing the fluorescein solution with a 

levofloxacin solution in growth media, bacteria were able to be exposed to an oral 

levofloxacin PK curve (Cmax = 12.4 ± 3.0 µM; tmax = 1 hour; half-life = 5.2 ± 0.5 hours). A 

kanamycin-resistant strain of Escherichia coli was determined to have a statistical 

difference in the ATP/OD600 when exposed to a levofloxacin concentration of 3.5 ± 1.3 

µM in the secondary compartment while a chloramphenicol-resistant strain of Bacillus 

subtilis showed a statistical difference at a concentration of 4.8 ± 1.8 µM. 

 

 

 

ACKNOWLEDGEMENTS 

 

 

I would like to start by thanking Michigan State University and the National 

Institutes of Health.  Their funding made this work possible.  I would also like to thank 

the College of Natural Science for travel funding, which allowed me to attend various 

conferences to present my work.   

 

I would like to thank the current members of guidance committee: Dr. Christopher 

Waters, Dr. Aaron Odom, Dr. Gary Blanchard. I would also like to thank the previous 

member of my guidance committee, Dr. Merlin Bruening.  Their guidance and 

recommendations concerning my research has been very helpful and is appreciated.  I 

would like to especially thank Dr. Waters for providing me with the instruments and 

research space required to perform my research until our lab could acquire the needed 

resources. 

 

I would like to thank my advisor, Dr. Dana Spence, who took a chance on an 

unfamiliar project and gave me the support and freedom to develop the project. I would 

also like to thank him for his support in me pursuing my Master of Science in 

Pharmacology & Toxicology, which provided a lot of the academic material required for 

this project.  His advice, support, and positive nature provided me with a comfortable 

environment where I could develop as a scientist.   

 

I would like to thank the past and present members of the Spence group.  I would 

like to thank Dr. Jayda Meisel and Dr. Sarah Lockwood for their previous research, 

which led to the development of my research.  I would like to especially thank Dr. Cody 

Pinger, Dr. Andre Castiaux, and DeMarcus Bunn.  The procedures that we all 

 

iv 

developed in order to incorporate membranes into 3D printed devices was the key to 

completing my research.  I would also like to thank all other current members of the 

Spence Lab who have provided much needed advice and encouragement: Dr. Suzanne 

Summers, Morgan Geiger, Lisa Meints, Monica Jacobs, and Samantha Winther. 

 

I would like to thank my family and friends for their support throughout my 

graduate experience. I would like to give a special thank you to my wife, Amanda 

LaFay.  No one has provided me with more support than her.  She was right there 

during the stress of my literature seminar, my second year oral, my Master’s Thesis, my 

dissertation and defense.  She had to endure listening to my presentations multiple 

times, both directly and indirectly as I practiced in another room. She could probably 

explain my research as well as I can. 

 

 

 

v 

TABLE OF CONTENTS 

 
 
 

LIST OF TABLES ............................................................................................................ ix 
 
LIST OF FIGURES .......................................................................................................... x 
 
CHAPTER 1  
INTRODUCTION ............................................................................................................. 1 
1.1 Introduction ..................................................................................................... 1 
1.2 Introduction to Bacteria, Antibiotics, and Antibiotic Resistant Bacteria ........... 2 
1.2.1  Bacteria and Classification .............................................................. 2 
1.2.2  Antibiotic Mechanism of Action ....................................................... 5 
1.2.3  Bacterial Mechanism of Resistance ................................................ 9 
1.2.4  Antibiotic Resistant Bacteria ......................................................... 13 
1.2.5  The Decrease of Antibiotic Development and Government 

Intervention ................................................................................... 15 
1.3 Introduction to Drug Development ................................................................ 20 
1.3.1  Pharmacokinetics and Pharmacodynamics .................................. 20 
1.3.2  In Vitro PK and PD Evaluation of Antibiotics ................................. 25 
1.3.3  In Vivo Animal PK/PD Evaluation of Antibiotics ............................ 30 
1.3.4  In Vivo Human PK/PD Evaluation of Antibiotics ............................ 32 
1.4 Current State of Affairs ................................................................................. 35 
REFERENCES .............................................................................................................. 36 
 
CHAPTER 2  
THE DEVELOPMENT OF A RAPID ASSAY FOR THE DETERMINATION OF 
ANTIBIOTIC SUSCEPTIBILITY IN GRAM-POSITIVE AND GRAM-NEGATIVE 
BACTERIAL STRAINS SHORTLY AFTER ANTIBIOTIC ADMINISTRATION ............... 48 
2.1 Introduction ................................................................................................... 48 

2.1.1  Current State of Drug of Antibiotic Development and Antibiotic 

Resistance .................................................................................... 48 
2.1.2  Antibiotic Susceptibility Testing ..................................................... 49 
2.1.3  Extracellular Adenosine Triphosphate during Bacterial Growth .... 52 
2.2  Materials and Methods ................................................................................. 54 
2.2.1  Growth Media & Agar Plate Preparation ....................................... 54 
2.2.2  Antibiotic Reagents ....................................................................... 54 
2.2.3  Bacterial Strains ............................................................................ 55 
2.2.4  Sample Preparation ...................................................................... 56 
2.2.5  Determination of OD600 and Extracellular ATP ............................ 59 
2.3  Results ......................................................................................................... 61 
2.3.1  ATP/OD600 of Growing Gram-Negative, KanR E. coli .................. 61 
2.3.2  Procedural Filtering and Centrifugation Compared to Just 

Centrifugation ................................................................................ 66 
2.3.3  Growing KanR E. coli Challenged with Antibiotic .......................... 68 

 

vi 

2.3.4  Growing KanLevR E coli Challenged with Antibiotic ..................... 74 
2.3.5  ATP/OD600 of Growing Gram-Positive, CmpR B. subtilis ............ 76 
2.3.6  Growing CmpR B. subtilis Challenged with Antibiotic ................... 80 
2.3.7  The Effect of One or Two Antibiotics on a Combined Bacterial 

Mixture .......................................................................................... 82 
2.4  Discussion .................................................................................................... 84 
REFERENCES .............................................................................................................. 88 
 
CHAPTER 3  
THE DEVELOPMENT OF A 3D PRINTED, TWO COMPARTMENT MODEL FOR PK 
MODELING OF VARIOUS ADMINISTRATIONS WITH NOVEL 3D PRINTED INSERTS 
CONTAINING INTEGRATED MEMBRANES ................................................................ 94  
3.1 Introduction to 3D Printing ............................................................................. 94 
3.1.1  Fused Deposition Modeling ........................................................... 97 
3.1.2  Polyjet Printing .............................................................................. 98 
3.1.3  Print-Pause-Print ........................................................................... 99 
3.1.4  Z-axis Drop Method ..................................................................... 102 
3.2  Materials and Methods ............................................................................... 105 
3.2.1  Design, Printing, and Assembly of 3D Printed Parts ................... 105 
3.2.2  Design and Printing of Membrane Inserts ................................... 110 
3.2.3  Characterization of the Device and Inserts with Fluorescein ....... 119 
3.3  Results ....................................................................................................... 123 
3.3.1  Fluorescein Leakage Testing of Both Inserts .............................. 123 
3.3.2  Fluorescein Leakage Testing of the Foldable Inserts .................. 125 
3.3.3  Oral PK Model using Fluorescein with the Foldable Inserts ........ 131 
3.3.4  Various Administration PK Models using Fluorescein with the 

Nonfoldable Inserts ..................................................................... 133 
3.4  Discussion .................................................................................................. 139 
REFERENCES ............................................................................................................ 144 
 
CHAPTER 4  
ANTIBIOTIC SUSCEPTIBILITY TESTING OF BACTERIA VIA A DYNAMIC 
ADMINISTRATION OF ANTIBIOTICS USING A 3D PRINTED TWO COMPARTMENT 
PK/PD MODEL  ........................................................................................................... 148  
4.1 Introduction ................................................................................................. 148 

4.1.1  Current State of Antibiotic Resistance, Development, and 

Susceptibility Testing .................................................................. 148 
4.1.2  Overview of Research from Previous Chapters .......................... 149 
4.2  Material and Methods ................................................................................. 152 

4.2.1  Oral Pharmacokinetic Profile of Levofloxacin  

Using Fluorescein ....................................................................... 152 
4.2.2  Growth Media and Agar Plates ................................................... 156 
4.2.3  Antibiotic Reagents ..................................................................... 156 
4.2.4  Bacterial Strains .......................................................................... 157 
4.2.5  Sample Preparation .................................................................... 157 
4.2.6  Determination of Extracellular ATP and OD600 .......................... 159 

 

vii 

4.3  Results ....................................................................................................... 161 

4.3.1  Oral Pharmacokinetic Profile of Levofloxacin Using Fluorescein 

 .................................................................................................... 161 

4.3.2  Bacterial Strains Challenged by an Oral PK Curve of Levofloxacin 

 .................................................................................................... 163 
4.4  Discussion .................................................................................................. 167 
REFERENCES ............................................................................................................ 170 
 
CHAPTER 5  
OVERALL CONCLUSIONS & FUTURE WORK  ........................................................ 175  
5.1 Overall Conclusions .................................................................................... 175 
5.2  Future Work ............................................................................................... 179 
5.2.1  More Antibiotics and Bacterial Strains ......................................... 179 
5.2.2  More Human-Like Pharmacokinetics .......................................... 180 
5.2.3  Device Alterations ....................................................................... 183 
REFERENCES ............................................................................................................ 186 
 

 

 

viii 

LIST OF TABLES 

 

Table 1.1: The 5 Main Goals of the National Action Plan for Combating Antibiotic-
Resistant Bacteria ......................................................................................................... 19 
 

 

 

ix 

 

LIST OF FIGURES 

Figure 1.1: The orientation of cell membrane(s) and the peptidoglycan layer of Gram-
positive and Gram-negative bacteria. .............................................................................. 4 
 
Figure 1.2: The mechanisms of action of six common classes of antibiotics ................... 6 
 
Figure 1.3: The mechanism of action of an antibiotic and possible mechanisms of 
resistance. ..................................................................................................................... 12 
 
Figure 1.4: A brief history of the decreasing time frame between the introduction of 
some antibiotics and the first discovery of resistance to that antibiotic. ........................ 14 
 
Figure 1.5: New Antibiotics approved by the FDA. ........................................................ 17 
 
Figure 1.6: An example pharmacokinetic curve. ........................................................... 22 
 
Figure 1.7: Example pharmacodynamic curves for efficacy (blue) and toxicity 
(orange). ........................................................................................................................ 24 
 
Figure 1.8: Diagrams of the compartmental models and in vitro models consisting of 
static, dilution-based, or diffusion-based models ........................................................... 29 
 
Figure 2.1: The most common susceptibility tests. ........................................................ 51 
 
Figure 2.2: The procedure for the susceptibility assay. ................................................. 58 
 
Figure 2.3: The ATP/OD600 growth curve of KanR E. coli. ........................................... 62 
 
Figure 2.4: The individual extracellular ATP and OD600 curves during growth of KanR 
E. coli. ........................................................................................................................... 63 
 
Figure 2.5: The ATP/OD600 curve with removal of the background absorbance and an 
overlay with the same data with the background absorbance. ...................................... 65 
 
Figure 2.6: A comparison of normalized ATP/OD600 of growing KanR E. coli samples 
that were filtered and centrifuged or just centrifuged prior to extracellular ATP 
determination. ............................................................................................................... 67 
 
Figure 2.7: The normalized ATP/OD600 of growing KanR E.coli challenged with 
levofloxacin. .................................................................................................................. 69 
 
Figure 2.8: The effects of levofloxacin on the normalized ATP/OD600 of growing KanR 
E. coli with only a 30 minute preincubation. .................................................................. 71 

 

x 

Figure 2.9: The effects of tetracycline on the normalized ATP/OD600 of growing KanR 
E.coli. ............................................................................................................................ 73 
 
Figure 2.10: A growing KanLevR E. coli culture was challenged with levofloxacin and 
gentamicin. .................................................................................................................... 75 
 
Figure 2.11: The ATP/OD600 growth curve of CmpR B. subtilis. .................................. 77 
 
Figure 2.12: The individual extracellular ATP and OD600 curves of B. subtilis ............. 79 
 
Figure 2.13: The normalized ATP/OD600 of growing CmpR B.subtilis challenged with 
kanamycin ..................................................................................................................... 81 
 
Figure 2.14: The single- or dual-antibiotic effect on a 50/50 bacterial mixture of KanR E. 
coli and CmpR B. subtilis .............................................................................................. 83 
 
Figure 3.1: The process of 3D printing a device. ........................................................... 95 
 
Figure 3.2: The Print-Pause-Print method for integrating membranes in 3D printed 
objects using a Stratasys J750 printer......................................................................... 101 
 
Figure 3.3: The Z-axis drop method for integrating membranes into 3D printed objects 
using a Stratasys J750 printer. .................................................................................... 103 
 
Figure 3.4: The layout of the main device. .................................................................. 106 
 
Figure 3.5: The dimension of the main device. ............................................................ 107 
 
Figure 3.6: The dimensions of the slide shield and exhaust. ....................................... 109 
 
Figure 3.7: The dimensions of the parts required to create the membrane inserts. .... 111 
 
Figure 3.8: The assembly procedure of the device and foldable inserts. .................... 114 
 
Figure 3.9: The procedure for printing the non-foldable insert. .................................... 117 
 
Figure 3.10: The assembly procedure of the device and nonfoldable inserts. ............. 118 
Figure 3.11: The normalized fluorescent signal in the secondary compartment of a 
device with a block in the primary compartment to test for leakage of the device-insert 
joint. ............................................................................................................................. 124 
 
Figure 3.12: The normalized fluorescent signal in both compartments of a device with a 
foldable insert with no openings or membranes in the primary compartment to test the 
design of the foldable inserts for leakage at the seams............................................... 126 

 

xi 

Figure 3.13: The normalized fluorescent signal in both compartments of a device with a 
foldable insert with nonporous membranes in the primary compartment to test the 
integration of the membranes into the insert for undesired leakage. ........................... 128 
 
Figure 3.14: The normalized fluorescent signal in the secondary compartment of a 
device with a foldable insert with nonporous membranes containing fluorescein in the 
primary compartment to test the integration of the side membrane into the insert for 
undesired leakage. ...................................................................................................... 130 
 
Figure 3.15: The PK curves for both the primary and secondary compartment for an oral 
model of levofloxacin using fluorescein with the foldable inserts. ................................ 132 
 
Figure 3.16: The PK curves for both the primary and secondary compartment for an oral 
model of levofloxacin using fluorescein with the nonfoldable inserts. .......................... 134 
 
Figure 3.17: The PK curves for both the primary and secondary compartment for an 
intermittent IV model using fluorescein with the nonfoldable inserts. .......................... 136 
 
Figure 3.18: The PK curves for both the primary and secondary compartments for a 
continuous IV model using fluorescein with the nonfoldable inserts. ........................... 138 
 
Figure 4.1: The adaption of the orbital shaker and incubator to allow for a syringe pump 
to be attached to the devices. ..................................................................................... 154 
 
Figure 4.2: Dimensions of custom lab ware and custom modifications ....................... 155 
 
Figure 4.3: The PK curves for both the primary and secondary compartment for an oral 
model of levofloxacin using fluorescein. ...................................................................... 162 
 
Figure 4.4: Normalized ATP/OD600 signals of a kanamycin-resistant E. coli strain 
exposed to a simulated levofloxacin oral administration PK curve. ............................. 164 
 
Figure 4.5: Normalized ATP/OD600 signals of a chloramphenicol-resistant B. subtilis 
strain exposed to a simulated levofloxacin oral administration PK curve. ................... 166 
 
Figure 5.1: Designs of alternations that could benefit the PK/PD data. ....................... 185 
 

 

xii 

CHAPTER 1 

INTRODUCTION 

 

1.1 Introduction 

The rise of antibiotic resistance in bacteria and the loss of development of 

antibiotics by pharmaceutical companies has created a situation where some strains of 

bacteria have developed resistance to multiple, if not all, antibiotics currently available.  

In order to make antibiotics a more worthy investment to pharmaceutical companies, the 

cost of development of an antibiotic needs to decrease.  The best route to decrease the 

cost of antibiotic development is to improve the success rate of antibiotics that enter 

clinical trials by providing more clinically-relevant data earlier in the developmental 

process. This is best achieved by challenging bacteria to the pharmacokinetics (PK) of 

an antibiotic that it will experience in a human during in vitro testing in order to get 

pharmacodynamic (PD) data that may better correlate with data seen in a human; 

thereby, decreasing the reliance on costly animal models and increasing the success 

rate in clinical trials.  However, an understanding of bacteria, antibiotics, and resistance 

of these antibiotics by the pathogen under investigation by the pathogen under 

investigation is required. 

 

 

 

 

 

 

1 

1.2  Introduction to Bacteria, Antibiotics, and Antibiotic Resistant Bacteria 

1.2.1 Bacteria and Classification 

Bacteria are classified as prokaryotes. Prokaryotes are single celled organisms 

that do not have membrane-bound organelles such as the nucleus, mitochondria, or 

lysosomes. Pathogenic bacteria are one of the main causes of infections; however, not 

all bacteria are pathogenic.1 There are multiple strains of bacteria that colonize the gut 

and skin of humans without causing harm. In fact, bacteria of the gut and skin can 

provide protection against pathogenic bacteria and other microbes.2-3 Nonetheless, if a 

pathogenic strain of bacteria is able to enter the body and reproduces, the bacteria can 

cause illnesses by various mechanisms that are unique to the strain of bacteria.4  

Bacteria can be further classified as Gram-positive or Gram-negative, which is 

determined by the ability of a bacterial cell to retain a crystal violet-iodine complex 

(Gram-stain). The retention of the Gram-stain is caused by the thickness of the 

peptidoglycan layer.5 Peptidoglycan is a polymer that forms in chains and consists of 

two alternating monomer subunits, namely N-acetylglucosamine and N-acetylmuramic 

acid. Enzymes stack individual strains of peptidoglycan to create layers with other 

peptidoglycan chains. Peptidoglycan layers provide support to the cell structure by 

counteracting the pressure on the cytoplasmic membrane caused by the cytoplasm. 

Gram-positive bacteria retain the Gram-stain because these bacteria have a thick 

peptidoglycan layer to provide support to the single phospholipid bilayer membrane that 

encapsulates the cell (Figure 1.1 A). Gram-negative bacteria do not retain the Gram-

stain due to a thinner layer of peptidoglycan. The peptidoglycan layer is thinner because 

it is surrounded by a second phospholipid bilayer that provides support (Figure 1.1 B). 

 

2 

The thickness of the peptidoglycan layer in Gram-negative bacteria is not adequate to 

retain the Gram-stain and is therefore easily washed away during the staining 

procedure.6 The second phospholipid bilayer membrane of Gram-negative bacteria 

sometimes requires more sophisticated antibiotics in order to gain entry into the 

bacterial cell.  

 

 

 

3 

Figure 1.1: The orientation of cell membrane(s) and the peptidoglycan layer of 
Gram-positive and Gram-negative bacteria. (A) Gram-positive bacteria have a thick 
peptidoglycan layer (green) that counteracts the pressure caused by the cytoplasm 
(light blue) on the cytoplasmic membrane (darker blue). (B) Gram-negative bacteria 
have a thin peptidoglycan layer due to additional support provided by the outer 
membrane (red). The thin peptidoglycan layer of Gram-negative bacteria does not retain 
the Gram-stain like the thick peptidoglycan layer of Gram-positive bacteria. 

 

 

4 

1.2.2 Antibiotic Mechanisms of Action 
 

Antibiotics are a group of drugs that either cause death (bactericidal) or inhibit 

growth and reproduction (bacteriostatic) of bacterial cells. These two types of antibiotics 

are not mutually exclusive as bacteriostatic antibiotics can become bactericidal if 

administered at high enough concentrations. There are numerous classes of antibiotics 

including beta-lactams, sulfonamides, aminoglycosides, quinolones, tetracyclines, 

macrolides, and quinolones. The mechanism of action of these classes of antibiotic are 

summarized in Figure 1.2.  

 

 

 

5 

Figure 1.2: The mechanisms of action of six common classes of antibiotics. Beta 
lactam antibiotics, like penicillins, affect cell wall synthesis by inhibiting enzymes 
required to layer peptidoglycan, which can lead to lysis due to weakened cell walls 
during cell replication that cannot counteract the pressure on the cytoplasmic 
membrane. Sulfonamides and quinolones inhibit nucleic acid synthesis by competing for 
dihydrofolate reductase or type II topoisomerase, respectively. Tetracycline and 
aminoglycosides bind to the 30S subunit while macrolides bind to the 50S subunit of the 
70S ribosome. These three classes all inhibit protein synthesis. 

 

 

6 

Beta-lactams were the first class of antibiotics with the discovery of penicillin by 

Alexander Fleming in 1928.7 Howard Florey and Ernst Chain were successfully able to 

isolate penicillin in 1939. The combined efforts of Fleming, Florey and Chain in the 

development of penicillin resulted in them being awarded the 1945 Nobel Prize in 

Physiology or Medicine.8 Beta-lactam antibiotics are bactericidal and their mechanism 

of action consists of inhibition of the enzymes that are responsible for the layering of 

peptidoglycan strains. During cell replication, the peptidoglycan cannot be layered 

leading to weakened cell walls that are unable counteract the internal pressure on the 

cytoplasmic membrane, these resulting in lysis.9-10  

The next class of antibiotics is the sulfonamides. Gerhard Domagk developed the 

first sulfonamide, Prontosil, between the discovery and isolation of penicillin. Domagk’s 

work in the discovery of the first sulfonamide resulted in him being awarded the 1939 

Nobel Prize in Physiology or Medicine.11-12 Sulfonamides, which are considered to be 

bacteriostatic antibiotics, compete for the enzyme dihydrofolate reductase. This enzyme 

plays  a crucial role in the synthesis of the purine nucleotides: adenosine and guanine. 

Competition of this enzyme results in decreased synthesis of nucleic acids needed for 

the bacteria cell to grow and replicate.13-15 

 Aminoglycosides were the next class of antibiotics discovered. Streptomycin 

was the first aminoglycoside and was discovered by Selman Waksman in 1944, for 

which he was awarded the 1952 Nobel Prize in Physiology or Medicine.16 

Aminoglycosides are considered bactericidal antibiotics and bind to the 30S subunit of 

the 70S ribosome, which is a major site of messenger ribonucleic acid (mRNA) 

translation. The binding of aminoglycosides to the 30S subunit inhibits the elongation of 

 

7 

the peptide chain during protein formation. The aminoglycoside binding also disrupts the 

proofreading process of the peptide chain leading to more errors in the peptide 

sequence, thus helping to inhibit cell replication.17 

The tetracylines were the next antibiotic class to be discovered. The first 

tetracycline developed was aureomycin. Aueromycin was developed by Benjamin 

Duggar in 1945.18 Tetracylines are considered to be bacteriostatic antibiotics and  also 

bind to the 30S subunit of the 70S ribosome. However, tetracycline inhibits the binding 

of transfer RNA (tRNA) to the ribosome resulting in the peptide chain not being 

elongated because amino acids are not being added.19-20 

The next class of antibiotics were the macrolides, which includes Erythromycin 

developed in 1952 by James McGuire.21 Macrolides are considered to be bacteriostatic 

antibiotics and also bind to the 70S ribosome. Unlike aminoglycosides and tetracylcines, 

macrolides bind to the 50S subunit of the ribosome. The binding of the macrolide blocks 

the exit channel of the ribosome. This blockage inhibits elongation of the peptide chain, 

a crucial step in bacterial reproduction.22-23 

 

The final class of antibiotics are the quinolones, the first of which, nalidixic acid, 

was discovered in 1962 by George Lesher and coworkers.24 Quinolones are bactericidal 

antibiotics that target type II topoisomerase to elicit the mechanism of action. Type II 

topoisomerase is an enzyme that unwinds the coiled bacterial deoxyribonucleic acid 

(DNA), cleaves, and religates the strand after DNA replication. Quinolones bind to the 

topoisomerase and the DNA strands, thereby inhibiting the topoisomerase from leaving. 

This results in a permanent break in the DNA strand that may eventually lead to 

regulated cell death.25 The six discussed antibiotic classes have mechanisms of action 

 

8 

that can be described as either inhibition of cell wall synthesis, inhibition of protein 

synthesis, or inhibition of nucleic acid synthesis. While these six classes of antibiotics 

are not the only classes of antibiotics,  the majority of antibiotics are a part of, or derived 

from, these six classes. 

 

1.2.3  Bacterial Mechanism of Resistance 

One characteristic of bacteria that enables resistance to antibiotics is their ability 

to evolve rapidly by a process known as binary fission. Binary fission is the process 

where bacteria grow in size while replicating DNA and other cellular components 

needed for survival before splitting into essentially two identical cells. “Essentially” is 

used because during DNA replication there could be errors in the nucleotide sequence, 

a phenomenon known as random mutagenesis. Random mutations in the DNA can 

result in a multitude of outcomes ranging from no observable changes to lethal 

mutations. A mutation could result in a bacterium that can survive and propagate in an 

environment that has a higher antibiotic concentration in comparison to a wild-type 

bacterium of the same species.26  The use of antibiotics provides a selective pressure 

on bacteria. Antibiotics will affect all bacteria that are susceptible to the antibiotic. 

Conversely, the mutated bacteria that can propagate in the presence of that antibiotic 

can continue to multiply leading to an increased number of bacteria that have that 

antibiotic resistant mutation.  

Bacteria also possess the ability to transfer genes to other bacteria, a process 

known as horizontal gene transfer. Antibiotic resistant genes can be horizontally 

transferred directly from the donor bacterium to the recipient cell through a sex pilus, a 

 

9 

form of transfer known as conjugation. Antibiotic resistant genes can also be transferred 

through the help of a bacteriophage, which will attach to and transfer the donor cell’s 

gene to a recipient cell. This transfer is known as transduction. The final form of 

horizontal gene transfer is transformation, where a recipient bacterium uptakes DNA 

that is found in the environment, possibly from a lysed bacterium that carried a resistant 

gene. Once inside the bacterium cell, the resistant gene could exist as a circular 

plasmid of DNA or recombine with the bacterium’s DNA. The bacterium’s replication 

mechanism will replicate both forms of DNA.26 In nature, the formation of resistance is 

natural through random mutagenesis, but the misuse and overuse of antibiotics by 

humans has provided the selective pressure to increase the number of existing 

antibiotic resistant bacteria.27-29 

Mutations and horizontal gene transfer can primarily cause antibiotic resistance 

by three mechanisms (Figure 1.3). In a normal, non-antibiotic resistant bacterium, the 

antibiotic would enter the cell, perhaps through a porin.30 Once inside the cell, the 

antibiotic would bind to its receptor and elicit its effect. In the case of beta-lactam 

antibiotics, the antibiotic inhibits enzymes involved in peptidoglycan layering and breaks 

in the wall occur during cell replication (Figure 1.3A). One mechanism that results in 

antibiotic resistance is the formation or gain of metabolic enzymes that can bind and 

metabolize the antibiotic by adding large functional groups to the antibiotic’s structure, 

thereby reducing the efficacy (Figure 1.3B).17 Another mechanism that may result in 

antibiotic resistance is the increase in the number of efflux pumps or the decrease in 

number of porins (Figure 1.3C). These changes in efflux pumps and porins will cause a 

lower internal concentration of antibiotics and if the minimum inhibitory concentration 

 

10 

(MIC) of the antibiotic is not reached, the bacteria will be able to continue to survive and 

propagate.24, 31 The final discussed mechanism that can cause antibiotic resistance is 

the alteration of the receptor protein (Figure 1.3D). A simple mutation in the DNA of the 

bacteria could lead to a change in the amino acid sequence, which could ultimately 

affect the protein sequence and the subsequent shape of the receptor. A change in the 

shape of the receptor protein could result in a decreased affinity for the antibiotic and a 

downstream reduced effect of the antibiotic on the bacterium.13, 22 A major concern is 

that a bacterium may contain two or three of these mechanisms of resistance, which 

makes designing an effective antibiotic difficult.20, 25, 32-34  

 

 

11 

Figure 1.3: The mechanism of action of an antibiotic and possible mechanisms of 
resistance. (A) An antibiotic will enter a bacterium, possibly through a porin, bind to its 
receptor, and elicit its effect. (B) Metabolic enzymes could form or be gained from 
another bacterium leading to the antibiotic being deactivated and unable to elicit its 
response. (C) The gain of efflux pumps and the loss of porins will cause the internal 
concentration of the antibiotic to never reach the minimum inhibitory concentration to 
affect the bacterium. (D) The mutation of the receptor could lead to a decreased affinity 
for the antibiotic resulting in a diminished effect.  

 

 

12 

1.2.4  Antibiotic Resistant Bacteria 

While antibiotic resistant bacteria survive and propagate in the presence of an 

antibiotic, it is important to note that these antibiotics were once effective in the 

treatment of the bacterial infection.35 In fact, bacteria can be classified based on 

whether they are resistant to one or multiple antibiotics. Multidrug resistance (MDR), 

extensive drug resistance (XDR), and pandrug (PDR) resistance are classifications for 

bacteria that are resistant to multiple, most, or all currently available antibiotics, 

respectively.36  

A brief history of the discovery of antibiotic resistance compared to the 

introduction of the antibiotic is illustrated in Figure 1.4. Antibiotic resistance has been 

known to exist since ~1940 when a penicillin-resistant strain of Staphylococcus was 

discovered. After the introduction of penicillin as a therapeutic in 1943, many antibiotics 

like tetracycline and vancomycin were used for about a decade or more before 

resistance to those antibiotics were first discovered. However, as antibiotic use became 

more ubiquitous, the time period between the introduction of the antibiotic and the 

discovery of resistance diminished until 1996, when levofloxacin was released as an 

antibiotic therapy. That same year, levofloxacin-resistance was discovered. The 

evolution of antibiotic resistant bacteria continued into the millennium when XDR 

tuberculosis was discovered. By 2005, a strain of Pseudomonas was discovered that at 

that time was resistant to all currently available antibiotics.37 Antibiotic resistant bacteria 

and the infections they cause are now a global issue because many bacterial infections 

contracted from either hospitals or the community that were once treatable are now 

more difficult, if not impossible to cure.38-41 

 

13 

Figure 1.4: A brief history of the decreasing time frame between the introduction 
of some antibiotics and the first discovery of resistance to that antibiotic. 
Penicillin resistance was first discovered prior to the use of penicillin as a therapy most 
likely due to extensive development period and primitive antibiotic research methods. 
Tetracycline and Vancomycin therapies existed for close to a decade or longer before 
the emergence of resistance. Levofloxacin was first used as a therapy the same year 
the resistance was first discovered. In the new millennium, the emergence of XDR and 
PDR bacterial strains occurred.  

 

 

14 

The Centers for Disease Control and Prevention (CDC) estimates that the 

number of illnesses caused by antibiotic resistant bacteria in the United States is over 2 

million per year and the number of deaths is conservatively estimated to be ~23,000.37 

This number is projected to grow as bacteria continue to evolve and the incidence of 

bacterial infections caused by resistant strains increases. In 2008, more than 50% of 

Staphylococcus aureus infections were methicillin-resistant, about 30% of enterococci 

infections were vancomycin-resistant, and over 20% of Pseudomonas aeruginosa 

infections were fluoroquinolone-resistant.4 In 2013, these infections accounted for 

approximately 80,500 illnesses and 11,300 deaths; 20,000 illnesses and 1,300 deaths; 

and 6,700 illnesses and 440 deaths, respectively.37 There are 15 other bacteria that are 

considered either an urgent threat, a serious threat, or of concern with the greatest 

threat being Clostridium difficile, which is estimated to cause 250,000 infections and 

upwards of 14,000 deaths per year in the United States as of 2015.42  

 

1.2.5 The Decrease of Antibiotic Development and Government Intervention  

 Another major problem with antibiotics, along with the emergence of antibiotic 

resistant bacteria, is decreasing number of antibiotics being approved by the U.S. Food 

& Drug Administration (FDA) (Figure 1.5).4, 43 A primary reason for this shortage is that 

many pharmaceutical companies are not pursuing the development of new antibiotics 

due to the miniscule return on investment. The return on investment has been 

decreasing due to the increasing cost of drug development and the unwillingness of the 

market to pay the high prices for novel antibiotics. Many antibiotics being approved by 

the FDA are just modified compounds belonging to the six classes discussed in Section 

 

15 

1.2. These modified antibiotics do not have a new mechanism of action and therefore 

will only provide a short term solution, as antibiotic resistant bacteria will eventually 

evolve to be resistant to the modified antibiotic.43-44 As of 2017, there were 51 antibiotics 

in the clinical phase of drug development. Of those 51 antibiotics, only eight had novel 

mechanisms of action.44 There is no guarantee that these eight novel antibiotics will be 

approved by the FDA as they are still being evaluated in humans for safety and efficacy.  

 

16 

Figure 1.5: New Antibiotics approved by the FDA. There was an observed decrease 
in the number of new antibiotics approved by the FDA from 1983 to 2011. This 
decrease in new antibiotics has been contributed to rising developmental costs of drugs 
and the market unwilling to pay high prices for novel antibiotics. Adapted from 
references 4 and 43. 

 

 

17 

The increased incidence of antibiotic resistant bacterial infections and the 

decrease in new antibiotics developed led to the September 2014 Executive Order from 

then current U.S. president, Barack Obama. In this Executive Order, President Obama 

called for immediate action to be taken in order to slow the emergence and spreading of 

antibiotic resistant bacteria.45 This Executive Order led to the Nation Strategy 

(September 2014) and the Nation Action Plan (March 2015). Members of the CDC, 

FDA, and the National Institutes of Health (NIH) developed the Nation Action Plan, 

which contained an 84 point plan based around the 5 goals in Table 1.1. These goals 

were to be accomplished by 2020 and, as of 2017, progress had been made in all 

goals. The 3rd and 4th goals focus on the development of new strategies to quickly 

detect antibiotic resistant bacteria and new therapies for treatment of these bacteria.42, 

46 These two goals can be accomplished through research by industrial or educational 

establishments and have primarily made progress in the distribution of funds and the 

formation of partnerships for research and development.47 Importantly, there has been 

an increase in approved antibiotics. From 2015 to 2018, there were seven antibiotics 

approved by the FDA, with three being approved in 2017 and three being approved in 

2018, although not all of these antibiotics have novel mechanisms of action.48 

 

 

18 

Table 1.1: The 5 Main Goals of the National Action Plan for Combating Antibiotic-
Resistant Bacteria 

White House Five Main Goals by 2020 

1) Slow the Development of Resistant Bacteria and Prevent the Spread of 
Resistant Infections 

2) Strengthen National One-Health Surveillance Efforts to Combat Resistance 

3) Advance Development and Use of Rapid and Innovative Diagnostic Tests for 
Identification and Characterization of Resistant Bacteria 

4) Accelerate Basic and Applied Research and Development for New Antibiotics, 
Other Therapeutics, and Vaccines 

5) Improve International Collaboration and Capacities for Antibiotic-resistant 
Prevention, Surveillance Control, and Antibiotic Research and Development 

 

19 

1.3 Introduction to Drug Development 

 

The average cost and a time required for a pharmaceutical company to develop 

a drug through FDA approval is $2.558 billion and 128 months, respectively.49 For an 

antibiotic, this lengthy period of development could result in antibiotic resistance 

developing in bacteria soon after FDA approval. The threat of resistance coupled with 

the high cost to develop antibiotics makes for a risky investment for pharmaceutical 

companies, as the company may not make a profit before antibiotic resistance makes 

their antibiotic obsolete.43, 50 Developing a drug through pre-clinical development costs 

an average $1.098 billion and requires about 2.5 years (31.2 months). Clinical trials cost 

an average of $1.460 billion and take nearly 7 years to complete (80.8 months).  The 

time from the end of clinical trials to FDA approval requires an average of 16 months.49 

During pre-clinical development, a potential antibiotic is evaluated for safety and 

efficacy outside of humans by in vitro experiments using cells and tissues or by in vivo 

animal models. Clinical trials are used to evaluate the safety and efficacy of the 

antibiotic in humans.51 These evaluations are achieved by measuring the 

pharmacokinetics and pharmacodynamics of the potential antibiotic. 

 

1.3.1 Pharmacokinetics and Pharmacodynamics 

 

Pharmacokinetics (PK) is the study of how the body affects a molecule. When an 

antibiotic is administered, the antibiotic will undergo absorption, distribution, 

metabolism, and excretion (ADME). Collectively, these events result in a PK profile. 

Absorption is the movement of a drug from the administration site into the bloodstream 

and can be different for each drug depending on the method of administration. For 

 

20 

example, intravenous administration bypasses absorption and results in an immediate 

maximum plasma concentration (Cmax) while an oral administration takes longer to 

reach Cmax because the molecule needs time to diffuse out of the gastrointestinal tract 

and absorb into the bloodstream.52-53 Distribution is the movement of a drug from the 

bloodstream into the various tissues of the body. Metabolism is the enzymatic reaction 

of the drug to make the molecule more water soluble and more easily excreted. 

Excretion is the removal of the drug and its metabolites by the urine, bile, or other 

excretion routes.54  

A PK curve consists of the concentration of a molecule in the plasma versus time 

as it undergoes ADME in the body (Figure 1.6). A PK curve can provide data including 

the maximum plasma concentration (Cmax), the time to reach Cmax (tmax), the absorption 

rate constant, the elimination rate constant, and the area under the curve (AUC) or the 

total exposure to the drug. The use of these data can be used to determine other PK 

parameters, such as the half-life, bioavailability, the volume of distribution, and the 

clearance rate.55 The addition of early pharmacodynamic (PD) data, such as the 

minimum toxic concentration (MTC) and the minimum inhibitory concentration (MIC) of 

an antibiotic can be used with a PK curve to determine the total time that a drug is 

effective (duration of action) or the concentration range where the drug is safe and 

effective (therapeutic range) for use in vivo. The therapeutic range is the concentration 

range between the minimum toxic concentration (MTC) and the minimum effective 

concentration (MEC).56 For antibiotics, the MEC is referred to as the minimum inhibitory 

concentration (MIC) or the minimum bactericidal concentration (MBC).57 

 

21 

Figure 1.6: An example pharmacokinetic curve. A pharmacokinetic curve is a 
comparison of plasma concentration of a molecule versus time and can provide data 
such as the Cmax, tmax, the absorption rate constant, the elimination rate constant, and 
the area under the curve (AUC). With preliminary PD data such as the MTC and MEC, 
the duration of action and the therapeutic range can also be determined from a PK 
curve. 

 

 

22 

Pharmacodynamics is the study of how a molecule affects the body and often 

provides information about the molecule’s efficacy and toxicity. A PD curve consists of 

the effect of the drug versus the concentration. For an antibiotic, this curve could be the 

percentage of cell death as a function of antibiotic concentration.57 A PD curve can be 

used for both efficacy and toxicity data (Figure 1.7). For efficacy, a PD profile will 

measure the physiological effect at various concentrations and can provide information 

such as the maximum effective concentration, the minimum effective concentration, the 

efficacy (Emax), and the potency (ED50). A drug that does not cause a 100% effect 

would have a lower efficacy than a drug that does. A drug that has a lower ED50 is 

considered more potent than a drug with a higher ED50. For toxicity, a PD profile will 

measure the adverse effect at multiple concentrations and provide information such as 

the no observable adverse effect limit (NOAEL), the lowest observable adverse effect 

limit (LOAEL), and the lethal dose for 10%, 50%, and 95% of an animal population 

(LD10, LD50, LD95 respectively). Toxic doses are used for human studies where 

lethality is not a feasible outcome. For example, a toxic dose of 50% (TD50) in a human 

population means that 50% of tested human subjects exhibited the toxic effect. The 

ratio of the TD50 or LD50 to the ED50 is called the therapeutic index. A drug with a 

larger therapeutic index is preferred because there is a wider concentration range 

demonstrating efficacy without associated issues with toxicity.58-59 

 

 

 

23 

Figure 1.7: Example pharmacodynamic curves for efficacy (blue) and toxicity 
(orange). A PD curve for efficacy can be used to determine the minimum and maximum 
effective concentration, the potency (ED50), and the efficacy of a drug. A PD curve for 
toxicity can be used to determine the NOAEL, the LOAEL, and the TD50. Taking the 
ratio of the TD50 to the ED50 is used to calculate the therapeutic index of a drug. 

 

 

24 

Pharmacokinetics and pharmacodynamics (PK/PD) are often discussed together 

because for a drug to be effective, it needs to reach the proper concentration.57 Both PK 

and PD are determined for antibiotics by in vitro and in vivo methods. Currently, 35.5% 

of potential drugs that are approved for clinical trials fail the Phase I clinical trial where 

safety in humans is determined. In fact, only 10.4% of drugs that are approved for 

clinical trials result in FDA approval. The high failure rate of the potential drugs in the 

human clinical trials may be in part due to differences in PK/PD of the potential drugs 

when evaluated in animals and in vitro compared to when evaluated in humans. 51, 60-61   

 

1.3.2 In Vitro PK and PD Evaluation of Antibiotics 

 

In vitro determination are those methods performed in a simplified system 

outside of a living mammal, usually using cells or tissue slices that attempt to mimic 

conditions in vivo. The benefit of in vitro methods is that they are less expensive, easier, 

and safer to use than in vivo methods. For antibiotics, in vitro methods are used for both 

PK and PD evaluations; although, many of these in vitro experiments are static and only 

evaluate a part of PK/PD. In a static experiment, the cells of interest only experience 

one concentration of drug throughout the course of the experiment. The concentration 

of the drug does not fluctuate over time as in a dynamic system like a living organism. In 

other words, the cells are not experiencing the PK profile of the drug that would be  

experienced by a living organism51 Some absorption parameters of a drug can be 

determined with assays using cells such as the Caco-2 line.62 Some distribution 

parameters of a drug can be determined with assays utilizing P-glycoproteins or other 

plasma proteins.63 Some metabolism and excretion parameters of a drug can be 

 

25 

determined in assays that utilizes isolated cytochrome P450 enzymes or liver 

microsomes.64-65 Toxicity data can be determined using isolated hepatocytes or other 

cell lines.61, 66 Efficacy data for antibiotics, such as the MIC or MBC, can be determined 

using bacterial cells like the broth microdilution assay.67 The major drawback of static in 

vitro methods are that the cells are experiencing a constant concentration of the drug 

and not the PK that would be experience in a living organism.57 In order to incorporate 

PK, the use of filters/membranes and multiple compartments is required. 

The human body itself can be thought of as having multiple compartments. 

These compartments are divided by some form of barrier that allows distribution at 

various rates depending on the location. A one-compartment model would be used for 

an area where the distribution of a drug is considered rapid and equilibrium is achieved 

almost simultaneously with the plasma (Figure 1.8A).68 These areas can include the 

heart, kidneys, liver, or other areas with high perfusion.69 A two-compartment model 

would be used when the distribution of the drug occurs in two phases (Figure 1.8B). The 

concentration change in the second compartment would be slower than in the first 

compartment. This type of model could be used for areas with slower diffusion such as 

fat or muscle. A multicompartment model would be used when the distribution of the 

drug occurs in more than two phases (Figure 1.8C), for example when diffusion occurs 

from the plasma, into muscle, and then into bone.68  

Various methods have been employed in order to incorporate multiple 

compartments into an in vitro device to replicate antibiotic PK for bacterial PD analysis. 

If the concentration of antibiotic does not change over time during the experiment, it is 

considered a static experiment (Figure 1.8D), which are useful for obtaining some 

 

26 

PK/PD data; however, using static antibiotic concentrations for PD analysis may not be 

providing relevant data because the total exposure to the antibiotic (AUC) is different 

than what is represented by static measures. A dynamic model, where the 

concentration of antibiotic varies over time, can be created by diluting the sample or 

using diffusion to distribute the antibiotic.  

Additional media can be added to a bacterial and antibiotic solution to alter the 

concentration of antibiotic and mimic the PK of the antibiotic. However, this would result 

in an increase in total liquid volume. In order to maintain a constant volume, media is 

removed from the antibiotic and bacterial solution, thus creating a one-compartment 

dilution-based model (Figure 1.8E).70 This design can be adapted to create a two-

compartment model.71 However, the major weakness of this design is that dilution of the 

antibiotic by the addition and removal of media also results in the dilution of the 

bacteria. Thus, bacterial counts may be reduced but the source of the reduction 

(antibiotic activity by the drug or reduced counts due to dilution) is not completely 

understood. This requires mathematical correction to be used in order to account for the 

dilution of the bacteria.72-73 The use of membrane filters have been used to avoid the 

loss of bacteria.74-76 In this method, a filter is placed in front of the waste stream to only 

remove media and antibiotic (Figure 1.8F). The weakness of using membrane filters is 

that they can become blocked during longer experiments.74, 77  

Instead of using a filter, a porous membrane can be used to create a diffusion 

barrier that allows antibiotic to diffuse between compartments, but the bacteria are 

localized to only one compartment. Diffusion-based models usually consist of two or 

more compartments separated by porous membranes. The central compartment, like 

 

27 

previous designs, is where the antibiotic is controlled by the addition and removal of 

antibiotic in media. The concentration of antibiotic in the peripheral compartment can 

only be altered by diffusion through the membrane, which is based on the concentration 

gradient across the porous membrane. Bacteria would be isolated in the peripheral 

compartment. Manipulation of the concentration in the central compartment can result in 

a more human-like PK profile of the antibiotic in the peripheral compartment resulting in 

more relevant PD data.77 The orientation of the two compartments can vary depending 

on the design. The peripheral compartment could be oriented directly next to the central 

compartment (Figure 1.8G) or in the middle of the central compartment (Figure 1.8H).78-

79 The use of many hollow capillaries acting as a central compartment running the 

length of the peripheral compartment is the basis for the hollow fiber chamber reactor 

(Figure 1.8I). Cells are in the peripheral compartment, while media and antibiotic are 

exchanged through the fibers. Diffusion can occur at any point along the fibers, which 

are acting as the diffusion barriers that would be found in the body. 80-81 The problem 

with diffusion-based in vitro models are that, like dilution-based models, the membranes 

can become blocked.77 

 

28 

 

Figure 1.8: Diagrams of the compartmental models and in vitro models consisting 
of static, dilution-based, or diffusion-based models. A one compartment model (A) 
is used to replicate rapid diffusion. A two compartment model (B) is used when diffusion 
occurs in two different phases. A multi-compartment model (C) consists of more than 
two compartments and is used when there are multiple diffusion phases. In a static 
model, the drug concentration is not altered throughout the experiment (D). In dilution-
based models, the drug concentration is changed by adding media into the 
compartment to dilute the sample and removing media in order to keep a constant 
volume (E). Filters (thick black line) can be used to avoid loss of bacteria (F). In 
diffusion-based models, the concentration of drug is changed by using membranes in 
the device to create other compartments. These membranes can be incorporated to 
create a peripheral compartment (purple) that is adjacent to (G) or in the middle of (H) 
the central compartment (white). A hollow fiber chamber reactor uses hollow capillaries 
as the central compartment (I). These capillaries allow diffusion into the peripheral 
compartment. Diffusion into compartments is designated by a blue arrow while diffusion 
out of the compartments is designated by a red arrow. The differences in the thickness 
of the arrows signifies a different phase of diffusion. Adapted from references 52 and 
62. 

 

29 

 

The major limitation of current in vitro PK/PD devices is that the data collected 

does not always correlate to the data collected from a living human. This difference is 

caused by the simplification of in vitro experimental designs to evaluate only part of 

PK/PD.51 Current technology including in silico simulations, 3D printing, and tissue 

engineering have been providing new methods to address this data correlation 

limitation. In silico simulations use structure- or data-based relationships between 

databases of molecules to determine how a potential drug would act in the body, which 

could generate PK data that can be used along with in vitro methods to create more 

clinically-relevant PD data.51, 61 However, in silico methods rely on databases of similar 

molecules to generate data and the value of in silico methods in the development of 

new antibiotics with novel mechanisms of action is still unknown.  Tissue engineering 

has been used to generate devices with functional tissue that can be used to mimic an 

organ in an in vitro model. These devices are known as organ-on-a-chip devices.82-87 

The use of 3D printing technology has allowed for rapid prototyping and new diffusion-

based dynamic PK/PD models.62, 88-89 The use of 3D printing for the design of PK/PD 

models will be discussed in more detail in Chapter 3.  

 

1.3.3 In Vivo Animal PK/PD Evaluation of Antibiotics 

In vivo PK/PD experiments evaluate the PK/PD of a potential drug inside of a 

living mammal. The potential drug will experience all PK properties because the 

potential drug is being tested in a living system. In vivo testing occurs in either animals 

or humans, with the purpose of testing in animals is to ensure safety and efficacy before 

testing in humans. Rodents are the most popular mammal used for animal testing and 

 

30 

include mice, rats, or guinea pigs. Larger mammals such as primates, rabbits, canines, 

or sheep are also routinely used for in vivo studies. Rodents are usually used as the 

starting organism because they are relatively cheap and easy to work with compared to 

larger mammals.90 However, larger mammals are needed for testing because they have 

more comparable metabolism with humans, a similar organ and body size, and similar 

toxicity responses to humans in comparison to rodents.91-94 For example, small rodents 

usually metabolize and excrete a drug faster than a human and this difference in 

clearance has been associated with relatively larger organ sizes in smaller mammals.92, 

95-96 Mice are the most common animals used because they have the benefit of being 

genetically modified to either remove or add the expression of certain proteins.97-98 

Some mice have their DNA modified so they can express human proteins, making the 

collected data relevant to humans.51, 61, 99-100 

The PK of a potential drug can be measured in animal models the same way that 

it is measured in humans. The potential drug and its metabolites’ concentrations can be 

quantitatively measured in plasma, bile, urine, feces, and tissue.  Samples are taken at  

various time points and typically analyzed by high-performance-liquid chromatography 

coupled with mass spectrometry.101  A non-invasive quantitation method also exists that 

involves the attachment of a radiolabel to the potential drug, such as Hydrogen-3 or 

Carbon 14.  The potential drug and its metabolites can then be monitored as they travel 

through the various compartments of the body.102-103  The experiment for measuring the 

PD of a potential drug in an animal model is dependent on the information trying to be 

gained.  For antibiotics, there are animal models to determine the efficacy of an 

antibiotic in various infections.  These infection models include, but are not limited to 

 

31 

endocarditis, meningitis, peritonitis, pneumonia, septicemia, and urinary tract 

infections.104-110 In most of these models, bacteria is implanted into the tissue and 

allowed to cause an infection.  Sometimes this involves compromising the immune 

system of the animal.  After an infection occurs, the antibiotic is administered and, after 

a predetermined amount of time, the amount of bacteria is quantified from bodily fluids 

or from homogenized tissue.107-108, 110-111  

In vivo animal models offer more information than in vitro models because 

animals are complex organisms that have all parts of PK present.  The drawback of 

animal models is that the data collected from PK/PD experiments do not always 

correlate to the data that is seen in humans due to differences in physiology. There are 

multiple methods to overcome some of these differences in physiology, such as genetic 

modifications, changing the dosage regimen, or chemically altering the PK of the 

animal.96, 104, 112-115 However, these animal models are still not human and the 

differences between species can still cause unforeseen differences in PK/PD. 

 

1.3.4 In Vivo Human PK/PD Evaluation of Antibiotics 

 

After multiple animal species testing, all PK/PD, toxicology, quality control, and 

drug design data and a plan for clinical trials study design must be sent to the FDA in 

the form of an investigation new drug (IND) submission.  The FDA has 30 days to 

address any comments or concerns before the potential drug can be tested in 

humans.116 Clinical trials are the most expensive and longest stage of drug 

development.  There are three phases to clinical trials. Phase I takes an average of 33.1 

 

32 

months and costs $49.6 million.  Phase II takes an average of 37.9 months and costs 

$95.3 million.  Phase III takes an average of 45.1 months and costs $314 million.60 

 

Phase I clinical trials involve a small group of healthy patients (10-100).  These 

trials are concerned with safety, determining adverse effects, determining safe dosage 

levels, and PK. These PK trials can be determine by a standard PK study design or a 

population PK study design.  A standard PK study design involves a minimum number 

of healthy patients (<12).  The drug can be administered one time or multiple times.  

Single administration can occur at concentrations slightly above the estimated safe 

dose to observe and record adverse effects.  Bodily fluids are collected from each 

individual at multiple time points and the amount of drug is quantified.117-119 Population 

PK study designs involve more patients (40+) and provides less stress to patients due 

to fewer samples being taken from each individual.  This study design provides average 

PK parameters, information about variation between individuals, and can reveal 

subgroups that respond differently than other groups.119-123  Phase I takes an average of 

33.1 months and costs $49.6 million.  As of 2013, 64.5% of drugs that enter Phase I 

clinical trials proceed to Phase II.60 

 

Phase II clinical trials involve a group of up to several hundred patients with the 

disease.  This phase is concerned with efficacy and further safety evaluation.  Patients 

are randomly assorted into groups that are administered either the potential drug or a 

placebo to determine if the efficacy of the drug is greater than the placebo effect.  These 

trials can be either single or double blinded to ensure minimum bias is present.58-59 

Phase II takes an average of 37.9 months and costs $95.3 million.  As of 2013, 32.4% 

of potential drugs that enter Phase II proceed to Phase III.60 

 

33 

 

Phase III clinical trials involve several hundred to several thousand patients with 

the disease.  Phase III trials are concerned with efficacy and safety in a larger group of 

people to determine if any subgroups exists or if any side effects were missed in 

previous studies.  The designs of this study are similar to those in Phase II except for a 

longer observation period.58-59  Phase III takes an average of 45.1 months and costs 

$314 million.60 As of 2013, 60.1% of potential drugs that enter phase III proceed to a 

new drug application (NDA) being submitted to the FDA.  Of the NDAs submitted, 

83.2% are approved.  These approval percentages result in only 10.4% of potential 

drugs that enter clinical trials being successful and approved by the FDA.60 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

34 

1.4 Current State of Affairs  

 

The overuse and misuse of antibiotics has led to a critical situation for humanity. 

Several bacterial strains are capable of causing infections that are resistant to many if 

not all of the currently available antibiotics.  Along with this threat, the time and cost of 

developing a new drug, including antibiotics, has led many pharmaceutical companies 

to no longer pursue antibiotic development.  A potential reason for the high cost of drug 

development, is the high failure rate of drugs due to poor correlation between in vitro/in 

vivo animal data and in vivo human data. Creating more human-like PK in vitro 

platforms may have the potential to provide more clinically relevant PD data. More 

clinically-relevant PD data could result in less research being needed at the more 

expensive stages of drug development like animal testing. 

 

In this dissertation, the above problem of decreased pharmaceutical interested in 

antibiotic development due to increasing time and cost of development is addressed by 

the development of two novel biotechnological methods.  The first method creates an 

antibiotic susceptibility assay that provides more rapid and quantitative information than 

current susceptibility methods.  The second method adapts this antibiotic susceptibility 

assay to create a device that challenges the bacteria to a PK profile similar to in vivo 

conditions.  These methods provide rapid and more clinically-relevant information that 

could decrease the time required to perform early in vitro tests or provide data that 

correlates better with humans than animal models do, resulting in less dependency on 

animal models and decreasing the cost of the preclinical phase of development. 

 

 

 

 

35 

REFERENCES 

 

36 

 

REFERENCES 

Taubes, G., The Bacteria Fight Back. Science 2008, 321 (5887), 356-361. 

Moyes, R. B.; Reynolds, J.; Breakwell, D. P., Differential Staining of Bacteria: 

Meroueh, S. O.; Bencze, K. Z.; Hesek, D.; Lee, M.; Fisher, J. F.; Stemmler, T. L.; 

Fleming, A., On the Antibacterial Action of Cultures of a Penicillium, with Special 

Ubeda, C.; Djukovic, A.; Isaac, S., Roles of the intestinal microbiota in pathogen 

Nakatsuji, T.; Chen, T. H.; Butcher, A. M.; Trzoss, L. L.; Nam, S. J.; Shirakawa, 

 
 
 
1. 
Chart, H.; Smith, H. R.; La Ragione, R. M.; Woodward, M. J., An investigation 
into the pathogenic properties of Escherichia coli strains BLR, BL21, DH5α and EQ1. 
Journal of Applied Microbiology 2000, 89 (6), 1048-1058. 
 
2. 
protection. Clinical & translational immunology 2017, 6 (2), e128-e128. 
 
3. 
K. T.; Zhou, W.; Oh, J.; Otto, M.; Fenical, W.; Gallo, R. L., A commensal strain of 
Staphylococcus epidermidis protects against skin neoplasia. Sci Adv 2018, 4 (2), 
eaao4502. 
 
4. 
 
5. 
Gram Stain. In Current Protocols in Microbiology, John Wiley & Sons, Inc.: 2005. 
 
6. 
Mobashery, S., Three-dimensional structure of the bacterial cell wall peptidoglycan. 
Proceedings of the National Academy of Sciences of the United States of America 
2006, 103 (12), 4404-4409. 
 
7. 
Reference to their Use in the Isolation of B. influenzæ. British journal of experimental 
pathology 1929, 10 (3), 226-236. 
 
8. 
therapeutic outcomes. Applied Microbiology and Biotechnology 2011, 92 (4), 677-687. 
 
9. 
journal of medicine 1988, 318 (7), 419-26. 
 
10.  Yocum, R. R.; Rasmussen, J. R.; Strominger, J. L., The mechanism of action of 
penicillin. Penicillin acylates the active site of Bacillus stearothermophilus D-alanine 
carboxypeptidase. Journal of Biological Chemistry 1980, 255 (9), 3977-86. 
 
11.  Otten, H., Domagk and the development of the sulphonamides. Journal of 
Antimicrobial Chemotherapy 1986, 17 (6), 689-690. 
 

Donowitz, G. R.; Mandell, G. L., Beta-Lactam antibiotics (1). The New England 

Kardos, N.; Demain, A. L., Penicillin: the medicine with the greatest impact on 

 

37 

12.  Bosch, F.; Rosich, L., The Contributions of Paul Ehrlich to Pharmacology: A 
Tribute on the Occasion of the Centenary of His Nobel Prize. Pharmacology 2008, 82 
(3), 171-179. 
 
13.  Sköld, O., Sulfonamide resistance: mechanisms and trends. Drug Resistance 
Updates 2000, 3 (3), 155-160. 
 
14.  Choi, S.-W.; Mason, J. B., Folate Status: Effects on Pathways of Colorectal 
Carcinogenesis. The Journal of Nutrition 2002, 132 (8), 2413S-2418S. 
 
15.  Crider, K. S.; Yang, T. P.; Berry, R. J.; Bailey, L. B., Folate and DNA Methylation: 
A Review of Molecular Mechanisms and the Evidence for Folate’s Role. Advances in 
Nutrition 2012, 3 (1), 21-38. 
 
16.  Woodruff, H. B., Selman A. Waksman, Winner of the 1952 Nobel Prize for 
Physiology or Medicine. Applied and Environmental Microbiology 2014, 80 (1), 2-8. 
 
17.  Mingeot-Leclercq, M.-P.; Glupczynski, Y.; Tulkens, P. M., Aminoglycosides: 
Activity and Resistance. Antimicrobial Agents and Chemotherapy 1999, 43 (4), 727-737. 
 
18.  Duggar, B. M., AUREOMYCIN: A PRODUCT OF THE CONTINUING SEARCH 
FOR NEW ANTIBIOTICS. Annals of the New York Academy of Sciences 1948, 51 (2), 
177-181. 
 
19.  Smilack, J. D., The Tetracyclines. Mayo Clinic Proceedings 1999, 74 (7), 727-
729. 
 
20.  Chopra, I.; Roberts, M., Tetracycline Antibiotics: Mode of Action, Applications, 
Molecular Biology, and Epidemiology of Bacterial Resistance. Microbiology and 
Molecular Biology Reviews 2001, 65 (2), 232-260. 
 
21.  Bunch, R. L.; McGuire, J. M. Erythromycin, Its Salts, and Method of Preparation. 
1953. 
 
22.  Marne, G.; Alexander, S. M., Macrolide Antibiotics: Binding Site, Mechanism of 
Action, Resistance. Current Topics in Medicinal Chemistry 2003, 3 (9), 949-960. 
 
23. 
Lincosamides and Streptogramin B Reveals the Nascent Peptide Exit Path in the 
Ribosome. Journal of Molecular Biology 2003, 330 (5), 1005-1014. 
 
24.  Andriole, V. T., The Quinolones: Past, Present, and Future. Clinical Infectious 
Diseases 2005, 41 (Supplement 2), S113-S119. 
 
25.  Aldred, K. J.; Kerns, R. J.; Osheroff, N., Mechanism of Quinolone Action and 
Resistance. Biochemistry 2014, 53 (10), 1565-1574. 

Tenson, T.; Lovmar, M.; Ehrenberg, M., The Mechanism of Action of Macrolides, 

 

38 

James, C. E.; Mahendran, K. R.; Molitor, A.; Bolla, J.-M.; Bessonov, A. N.; 

Fisher, J. F.; Meroueh, S. O.; Mobashery, S., Bacterial Resistance to β-Lactam 

Levy, S. B.; Marshall, B., Antibacterial resistance worldwide: causes, challenges 

26.  Culyba, M. J.; Mo, C. Y.; Kohli, R. M., Targets for Combating the Evolution of 
Acquired Antibiotic Resistance. Biochemistry 2015, 54 (23), 3573-3582. 
 
27. 
and responses. Nat Med 2004. 
 
28.  Butaye, P.; Devriese, L. A.; Haesebrouck, F., Antimicrobial Growth Promoters 
Used in Animal Feed: Effects of Less Well Known Antibiotics on Gram-Positive 
Bacteria. Clinical Microbiology Reviews 2003, 16 (2), 175-188. 
 
29.  Castanon, J. I. R., History of the Use of Antibiotic as Growth Promoters in 
European Poultry Feeds. Poultry Science 2007, 86 (11), 2466-2471. 
 
30. 
Winterhalter, M.; Pagès, J.-M., How β-Lactam Antibiotics Enter Bacteria: A Dialogue 
with the Porins. PLoS ONE 2009, 4 (5), e5453. 
 
31. 
Antibiotics:  Compelling Opportunism, Compelling Opportunity. Chemical Reviews 2005, 
105 (2), 395-424. 
 
32.  Ruiz, J.; Pons, M. J.; Gomes, C., Transferable mechanisms of quinolone 
resistance. International Journal of Antimicrobial Agents 2012, 40 (3), 196-203. 
 
33.  Davies, J.; Davies, D., Origins and Evolution of Antibiotic Resistance. 
Microbiology and Molecular Biology Reviews : MMBR 2010, 74 (3), 417-433. 
 
34. 
American Journal of Medicine 2006, 119 (6), S3-S10. 
 
35.  Nemeth, J.; Oesch, G.; Kuster, S. P., Bacteriostatic versus bactericidal antibiotics 
for patients with serious bacterial infections: systematic review and meta-analysis. 
Journal of Antimicrobial Chemotherapy 2015, 70 (2), 382-395. 
 
36. 
Drug Resistance (XDR), and Multidrug Resistance (MDR) among Gram-Negative 
Bacilli: Need for International Harmonization in Terminology. Clinical Infectious 
Diseases 2008, 46 (7), 1121-1122. 
 
37.  Antibiotic Resistance Threats in the United States, 2013. CDC, Ed. CDC: Atlanta, 
2013. 
 
38. 
ESKAPEing the labyrinth of antibacterial discovery. Nat Rev Drug Discov 2015, 14 (8), 
529-542. 
 

Falagas, M. E.; Karageorgopoulos, D. E., Pandrug Resistance (PDR), Extensive 

Tenover, F. C., Mechanisms of Antimicrobial Resistance in Bacteria. The 

Tommasi, R.; Brown, D. G.; Walkup, G. K.; Manchester, J. I.; Miller, A. A., 

 

39 

Laxminarayan, R.; Duse, A.; Wattal, C.; Zaidi, A. K. M.; Wertheim, H. F. L.; 

Liu, Y.-Y.; Wang, Y.; Walsh, T. R.; Yi, L.-X.; Zhang, R.; Spencer, J.; Doi, Y.; Tian, 

39. 
Sumpradit, N.; Vlieghe, E.; Hara, G. L.; Gould, I. M.; Goossens, H.; Greko, C.; So, A. D.; 
Bigdeli, M.; Tomson, G.; Woodhouse, W.; Ombaka, E.; Peralta, A. Q.; Qamar, F. N.; 
Mir, F.; Kariuki, S.; Bhutta, Z. A.; Coates, A.; Bergstrom, R.; Wright, G. D.; Brown, E. D.; 
Cars, O., Antibiotic resistance - the need for global solutions. The Lancet Infectious 
Diseases 2013, 13 (12), 1057-1098. 
 
40.  Rice, L. B., Antimicrobial Resistance in Gram-Positive Bacteria. The American 
Journal of Medicine 2006, 119 (6), S11-S19. 
 
41. 
G.; Dong, B.; Huang, X.; Yu, L.-F.; Gu, D.; Ren, H.; Chen, X.; Lv, L.; He, D.; Zhou, H.; 
Liang, Z.; Liu, J.-H.; Shen, J., Emergence of plasmid-mediated colistin resistance 
mechanism MCR-1 in animals and human beings in China: a microbiological and 
molecular biological study. The Lancet Infectious Diseases 2016, 16 (2), 161-168. 
 
42.  National Action Plan for Combating Antibiotic-Resistant Bacteria. The White 
House: Washington D.C., 2015. 
 
43.  Sukkar, E., Why are there so few antibiotics in the reserach and development 
pipeline? The Pharmaceutical Journal 2013, 291, 520. 
 
44.  Kmietowicz, Z., Few novel antibiotics in the pipeline, WHO warns. BMJ 2017, 
358, j4339. 
 
45.  Obama, B., Executive Order - Combating Antibiotic-Resistant Bacteria. The 
White House: Washington D.C., 2014. 
 
46.  National Strategy for Combat Antibiotic-Resistant Bacteria. The White House: 
Washington D.C., 2014. 
 
47.  National Action Plan For Combating Antibiotic-Resistant Bacteria: Progress 
Reprot for Years 1 and 2. The White House: Washington D.C., 2017. 
 
48.  New Drugs at FDA: CDER's New Moleuclar Entities and New Therapeutic 
Biological Products. FDA: Silver Spring, MD, 2018. 
 
49.  DiMasi, J. A.; Grabowski, H. G.; Hansen, R. W., Innovation in the pharmaceutical 
industry: New estimates of R&D costs. Journal of Health Economics 2016, 47, 20-33. 
 
50.  Ventola, C. L., The Antibiotic Resistance Crisis: Part 1: Causes and Threats. 
Pharmacy and Therapeutics 2015, 40 (4), 277-283. 
 
51.  Heller, A. A.; Lockwood, S. Y.; Janes, T. M.; Spence, D. M., Technologies for 
Measuring Pharmacokinetic Profiles. Annual Review of Analytical Chemistry 2018, 11 
(1), 79-100. 

 

40 

52.  Khojasteh, S. C.; Wong, H.; Hop, C. E. C. A., Pharmacokinetics. In Drug 
Metabolism and Pharmacokinetics Quick Guide, Springer New York: New York, NY, 
2011; pp 1-15. 
 
53.  Bonate, P. L., Pharmacokinetics. Wiley Interdisciplinary Reviews: Computational 
Statistics 2011, 3 (4), 332-342. 
 
54. 
87 (1), 93-120. 
 
55.  Urso, R.; Blardi, P.; Giorgi, G., A short introduction to pharmacokinetics. 
European review for medical and pharmacological sciences 2002, 6 (2-3), 33-44. 
 
56.  Byers, J. M., What Is a Therapeutic Range? Laboratory Medicine 1980, 11 (12), 
784-787. 
 
57.  Vaddady, P. K.; Lee, R. E.; Meibohm, B., In vitro 
pharmacokinetic/pharmacodynamic models in anti-infective drug development: focus on 
TB. Future Medicinal Chemistry 2010, 2 (8), 1355-1369. 
 
58.  Golan, D. E.; Tashjian, A. H.; Armstrong, E. J.; Armstrong, A. W., Principles of 
Pharmacology: The Pathophysiologic Basis of Drug Therapy 3ed.; Lippincott Williams & 
Wilkins: Philadelphia, PA, 2012. 
 
59. 
Pharmacokinetic/Pharmacodynamic Models. Ph.D. Disseration, Michigan State 
University, East Lansing, MI, 2014. 
 
60.  Hay, M.; Thomas, D. W.; Craighead, J. L.; Economides, C.; Rosenthal, J., 
Clinical development success rates for investigational drugs. Nature biotechnology 
2014, 32 (1), 40. 
 
61.  Heller, A. A. In Silico Pharmacokinetics and the Potential Effects on 
Pharmacodynamics: The Missing Method in Antibiotic Development. Michigan State 
Unviersity, East Lansing, MI, 2018. 
 
62. 
Drug penetration and metabolism in 3D cell cultures treated in a 3D printed fluidic 
device: assessment of irinotecan via MALDI imaging mass spectrometry. Proteomics 
2016, 16 (11-12), 1814-21. 
 
63. 
Troutman, M. D.; de Morais, S. M., In Vitro P-glycoprotein Assays to Predict the in Vivo 
Interactions of P-glycoprotein with Drugs in the Central Nervous System. Drug 
Metabolism and Disposition 2008, 36 (2), 268-275. 
 

LaBonia, G. J.; Lockwood, S. Y.; Heller, A. A.; Spence, D. M.; Hummon, A. B., 

Lockwood, S. Y. Utilizing Fluidic Platforms for the Development of In Vitro 

Fan, J.; de Lannoy, I. A. M., Pharmacokinetics. Biochemical Pharmacology 2014, 

Feng, B.; Mills, J. B.; Davidson, R. E.; Mireles, R. J.; Janiszewski, J. S.; 

 

41 

Joo, J.; Wu, Z.; Lee, B.; Shon, J. C.; Lee, T.; Lee, I.-K.; Sim, T.; Kim, K.-H.; Kim, 

64.  Riede, J.; Poller, B.; Umehara, K.-i.; Huwyler, J.; Camenisch, G., New IVIVE 
method for the prediction of total human clearance and relative elimination pathway 
contributions from in vitro hepatocyte and microsome data. European Journal of 
Pharmaceutical Sciences 2016, 86, 96-102. 
 
65. 
N. D.; Kim, S. H.; Liu, K.-H., In vitro metabolism of an estrogen-related receptor γ 
modulator, GSK5182, by human liver microsomes and recombinant cytochrome P450s. 
Biopharmaceutics & Drug Disposition 2015, 36 (3), 163-173. 
 
66.  Soldatow, V. Y.; Lecluyse, E. L.; Griffith, L. G.; Rusyn, I., In vitro models for liver 
toxicity testing. Toxicology research 2013, 2 (1), 23-39. 
 
67. 
Jorgensen, J. H.; Ferraro, M. J., Antimicrobial susceptibility testing: a review of 
general principles and contemporary practices. Clinical infectious diseases : an official 
publication of the Infectious Diseases Society of America 2009, 49 (11), 1749-55. 
 
68.  Dhillon, S.; Kostrzewski, A., Clinical Pharmacokinetics. Pharmaceutical Press: 
London, UK, 2006. 
 
69.  DiPiro, J. T.; Spruill, W. J.; Wade, W. E.; Blouin, R. A.; Pruemer, J. M., Concepts 
in Clinical Pharmacokinetics. 4 ed.; American Society of Health-System Pharmacists: 
Bethesda, Maryland, 2005. 
 
70.  White, R. L., What In Vitro Models of Infection Can and Cannot Do. 
Pharmacotherapy: The Journal of Human Pharmacology and Drug Therapy 2001, 21 
(11P2), 292S-301S. 
 
71.  Murakawa, T.; Sakamoto, H.; Hirose, T.; Nishida, M., New in vitro kinetic model 
for evaluating bactericidal efficacy of antibiotics. Antimicrobial Agents and 
Chemotherapy 1980, 18 (3), 377-381. 
 
72.  Keil, S.; Wiedemann, B., Mathematical corrections for bacterial loss in 
pharmacodynamic in vitro dilution models. Antimicrobial Agents and Chemotherapy 
1995, 39 (5), 1054-1058. 
 
73.  White, C. A.; Toothaker, R. D.; Smith, A. L.; Slattery, J. T., Correction for 
bacterial loss in in vitro dilution models. Antimicrobial Agents and Chemotherapy 1987, 
31 (11), 1859-1860. 
 
74.  Budha, N. R.; Lee, R. B.; Hurdle, J. G.; Lee, R. E.; Meibohm, B., A Simple in vitro 
PK/PD Model System to Determine Time-Kill Curves of Drugs against Mycobacteria. 
Tuberculosis (Edinburgh, Scotland) 2009, 89 (5), 378-385. 
 

 

42 

Toothaker, R. D.; Welling, P. G.; Craig, W. A., An in vitro model for the study of 

Zinner, S. H.; Husson, M.; Klastersky, J., An artificial capillary in vitro kinetic 

Löwdin, E.; Odenholt, I.; Bengtsson, S.; Cars, O., Pharmacodynamic effects of 
75. 
sub-MICs of benzylpenicillin against Streptococcus pyogenes in a newly developed in 
vitro kinetic model. Antimicrobial Agents and Chemotherapy 1996, 40 (11), 2478-2482. 
 
76.  Grasso, S.; Meinardi, G.; De Carneri, I.; Tamassia, V., New In Vitro Model to 
Study the Effect of Antibiotic Concentration and Rate of Elimination on Antibacterial 
Activity. Antimicrobial Agents and Chemotherapy 1978, 13 (4), 570-576. 
 
77.  Gloede, J.; Scheerans, C.; Derendorf, H.; Kloft, C., In vitro pharmacodynamic 
models to determine the effect of antibacterial drugs. Journal of Antimicrobial 
Chemotherapy 2010, 65 (2), 186-201. 
 
78. 
antibacterial dosage regimen design. J Pharm Sci 1982, 71 (8), 861-4. 
 
79.  Shah, P. M., Activity of imipenem in an in-vitro model simulating pharmacokinetic 
parameters in human blood. Journal of Antimicrobial Chemotherapy 1985, 15 (suppl_A), 
153-157. 
 
80.  Gumbo, T.; Louie, A.; Deziel, M. R.; Parsons, L. M.; Salfinger, M.; Drusano, G. 
L., Selection of a Moxifloxacin Dose That Suppresses Drug Resistance in 
Mycobacterium tuberculosis, by Use of an In Vitro Pharmacodynamic Infection Model 
and Mathematical Modeling. The Journal of Infectious Diseases 2004, 190 (9), 1642-
1651. 
 
81. 
model of antibiotic bactericidal activity. The Journal of infectious diseases 1981, 144 (6), 
583-7. 
 
82. 
Cirit, M., Integrated Gut and Liver Microphysiological Systems for Quantitative In Vitro 
Pharmacokinetic Studies. The AAPS Journal 2017, 19 (5), 1499-1512. 
 
83.  Huh, D.; Torisawa, Y.-s.; Hamilton, G. A.; Kim, H. J.; Ingber, D. E., 
Microengineered physiological biomimicry: organs-on-chips. Lab on a Chip 2012, 12 
(12), 2156-2164. 
 
84.  Wikswo, J. P., The relevance and potential roles of microphysiological systems in 
biology and medicine. Experimental Biology and Medicine 2014, 239 (9), 1061-1072. 
 
85.  Murphy, S. V.; Atala, A., 3D bioprinting of tissues and organs. Nat Biotech 2014, 
32 (8), 773-785. 
 
86.  Kim, H. J.; Ingber, D. E., Gut-on-a-Chip microenvironment induces human 
intestinal cells to undergo villus differentiation. Integrative Biology 2013, 5 (9), 1130-
1140. 

Tsamandouras, N.; Chen, W. L. K.; Edington, C. D.; Stokes, C. L.; Griffith, L. G.; 

 

43 

87.  Chang, R.; Nam, J.; Sun, W., Direct cell writing of 3D microorgan for in vitro 
pharmacokinetic model. Tissue Engineering Part C: Methods 2008, 14 (2), 157-166. 
 
88.  Pinger, C. W.; Heller, A. A.; Spence, D. M., A Printed Equilibrium Dialysis Device 
with Integrated Membranes for Improved Binding Affinity Measurements. Analytical 
Chemistry 2017, 89 (14), 7302-7306. 
 
Lockwood, S. Y.; Meisel, J. E.; Monsma, F. J.; Spence, D. M., A Diffusion-Based 
89. 
and Dynamic 3D-Printed Device That Enables Parallel in Vitro Pharmacokinetic Profiling 
of Molecules. Analytical Chemistry 2016, 88 (3), 1864-1870. 
 
90.  Cryan, S. A.; Sivadas, N.; Garcia-Contreras, L., In vivo animal models for drug 
delivery across the lung mucosal barrier. Advanced Drug Delivery Reviews 2007, 59 
(11), 1133-51. 
 
91.  Sakai, C.; Iwano, S.; Yamazaki, Y.; Ando, A.; Nakane, F.; Kouno, M.; Yamazaki, 
H.; Miyamoto, Y., Species Differences in the Pharmacokinetic Parameters of 
Cytochrome P450 Probe Substrates between Experimental Animals, such as Mice, 
Rats, Dogs, Monkeys, and Microminipigs, and Humans. J Drug Metab Toxicol 2015, 5 
(173). 
 
92.  Boxenbaum, H., Interspecies scaling, allometry, physiological time, and the 
ground plan of pharmacokinetics. Journal of Pharmacokinetics and Biopharmaceutics 
1982, 10 (2), 201-227. 
 
93.  Pritchard, J. F.; Jurima-Romet, M.; Reimer, M. L. J.; Mortimer, E.; Rolfe, B.; 
Cayen, M. N., Making Better Drugs: Decision Gates in Non-Clinical Drug Development. 
Nature Reviews Drug Discovery 2003, 2, 542. 
 
94.  Greaves, P.; Williams, A.; Eve, M., First dose of potential new medicines to 
humans: how animals help. Nature Reviews Drug Discovery 2004, 3, 226. 
 
95.  Dedrick, R. L., Animal scale-up. Journal of Pharmacokinetics and 
Biopharmaceutics 1973, 1 (5), 435-461. 
 
96.  Drusano, G. L., Role of pharmacokinetics in the outcome of infections. 
Antimicrobial Agents and Chemotherapy 1988, 32 (3), 289-297. 
 
97.  Uno, S.; Dalton, T. P.; Dragin, N.; Curran, C. P.; Derkenne, S.; Miller, M. L.; 
Shertzer, H. G.; Gonzalez, F. J.; Nebert, D. W., Oral benzo[a]pyrene in Cyp1 knockout 
mouse lines: CYP1A1 important in detoxication, CYP1B1 metabolism required for 
immune damage independent of total-body burden and clearance rate. Molecular 
pharmacology 2006, 69 (4), 1103-14. 
 
 

 

44 

Zhang, Q. Y.; Gu, J.; Su, T.; Cui, H.; Zhang, X.; D'Agostino, J.; Zhuo, X.; Yang, 

98. 
W.; Swiatek, P. J.; Ding, X., Generation and characterization of a transgenic mouse 
model with hepatic expression of human CYP2A6. Biochemical and biophysical 
research communications 2005, 338 (1), 318-24. 
 
99.  Cheung, C.; Yu, A.-M.; Chen, C.-S.; Krausz, K. W.; Byrd, L. G.; Feigenbaum, L.; 
Edwards, R. J.; Waxman, D. J.; Gonzalez, F. J., Growth Hormone Determines Sexual 
Dimorphism of Hepatic Cytochrome P450 3A4 Expression in Transgenic Mice. Journal 
of Pharmacology and Experimental Therapeutics 2006, 316 (3), 1328-1334. 
 
100.  Zhang, D.; Luo, G.; Ding, X.; Lu, C., Preclinical experimental models of drug 
metabolism and disposition in drug discovery and development. Acta Pharmaceutica 
Sinica B 2012, 2 (6), 549-561. 
 
101.  Busch, U.; Schmid, J.; Heinzel, G.; Schmaus, H.; Baierl, J.; Huber, C.; Roth, W., 
Pharmacokinetics of Meloxicam in Animals and The Relevance to Humans. Drug 
Metabolism and Disposition 1998, 26 (6), 576-584. 
 
102.  Antunes, I. F.; Willemsen, A. T. M.; Sijbesma, J. W. A.; Boerema, A. S.; van 
Waarde, A.; Glaudemans, A.; Dierckx, R.; de Vries, E. G. E.; Hospers, G. A. P.; de 
Vries, E. F. J., In Vivo Quantification of ERbeta Expression by Pharmacokinetic 
Modeling: Studies with (18)F-FHNP PET. Journal of nuclear medicine : official 
publication, Society of Nuclear Medicine 2017, 58 (11), 1743-1748. 
 
103.  Wooten, D. W.; Guehl, N. J.; Verwer, E. E.; Shoup, T. M.; Yokell, D. L.; Zubcevik, 
N.; Vasdev, N.; Zafonte, R. D.; Johnson, K. A.; El Fakhri, G.; Normandin, M. D., 
Pharmacokinetic Evaluation of the Tau PET Radiotracer 18F-T807 (18F-AV-1451) in 
Human Subjects. Journal of Nuclear Medicine 2017, 58 (3), 484-491. 
 
104.  Craig, W. A.; Redington, J.; Ebert, S. C., Pharmacodynamics of amikacin in vitro 
and in mouse thigh and lung infections. Journal of Antimicrobial Chemotherapy 1991, 
27 (suppl_C), 29-40. 
 
105.  Zhao, M.; Lepak, A. J.; Andes, D. R., Animal models in the 
pharmacokinetic/pharmacodynamic evaluation of antimicrobial agents. Bioorganic & 
Medicinal Chemistry 2016, 24 (24), 6390-6400. 
 
106.  Frimodt-Moller, N., Correlation between pharmacokinetic/pharmacodynamic 
parameters and efficacy for antibiotics in the treatment of urinary tract infection. Int J 
Antimicrob Agents 2002, 19 (6), 546-53. 
107.  Velkov, T.; Bergen, P. J.; Lora-Tamayo, J.; Landersdorfer, C. B.; Li, J., PK/PD 
models in antibacterial development. Current opinion in microbiology 2013, 16 (5), 573-
9. 
 
108.  Nielsen, E. I.; Friberg, L. E., Pharmacokinetic-pharmacodynamic modeling of 
antibacterial drugs. Pharmacological reviews 2013, 65 (3), 1053-90. 

 

45 

109.  Selbie, F. R., Comparison of sodium and procaine penicillin in treatment of 
experimental staphylococcal infection in mice. British medical journal 1954, 1 (4875), 
1350-3. 
 
110.  Reyes, N.; Aggen, J. B.; Kostrub, C. F., In vivo efficacy of the novel 
aminoglycoside ACHN-490 in murine infection models. Antimicrob Agents Chemother 
2011, 55 (4), 1728-33. 
 
111.  Brunetti, J.; Falciani, C.; Roscia, G.; Pollini, S.; Bindi, S.; Scali, S.; Arrieta, U. C.; 
Gómez-Vallejo, V.; Quercini, L.; Ibba, E.; Prato, M.; Rossolini, G. M.; Llop, J.; Bracci, L.; 
Pini, A., In vitro and in vivo efficacy, toxicity, bio-distribution and resistance selection of 
a novel antibacterial drug candidate. Scientific Reports 2016, 6, 26077. 
 
112.  Gerber, A. U.; Brugger, H.-P.; Feller, C.; Stritzko, T.; Stalder, B., Antibiotic 
Therapy of Infections Due to Pseudomonas aeruginosa in Normal and 
Granulocytopenic Mice: Comparison of Murine and Human Pharmacokinetics. The 
Journal of Infectious Diseases 1986, 153 (1), 90-97. 
 
113.  Keel, R. A.; Crandon, J. L.; Nicolau, D. P., Efficacy of human simulated 
exposures of ceftaroline administered at 600 milligrams every 12 hours against 
phenotypically diverse Staphylococcus aureus isolates. Antimicrob Agents Chemother 
2011, 55 (9), 4028-32. 
 
114.  Lin, J. H.; Lin, T. H., Renal handling of drugs in renal failure. I: Differential effects 
of uranyl nitrate- and glycerol-induced acute renal failure on renal excretion of TEAB 
and PAH in rats. The Journal of pharmacology and experimental therapeutics 1988, 246 
(3), 896-901. 
 
115.  Mattoes, H. M.; Banevicius, M.; Li, D.; Turley, C.; Xuan, D.; Nightingale, C. H.; 
Nicolau, D. P., Pharmacodynamic Assessment of Gatifloxacin against Streptococcus 
pneumoniae. Antimicrobial Agents and Chemotherapy 2001, 45 (7), 2092-2097. 
 
116.  Giering, L. P.; Smith, J. E., Industrial perspective: from IND to NDA. Journal of 
Nuclear Cardiology 1995, 2 (1), 66-70. 
 
117.  Sagcal-Gironella, A. C. P.; Sherwin, C. M. T.; Tirona, R. G.; Phm, B.; Rieder, M. 
J.; Brunner, H. I.; Vinks, A. A., Open-label, single dose pharmacokinetic study of 
prednisolone at steady state in young patients with SLE on prednisone therapy. Clinical 
Therapeutics 2011, 33 (10), 1524-1536. 
 
118.  Zain-Hamid, R.; Ismail, Z.; Mahendra Raj, S.; Shuaib, I. L.; Mohsin, S. S. J., The 
Pharmacokinetics of Single Dose vs Steady-State Doses of Propranolol in Cirrhotic 
Malay Patients. The Malaysian Journal of Medical Sciences : MJMS 2002, 9 (1), 16-20. 
 
119.  Ministry of Health, L., and Welfare, Clinical Pharmacokinetic Studies of 
Pharmaceuticals. Ministry of Health, Labour, and Welfare: Japan, 2001. 

 

46 

120.  Garonzik, S. M.; Li, J.; Thamlikitkul, V.; Paterson, D. L.; Shoham, S.; Jacob, J.; 
Silveira, F. P.; Forrest, A.; Nation, R. L., Population Pharmacokinetics of Colistin 
Methanesulfonate and Formed Colistin in Critically Ill Patients from a Multicenter Study 
Provide Dosing Suggestions or Various Categories of Patients. Antimicrobial Agents 
and Chemotherapy 2011, 55 (7), 3284-3294. 
 
121.  Peloquin, C. A.; Hadad, D. J.; Molino, L. P. D.; Palaci, M.; Boom, W. H.; Dietze, 
R.; Johnson, J. L., Population Pharmacokinetics of Levofloxacin, Gatifloxacin, and 
Moxifloxacin in Adults with Pulmonary Tuberculosis. Antimicrobial Agents and 
Chemotherapy 2008, 52 (3), 852-857. 
 
122.  Charles, B., Population pharmacokinetics: an overview. Australian Prescriber 
2014, 37 (6), 210-213. 
 
123.  Kloft, C.; Wallin, J.; Henningsson, A.; Chatelut, E.; Karlsson, M. O., Population 
pharmacokinetic-pharmacodynamic model for neutropenia with patient subgroup 
identification: comparison across anticancer drugs. Clinical cancer research : an official 
journal of the American Association for Cancer Research 2006, 12 (18), 5481-90. 

 

 

 

47 

CHAPTER 2 

 

THE DEVELOPMENT OF A RAPID ASSAY FOR THE DETERMINATION OF 

ANTIBIOTIC SUSCEPTIBILITY IN GRAM-POSITIVE AND GRAM-NEGATIVE 

BACTERIAL STRAINS SHORTLY AFTER ANTIBIOTIC ADMINISTATION 

2.1 Introduction 

 

2.1.1 Current State of Drug and Antibiotic Development and Antibiotic Resistance 

 

As discussed in Chapter 1, drug development costs a pharmaceutical company 

greater than $2.5 billion and can take 10 years or more to complete development.1  This 

high cost is associated with about 90% of drugs failing after entering clinical trials.2  In 

order to reach the clinical stage, drugs must have completed preclinical trials, which 

individually costs about $1.1 billion and can take over 30 months to complete.1 The high 

cost may also be associated with increasing requirements in clinical trials to ensure 

safety and efficacy in humans3. Regardless of the reason for continually increasing 

costs, potential drugs that progress further in development before failure in the clinical 

stage cost companies more time and capital.4-5  The high cost of development, the high 

failure rate in clinical trials, and the threat of antibiotic resistance has led to many 

companies not pursuing the development of novel antibiotics.6-10  This loss of antibiotic 

development has resulted in fewer antibiotics reaching the market.11 Novel antibiotics 

with new mechanisms of action are needed as antibiotic resistance continues  to 

increase.12-13 

 

48 

 

Recently, there have been multiple discoveries of pandrug resistant bacterial 

strains in the community and in hospitals.14-16  These strains have been able to develop 

pandrug resistance due to the misuse and overuse of antibiotics, thereby providing 

selective pressure on the bacteria to evolve resistance by random mutagenesis and 

horizontal gene transfer.17-20  The threat of antibiotic resistance led to the U.S. 

government intervention under the presidential administration of Barack Obama, which 

resulted in an 84 point National Action plan.  This action plan included necessary 

development of rapid diagnostics and new antibiotics or therapeutics, which both rely on 

the use of susceptibility tests.21-24 

 

2.1.2 Antibiotic Susceptibility Testing 

 

Modern susceptibility testing involves adding an antibiotic to a bacterial culture 

and determining if the antibiotic is inhibiting the growth of the bacteria.  The most 

common susceptibility tests used today are the broth microdilution assay, the disk 

diffusion method, and the Etest.25  The most common method of the three is the 

microdilution assay (Figure 2.1A), which involves creating a series of dilutions with 

antibiotics in bacterial growth media, usually in a 96 well plate.  Bacteria are added to 

the dilutions and incubated overnight followed by growth determination based on the 

turbidity (optical density) of the bacterial culture.  This assay can be useful for 

determining the best antibiotic, and the best concentration of antibiotic, for the tested 

bacterial strain.25 The microdilution assay has been further adapted for colorimetric 

measurements for some antimicrobial species and has also been adapted to allow for 

high-throughput analysis. The disadvantage of the broth microdilution assay is that the 

 

49 

assay can take 24-48 hours for some bacterial strains to grow to detectable turbidity.26-

28 

 

The disk diffusion method is a common susceptibility test that is easy to use and 

inexpensive (Figure 2.1B). In the disk diffusion method, antibiotics are placed in wells 

that are inserted into an agar plate containing nutrients for bacteria to grow.  The 

antibiotics diffuse into the agar and decrease in concentration as the distance from the 

well increases.  Bacteria are then grown on the plate and the size of the ring around the 

well that does not allow bacteria to grow gives information about the susceptibility of the 

tested antibiotic to the tested strain of bacteria. The disk diffusion method usually takes 

16-24 hours to allow the bacteria to grow on the agar, but the procedure can be 

automated to take as little as 6 hours.25, 29-31  

The Etest is a commercially available plastic strip that has a gradient of 

antibiotics that decreases in concentration down the strip (Figure 2.1C).  The Etest 

works in a similar way to the disk diffusion method.  The plastic strips are place on an 

agar plate and bacteria are grown.  Specifically, the antibiotic on the strips diffuses out 

into the agar and can inhibit growth leaving areas with no growth around the strips. This 

method provides better precision between laboratories and is more quantitative than the 

previous method, but a similar time to the disk diffusion method is still required for the 

bacteria to grow.25, 32-33 The three susceptibility tests mentioned above provide valuable 

information regarding the type and concentration of antibiotics that should be used for 

treatment of a specific bacterial strain.  However, there are many disadvantages to 

these tests including semi-qualitative readouts, low precision between methods, and 

long incubation times.27, 34-35 

 

50 

 

Figure 2.1: The most common susceptibility tests. (A) The disk diffusion method 
utilizes a series of antibiotic dilutions to determine the minimum concentration of 
antibiotics needed to inhibit the growth of bacteria (MIC).  (B) The disk diffusion method 
utilizes the diffusion of antibiotic through agar in order to create an antibiotic gradient 
that gets lower as the distance from source increases. (C) The Etest is a quantitative 
version of the disk diffusion method that involves the use of plastic strips that have an 
antibiotic gradient that decreases down the strip.  These strips are placed in an agar 
plate and the antibiotic diffuses into the agar.  For both (B) and (C), the diameter or 
width of the area where bacteria does not grow provides information about the MIC. 

 

51 

2.1.3 Extracellular Adenosine Triphosphate during Bacterial Growth 

 

Adenosine triphosphate (ATP) is the molecule most often associated with cellular 

energy and participates in cell-to-cell communication as a signaling molecule.  ATP has 

also been used for detection of bacteria on food and medical supplies.36-38  Intracellular 

ATP has been used for susceptibility testing because intact bacteria have an 

intracellular concentration of 1-5 mM, but to measure this intracellular concentration, the 

bacterial cells must be lysed.39-41 Extracellular ATP concentrations of various bacteria 

have been quantified from 15 nM to 1.9 mM.42-45 

The rapid assay for susceptibility testing in this chapter is developed utilizing the 

extracellular ATP released by bacteria during the earlier stages of the growth phase. 

The early part of growth phase consists of three phases, namely, the lag phase, the 

logarithmic phase, and the stationary phase.  The lag phase is the phase where 

bacteria are adjusting to the environment and occurs prior to the logarithmic phase 

where the bacterial cells start dividing at an exponential rate.  The stationary phase is 

when the bacterial population is no longer increasing.24 Mempin et al. (2013) reported 

that the ratio of extracellular ATP to the optical density at 600 nm (OD600) increased 

during logarithmic growth until reaching the stationary phase where the ATP/OD600 

decreased.39  The OD600 is used as a measure of the relative number of bacterial cells.  

Interestingly, the decrease in ATP/OD600 was only seen when the bacteria were 

alive.39 It is hypothesized that monitoring the ATP/OD600 at a point after it reached a 

maximum could provide information about antibiotic susceptibility. Specifically, a 

decreasing ATP/OD600 would be indicative of a living culture while a constant or 

increasing ATP/OD600 would be indicative of not growing, dying or dead bacteria, like if 

 

52 

in the presence of an antibiotic.  This hypothesis is the basis to determine if a Gram-

positive, Gram-negative, or mixture bacterial strain is susceptible to a bacteriostatic or 

bactericidal antibiotic during growth within an hour of adding the antibiotic, creating a 

rapid, quantitative assay. 

 

 

 

53 

2.2 Materials and Methods 

2.2.1 Growth Media & Agar Plate Preparation 

Growth media was prepared by dissolving 3.0 g of Lysogeny broth – Lennox 

pellets (LB, EMD Chemicals, Darmstadt, Germany) in 150 mL of distilled and deionized 

water (DDW).  The solution was covered and mixed thoroughly before being autoclaved 

for 45 minutes at 121 ºC. The solution was cooled until it could be handled without the 

use of heat resistant gloves, after which the antibiotic(s) of interest were added to the 

desired concentration. The solution was then transferred to 50 mL centrifuge tubes and 

stored at 4 ºC.  The same procedure was used to create agar plates, except 2.25 g of 

agar (laboratory grade, Fisher Scientific, Fair Lawn, NJ) were added before the 

autoclaving step.   

 

2.2.2 Antibiotic Reagents   

Gibco  (Life Technologies, Grand Island, NY) was the supplier of kanamycin 

sulfate (USP grade) and 10 mg/mL gentamicin reagent solution.  Sigma Life Science 

(St. Louis, MO) was the supplier of levofloxacin (HPLC, ≥ 98.0%), chloramphenicol 

(water soluble) and tetracycline hydrochloride. Growth media for kanamycin resistant 

(KanR) Escherichia  coli (E. coli) was created by adding 50 mg of kanamycin to each 

liter of LB solution, resulting in a 100 μM solution of kanamycin in LB. Growth media for 

chloramphenicol resistant (CmpR) Bacillus subtilis (B. subtilis) was prepared by adding 

5 mg of chloramphenicol to each liter of LB solution, resulting in a 15 μM  solution of 

chloramphenicol in LB.   These growth media are referred to as the selective media or 

 

54 

selective agar and are used to ensure only growth of the antibiotic-resistant bacteria of 

interest. 

    

2.2.3 Bacterial Strains  

Michigan State University professors, Dr. David P. Weliky and Dr. Lee R. Kroos 

provided the KanR E. coli [BL21 (DE3)] and CmpR B. subtilis (BSL51), respectively.46-47 

The KanR E. coli was challenged to become levofloxacin-resistant. KanR E. coli was 

grown in 10 mL of selective LB containing 50 mg/L kanamycin for 6 hours.  Then, 1 mL 

of the E. coli culture was plated onto a selective agar plate containing kanamycin.  A P5 

filter paper disk (Fisherbrand, Pittsburgh, PA) was cut into smaller 22 mm disks and 

placed in a 25 mL of a 10 mg/L levofloxacin solution for a minimum of 2 hours.  A filter 

paper disk was removed from the levofloxacin solution and positioned at the center of 

the agar plate and the agar plate was incubated at 37 ºC overnight to promote growth.  

The filter paper disk, like the wells in the disk diffusion method, create a gradient of 

levofloxacin that decrease in concentration as the distance from the filter paper disk 

increases.  After overnight growth, the bacteria that were able to grow at the highest 

concentration of levofloxacin (closest to the disk) were transferred by a sterile 

inoculating loop to a new 10 mL solution of selective LB and grown for 6 more hours.  

This procedure was repeated until the bacteria could survive and propagate in a 

solution of 5 mg/L levofloxacin.  A 20% glycerol stock solution of the bacterial culture 

was prepared and stored at -80 ºC.  The glycerol stock solution was created by 

preparing an 80% (v/v) glycerol solution in DDW using glycerol (spectrophotometric 

grade, 99.5+%) purchased from Aldrich (Milwaukee, WI) and then incubated for 45 

 

55 

minutes at 121 ºC.  The bacterial culture of the levofloxacin and kanamycin resistant 

(KanLevR) E. coli was plated and grown overnight to create colonies. A single colony 

was transferred from the agar plate and grown for five hours to mid-logarithmic phase in 

10 mL of selective LB. A bacterial stock solution was created in a sterile tube by adding 

200 µL of the 80% glycerol solution and 800 µL of the mid-logarithmic phase bacteria.  

The bacterial stock was then mixed by vortexing and stored at -80 ºC. 

 

2.2.4 Sample Preparation 

This procedure (Figure 2.2) was adapted from Mempin et al.39 Bacteria from a 

glycerol stock were transferred to an agar plate containing the antibiotic that the 

bacteria are resistant via an inoculating loop. The plate is incubated overnight for E. coli 

or 48 hours for B. subtilis. A single colony was transferred using a pipette tip to a culture 

tube containing 2 mL of selective LB so only the bacteria of interest could grow.  This 

tube was then incubated at 210 rpm and 37 ºC overnight on a Talboys Professional 

Incubating Orbital Minishaker (Talboys, Thorofare, NJ).  The next morning, the OD600 

of the overnight culture was measured. A diluted culture with an OD600 ~ 0.005 was 

prepared and grown for 2 hours at 210 rpm and 37 ºC.  After, 1 mL of culture was 

removed for use as a 0 minutes sample and the drug of interest was added.  The 

concentrations of the drugs of interest were determined from pharmacological 

concentration (5 mg/L for levofloxacin and tetracycline; 6-10 mg/L for gentamicin).48-50  

Concentrations for kanamycin (50 mg/L) and chloramphenicol (5 mg/L) were 

determined by the inhibitory concentration of the bacterial stocks. The culture resumed 

incubation (210 rpm and 37 ºC) and additional 1 mL samples were aliquoted every 

 

56 

twenty minutes after the introduction of the drug of interest for one hour. The aliquots 

were stored at 4 ºC until the OD600 and extracellular ATP were quantified.   

 

57 

 

Figure 2.2: The procedure for the susceptibility assay. Bacteria from a glycerol 
stock are grown on a selective agar plate.  A colony of the bacteria is transferred from 
the agar plate and is grown overnight in selective media.  The OD600 is measured and 
a diluted culture with an OD ~ 0.005 is prepared in 10 mL of selective media.  This 
culture is preincubated for 2 hours or until after the maximum ATP/OD600 with shaking 
and then 1 mL of the culture is removed.  The antibiotic of interest is then added and 
incubation resumes.  Aliquots are taken every 20 minutes for one hour after the 
antibiotic has been added. Some of the culture is used for OD600 determination and the 
rest is centrifuged at 30,000g for 30 seconds.  The supernatant is removed and 
transferred to a black bottom 96 well plate to measure extracellular ATP concentration 
via the luciferin luciferase assay. Bacteria are illustrated as blue circles and ATP 
molecules are illustrated as red circles. 

 

58 

2.2.5 Determination of OD600 and Extracellular ATP 

The OD600 was measured to determine the relative amount of bacteria in 

solution. An OD600 is an absorbance measurement at a wavelength of 600 nm.  A 

volume of 150 µL was transferred from the collected aliquots to a clear bottom 96 well 

plate.  The remaining volume of the aliquots was centrifuged for 30 seconds at 30,000 

g.  The supernatant was transferred to a new tube and stored at 4 ºC until ATP 

measurements were performed.  In earlier procedures, the supernatant was filtered 

through a sterile, 0.45 µm, mixed cellulose ester membrane syringe filter (Merck 

Millipore, Tullagreen, Carrigtwohill, Co. Cork, IRL) and then centrifuged for 30 seconds 

at 30,000 g before being stored at room temperature. A luciferin-luciferase assay was 

used to measure the ATP concentrations of each aliquot. The luciferin-luciferase assay 

is a chemiluminescence measurement where the concentration of ATP is the limiting 

reagent in a reaction creating an excited oxyluciferin molecule that produces light.51  

The amount of light produced by the reaction is directly correlated with the 

concentration of ATP allowing for quantitative determination. The luciferin-luciferase 

solution was prerpared by combining 2 mg of potassium luciferin (Gold Biotechnology, 

St. Louis, MO) and 10 mg of firefly lantern extract (Sigma) into 5 mL of DDW and 

vortexing.  The luciferin-luciferase reagent was aliquoted into 1 mL storage vials and 

stored at -20 ºC.  To measure the ATP concentration, 150 µL of sample were added to 

a black bottom 96 well plate and 20 µL of the luciferin-luciferase solution were added.  

The samples were mixed immediately before measuring the chemiluminescense.  The 

concentration of ATP in the sample was quantified by comparing with the 

chemiluminescense signal of ATP standards.  An ATP working solution was prepared 

 

59 

using standard ATP (Adenosine 5’-triphosphatte disodium salt hydrate (Grade I, ≥ 99%), 

Sigma) in DDW with a final concentration of 1 µM.  Serial dilutions were performed in 

selective LB to create the samples used for the standard curve.  The concentration of 

extracellular ATP is dependent on the number of bacteria cells present in the culture. 

Therefore, the ATP/OD600 ratio was used to indicate if the bacteria were alive.  The 

ATP/OD600 ratio can vary due to the age of the samples such as the bacteria or the 

growth mediums (agar plates and LB growth media).  The ATP/OD600 ratios were 

normalized against the ATP/OD600 ratio of the control before the drug of interest was 

added to account for the differences in the samples.  The ATP/OD600 ratio of the 

antibiotic sample at a given time point was compared against the ATP/OD600 ratio of 

the control at the same time point to determine if there was a significant difference (α = 

0.05)  A Flexstation 3 microplate reader (Molecular Devices, Sunnyvale, CA) was used 

for all ATP and OD600 measurements. A Sorvall ST 8R Centrifuge (Thermo Scientific, 

Waltham, MA) was used for centrifugation.   

 

 

 

 

60 

2.3 Results 

2.3.1 ATP/OD600 of Growing Gram-Negative, KanR E. coli 

 

The trends published by Mempin et al. were confirmed when monitoring the 

ATP/OD600 ratio of growing KanR E.coli over the course of 7 hours.  As shown in 

Figure 2.3, the ATP/OD600 ratio increased until 180 minutes before being depleted. 

The individual curves of extracellular ATP and OD600 are shown in Figure 2.4.  The 

OD600 does not include the background absorbance of the plate and media.  The 

OD600 values with the background absorbance were used to generate Figure 2.3. The 

peak in the ATP/OD600 in Figure 2.3 corresponds with the peak of extracellular ATP in 

Figure 2.4 A.  The decrease in the ATP/OD600 curve in Figure 2.3 also corresponds 

with the OD600 no longer increasing logarithmically as seen in Figure 2.4 B.  

 

 

61 

Figure 2.3: The ATP/OD600 growth curve of KanR E. coli.  After diluting the 
overnight bacterial culture in new selective media, the ATP/OD600 increased until 180 
minutes.  Samples taken after 180 minutes showed a rapid decrease in the 
ATP/OD600. (n = 3; error = standard deviation) 

 

 

62 

Incubation Time (minutes)0100200300400500ATP (nM)/OD600050100150200250300350Figure 2.4: The individual extracellular ATP and OD600 curves during growth of 
KanR E. coli.  (A) The extracellular ATP curve shows a maximum at 180 minutes, 
which coincides with the maximum seen in the ATP/OD600 of Figure 2.3. (B) The 
OD600 graphs shows that at this same time point, the OD600 is no longer increasing 
logarithmically and appears to be starting to plateau. (n = 3; error = standard deviation) 

 

 

63 

 

As noted in the previous paragraph, the OD600 curve in Figure 2.4 B does not 

include the background absorbance associated with the 96 well-plate and the growth 

media.  When the background absorbance is removed, the resulting ATP/OD600 curve 

resembles that shown in Figure 2.5 A.  The maximum in the ATP/OD600 ratio is actually 

at 30 minutes.  Overlaying Figure 2.5 A with Figure 2.3 gives the graph in Figure 2.5 B.  

The maximum at 30 minutes is much lower in the sample with background absorbance 

due to relatively higher OD600 values caused by the background absorbance, which 

masks small changes in the OD600 during time points when a lot of extracellular ATP 

were being produced by relatively few bacteria.  The extracellular ATP peak may be 

indicative of the bacteria entering logarithmic phase, as the peak of the ATP/OD600 

appears to be occurring at the start of the logarithmic growth phase, not at the start of 

the stationary phase.  

 

64 

 

Figure 2.5: The ATP/OD600 curve with removal of the background absorbance 
and an overlay with the same data with the background absorbance. (A) Removing 
the background absorbance of the plate and media causes the maximum in the 
ATP/OD600 to shift to 30 minutes. (B) Overlaying the ATP/OD600 data without the 
background absorbance (black circles) with the data with the background absorbance 
(white circles) shows that the small changes in the OD600 were masked by the 
background absorbance and results in much lower ATP/OD600 values. (n = 3; error = 
standard deviation.) 

 

65 

2.3.2 Procedural Filtering and Centrifugation Compared to Just Centrifugation 

 

In section 2.2.5, it was mentioned that in preparation for measuring the 

extracellular ATP concentration of samples, the samples were either filtered and 

centrifuged, or just centrifuged to remove the bacterial cells from the culture.  In the 

original procedure, the bacterial culture was filtered through a membrane syringe filter 

and then centrifuged.  However, the membrane syringe filters were quite expensive at 

over $2 per filter and a new filter was needed for every aliquot.  A 3 hour growth curve 

for KanR E. coli was performed and the aliquots at each time point were prepared by 

either filtering and centrifugation or just centrifugation.  The normalized ATP/OD600 

values are shown in Figure 2.6.  The use of the filter did not provide any significant 

benefit (α = 0.05); therefore, only centrifugation was used allowing for a less expensive 

procedure per sample. 

 

 

66 

Figure 2.6: A comparison of normalized ATP/OD600 of growing KanR E. coli 
samples that were filtered and centrifuged or just centrifuged prior to 
extracellular ATP determination. There was no significant difference (α = 0.05) 
determined in the normalized ATP/OD600 ratio between samples that were filtered and 
centrifuged (white bars) and samples that were just centrifuged (black bars) at any point 
during a three hour growth curve of KanR E. coli. (n = 3; error = standard deviation). 

 

 

67 

2.3.3 Growing KanR E. coli Challenged with Antibiotic 

 

Kanamycin-resistant E. coli was preincubated for 2 hours before being 

challenged with the fluoroquinolone antibiotic, levofloxacin, at a concentration of 5 mg/L.  

The effect of levofloxacin on the normalized ATP/OD600 was determined at 20 minute 

intervals for one hour after adding the levofloxacin.  As shown in Figure 2.7, the 

samples dosed with levofloxacin had significantly different normalized ATP/OD600 

values in as little as 20 minutes after adding the antibiotic. This difference is the result of 

the OD600 of levofloxacin-dosed samples staying relatively constant and the 

extracellular ATP increasing due to cell death while the control samples, which were not 

dosed with levofloxacin, performed similarly to the samples in section 2.3.1. 

 

68 

Figure 2.7: The normalized ATP/OD600 of growing KanR E.coli challenged with 
levofloxacin. After adding levofloxacin (5 mg/L) to a 2 hour growing culture of KanR 
E.coli, a significant difference was determined at all time intervals between the control 
which was not challenged with antibiotic (white bars) and those samples that were 
challenged with the antibiotics (black bars).  The earliest time interval with a significant 
difference was at 20 minutes after adding the antibiotic. (n=3; error = standard 
deviation; * p < 0.05)  

 

 

69 

 

While the effect of levofloxacin was detectable in as little as 20 minutes after 

adding the antibiotic, the total time of the assay with the preincubation step to detection 

was 2 hours and 20 minutes. The procedure was performed again, this time with only a 

30 minute preincubation.  The time of 30 minutes was chosen because in Figure 2.5 A, 

the peak ATP/OD600 occurs after 30 minutes of growth. The results of the assay with 

only a 30 minute preincubation are shown in Figure 2.8.  A significant difference was 

determined within one hour between the controls and the sample dosed with 

levofloxacin. A significant difference was calculated, via a paired t-test, at 20 and 60 

minutes post-antibiotic (p < 0.05) and 40 minutes post-antibiotic (p < 0.10).  A 

conservative estimate for the amount of time required to perform this assay would be 2 

hours for this strain of bacteria. It is important to note that the ATP/OD600 of the 

controls after the 30 minute preincubation decrease much quicker than the 2 hour 

preincubation. It is also important to note that the normalized ATP/OD600 of the 

samples challenged with an antibiotic do not increase as fast.  This trend could be due 

to fewer bacterial cells being present; therefore, there is less ATP that is released from 

dying cells. 

 

70 

Figure 2.8: The effects of levofloxacin on the normalized ATP/OD600 of growing 
KanR E. coli with only a 30 minute preincubation. After only growing the KanR E. 
coli culture for 30 minutes, levofloxacin (5 mg/L) was added, a significant difference was 
determined at all time intervals between the control which was not challenged with 
antibiotic (white bars) and those samples that were challenged with the antibiotics 
(black bars).  The earliest time interval with a significant difference was at 20 minutes 
after adding the antibiotic. (n=3; error = standard deviation; * p < 0.05; ** p < 0.10) 

 

 

71 

 

Levofloxacin is a bactericidal antibiotic.  The bactericidal nature of the antibiotic 

may be the reason that the ATP/OD600 is increasing.  The procedure was performed 

again with a 2 hour preincubation, but this time the bacterial culture was dosed with the 

bacteriostatic antibiotic tetracycline at a bacteriostatic pharmacological concentration, 5 

mg/L. The results of this test are shown in Figure 2.9.  A significant difference was 

calculated 60 minutes after the antibiotic was added.  This increase was caused by the 

OD600 remaining relatively constant, as expecticed, and the extracellular ATP 

increasing.  This was surprising because a bacteriostatic antibiotic was not expected to 

cause lysis and an increase in extracellular ATP at these concentrations. So far, the 

assay has shown to be capable of determining if a bacterial strain is susceptible to both 

a bactericidal and bacteriostatic antibiotic in at least one hour of adding the antibiotic 

and there is potential for the assay to require a much smaller preincubation time leading 

to shorter total time required to perform the complete assay. 

 

 

 

 

 

 

72 

 
Figure 2.9: The effects of tetracycline on the normalized ATP/OD600 of growing 
KanR E.coli. After adding tetracycline (5 mg/L) to a 2 hour growing culture of KanR 
E.coli, a significant difference was determined at the 60 minute time intervals between 
the control which was not challenged with antibiotic (white bars) and those samples that 
were challenged with the antibiotics (black bars).  (n=3; error = standard deviation; * p < 
0.05) 

 

 

73 

2.3.4 Growing KanLevR E. coli Challenged with Antibiotic 

 

The previous section showed that the increase in ATP/OD600 occurs when 

KanR E. coli was challenged with an antibiotic that the bacteria was susceptible. To 

ensure that the antibiotic’s efficacy is causing the increase and not just the presence of 

the antibiotic, a levofloxacin resistant strain of KanR E. coli was created.  The assay 

was performed again with a two hour preincubation using the KanLevR E.coli and the 

bacteria was challenged with either levofloxacin or the aminoglycoside antibiotic, 

gentamicin at pharmacological concentrations.  The KanLevR E. coli should be 

susceptible to gentamicin. In Figure 2.10 A, the bacterial strain was challenged by 5 

mg/L levofloxacin.  There were no significant differences detected within one hour of 

adding the antibiotic.  In Figure 2.10 B, the bacterial strain was challenged by 6 mg/L 

gentamicin.  Significant differences were detected between the samples that were not 

challenged with antibiotic and those that were challenged at both 40 and 60 minutes 

after adding the antibiotic. This shows that the bacterial susceptibility to the antibiotic is 

the reason for the change in the ATP/OD600 level. 

 

74 

Figure 2.10: A growing KanLevR E. coli culture was challenged with levofloxacin 
and gentamicin. (A) Levofloxacin (5 mg/L) was added to the strain and no significant 
difference was detected between the culture that was not challenged (white bars) and 
the culture challenged by the antibiotic (black bars). (B) Gentamicin (6 mg/L) was added 
to the strain and significant differences were detected at both 40 and 60 minutes after 
adding the antibiotic. (n = 3; error = standard deviation; * p< 0.05) 

 

 

75 

2.3.5 ATP/OD600 of Growing Gram-Positive, CmpR B. subtilis 

 

In the previous three sections, the ATP/OD600 was analyzed for Gram-negative 

bacteria.  To ensure that this assay has the potential to work more than just Gram-

negative bacteria, the assay was tested on a Gram-positive bacterial strain.  Before the 

assay was performed, the ATP/OD600 was monitored on a fresh dilution (OD~0.005) of 

CmpR B. subtilis.  Background absorbance was removed from the OD600 before 

calculating the ATP/OD600 so the maximum ATP/OD600 could be detected. In Figure 

2.11, the same trend is seen that was seen in for E. coli. The ATP/OD600 quickly 

increases until 60 minutes before proceeding to decrease. 

 

 

 

76 

Figure 2.11: The ATP/OD600 growth curve of CmpR B. subtilis.  After diluting the 
overnight bacterial culture in new selective media, the ATP/OD600 increased until 60 
minutes.  Samples taken after 60 minutes showed a rapid decrease in the ATP/OD600. 
(n = 3; error = standard deviation) 

 

 

77 

Incubation Time (minutes)060120180240300360420ATP (nM)/OD600020040060080010001200 

Again, the individual extracellular ATP and OD600 curves were analyzed to 

determine when this peak is occurring. As seen in Figure 2.12 A, the maximum 

ATP/OD600 is not occurring when the extracellular ATP concentration is at a maximum, 

as the maximum extracellular ATP has not yet been reached. As shown in Figure 2.12 

B, the maximum ATP/OD600 appears to be occurring as the OD600 is entering 

logarithmic growth phase as the OD600 does not appear to be entering stationary 

phase in the 360 minutes that the bacteria were monitored. This agrees with the 

conclusions from the Gram-negative tests that the maximum ATP/OD600 occurs when 

the bacterial culture is entering the logarithmic growth phase. 

 

 

 

78 

Figure 2.12: The individual extracellular ATP and OD600 curves of B. subtilis.    
(A) The extracellular ATP increases over the course of 360 minutes, the maximum is 
not apparent. (B) The OD600 is also increasing over the course of 360 minutes, 
indicating that stationary phase has not been reached. (n = 3; error = standard 
deviation) 

 

 

79 

2.3.6 Growing CmpR B. subtilis Challenged with Antibiotic 

 

The Gram-positive, CmpR B. subtilis showed a similar ATP/OD600 curve to E. 

coli. Since the maximum ATP/OD600 of B subtilis occurred at one hour, the procedure 

was unchanged and only a two hour preincubation was used before challenging the 

growing culture with the bactericidal aminoglycoside antibiotic, kanamycin at 50 mg/L.  

The kanamycin challenged samples showed a statistical difference in the normalized 

ATP/OD600 at all time points measured, with the earliest being at 20 minutes after 

adding the antibiotic (Figure 2.13).  This shows that this assay can work on both Gram-

positive and Gram-negative bacteria, provided that the bacteria in question have the 

ATP/OD600 trend during growth.   

 

80 

Figure 2.13: The normalized ATP/OD600 of growing CmpR B.subtilis challenged 
with kanamycin. After adding kanamycin (50 mg/L) to a 2 hour growing culture of 
CmpR B. subtilis, a significant difference was determined at all time intervals between 
the control which was not challenged with antibiotic (white bars) and those samples that 
were challenged with the antibiotics (black bars).  The earliest time interval with a 
significant difference was at 20 minutes after adding the antibiotic. (n=3; error = 
standard deviation; * p < 0.05) 

 

 

81 

2.3.7 The Effect of One or Two Antibiotics on a Combined Bacterial Mixture 

 

So far, the assay has been tested on Gram-positive and Gram-negative bacteria 

as well as bactericidal and bacteriostatic antibiotics.  These experimental designs 

benefit research by determining the antibiotic that is effective in treatment of a bacterial 

strain with unknown resistance and in testing new antibiotics against bacterial strains of 

known resistance.  However, to have use in a hospital or doctor’s office, the assay 

would be more beneficial in determining what antibiotic would be effective against a 

bacterial infection of unknown composition.  The assay was performed by growing a 

50/50 mixture of KanR E. coli and CmpR B. subitlis in growth media that contains no 

antibiotic to ensure both strains can grow uninhibited.  After 2 hours, the samples were 

challenged with either the macrolide-like bacteriostatic antibiotic, chloramphenicol (5 

mg/L); the bactericidal aminoglycoside, kanamycin (50 mg/L); or the bactericidal 

aminoglycoside, gentamicin (10 mg/L). Both strains of bacteria are resistant to either 

kanamycin or chloramphenicol, but both are susceptible to gentamicin.  Within 20 

minutes of adding the antibiotic, only gentamicin resulted in a statistical difference in the 

normalized ATP/OD600.  After 40 minutes and 60 minutes, all antibiotics caused a 

statistical difference in the normalized ATP/OD600, but gentamicin’s effect on the 

normalized ATP/OD600 was statistically different from the effect of the other drugs on 

the normalized ATP/OD600 (Figure 2.14).  This shows that the antibiotic with the 

greatest efficacy in killing the bacterial culture causes the greatest increase in the 

ATP/OD600. 

 

82 

 

Figure 2.14: The single- or dual-antibiotic effect on a 50/50 bacterial mixture of 
KanR E. coli and CmpR B. subtilis.  A mixture of KanR E. coli and CmpR B. subtilis 
was preincubated for two hours with no antibiotic present.  Either no antibiotic (controls; 
white bars), chloramphenicol (5 mg/L; striped bars), kanamycin (50 mg/L; checkered 
bars), or gentamicin (10 mg/L; black bars) was added.  Gentamicin caused a statistical 
difference from the controls at all time points after the antibiotic was added, with the 
earliest occurring 20 minutes after adding the antibiotic.  Chloramphenicol and 
kanamycin did not show statistical differences from the controls until 40 minutes after 
adding the antibiotic. (n = 3; error = standard deviation; * p < 0.05) 

 

83 

 2.4 Discussion 

 

The trend described my Mempin et al. (2013) concerning the ATP/OD600 ratio of 

growing bacteria was seen both in Gram-positive B. subtilis and Gram-negative E. coli.  

These strains were chosen purely for convenience due to them being readily available 

and non-pathogenic.  By removing the background absorbance of the 96 well-plate and 

the growth media in the OD600 measurement, the ATP/OD600 peak shifted to a much 

earlier time: 30 minutes for E. coli and 60 minutes for B. subtilis.  Mempin et al. showed 

that the maximum ATP/OD600 was indicative of the bacteria entering stationary phase.  

However, without the background absorbance masking the small changes in the 

OD600, the maximum ATP/OD600 appears to be occurring at the time when the 

bacteria are entering logarithmic growth, not exiting.  At these time points, very few 

bacteria are causing a relatively large increase in extracellular ATP.  Mempin et al. 

showed that only living cells cause the decrease in the ATP/OD600 ratio.39  

 In the previous section, adding an antibiotic after the maximum ATP/OD600 

caused the ATP/OD600 to not decrease with the controls (bacteria with no antibiotic).  

In fact, the ATP/OD600 increased due to the OD600 remaining constant and the 

extracellular ATP increasing most likely due to the death and leakage of intracellular 

ATP from the bacteria into the extracellular matrix.  This trend was shown with both 

bactericidal (levofloxacin, gentamicin, and kanamycin) and bacteriostatic antibiotics 

(tetracycline and chloramphenicol). In all cases, the susceptibility of the antibiotic on the 

bacteria was detectable by a significant difference between the ATP/OD600 of the 

antibiotic challenged samples and the controls 60 minutes after adding the antibiotic, 

with the earliest detection being 20 minutes after adding the antibiotic. The difference in 

 

84 

time for a statistical difference to be measured may be dependent on the mechanism of 

action of the antibiotic. The preincubation is important because the bacteria need to be 

incubated long enough for the bacteria to reach their maximum ATP/OD600.  However, 

the assay was shown to work directly after the maximum ATP/OD600 or 1.5 hours after.  

With the minimum preincubation time, the time required to grow and challenge the 

bacteria can take as little as 1.5 hours depending on the strain of bacteria. Using the 

assay in this way could be beneficial for research regarding development of new 

antibiotics or creating profiles for bacteria with unknown resistance.   

The assay can also show which antibiotics are causing more cell death in a 

mixed bacterial culture because the antibiotic or combination of antibiotics that kills the 

most bacteria causes the greatest ATP/OD600 when compared with controls.  The 

antibiotic gentamicin had a statistically different ATP/OD600 to controls within 20 

minutes of being administered to a mixture of KanR E.coli and CmpR B. subtilis. The 

antibiotics of chloramphenicol and kanamycin were only effective in treating one of the 

bacterial strains in the mixture and statistical differences were not determined until 40 

minutes after the antibiotic was added and the increase in the ATP/OD600 ratio was 

much smaller than the ATP/OD600 caused by gentamicin at the same time points. This 

experimental design could be beneficial for determining which antibiotic or combination 

of antibiotics should be administered. 

The limitation of this assay is that it is dependent on the release of ATP during 

bacterial growth.  As the exact reason for the increase in extracellular ATP is unknown, 

bacteria will have to be tested prior to the assay to determine if they follow the 

ATP/OD600 trend described here and when the maximum ATP/OD600 occurs.  This 

 

85 

assay utilizes a chemiluminescense assay to measure the extracellular ATP. As some 

bacteria naturally luminesce, this can interfere with the measurement of extracellular 

ATP and another quantification method may be required.  The bacteria used in this 

assay are relatively fast growing, a slower growing bacteria may require more time, but 

if the ATP/OD600 trend is seen, the assay should work. Another limitation is that the 

bacteria are being tested against a static (constant) concentration of the antibiotic. 

While this allows the assay to test for efficacy of an antibiotic, it does not take into 

account the PK/PD of the human body, which will be altering the concentration of the 

antibiotic and, in turn, the efficacy.  Pharmacological concentrations were used to 

challenge the bacteria; however, inside the body these concentrations are found in the 

blood, not in the tissues where the bacteria are found.  This could explain why the 

bacteriostatic antibiotics were increasing the extracellular ATP concentration since a 

higher concentration of a bacteriostatic antibiotic could lead to bactericidal results.  This 

assay should work for a bacteriostatic antibiotic, but the amount of time to measure a 

statistical difference in the ATP/OD600 may take longer since the extracellular ATP 

should not increase, and the OD600 would remain the same leading to a constant 

ATP/OD600.  The difference would not be measurable until the viable bacteria in the 

controls deplete the extracellular ATP and increase in OD600 enough to decrease the 

ATP/OD600 of the controls.  

This assay could be more beneficial if developed to work for a dynamic, two-

compartment model instead of a static model like here. The first compartment could 

have a controlled PK curve of the antibiotic to pharmacological concentrations. The 

second compartment separated by a diffusion membrane would allow for lower 

 

86 

antibiotic concentrations.  These secondary compartment concentrations may be more 

representative to the antibiotic concentrations that the bacteria experience in the tissue.   

Overall, this assay has been shown to provide a rapid determination of antibiotic 

efficacy for both Gram-positive and Gram-negative bacteria as well as bactericidal 

antibiotics.  The amount of time required to complete the assay is less than the reported 

time required to grow the bacteria required to determine the results from the current 

susceptibility methods.  The total cost of this assay is comparable to that of the broth 

microdilution method as it requires all of the same equipment with the addition of the 

luciferin-luciferase assay, which only constitutes a $0.19 increase/sample. This method 

also has the potential to be adapted for high-throughput analysis and can be adapted 

for multi-compartment dynamic PK/PD systems. 

 

 

 

 

 

 

 

 

 

87 

REFERENCES

 

88 

REFERENCES 

Getz, K. A.; Wenger, J.; Campo, R. A.; Seguine, E. S.; Kaitin, K. I., Assessing the 

DiMasi, J. A.; Feldman, L.; Seckler, A.; Wilson, A., Trends in risks associated 

Heller, A. A. In Silico Pharmacokinetics and the Potential Effects on 

DiMasi, J. A.; Grabowski, H. G.; Hansen, R. W., Innovation in the pharmaceutical 

Hay, M.; Thomas, D. W.; Craighead, J. L.; Economides, C.; Rosenthal, J., 

 
 
 
1. 
industry: New estimates of R&D costs. Journal of Health Economics 2016, 47, 20-33. 
 
2. 
Clinical development success rates for investigational drugs. Nature biotechnology 
2014, 32 (1), 40. 
 
3. 
impact of protocol design changes on clinical trial performance. American journal of 
therapeutics 2008, 15 (5), 450-457. 
 
4. 
with new drug development: success rates for investigational drugs. Clinical 
pharmacology and therapeutics 2010, 87 (3), 272-7. 
 
5. 
Pharmacodynamics: The Missing Method in Antibiotic Development. Michigan State 
Unviersity, East Lansing, MI, 2018. 
 
6. 
Discov 2009, 8 (12), 959-68. 
 
7. 
in pharmaceutical R&D efficiency. Nature reviews Drug discovery 2012, 11 (3), 191. 
 
8. 
R. M.; Pairaudeau, G.; Pennie, W. D.; Pickett, S. D.; Wang, J., An analysis of the 
attrition of drug candidates from four major pharmaceutical companies. Nature reviews 
Drug discovery 2015, 14 (7), 475. 
 
9. 
Berndt, E. R.; Nass, D.; Kleinrock, M.; Aitken, M., Decline in economic returns 
from new drugs raises questions about sustaining innovations. Health affairs (Project 
Hope) 2015, 34 (2), 245-52. 
 
10.  Sukkar, E., Why are there so few antibiotics in the reserach and development 
pipeline? The Pharmaceutical Journal 2013, 291, 520. 
 
11. 
 
12. 
Mechanisms of Bacterial Resistance to Antimicrobial Agents. Microbiology spectrum 
2018, 6 (1). 

Munos, B., Lessons from 60 years of pharmaceutical innovation. Nat Rev Drug 

Waring, M. J.; Arrowsmith, J.; Leach, A. R.; Leeson, P. D.; Mandrell, S.; Owen, 

Taubes, G., The Bacteria Fight Back. Science 2008, 321 (5887), 356-361. 

van Duijkeren, E.; Schink, A. K.; Roberts, M. C.; Wang, Y.; Schwarz, S., 

Scannell, J. W.; Blanckley, A.; Boldon, H.; Warrington, B., Diagnosing the decline 

 

89 

Zowawi, H. M.; Forde, B. M.; Alfaresi, M.; Alzarouni, A.; Farahat, Y.; Chong, T.-

13.  Blair, J. M. A.; Webber, M. A.; Baylay, A. J.; Ogbolu, D. O.; Piddock, L. J. V., 
Molecular mechanisms of antibiotic resistance. Nature Reviews Microbiology 2014, 13, 
42. 
 
14. 
M.; Yin, W.-F.; Chan, K.-G.; Li, J.; Schembri, M. A., Stepwise evolution of pandrug-
resistance in Klebsiella pneumoniae. Scientific reports 2015, 5, 15082. 
 
15.  Meir-Gruber, L.; Manor, Y.; Gefen-Halevi, S.; Hindiyeh, M. Y.; Mileguir, F.; Azar, 
R.; Smollan, G.; Belausov, N.; Rahav, G.; Shamiss, A., Population screening using 
sewage reveals pan-resistant bacteria in hospital and community samples. PloS one 
2016, 11 (10), e0164873. 
 
16.  Ghafur, A.; Lakshmi, V.; Kannain, P.; Thirunarayan, M., Emergence of Pan drug 
resistance amongst gram negative bacteria! The First case series from India. Journal of 
Microbiology and Infectious Diseases 2014, 4 (03). 
 
17.  Botts, R. T.; Apffel, B. A.; Walters, C. J.; Davidson, K. E.; Echols, R. S.; Geiger, 
M. R.; Guzman, V. L.; Haase, V. S.; Montana, M. A.; La Chat, C. A.; Mielke, J. A.; 
Mullen, K. L.; Virtue, C. C.; Brown, C. J.; Top, E. M.; Cummings, D. E., Characterization 
of Four Multidrug Resistance Plasmids Captured from the Sediments of an Urban 
Coastal Wetland. Frontiers in Microbiology 2017, 8 (1922). 
 
18. 
Microbiology 2015, 41 (1), 101-108. 
 
19. 
determining minimal concentrations of antibiotics that drive horizontal transfer of 
resistance. Science of The Total Environment 2016, 548-549, 131-138. 
 
20.  Miller, J. H.; Novak, J. T.; Knocke, W. R.; Pruden, A., Survival of Antibiotic 
Resistant Bacteria and Horizontal Gene Transfer Control Antibiotic Resistance Gene 
Content in Anaerobic Digesters. Frontiers in Microbiology 2016, 7 (263). 
 
21. 
agricultural antibiotic use sufficiently addressed? Journal of Law and the Biosciences 
2015, 2 (1), 129-138. 
 
22.  Obama, B., Executive Order - Combating Antibiotic-Resistant Bacteria. The 
White House: Washington D.C., 2014. 
 
23.  National Action Plan for Combating Antibiotic-Resistant Bacteria. The White 
House: Washington D.C., 2015. 
 
24.  Heller, A. A.; Spence, D. M., A rapid method for post-antibiotic bacterial 
susceptibility testing. PLOS ONE 2019, 14 (1), e0210534. 

Jooma, S., Executive action to combat the rise of drug-resistant bacteria: is 

Juhas, M., Horizontal gene transfer in human pathogens. Critical Reviews in 

Jutkina, J.; Rutgersson, C.; Flach, C.-F.; Joakim Larsson, D. G., An assay for 

 

90 

Jorgensen, J. H.; Ferraro, M. J., Antimicrobial susceptibility testing: a review of 
25. 
general principles and contemporary practices. Clinical infectious diseases : an official 
publication of the Infectious Diseases Society of America 2009, 49 (11), 1749-55. 
 
26. 
Leong, C.; Buttafuoco, A.; Glatz, M.; Bosshard, P. P., Antifungal susceptibility 
testing of Malassezia spp. with an optimized colorimetric broth microdilution method. 
Journal of clinical microbiology 2017, 55 (6), 1883-1893. 
 
27.  Gupta, P.; Khare, V.; Kumar, D.; Ahmad, A.; Banerjee, G.; Singh, M., 
Comparative Evaluation of Disc Diffusion and E-test with Broth Micro-dilution in 
Susceptibility testing of Amphotericin B, Voriconazole and Caspofungin against Clinical 
Aspergillus isolates. Journal of Clinical and Diagnostic Research : JCDR 2015, 9 (1), 
DC04-DC07. 
 
28.  Prüller, S.; Frömke, C.; Kaspar, H.; Klein, G.; Kreienbrock, L.; Kehrenberg, C., 
Recommendation for a standardised method of broth microdilution susceptibility testing 
for porcine Bordetella bronchiseptica. PloS one 2015, 10 (4), e0123883. 
 
29.  Hombach, M.; Jetter, M.; Blöchliger, N.; Kolesnik-Goldmann, N.; Böttger, E. C., 
Fully automated disc diffusion for rapid antibiotic susceptibility test results: a proof-of-
principle study. Journal of Antimicrobial Chemotherapy 2017, 72 (6), 1659-1668. 
 
30.  Buzón-Durán, L.; Capita, R.; Alonso-Calleja, C., Antibiotic susceptibility of 
methicillin-resistant staphylococci (MRS) of food origin: A comparison of agar disc 
diffusion method and a commercially available miniaturized test. Food Microbiology 
2018, 72, 220-224. 
 
31.  Matuschek, E.; Brown, D. F. J.; Kahlmeter, G., Development of the EUCAST disk 
diffusion antimicrobial susceptibility testing method and its implementation in routine 
microbiology laboratories. Clinical Microbiology and Infection 2014, 20 (4), O255-O266. 
 
32.  Badiee, P.; Alborzi, A., Susceptibility of clinical Candida species isolates to 
antifungal agents by E-test, Southern Iran: A five year study. Iranian journal of 
microbiology 2011, 3 (4), 183. 
 
33.  Ranque, S.; Lachaud, L.; Gari-Toussaint, M.; Michel-Nguyen, A.; Mallié, M.; 
Gaudart, J.; Bertout, S., Inter-laboratory reproducibility of Etest amphotericin-B and 
caspofungin yeast susceptibility testing and comparison with the CLSI method. Journal 
of clinical microbiology 2012, JCM. 00490-12. 
 
34.  Ogata, S. K.; Gales, A. C.; Kawakami, E., Antimicrobial susceptibility testing for 
Helicobacter pylori isolates from Brazilian children and adolescents: comparing agar 
dilution, E-test, and disk diffusion. Brazilian Journal of Microbiology 2014, 45, 1439-
1448. 
 

 

91 

Luo, J.; Liu, X.; Tian, Q.; Yue, W.; Zeng, J.; Chen, G.; Cai, X., Disposable 

35.  Alberti, M. O.; Hindler, J. A.; Humphries, R. M., Performance of Etest for 
Antimicrobial Susceptibility Testing of Abiotrophia defectiva and Granulicatella Species. 
Journal of clinical microbiology 2016, 54 (8), 2194-2196. 
 
36. 
bioluminescence-based biosensor for detection of bacterial count in food. Analytical 
Biochemistry 2009, 394 (1), 1-6. 
 
37.  Bottari, B.; Santarelli, M.; Neviani, E., Determination of microbial load for different 
beverages and foodstuff by assessment of intracellular ATP. Trends in Food Science & 
Technology 2015, 44 (1), 36-48. 
 
38.  Amodio, E.; Dino, C., Use of ATP bioluminescence for assessing the cleanliness 
of hospital surfaces: A review of the published literature (1990–2012). Journal of 
Infection and Public Health 2014, 7 (2), 92-98. 
 
39.  Mempin, R.; Tran, H.; Chen, C.; Gong, H.; Kim Ho, K.; Lu, S., Release of 
extracellular ATP by bacteria during growth. BMC Microbiology 2013, 13, 301-301. 
 
40.  Yaginuma, H.; Kawai, S.; Tabata, K. V.; Tomiyama, K.; Kakizuka, A.; 
Komatsuzaki, T.; Noji, H.; Imamura, H., Diversity in ATP concentrations in a single 
bacterial cell population revealed by quantitative single-cell imaging. Scientific Reports 
2014, 4, 6522. 
 
41.  Schneider, D. A.; Gourse, R. L., Relationship between growth rate and ATP 
concentration in Escherichia coli: a bioassay for available cellular ATP. The Journal of 
biological chemistry 2004, 279 (9), 8262-8. 
 
42.  Hironaka, I.; Iwase, T.; Sugimoto, S.; Okuda, K.-i.; Tajima, A.; Yanaga, K.; 
Mizunoe, Y., Glucose Triggers ATP Secretion from Bacteria in a Growth-Phase-
Dependent Manner. Applied and Environmental Microbiology 2013, 79 (7), 2328-2335. 
 
43. 
Mizunoe, Y., Isolation and Identification of ATP-Secreting Bacteria from Mice and 
Humans. Journal of Clinical Microbiology 2010, 48 (5), 1949-1951. 
 
44. 
“Danger Signal”. Science Signaling 2009, 2 (56), pe6-pe6. 
 
45. 
level variations in heterotrophic bacteria during attachment on hydrophilic and 
hydrophobic surfaces. International microbiology : the official journal of the Spanish 
Society for Microbiology 2006, 9 (1), 37-46. 
 
 

Iwase, T.; Shinji, H.; Tajima, A.; Sato, F.; Tamura, T.; Iwamoto, T.; Yoneda, M.; 

Trautmann, A., Extracellular ATP in the Immune System: More Than Just a 

Ivanova, E. P.; Alexeeva, Y. V.; Pham, D. K.; Wright, J. P.; Nicolau, D. V., ATP 

 

92 

Fish, D. N.; Chow, A. T., The Clinical Pharmacokinetics of Levofloxacin. Clinical 

46.  Ratnayake, P. U.; Ekanayaka, E. A. P.; Komanduru, S. S.; Weliky, D. P., Full-
Length Trimeric Influenza Virus Hemagglutinin II Membrane Fusion Protein and Shorter 
Constructs Lacking the Fusion Peptide or Transmembrane Domain: 
Hyperthermostability of the Full-Length Protein and the Soluble Ectodomain and Fusion 
Peptide Make Significant Contributions to Fusion of Membrane Vesicles. Protein 
expression and purification 2016, 117, 6-16. 
 
47.  Halder, S.; Parrell, D.; Whitten, D.; Feig, M.; Kroos, L., Interaction of 
intramembrane metalloprotease SpoIVFB with substrate Pro-σK. Proceedings of the 
National Academy of Sciences 2017, 114 (50), E10677-E10686. 
 
48. 
Pharmacokinetics 1997, 32 (2), 101-119. 
 
49.  Agwuh, K. N.; MacGowan, A., Pharmacokinetics and pharmacodynamics of the 
tetracyclines including glycylcyclines. Journal of Antimicrobial Chemotherapy 2006, 58 
(2), 256-265. 
 
50.  Stickland, M. D.; Kirkpatrick, C. M. J.; Begg, E. J.; Duffull, S. B.; Oddie, S. J.; 
Darlow, B. A., An extended interval dosing method for gentamicin in neonates. Journal 
of Antimicrobial Chemotherapy 2001, 48 (6), 887-893. 
 
51.  Hirano, T.; Hasumi, Y.; Ohtsuka, K.; Maki, S.; Niwa, H.; Yamaji, M.; Hashizume, 
D., Spectroscopic Studies of the Light-Color Modulation Mechanism of Firefly (Beetle) 
Bioluminescence. Journal of the American Chemical Society 2009, 131 (6), 2385-2396. 
 

 
 

 

 

93 

CHAPTER 3 

 

THE DEVELOPMENT OF A 3D PRINTED, TWO COMPARTMENT MODEL FOR PK 

MODELING OF VARIOUS ADMINISTRATIONS WITH NOVEL 3D PRINTED INSERTS 

CONTAINING INTEGRATED MEMBRANES 

 

3.1 Introduction to 3D Printing 

Three-dimensional (3D) printing is a relatively new technology that can be used 

to rapidly make prototypes of a design. It allows for small changes in each iteration until 

a desired outcome is reached.1  Printing in 3D works by initially designing the part on a 

computer using computer aided design (CAD) software.  The part is then saved as a 

stereolithography (.stl) file, which divides the part into a series of layers made up of 

triangular sections.  The number of triangular sections determines the resolution of the 

print. The vertices of these triangles are stored on the .stl file and can be interpreted by 

a 3D printer to print 2D, horizontal sections of the part, which are printed layer by layer 

to create the final 3D product (Figure 3.1).1-2  

 

 

 

 

94 

Figure 3.1: The process of 3D printing a device. (Left) First, the device is designed to 
desired dimensions using CAD software.  Then, the part file is converted to an .stl file so 
the 3D printer can interpret the part.  (Center) The part is then printed layer by layer by 
the 3D printer using the desired material. (Right) The device is then removed from the 
printer and cleared of all support material, which is sacrificial material used to create 
voids and channels, to provide the finished device. 

 

 

95 

The use of 3D printing has benefited many fields of research including, but not 

limited to, custom labware development, biology, medicine, and fluidics.2-7  Prior to the 

utilization of 3D printing in the fabrication of fluidic and microfluidic devices, 

polydimethylsiloxane (PDMS) was used as the fabrication material. The utilization of 3D 

printing for fabrication of these devices is advantageous due to easier fabrication, 

integration of multiple materials, integration of multiple 3D printed components, better 

inter-laboratory standardization, and constant development of new and improved 

materials.1-2  3D printing has been successfully used in the fabrication of fluidic 

pharmacokinetic/pharmacodynamics (PK/PD) devices.8-11  These devices utilize 

commercial membrane inserts to mimic absorption, distribution, and/or excretion of 

molecules to gather PK information.  Integration of multiple parts of the PK analysis has 

enabled a PK curve in a single compartment for either an oral or continuous intravenous 

(IV) drug administration.  By recreating how the molecule should react in vivo, more 

clinically-relevant PD data can be obtained.8-9 The disadvantage to the current 3D 

printed, multicompartment models is that they are limited to only actively measuring one 

compartment.  This limitation is due to the reliance on commercially available inserts 

that are designed to only allow diffusion across one plane. In this chapter, fused 

deposition modeling (FDM) and polyjet printing were used to create a 3D printed insert 

that incorporates porous membranes.  These inserts allow diffusion across more than 

one plane in order to create a PK/PD device that enables active monitoring of more than 

one compartment. 

 

 

 

96 

3.1.1 Fused Deposition Modeling 

The FDM style of 3D printing is the most widely used form of 3D printing.1  FDM 

printing utilizes the semi-molten state of materials in order to manipulate the material so 

that it can be printed onto a stage, layer by layer, to create the designed model.12-14     

An FDM style 3D printer consists of a coil of filament of a material including, but not 

limited to, acrylonitrile butadiene styrene (ABS), polylactic acid (PLA), polycarbonate, 

polystyrene, and glass.15-19 FDM printers can also print filaments consisting of metal or 

ceramics as long as a binder is incorporated to create the filament.20-23 This filament is 

guided by two rollers to the nozzle of the print head.  At the print head, temperature 

control units increase the temperature of the thermoplastic filament until it reaches a 

semi-molten state.  The material is then released from the nozzle onto the stage in the 

shape of the design.  The stage then lowers and the next layer of material is added.  

This layer by layer printing continues until the 3D design is complete.1-2, 12, 14 Voids in 

the part can be filled with printed support material that can be removed by physical or 

chemical means.  Without the use of a support material, these voids would cause the 

object to collapse during printing. 

The benefit of FDM printing is that it can allow custom parts to be printed in 

remote locations where a machine shop is not viable.24 FDM printing also has the 

benefit of using materials that can be altered by organic solvents, such as acetone, to 

change the morphology of the 3D printing object.  For example, acetone can be used to 

make the surface smooth and water-tight.25 The disadvantage of FDM style printing is 

that the nature of the materials are prone to alterations of their structure at elevated 

temperature.  Also, organic solvents can alter device morphology, meaning that the 

 

97 

devices cannot be used long-term with organic molecules, as this will also change the 

structure. 2, 26-28 Another disadvantage to FDM, as well as other types of 3D printing 

modes, is that the resolution of the printer is related to the cost of the printer.29 Many 

affordable FDM printers do not have the resolution capabilities to create intricate 

devices. 

 

3.1.2 Polyjet Printing 

Polyjet printing uses liquid polymers that are photocurable under ultraviolet (UV) 

light.  The liquid polymer is transferred to the print heads, which lays the liquid polymer 

onto the stage in a thin layer before a UV light source passes over the liquid polymer 

and cures it.  The stage then lowers and the next layer is printed.2, 10  The use of a 

support material fills voids in the device so the liquid material does not fill the empty 

space during the print.  This support material can be removed at a later time by physical 

and/or chemical means.2 The advantage of polyjet printing is that the latest printer can 

print up to six materials in a single print job.  Additionally, there are many different types 

of materials for this style of printing including rigid and rubber-like material in various 

colors.2, 10  The disadvantages to this type of printing are that the printers are relatively 

expensive, large, and some of the materials are toxic to biological cells.30  This latter 

disadvantage limits the types of materials that can be used in biological applications. 

This style of printing has been used to develop methods to integrate membranes into 

3D printed devices.  These methods are the Print-Pause-Print method and the further 

adapted Z-axis drop method. 

 

 

98 

3.1.3 Print-Pause-Print 

 

The Print-Pause-Print method (Figure 3.2) for integrating membranes into a 3D 

printed device takes advantage of the error in the resolution of the 3D printer, which is 

27 µm for the printer used in the studies described in this dissertation.  Membranes 

around this size, like polycarbonate membranes, can be incorporated into the device 

during the print.  For the best results, the printer should be programmed to not print 

support material as the use of support material requires post-processing, which could 

damage the membranes.  Also, support material in contact with the other materials 

results in that material having a matte finish instead of a glossy finish, which is needed 

to create water-tight joints.  There are three settings on the Stratasys J750 printer that 

need to be set to zero in order to program the printer to not print support material unless 

absolutely necessary, such as voids in the device. These three parameters are the 

Carpet_height, Carpet-protectorZ, and ImproveSupport_thickOfPedestal.11  Once these 

parameters are set to zero, the print can begin.  The printer is allowed to print the part to 

the desired height of the membrane with an opening for the membrane.  The printer is 

then paused, the membrane is placed over the opening, and the printing is resumed to 

finish the print sealing over the membrane. The membrane is usually seated between 

layers of rubber-like material in order for the membrane to stick to the device, otherwise 

the printer may move the membrane during the print.10  One disadvantage to this 

method is that only membranes with thickness similar to the error can be used during 

the print.  A membrane that is too thick will cause the printer heads to collide with the 

part causing damage to both the part and the printer.  Another disadvantage is that the 

lack of support material makes it hard to remove the parts from the printer since the 

 

99 

material is now printed directly onto the stage instead of a thin layer of support material.  

The force needed to remove the parts can cause thin area of the device to distort in 

shape. 

 

 

100 

Figure 3.2: The Print-Pause-Print method for integrating membranes in 3D printed 
objects using a Stratasys J750 printer.  (A) The object with openings for the 
membrane is printed to the desired height of the membrane. (B) The membrane (light 
blue) is then placed onto the 3D printed object and secured above the openings.  (C) 
The printing is continued until the object is complete sealing in the membrane.   

 

 

101 

3.1.4 Z-axis Drop Method 

 

The Z-axis drop method (Figure 3.3) is similar to the Print-Pause-Print method, 

but the differences allow for the integration of membranes over 27 µm.  This method is 

also performed in support free mode for the best results.  The first part of the object with 

an opening for the membrane is printed onto the stage to the desired height of the 

membrane. The final layers of the print before adding the membrane are usually a 

rubber-like material to allow for the membranes to be secured before printing the 

remainder of the part.  Before starting the second part of the print, the stage is lowered, 

also known as dropping the Z-axis of the print, a total distance equal to the height of the 

first part and the membrane.  The second half of the print is started and the printer 

starts the print at the top of membrane sealing it into the part. The disadvantage to this 

method is that the printing of the material directly onto the stage requires a lot of force 

and careful finesse is required to remove the parts with damaging them. 

 

 

102 

Figure 3.3: The Z-axis drop method for integrating membranes into 3D printed 
objects using a Stratasys J750 printer. This is a modified method to the Print-Pause-
Print method for membranes with a thickness greater than 27 µm. The Z-axis is shown 
as a solid red line. (A) The first part is printed with an opening (dark gray).  (B) The 
membrane (light blue) is then placed and secured onto the 3D printed object above the 
opening. (C) The Z-axis or stage is dropped (red arrow) so the top of the membrane is 
now the starting point. (D) The second part is printed sealing in the membrane.   

 

 

103 

 

Here in, both the Print-Pause-Print method and the Z-axis Drop method were 

used to 3D print inserts with two membranes that allow diffusion of a molecule across 

two planes.  Previously, only diffusion across one plane was necessary for the 

applications and experimental procedures in our laboratory. These inserts are used with 

a 3D printed device to create a two-compartment model to mimic the PK of a molecule 

in humans from various types of administration. 

 

 

 

104 

3.2 Materials and Methods 

3.2.1 Design, Printing, and Assembly of 3D Printed Parts 

 

Autodesk Inventor Professional (San Rafael, CA), a CAD software, was used to 

design all devices and inserts. The files were exported as .stl files and printed on a 

Stratasys J750 Polyjet 3D printer (Eden Prairie, MN).  The main device is designed to 

create a PK two-compartment model (Figure 3.4).  Exact dimensions of the main device 

are shown in Figure 3.5. In this device, a channel that is a half-circle on bottom with a 

radius of 1 mm and a 1 x 2 mm rectangle on top runs the length of the device from the 

inlet to the outlet.  The primary compartment, which will represent the plasma 

compartment of the body, sits on top of the channel.  The plasma compartment slightly 

overlaps with the underlying channel.  This overlap is to ensure that there is contact 

between the flowing liquid through the channel and the membrane above the channel. 

The primary compartment is the compartment where the 3D printed membrane insert 

will sit, to create the diffusion barriers between compartments.  The secondary 

compartment, which will represent the interstitial fluid compartment of the body, is 

adjacent to the primary compartment.  

 

 

105 

 

Figure 3.4: The layout of the main device.  A channel runs through the device from 
the inlet to the outlet.  The plasma compartment is located directly above the channel 
(A) and the interstitial compartment (B) is located adjacent to A. The insert is implanted 
into A, which creates diffusion barriers between the channel and A as well as between A 
and B. Concentration gradients allow for diffusion of molecules from the channel into the 
plasma compartment and between compartments.  Rubber-like material (thick black 
border) is printed on the inside of the primary compartment in order to create a liquid-
tight joint between the device and the membrane insert, which is also coated with 
rubber-like material. 

 

106 

 

 

Figure 3.5: The dimension of the main device. The final device is shown in the top 
and center of the figure.  Parts in black are printed in rubber-like material and 
dimensions of those parts are shown on the right of the figure.  The main device is 
printed in Veroclear. 

 

107 

All devices in this chapter were printed in a rigid material (Veroclear or 

Verowhite), with some devices including sections of rubber-like material (Tango or 

Agilus). The materials can be printed with a glossy or matte finish , although glossy is 

always selected whenever possible. Tango and Agilus are useful materials because 

they allow for liquid-tight joints to be created between the devices and other parts, such 

as the membrane inserts.  These rubber-like materials are used on the inside of the 

primary compartment on the device to form a joint with the membrane insert.  The 

rubber-like material is also used on the outside of the outlet in order to create a liquid-

tight compartment with the exhaust.   

The exhaust, as well as the slide shield, were added during testing in order to 

address leaking issues.  The slide shield was specifically used to provide constant 

downward pressure on the membrane insert.  When the slide shield was put into place, 

the thickness of rigid material increases and comes in contact with the top of the 

membrane insert. Without this downward pressure, the pressure of the liquid flowing 

through the channel can push up on the insert leading to breaks in the liquid-tight seal.  

The exhaust was incorporated because, without it, the liquid level slowly drops due to 

the level of the outlet below the insert.  The exhaust addresses this issue by raising the 

outlet to the desired height of the liquid in the insert.  The exact dimension of the slide 

shield and the exhaust can be seen in Figure 3.6.   

 

108 

Figure 3.6: The dimensions of the slide shield and exhaust. The dimensions for the 
slide shield can be seen on the left side of the figure.  The dimensions for the exhaust 
can be seen on the right side of the figure. Parts in black are printed in rubber-like 
material and dimensions are shown on the bottom-right of the figure.   

 

109 

 

3.2.2 Design and Printing of Membrane Inserts 

 

Two types of inserts were printed for use in the two-compartment device.  The 

dimensions for all parts these inserts are provided in Figure 3.7. The first prototype was 

a foldable insert, which could be printed using a slightly modified Print-Pause-Print 

method to incorporate 0.4 µm polycarbonate membranes (< 27 µm, Sterlitech, Kent, 

WA).  The second prototype was a nonfoldable insert, which was printed using a 

modified Z-axis drop method to incorporate 0.45 µm polyethersulfone (PES) 

membranes (> 27 µm, Sterlitech).  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

110 

Figure 3.7: The dimensions of the parts required to create the membrane inserts. 
The foldable insert dimensions are on the left.  The nonfoldable insert parts and their 
dimensions can be found on the right. The central part of the foldable nonfoldable insert 
has a thin layer of rubber-like material on the sides and bottom. 

 

111 

 

The foldable insert was designed so that the membrane separating the channel 

from the primary compartment, and the membrane separating the primary compartment 

from the secondary compartment, could be incorporated into the print at the same time 

and require a single .stl file.  The device has 45º angles on the edges of each foldable 

face to ensure that the faces made a seam in the corners when folded. The printer was 

set to print support-free to avoid membrane damage and post-processing.  The .stl file 

was loaded onto GrabCAD Print (Stratasys) and the part was situated on the stage so 

the 45º angles sloped upward from the stage to ensure no support material was printed.  

The device was printed in a rigid material, but there is a coating option that allows the 

part to be coated in another material.  These devices were printed with a 0.3 mm 

coating of a rubber-like material.  This resulted in parts of the insert less than 0.6 mm to 

be printed in rubber-like material.  These areas on the insert create the hinges required 

to fold the faces of the insert to create a cube.   

Before printing, a thin transparency was attached to the stage of the printer with 

double-sided adhesive tape.  This transparency allows for the removal of the inserts 

from the stage without damage, leaving a smooth glossy finish on the outer side of the 

inserts.  The print is started and after 0.27 mm of the part is printed, the print is paused.  

At this point, only rubber-like material has been printed. The polycarbonate membranes, 

which were hand cut to cover the openings on the insert, are placed over the printed 

opening.  The print is resumed until the insert is finished.  The transparency is then 

removed from the stage.  A razor is used to slightly cut one corner of the insert from the 

transparency, which is then placed in a freezer (-20 ºC) for 5 minutes.  After 5 minutes, 

 

112 

the inserts should be able to be easily removed from the transparency.  If the inserts do 

not come off, the process of placing the transparency in the freezer is repeated.   

The overall assembly procedure of the foldable insert and the device is shown in 

Figure 3.8. The insert is folded into a cube creating two faces with an incorporated 

membrane on the side and bottom. A thin layer of vacuum grease is added to the outer 

faces that do not contain a membrane.  The insert is then placed into the primary 

compartment of the device so that the side membrane is facing the secondary 

compartment, using adhesive (Loctite 495 Super Bonder, Henkel Corp., Rocky Hill, CT) 

in the corners to seal the seams of the insert. The slide shield and exhaust are then 

attached and the device is ready for use.   

 

113 

Figure 3.8: The assembly procedure of the device and foldable inserts.  (A) The 
parts are designed using CAD software.  (B) The inserts are printed using a Stratasys 
J750 3D printer.  (C) The inserts are folded up creating a box with two membrane 
windows. (D) The insert is implanted into primary compartment of the device. (E) The 
slide shield and exhaust are equipped and the device is equipped to a syringe pump.  

 

 

114 

The nonfoldable insert is printed with a more complicated procedure compared to 

the foldable inserts.  The overall procedure is shown in Figure 3.9.  The first step is to 

print the central part of the device, which consists of a rigid box with a 0.5 µm layer of 

rubber-like material on all faces with no openings.  The bottom of the insert has a 0.24 

µm layer of rubber-like material.  The side face has no rubber-like material.  This central 

part is printed with support material to enable the bottom rubber-like material to have a 

glossy finish.  There is no orientation of this part that allows no support material to be 

used and end with a glossy finish on all faces. Only one of the membrane faces can 

have a glossy finish after the first step.  The central part is then cleaned of support 

material by physical and chemical means. The second step is to program the printer to 

print support free and then print a placeholder and a grid.  The purpose of the 

placeholder is to ensure that the coordinates of the printed parts do not change from 

one step to another as the printer will try to minimize the overall print time by moving all 

parts to the top-left of the stage. The grid shows stage printing location and where 

inserts need to be placed.  The grid has two different sized openings.  One opening is 

for the addition of the side membrane while the other opening is for the addition of the 

bottom membrane. The insert is measured with calipers and then placed into the 

opening so the side opening is facing upwards.  The stage is then dropped the height of 

the insert so the print will begin on the top of the insert.  In GrabCAD Print, the side part 

.stl file is uploaded and oriented so the next part is in the opening of the grid.  The grid 

is then deleted, but not the placeholder.  This ensures that the side part will print onto 

the insert in the desired area and orientation. The side part is then printed in rubber-like 

material in order to add a glossy finish to that face, since it could not be done in the 

 

115 

original print.  The PES membrane is then place over the opening and the stage is 

lowered again to take into account the rubber-like material layer and the membrane.  

The side part is printed again in the same orientation sealing in the membrane.  The 

device is then removed from the grid.  The next step is to add the membrane onto the 

bottom face of the insert. Since the central part was able to print with a glossy-finish on 

the bottom, 0.24 µm of rubber-like material was already added in the original print.  The 

PES membrane can be placed onto the bottom face of the insert and measured with 

calipers.  The stage is reconfigured so that it drops to the height of the insert with the 

membrane, which is actually less than the height of the insert for adding the side 

membrane so the stage actually rises, but is still lower than a normal print.  The insert is 

placed in the other size opening on the grid.  In GrabCAD, the grid can be brought back 

by clicking Control + Z. The bottom part .stl file is then oriented to print in the proper 

opening on the grid.  The grid is then deleted and the bottom part is printed onto the 

insert sealing in the bottom membrane.  The insert is now complete. The assembly of 

the nonfoldable insert is similar to that of the foldable insert and is shown in Figure 3.10. 

Adhesive is not required with the nonfoldable insert as there are no seams in the 

corners of the insert. 

 

 

 

 

 

 

 

116 

 

Figure 3.9: The procedure for printing the non-foldable insert. (A) The central part 
is printed out of rigid material (white/gray) with a rubber-like material (black) on the 
sides and bottom. Support material is required and is cleaned off. (B) In support-free 
mode, a placeholder (small red dot) and grid (blue) are printed. (C) The central part is 
placed into the opening so the side opening without any rubber-like material is oriented 
upwards. (D) The stage is dropped the height of the central part and the side part is 
printed in rubber-like material with a glossy finish.  (E) The membrane (yellow) is then 
placed over the opening. (F) The stage is dropped taking into account the additional 
height of the side part and membrane.  The side part is printed again sealing in the side 
membrane. (G) The central device is placed in the other opening on the grid so the 
bottom opening is oriented upwards.  The second membrane is placed over the 
opening. (H) The stage is dropped to the height of the insert and membrane and the 
bottom part is printed sealing in the bottom membrane.   

 

117 

Figure 3.10: The assembly procedure of the device and nonfoldable inserts.  (A) 
First, the parts are designed using CAD software. (B)The parts are printed using a 
Stratasys J750 with membranes being incorporated into the insert using the Z-axis drop 
method. (C) The insert is placed into the primary compartment of the device with the 
help of vacuum grease. (D) The slide shield is moved into place and the exhaust is 
attached. (E) The device is equipped to a syringe pump.  

 

 

118 

3.2.3 Characterization of the Device and Inserts with Fluorescein 

A fluorophore, fluorescein, was used to characterize the device.  Fluorescein was 

used because it has a similar molecular weight and diffusion coefficient in water to the 

antibiotic, levofloxacin.8, 31 All fluorescein solutions were made with DDW and the pH of 

the solution was adjusted to a pH of 7 – 8 using sodium hydroxide (ACS grade, Macron 

Fine Chemicals, Center Valley, PA).  The purpose of raising the pH was to minimize 

adsorption of molecules to the material of the device and to encourage leaching of 

molecules out of the device material.  A stock solution of fluorescein (36 µM) was used 

to test for leakage at the insert-device joint for both inserts and at areas of the foldable 

where unwanted diffusion may occur such as the corner or membrane seams.  

Solutions were delivered from 5 mL gastight syringes (Model 1005 TLL, PTFE Luer 

Lock, Hamilton Company, Reno, NV) to the devices by PEEK tubing (1/16” OD x 0.040” 

ID, IDEX Health & Sciences, Lake Forest, IL) driven by a Fusion 200 syringe pump 

(Chemyx,Inc., Stafford, TX) with the 10 syringe rack equipped.  

The insert-device joint was tested for leakage by placing a 3D printed block into 

the primary compartment.  Devices were stored in DDW (pH = 7-8) prior to use and 

dried for 10 – 15 minutes before running an experiment. The channel of the device was 

primed at 500 µL/min vertically with fluorescein to ensure no air bubbles formed 

underneath the membrane inserts as this would decrease the surface area of the 

membrane in contact with the fluorescein and decrease the diffusion rate. After priming, 

the flow rate was reduced to 5 µL/min and 300 µL of DDW were added into the 

secondary compartment.  The device was incubated at 37 ºC and 350 rpm on a Talboys 

Professional Incubating Orbital Minishaker (Talboys, Thorofare, NJ). Over the course of 

 

119 

6 hours, fluorescence was measured by sampling 5 µL from the secondary 

compartment and diluting into 100 uL of DDW and then exciting 100 µL of the solution 

at 494 nm and measuring the emission of light at 521 nm in a black bottom 96 well plate 

using a Flexstation 3 plate reader (Molecular Devices, Sunnyvale, CA).  The block in the 

primary compartment was a 10 x 10 x 7.9 mm block of rigid material coated with 0.6 mm 

rubber-like material on the bottom and 0.5 mm rubber-like material on all other sides. 

The final dimensions of the block were 11 x 11 x 8.5 mm.  

The design of the foldable insert was evaluated for leakage by printing a foldable 

insert with no openings or membranes.  The insert was folded and placed into the 

primary compartment. The device was primed with fluorescein at 500 µL/min and then 

reduced to 5 µL/min.  DDW (300 µL) was placed in both the primary and secondary 

compartment.  The device was incubated at 37 ºC and shaken at 350 rpm.  Samples of 

5 µL were removed over the course of 5 hours and diluted into 100 µL DDW and 

fluorescence was measured as described above.   

The integration of membranes into the inserts by the print-pause-print method 

was evaluated for leakage by printing the insert and integrating polycarbonate 

membranes with no pores (Sterlitech) into the insert. The insert was folded and placed 

into the primary compartment. Fluorescein was primed and pumped through the 

channel. DDW (300 µL) was placed in both the primary and secondary compartment.  

The device was shaken and incubated at 350 rpm and 37 ºC. Samples (5 µL) were 

removed over the course of 5 hours and diluted into 100 µL DDW followed by 

measurement of the fluorescein emission. To ensure neither membrane was leaking, as 

opposed to just the one above the channel, the same procedure was performed using 

 

120 

the inserts with non-porous membranes.  However, fluorescein was placed into the 

primary compartment and 300 µL DDW was placed in the secondary compartment with 

nothing in the channel. The device was incubated at 37 ºC and shaken at 350 rpm.  

Samples (5 µL) were removed and measured as described above.  All fluorescent 

signals were normalized against the fluorescent signal of the primary compartment at 

the 0 hour time point.  All time points were compared against the time point at 0 hours to 

determine if there was a significant difference (α = 0.05) in fluorescence signifying 

leakage. 

Both types of inserts with porous membranes were used to create oral 

administration PK curves for both the primary and secondary compartment.  The porous 

membrane insert was placed into the primary compartment and the devices were air 

dried for about 10 – 15 minutes.  The devices were primed vertically with fluorescein 

solutions (360 µM for the foldable, polycarbonate insert or 125 µM for the nonfoldable, 

PES insert; pH = 7-8) at 500 µL/min.  After the device was primed the flow rate was 

reduced to 5 µL/min and the primary and secondary compartments filled with 300 µL of 

DDW.  The device was shaken at 350 rpm and incubated at 37 ºC.  After 1 hour, the 

concentration gradient was reversed by priming the channel with DDW (pH = 7 – 8). 

About 500 µL of priming was required to thoroughly clear the channel of fluorescein.   

The flow rate was reduced to 5 µL/min and continued until the end of the experiment. 

Samples (5 µL) were taken from both compartments over the course of 5 – 7 hours and 

diluted into 100 µL DDW, and the fluorescein concentration was determined against 

incubation time to create a PK curve for both compartments.   

 

121 

The nonfoldable, porous membrane insert was used to replicate both intermittent 

and continuous IV administrations.  For the intermittent IV model, DDW was primed 

vertically through the channel and pumped for the entirety of the experiment.  For the 

continuous IV model, fluorescein (14 µM) was primed vertically through the channel and 

pumped for the entirety of the experiment.  In both models, 300 µL of fluorescein (14 

µM) were added to the primary compartment while 300 µL of DDW were added to the 

secondary compartment.  Priming rates, flow rates, sample collection, and evaluation 

were the same as the oral model.   

 

 

 

 

122 

3.3 Results 

3.3.1 Fluorescein Leakage Testing of Both Inserts 

 

The outer faces of the inserts and the inner faces of the primary compartment 

were printed with a rubber-like material in a glossy finish in order to create a liquid-tight 

joint with the device.  The rubber-like lining of the primary compartment in the device 

was 25 µm smaller in length and width than the insert, causing the rubber-like material 

to compress when the inset was placed creating the liquid-tight joint.  This joint was 

evaluated for leakage via the usage of a glossy finished, rubber-like material covered 

block with the same parameters as the finished foldable inserts. The block was placed 

in the primary compartment and the fluorescence of DDW in the secondary 

compartment was monitored over the course of 6 hours to determine if fluorescein was 

moving through the joint.  The results of the experiment show that at no point during the 

6 hour experiment was there a significant difference in the fluorescent signal of the 

DDW in the secondary compartment (Figure 3.11).  This shows that the design of the 

device and inserts are capable of creating a liquid-tight joint.  

 

 

 

 

 

123 

Figure 3.11: The normalized fluorescent signal in the secondary compartment of a 
device with a block in the primary compartment to test for leakage of the device-
insert joint.  The design of the joint between the insert and the device does not allow 
the passage of fluorescein molecules from the channel into the secondary compartment 
when a block with the same parameters as both inserts is placed in the primary 
compartment.  At no point during a 6 hour experiment, where fluorescein was flowing 
underneath a block, was there a significant difference in the fluorescent signal of DDW 
in the secondary compartment. (n =  3; error =standard deviation; α = 0.05) 

 

 

124 

3.3.2 Fluorescein Leakage Testing of the Foldable Insert 

Once the dimensions of the foldable insert seemed adequate to create a liquid-

tight joint between the compartments of the device, the design of the foldable inserts 

was then evaluated to determine if fluorescein could pass through the adhesive bonded 

seams when folded up inserts were in the primary compartment.  Without adhesive, the 

liquid level in the primary compartment decreased to nothing in an hour or two.  The use 

of the adhesive allowed for the seams to become impervious to liquid movement.  This 

was tested by monitoring the fluorescence of DDW over the course of 5 hours while 

flowing fluorescein through the channel.  At no point during the 5 hour experiment was 

there a statistical difference in the fluorescence of the DDW in either compartment 

(Figure 3.12).  This indicates that there was no leakage from the channel into either 

compartment in that time frame.   

 

 

 

 

 

 

125 

Figure 3.12: The normalized fluorescent signal in both compartments of a device 
with a foldable insert with no openings or membranes in the primary 
compartment to test the design of the foldable inserts for leakage at the seams.  
The design of the foldable insert does not allow passage of fluorescein into or around 
the insert with the help of adhesive reinforcement of the seams. There was no 
significant difference in the fluorescent signal of DDW in either the primary (black bars) 
or the secondary compartment (gray bars) over the course of 5 hours when fluorescein 
was flowing through the channel underneath the insert.  (n = 9; 3 devices, error = SEM, 
α = 0.05) 

 

126 

 

Both the design of the device-insert joint and the foldable insert have been 

shown to provide a liquid-tight barrier that does not allow movement of fluorescein into 

the primary or secondary compartment without the help of membranes.  To ensure that 

the integration of membranes into the foldable insert via the print-pause-print method 

does not result in undesired movement of molecules into the other compartments, 

nonporous, polycarbonate membranes were printed into the foldable insert and the 

insert was tested for leakage by folding the insert and placing it into the primary 

compartment.  Adhesive was used to seal the corner seams. The fluorescence of DDW 

was monitored over the course of 5 hours in both compartments while fluorescein was 

flowed through the channel beneath the insert.  There was no significant difference in 

the normalized florescent signal in either compartment over the 5 hours that the 

experiment was performed (Figure 3.13). This signifies that the integration of 

membranes via the print-pause-print method does not lead to undesired leakage of 

molecules around the diffusion barriers. 

 

 

 

 

127 

 

Figure 3.13: The normalized fluorescent signal in both compartments of a device 
with a foldable insert with nonporous membranes in the primary compartment to 
test the integration of the membranes into the insert for undesired leakage.  The 
integration of nonporous membranes into the insert via the print-pause-print method 
does not allow passage of fluorescein around the membrane above the channel. There 
was no significant difference in the fluorescent signal of DDW in either the primary 
(black bars) or the secondary compartment (gray bars) over the course of 5 hours when 
fluorescein was flowing through the channel underneath the insert.  (n = 9; 3 devices, 
error = SEM, α = 0.05) 

 

128 

 

The data from the previous figure shows that there was no change in the 

fluorescent signal of the DDW in either compartment.  However, if the integration of the 

bottom membrane was successful but not the side membrane, the outcome would look 

the same.  Fluorescein was placed into the primary compartment, which contained the 

foldable insert with nonporous membranes, and DDW was placed into the secondary 

compartment with nothing flowing through the channel.  The fluorescence of the DDW in 

the secondary compartment was monitored over the course of 5 hours to ensure that 

the integration of the membrane did not provide undesired movement of molecules.  At 

no point during the experiment was there a significant difference in the fluorescent 

signal of the DDW in the secondary compartment (Figure 3.14).  This shows that the 

integration of the side membrane also does not allow undesired movement of 

molecules.   

 

 

 

 

129 

 

Figure 3.14: The normalized fluorescent signal in the secondary compartment of a 
device with a foldable insert with nonporous membranes containing fluorescein 
in the primary compartment to test the integration of the side membrane into the 
insert for undesired leakage.  The integration of nonporous membranes into side of 
the foldable insert via the print-pause-print method did not allow passage of fluorescein 
around the membrane between the primary and secondary compartment. There was no 
significant difference in the fluorescent signal of DDW in the secondary compartment 
(gray bars) over the course of 5 hours when fluorescein was in the primary 
compartment.  (n ≤ 9; 3 devices, error = SEM, α = 0.05) 

 

130 

3.3.3 Oral PK Model using Fluorescein with the Foldable Inserts 
 
 

Overall, the above experiments show that the design of the device-insert joint, 

the design of the foldable insert with the help of adhesive, and the integration of the 

nonporous membranes all provide a liquid-tight seal that does not allow undesired 

movement of molecules across the barriers. The integration of porous membranes 

should allow diffusion of molecules that are only moving through the pores of the 

membrane.  The goal of this experiment was to use fluorescein to replicate oral 

administration of levofloxacin.  Levofloxacin usually reaches a maximum plasma 

concentration (Cmax) of 7.8 -14.4 µM in 1 – 2 hours and has a half-life of 6 – 8 hours in 

healthy patients.32  Polycarbonate membranes with 0.4 µm pores were integrated into 

the foldable insert.  Fluorescein (360 µM) was primed and flowed through the channel 

with DDW in both compartment. After one hour, the fluorescein in the channel was 

replaced with flowing DDW for the remainder of the experiment.  The concentration of 

fluorescein was quantified throughout the experiment and is shown in Figure 3.15. The 

concentration of fluorescein reached a Cmax of 9.1 ± 1.5 µM in 2 hours; however, the 

shape of the curve indicates that the concentration may have been higher prior to the 2 

hour time point. The half-life of fluorescein was 3.7 ± 0.3 hours.  The novel information 

collected from this device is the secondary compartment concentrations, which shows 

that the fluorescein concentration in the secondary compartment was reaching a steady 

state concentration at about 3.0 – 3.5 µM after 3 - 4 hours. 

 

131 

Figure 3.15: The PK curves for both the primary and secondary compartment for 
an oral model of levofloxacin using fluorescein with the foldable inserts.  In the 
primary compartment (top plot), the fluorescein concentration reached a Cmax of 9.1 ± 
1.5 µM in 2 hours with a half-life of 3.7 ± 0.3 hours.  In the secondary compartment 
(bottom plot), the fluorescein reached a steady state concentration of 3.0 – 3.5 µM after 
3 - 4 hours. (n = 3, error = standard deviation) 

 

 

132 

Time (hours)0123456Fluorescein Concentration (µM)024681012Time (hours)0123456Fluorescein Concentration (µM)0123453.3.4 Various Administration PK Models using Fluorescein with the Nonfoldable 

Inserts 

 

The nonfoldable insert has the same dimensions as the foldable insert and 

utilizes the same printer and material to integrate the membranes.  The nonfoldable 

insert does not have any seams like the foldable insert, so there was no need for 

additional leakage testing.  The nonfoldable inserts contained 0.45 µm PES 

membranes.  The nonfoldable inserts were used in the device to mimic PK profiles of 

molecules administered orally, by intermittent IV, or by continuous IV.  The oral 

administration was simulated to mimic an oral PK profile of levofloxacin like with the 

foldable insert.   

 

The oral administration of levofloxacin model was performed the same way as for 

the foldable insert.  The only difference is the concentration of fluorescein used was 

only 125 µM possibly due to difference in the percent porosity of the membranes. The 

results of this experiment are shown in Figure 3.16. In the primary compartment, the 

Cmax of fluorescein occurred at 14.4 ± 1.1 µM in 1 hour.  The half-life of fluorescein in 

the primary compartment was 5.0 ± 0.4 hours.  In the secondary compartment, the 

concentration of fluorescein approached a steady state concentration of about 8.0 – 8.5 

µM at 6 – 7 hours.   

 

133 

Figure 3.16: The PK curves for both the primary and secondary compartment for 
an oral model of levofloxacin using fluorescein with the nonfoldable inserts.  In 
the primary compartment (top plot), the fluorescein concentration reached a Cmax of 
14.4 ± 1.1 µM in 1 hour with a half-life of 5.0 ± 0.4 hours.  In the secondary 
compartment (bottom plot), the fluorescein reached a steady state concentration of 8.0 
– 8.5 µM after 6 – 7 hours. (n = 3, error = standard deviation) 

 

 

134 

Time (hours)02468Fluorescein Concentration (024681012141618Time (hours)02468Fluorescein Concentration (0246810 

For an intermittent or continuous IV administration, a loading dose of the drug is 

administered directly into the bloodstream.  In the device, this was simulated by adding 

the desired concentration of fluorescein (14 µM) into the primary compartment.  The 

secondary compartment was filled with DDW.  The channel was primed and flowed with 

DDW for the intermittent IV model or fluorescein for the continuous IV model.  These 

models are not related to any particular drug.  The purpose was to show the ability of 

the device to replicate other types of drug administration in the primary compartment 

and to determine how these different administrations compare in the secondary 

compartment concentrations. The results of the intermittent IV can be seen in Figure 

3.17. In the primary compartment, the half-life was calculated at 2.3 ± 0.1 hours. Just for 

reference, the antibiotic, gentamicin, which is administered intravenously has a half-life 

of 1-3 hours.33 Also, both the distribution (0-3 hours) and elimination phase (3+ hours) 

can be seen in the primary compartment PK curve.  The visualization of both phases is 

indicative of a two-compartment model. In the secondary compartment, the fluorescein 

concentration reached a steady state concentration of 5.5 µM in 4 -5 hours. 

 

135 

Figure 3.17: The PK curves for both the primary and secondary compartment for 
an intermittent IV model using fluorescein with the nonfoldable inserts.  In the 
primary compartment (top plot), the half-life of fluorescein was 2.3 ± 0.1 hours.  Both the 
distribution and elimination phase are visible indicative of a two-compartment model.  In 
the secondary compartment (bottom plot), the fluorescein reached a steady state 
concentration of 5.5 µM after 4 – 5 hours. (n = 3, error = standard deviation) 

 

 

136 

Time (hours)02468Fluorescein Concentration (0246810121416Time (hours)02468Fluorescein Concentration (01234567 

The results of the continuous IV can be seen in Figure 3.18.  The purpose of a 

continuous IV is to keep the concentration of the plasma at a steady state 

concentration. In the primary compartment, the initial dose of 14 µM decreased to 10.5 

µM over the course of 2 hours before entering a steady state.  In the secondary 

compartment, the fluorescein concentration increases faster than the intermittent IV 

model and was still continuing to rise at the 7 hour mark when the experiment ended. 

The decrease in concentration of the primary compartment while the channel had 

fluorescein of the same concentration indicates that the diffusion from the primary 

compartment into the secondary compartment is greater than the diffusion from the 

channel into the primary compartment. 

 

 

137 

Figure 3.18: The PK curves for both the primary and secondary compartments for 
a continuous IV model using fluorescein with the nonfoldable inserts.  In the 
primary compartment (top plot), the fluorescein concentration decreased to a steady 
state of 10.5 µM after 2 hours.  In the secondary compartment (bottom plot), the 
fluorescein increased faster than the intermittent IV and continued to increase until the 
end of the experiment (n = 3, error = standard deviation) 

 

 

138 

Time (hours)02468Fluorescein Concentration (024681012141618Time (hours)02468Fluorescein Concentration (0246810123.4 Discussion 

 

The Statasys J750, a polyjet style 3D printer, was a useful tool in the fabrication 

of a 3D printed two-compartment PK model.  The multiple materials that the Stratsys 

J750 can print allows for multiple functions.  The rigid material allows for a device that 

can be reused without degradation. The rubber-like material allows for flexible hinges 

and liquid-tight joints to be created due to its compressibility.  The liquid-tight joints allow 

for multiple parts to be assembled creating the device described above.  The ability to 

print support free and change the stage height has allowed for membranes to be 

integrated into the print job in one or multiple planes.  The ability to integrate 

membranes into multiple planes was the innovative feature that allowed for a two-

compartment model to be created with an exhaust to maintain proper liquid levels in the 

device and a slide shield to keep constant downward pressure on the insert. 

 

The main device was originally designed to be printed on a polyjet style 3D 

printer.  However, the addition of a rubber-like material into the primary compartment 

and on the exhaust to create a compressible joint that can create a liquid tight joint with 

another part made with a rubber-like material lining was not part of the original design.  

The key to creating a liquid-tight joint is to design the rubber-like material linings to 

overlap by at least 25 µm so the linings have to compress each other creating the joint.  

Printing the rubber-like material in a glossy finish when possible gives the added benefit 

of creating a smoother part, as well as stickier joint when in contact with another glossy 

finish rubber-like part.  Both the inner lining of the primary compartment and the outer 

lining of the insert are printed in a glossy finish, rubber-like material in order to create 

the liquid-tight joint. This joint did not allow the movement of fluorescein into the other 

 

139 

compartment when a glossy finish, rubber-like material lined block with the same 

dimensions as the membrane inserts was placed in the primary compartment.   

Before the discovery of the print-pause-print method and the ability to print 

support-free, the inserts were printed on an FDM style printer and the membranes were 

added later.  The benefit of this printer is that support material was not needed and the 

ABS material that the inserts were composed of became semisolid with the use of 

acetone, which allowed membrane to be attached to the device.  The insert was dipped 

in acetone and then pressed against the membranes and the inserts were allowed to 

dry.  However, the resolution of the FDM resulted in parts that did not have a smooth 

finish and could not be made liquid-tight.  Also, the use of acetone to attach the 

membranes resulted in parts having various final dimensions.  In the end, FDM was not 

used for the above reasons and it was more convenient to print both the device and 

inserts on the same type of printer instead of using two different printers located in two 

different buildings.  

 

As seen above, the print-pause-print and the further modified Z-axis drop method 

with use of the Stratasys J750 allowed for the creation of inserts with porous 

membranes in two planes allowing for the two-compartment model to be successful.  

The use of the Stratasys J750 was also beneficial because it allowed for the inserts to 

have an outer layer of rubber-like material with a glossy finish, which is required for the 

liquid-tight joint with the device. The sealing of membranes between layers of the 

rubber-like material created a seam that was impervious to the movement of molecules 

except through the pores of the membranes.  This was shown by integrating non-porous 

membranes into the inserts and measuring fluorescein movement in the various 

 

140 

compartments, which resulted in no detectable fluorescence in any compartments.  

Both the foldable insert containing porous polycarbonated membranes and the 

nonfoldable insert containing porous PES membranes were successfully capable of 

creating a PK curve for an oral model of levofloxacin by using fluorescein, which has a 

similar molecular weight and diffusion coefficient in water to levofloxacin.31  However, 

the PES membrane required a lower concentration of fluorescein to be flowed through 

the channel to create a similar curve and resulted in a longer half-life that was more 

human-like.  The difference in the curves may be due to the higher percent porosity, or 

greater number of pores, of PES membranes than polycarbonate membranes due to 

how the membranes are made. The higher percent porosity would explain the greater 

diffusion across the membranes.  The concentration of fluorescein in the secondary 

compartment equilibrates with the primary compartment, which is equilibrating with the 

channel that contains DDW with no fluorescein.  When the concentration of fluorescein 

in the secondary compartment is higher, the longer the half-life is due to the equilibrium 

amongst all compartments.  The device with the PES, nonfoldable insert was also 

capable of recreating a PK curve for an intermittent and continuous IV administration.  

The additional benefits of the nonfoldable, PES insert include that it does not require 

adhesive to seal the insert in the primary compartment and the nonmoving design of the 

insert allows the device to be removable and reusable.  The foldable, polycarbonate 

membrane requires adhesive to seal the corner seams when the insert is folded and the 

pure rubber-like hinges tend to tear when the insert is removed making the insert mostly 

one use.  The benefit of this two-compartment model is that even though it was created 

for antibiotic susceptibility determination, the two-compartment model can be used for 

 

141 

numerous other applications.  For example, toxicity or metabolism can be tested by 

placing liver cells in the secondary compartment.  Another benefit is that this device 

allows for active measurements to be performed for both compartments, which 

previously has not been achievable for 3D printed devices. 

 

One limitation of this device is that the current style of printing only allows for 

membranes to be printed with flat faces such as cubes.  Also the current design of the 

device has a cubic secondary compartment.  Square compartments are not the best 

configuration for mixing.  If the inserts and device had cylindrical compartments, the 

mixing would be better, due to centrifugal mixing, and the diffusion patterns could differ.  

Another limitation is that the half-life of fluorescein was shorter than the half-life of 

levofloxacin in humans.  This may be able to be addressed by adding a tertiary 

compartment, increasing the number of molecules in the secondary compartment by 

increasing the surface area of the membrane between the primary and secondary 

compartment, or decreasing the excretion of fluorescein from the primary compartment 

by slowing the transition from fluorescein to DDW in the channel.  All of these would 

help increase the half-life.  Another limitation is that during the continuous IV model, the 

concentration in the primary compartment decreased from the loading dose due to 

greater diffusion into the secondary compartment than diffusion into the primary 

compartment from the channel.  This may be addressed by increasing the surface area 

between the channel and membrane on the underside of the primary compartment.  

The final limitation being discussed is that this device provides information about the 

concentration of a free drug in the secondary or interstitial fluid compartment.  This type 

of data is not well known about humans as current PK/PD evaluation of drugs only 

 

142 

involve measuring plasma concentrations and determining clinical outcomes. However, 

all drugs do not act the same in the human body due to differences in PK that affect 

distribution and excretion such as plasma protein binding.34 To confirm the data 

obtained from the secondary compartment, other methods like in silico methods would 

be required. 

 

Overall, a 3D printed device has been created using a Stratasys J750 polyjet 

printer that allows for integration of membranes in two or more planes.  These inserts 

have allowed for diffusion of a molecule to occur into two different compartments 

creating a true two-compartment PK/PD model.  This model allows for constant 

monitoring of the concentration of the molecule of interest in both the primary and 

secondary compartment that represent the plasma and interstitial fluid compartments of 

the body, respectively.  Previous 3D printed, PK/PD models did not include a secondary 

compartment leading to a shorter half-life of the drug of interest and fewer 

administration scenarios that can be simulated. This model has been successful in 

simulating PK profiles of different drug administrations including oral, intermittent IV, 

and continuous IV. The benefit of the model being designed for 3D printing is that 

modification can be made depending on the personalized need of the scientist and 

these changes can be implemented rapidly as the device only takes a few hours to 

print.   

 

 

 
 
 
 

 

 

143 

REFERENCES

 

144 

 

REFERENCES 

Gross, B.; Lockwood, S. Y.; Spence, D. M., Recent Advances in Analytical 

Baden, T.; Chagas, A. M.; Gage, G. J.; Marzullo, T. C.; Prieto-Godino, L. L.; 

Lupton, D., Fabricated data bodies: Reflections on 3D printed digital body objects 

Au, A. K.; Huynh, W.; Horowitz, L. F.; Folch, A., 3D printed Microfluidics. 

Gross, B. C.; Erkal, J. L.; Lockwood, S. Y.; Chen, C.; Spence, D. M., Evaluation 

Ho, C. M.; Ng, S. H.; Li, K. H.; Yoon, Y. J., 3D printed microfluidics for biological 

1. 
of 3D Printing and Its Potential Impact on Biotechnology and the Chemical Sciences. 
Analytical Chemistry 2014, 86 (7), 3240-3253. 
 
2. 
Chemistry by 3D Printing. Anal Chem 2017, 89 (1), 57-70. 
 
3. 
applications. Lab Chip 2015, 15 (18), 3627-37. 
 
4. 
Euler, T., Open Labware: 3-D printing your own lab equipment. PLoS biology 2015, 13 
(3), e1002086. 
 
5. 
in medical and health domains. Social Theory & Health 2015, 13 (2), 99-115. 
 
6. 
Angewandte Chemie (International ed. in English) 2016, 55 (12), 3862-81. 
 
7. 
Yazdi, A. A.; Popma, A.; Wong, W.; Nguyen, T.; Pan, Y.; Xu, J., 3D printing: an 
emerging tool for novel microfluidics and lab-on-a-chip applications. Microfluidics and 
Nanofluidics 2016, 20 (3), 50. 
 
8. 
Lockwood, S. Y.; Meisel, J. E.; Monsma, F. J.; Spence, D. M., A Diffusion-Based 
and Dynamic 3D printed Device That Enables Parallel in Vitro Pharmacokinetic Profiling 
of Molecules. Analytical Chemistry 2016, 88 (3), 1864-1870. 
 
9. 
Drug penetration and metabolism in 3D cell cultures treated in a 3D printed fluidic 
device: assessment of irinotecan via MALDI imaging mass spectrometry. Proteomics 
2016, 16 (11-12), 1814-21. 
 
10.  Pinger, C. W.; Heller, A. A.; Spence, D. M., A Printed Equilibrium Dialysis Device 
with Integrated Membranes for Improved Binding Affinity Measurements. Analytical 
Chemistry 2017, 89 (14), 7302-7306. 
 
11.  Castiaux, A. D.; Pinger, C. W.; Spence, D. M., Ultrafiltration binding analyses of 
glycated albumin with a 3D printed syringe attachment. Analytical and bioanalytical 
chemistry 2018, 410 (29), 7565-7573. 
 

LaBonia, G. J.; Lockwood, S. Y.; Heller, A. A.; Spence, D. M.; Hummon, A. B., 

 

145 

Ziemian, C. W.; Crawn, P. M., Computer aided decision support for fused 

Zhong, W.; Li, F.; Zhang, Z.; Song, L.; Li, Z., Short fiber reinforced composites for 

Zein, I.; Hutmacher, D. W.; Tan, K. C.; Teoh, S. H., Fused deposition modeling of 

12.  Waldbaur, A.; Rapp, H.; Länge, K.; Rapp, B., Let there be chip - Towards rapid 
prototyping of microfluidic devices: One-step manufacturing processes. 2011; Vol. 3, p 
2681-2716. 
 
13.  Pham, D.; Gault, R., A Comparison of Rapid Prototyping Technologies. 1998; 
Vol. 38, p 1257-1287. 
 
14. 
deposition modeling. Rapid Prototyping Journal 2001, 7 (3), 138-147. 
 
15.  Klein, J.; Stern, M.; Franchin, G.; Kayser, M.; Inamura, C.; Dave, S.; Weaver, J. 
C.; Houk, P.; Colombo, P.; Yang, M.; Oxman, N., Additive Manufacturing of Optically 
Transparent Glass. 3D Printing and Additive Manufacturing 2015, 2 (3), 92-105. 
 
16.  Bhattacharjee, N.; Urrios, A.; Kang, S.; Folch, A., The upcoming 3D printing 
revolution in microfluidics. Lab on a chip 2016, 16 (10), 1720-1742. 
 
17. 
novel scaffold architectures for tissue engineering applications. Biomaterials 2002, 23 
(4), 1169-1185. 
 
18. 
fused deposition modeling. Materials Science and Engineering: A 2001, 301 (2), 125-
130. 
 
19.  Ambrosi, A.; Pumera, M., 3D printing technologies for electrochemical 
applications. 2016; Vol. 45. 
 
20.  Agarwala, M. K.; Jamalabad, V. R.; Langrana, N. A.; Safari, A.; Whalen, P. J.; 
Danforth, S. C., Structural quality of parts processed by fused deposition. Rapid 
Prototyping Journal 1996, 2 (4), 4-19. 
 
21.  Bandyopadhyay, A.; Panda, R. K.; Janas, V. F.; Agarwala, M. K.; Danforth, S. C.; 
Safari, A., Processing of Piezocomposites by Fused Deposition Technique. Journal of 
the American Ceramic Society 1997, 80 (6), 1366-1372. 
 
22.  Wu, G.; A. Langrana, N.; Sadanji, R.; Danforth, S., Solid freeform fabrication of 
metal components using fused deposition of metals. Materials & Design 2002, 23 (1), 
97-105. 
 
23. 
Jafari, M. A.; Han, W.; Mohammadi, F.; Safari, A.; Danforth, S. C.; Langrana, N., 
A novel system for fused deposition of advanced multiple ceramics. Rapid Prototyping 
Journal 2000, 6 (3), 161-175. 
 
24.  Wong, J. Y.; Pfahnl, A. C., 3D printing of surgical instruments for long-duration 
space missions. Aviation, space, and environmental medicine 2014, 85 (7), 758-63. 

 

146 

25.  McCullough, E. J.; Yadavalli, V. K., Surface modification of fused deposition 
modeling ABS to enable rapid prototyping of biomedical microdevices. Journal of 
Materials Processing Technology 2013, 213 (6), 947-954. 
 
26.  Gelber, M. K.; Bhargava, R., Monolithic multilayer microfluidics via sacrificial 
molding of 3D printed isomalt. Lab Chip 2015, 15 (7), 1736-41. 
 
27.  Kang, H.-W.; Lee, S. J.; Ko, I. K.; Kengla, C.; Yoo, J. J.; Atala, A., A 3D 
bioprinting system to produce human-scale tissue constructs with structural integrity. 
Nature Biotechnology 2016, 34, 312. 
 
28.  Miller, J. S.; Stevens, K. R.; Yang, M. T.; Baker, B. M.; Nguyen, D.-H. T.; Cohen, 
D. M.; Toro, E.; Chen, A. A.; Galie, P. A.; Yu, X.; Chaturvedi, R.; Bhatia, S. N.; Chen, C. 
S., Rapid casting of patterned vascular networks for perfusable engineered three-
dimensional tissues. Nature Materials 2012, 11, 768. 
 
29.  Macdonald, N. P.; Cabot, J. M.; Smejkal, P.; Guijt, R. M.; Paull, B.; Breadmore, 
M. C., Comparing Microfluidic Performance of Three-Dimensional (3D) Printing 
Platforms. Analytical Chemistry 2017, 89 (7), 3858-3866. 
 
30.  Walsh, M. E.; Ostrinskaya, A.; Sorensen, M. T.; Kong, D. S.; Carr, P. A., 3D-
Printable Materials for Microbial Liquid Culture. 3D Printing and Additive Manufacturing 
2016, 3 (2), 113-118. 
 
31.  Vrany, J. D.; Stewart, P. S.; Suci, P. A., Comparison of recalcitrance to 
ciprofloxacin and levofloxacin exhibited by Pseudomonas aeruginosa bofilms displaying 
rapid-transport characteristics. Antimicrobial agents and chemotherapy 1997, 41 (6), 
1352-1358. 
32. 
Pharmacokinetics 1997, 32 (2), 101-119. 
 
33.  Smith, A. L.; Smith, D. H.; Siber, G. R.; Paisley, J. W.; Echeverria, P., 
Pharmacokinetics of Gentamicin in Children and Adults. The Journal of Infectious 
Diseases 1975, 132 (6), 637-651. 
 
34. 
Clinics of North America 2003, 17 (3), 599-613. 

Fish, D. N.; Chow, A. T., The Clinical Pharmacokinetics of Levofloxacin. Clinical 

Liu, P.; Derendorf, H., Antimicrobial tissue concentrations. Infectious Disease 

 

 

 
 

 

147 

CHAPTER 4 

 

ANTIBIOTIC SUSCEPTIBILITY TESTING OF BACTERIA VIA A DYNAMIC 

ADMINISTRATION OF ANTIBIOTICS USING A 3D PRINTED TWO COMPARTMENT 

PK/PD MODEL 

 

4.1 Introduction 

4.1.1 Current State of Antibiotic Resistance, Development, and Susceptibility 

Testing 

 

As previously discussed in Chapters 1 and 2, the ability of bacteria to develop 

antibiotic resistance by random mutagenesis or horizontal gene transfer while under the 

selective pressure of antibiotics has led to the emergence of bacterial strains that are 

resistant to multiple currently available antibiotic.1-2 New antibiotics are required to treat 

some of these bacterial strains, as current antibiotics are ineffective.  However, the 

current investments required to develop a drug is in excess of $2.5 billion and 10 

years.3 Along with this major investment, the low success rate of drugs in clinical trials 

(10.4%), the rapid evolution of bacteria to develop resistance to antibiotics, and the 

unwillingness of the public to pay the high price to produce these novel antibiotics has 

created a problematic situation where pharmaceutical companies are not pursuing the 

development of new antibiotics because of fears of not making a profit.4-10 The 

increased incidence of antibiotic resistance and the decrease in antibiotic development 

led to the National Action plan, which discusses the need for the development of rapid 

diagnostic tools and the development of new antibiotics and therapeutics.11-13 Current 

 

148 

diagnostic tools, which are used in parts of the early developmental process of 

antibiotics, primarily consist of susceptibility tests. 

 

The most common susceptibility tests include the broth microdilution method, the 

disk diffusion method, and the Etest.14 The procedure for these susceptibility tests 

involve challenging a bacterial sample with an antibiotic to determine if the antibiotic is 

effective in inhibiting the growth of the bacteria or causing cell death.15 These tests are 

also useful in determining the minimum concentration that causes these outcomes, 

which is valuable pharmacodynamic (PD) information.16 These susceptibility tests come 

with many disadvantages that have been previously discussed; however, the major 

limitation is that the bacteria is challenged against a static concentration of antibiotic, as 

opposed to the dynamic concentration profiles that occur in vivo.17 

 

4.1.2 Overview of Research from Previous Chapters 

 

In order to address the issues discussed in the previous section, the overall 

viewpoint has been to decrease the time required to perform a susceptibility test and to 

develop a dynamic susceptibility test where the concentration of antibiotic follows a 

human pharmacokinetic (PK) curve.  These areas of focus were chosen because in 

vitro tests are usually the earliest, cheapest, and easiest tests to perform.  In vitro 

experiments are part of preclinical trials, which cost about $1.1 billion compared to the 

$1.4 billion of clinical trials.3  The FDA regulates what is required in clinical trials. 

However, determining that a potential drug will not function effectively in humans earlier 

in the drug development process has the potential to save significant amounts of money 

as the investment increases the further the potential drug progresses in development.18-

 

149 

19  Only 64.5% of potential drugs that enter clinical trials pass Phase I where safety is 

assessed.6  The high failure rate may be due to disharmony between data collected 

from animals and humans due to difference in the PK.20-22  By creating more complex in 

vitro tests that replicate human PK, the resulting PD data may be more clinically 

relevant and result in less of a reliance on the more expensive forms of preclinical tests, 

such as animal testing. 

 

In Chapter 2, an antibiotic susceptibility assay was created in order to determine 

the efficacy of an antibiotic within one hour of challenging the bacteria with an antibiotic. 

This was achieved by monitoring the ATP/OD600 of living, growing bacteria, which 

during growth increases until early logarithmic growth and then decreases over the 

logarithmic growth phase.  After the peak ATP/OD600, if an effective antibiotic were 

added, a statistical difference in the ATP/OD600 could be measured compared to a 

healthy growing bacteria within one hour of adding the antibiotic because dead or dying 

bacteria do not decrease the ATP/OD600, they actually increase the ratio due to lysing 

of the bacterial cells.  This procedure was performed with a Gram-negative and Gram-

positive strain of bacteria as well as multiple classes of antibiotics.17   

One problem that did occur with these experiments was that when using a 

bacteriostatic antibiotic, the ATP/OD600 increased.  This was not expected because a 

bacteriostatic antibiotic is supposed to inhibit cellular growth and replication, not cause 

lysis.  However, for these experiments, the static concentration used was the maximum 

plasma concentration of the antibiotic given to humans not the minimum inhibitory 

concentration.  Since bacterial infections usually occur in the interstitial compartment 

and not the plasma compartment, the bacteria was subjected to a higher concentration 

 

150 

of the bacteriostatic antibiotic, which can cause bactericidal results.23 This outcome 

provided the idea to create a more complex assay that can provide PD data such as the 

minimum effective concentration while undergoing a human PK curve of that antibiotic.  

This required the development of a 3D printed, two-compartment model. 

 

In Chapter 3, the 3D printed, two-compartment model was successfully created 

using a Stratasys J750 Polyjet style 3D printer.  Using this printer, an insert containing 2 

porous membranes, one on the bottom and one on the side, was created allowing for a 

two-compartment model to be created.  This model can mimic the absorption, 

distribution, and excretion profiles of a drug molecules similar to those observed in 

humans.  The model of an oral administration of fluorescein showed similar PK to the 

antibiotic levofloxacin and is further investigated herein.  This chapter will combine 

techniques from the previous two chapters to challenge bacteria to a human-like PK 

profile of the antibiotic levofloxacin administered orally and monitor the effect of the 

antibiotic on the bacteria in the secondary compartment by monitoring the ATP/OD600. 

 

 

 

 

151 

4.2 Materials and Methods 

4.2.1 Oral Pharmacokinetic Profile of Levofloxacin Using Fluorescein 

The procedure for the oral PK profile of levofloxacin using fluorescein is similar to 

the procedure in Chapter 3.  Fluorescein was used in place of levofloxacin because it 

has a similar molecular weight and diffusion coefficient in water.24 Fluorescein is also 

cheaper than levofloxacin and quantifiable on a plate reader via fluorescence 

spectroscopy.  The oral PK profile was performed using the device and nonfoldable 

membrane insert with polyethersulfone membranes as described in Chapter 3.  Prior to 

use, the devices were soaked in distilled and deionized water (DDW) with a pH = 7 – 8. 

Devices were air dried 10 – 15 minutes before use.  A stock solution of 125 µM 

fluorescein was made and delivered by 5 mL gastight syringes (Model 1005 TLL, PTFE 

Luer Lock, Hamilton Company, Reno, NV) to the devices by PEEK tubing (1/16” OD x 

0.040” ID, IDEX Health & Sciences, Lake Forest, IL) driven by a Fusion 200 syringe 

pump (Chemyx,Inc., Stafford, TX) with the 10 syringe rack equipped. The channel in the 

devices were primed vertically with the fluorescein solution at 500 µL/min.  After the 

device was primed, the flow rate was reduced to 5 µL/min and the devices were placed 

onto a modified Talboys Professional Incubating Orbital Minishaker (Talboys, Thorofare, 

NJ). The orbital shaker is modified using 3D printed parts to allow the peek tubing to 

enter the covering of the shaker incubator, as shown in Figure 4.1.  Also, a custom 

holder and cap was 3D printed in order to secure the devices to the orbital shaker.  The 

dimensions of these custom parts are shown in Figure 4.2.  The primary and secondary 

compartments were filled with 300 µL of DDW.  The device was shaken at 350 rpm and 

incubated at 37 ºC.  After 1 hour, the concentration gradient was reversed by priming 

 

152 

the channel with DDW (pH = 7 – 8). About 500 µL of priming is required to thoroughly 

clear the channel of fluorescein.   The flow rate was reduced to 5 µL/min and continued 

until the end of the experiment. Samples (5 µL) were taken from both compartments 

over the course of 7 hours and diluted into 100 µL DDW and the fluorescein 

concentration was quantified and graphed against the incubation time to create a PK 

curve for both compartments.   

 

 

153 

Figure 4.1: The adaption of the orbital shaker and incubator to allow for a syringe 
pump to be attached to the devices. The 3D printed cover for the shaker/incubator 
has 6 holes in the side in order for PEEK tubing to cross the cover and attach the 
devices to 5 mL syringes on a syringe pump.  The syringe pump flows either a 
fluorescein or antibiotic solution through the channel of the devices and allows diffusion 
to occur.  The devices can be shaken and incubated without the need of having the 
syringe pump on the shaker/incubator.  

 

 

154 

 

Figure 4.2: Dimensions of custom lab ware and custom modifications.  (Left) The 
dimensions for the shaker adapter to hold the devices to the platform.  A bracket (not 
shown) is required to secure the adapter to the platform. (Center) The dimensions for a 
custom cover for the orbital shaker to allow for the devices to be attached to a syringe 
pump. (Right Top) The dimensions for the cover for the adapter to hold the devices in 
the adapter during shaking. (Right Middle and Bottom) The dimensions for a 10 mL 
Petri dish and cover to preincubate the bacteria prior to antibiotic testing. 

 

155 

4.2.2 Growth Media and Agar Plates 

The procedure for the preparation of growth media and agar plates is the same 

as in Chapter 2. Growth media was prepared by dissolving 3.0 g of Lysogeny broth – 

Lennox pellets (LB, EMD Chemicals, Darmstadt, Germany) in 300 mL of distilled and 

deionized water (DDW).  The solution was covered and mixed thoroughly before being 

autoclaved for 45 minutes at 121 ºC. The solution was cooled until it could be handled 

without the use of heat resistant gloves and then the antibiotics were added to the 

desired concentration. The solution was then transferred to 50 mL centrifuge tubes and 

stored at 4 ºC until required.  The same procedure was used to create agar plates, but 

2.25 g of agar (laboratory grade, Fisher Scientific, Fair Lawn, NJ) were added before 

the autoclaving step.   

 

4.2.3 Antibiotic Reagents 

The antibiotic reagents used in this chapter are similar or the same as those 

used in Chapter 2. Gibco by Life Technologies (Grand Island, NY) was the supplier of 

kanamycin sulfate (USP grade). Sigma Life Science (St. Louis, MO) was the supplier of 

levofloxacin (HPLC, ≥ 98.0%) and chloramphenicol (water soluble). Growth media for 

kanamycin resistant (KanR) Escherichia  coli (E. coli) was created by adding 40 mg of 

kanamycin to each liter of LB solution, resulting in a 100 μM solution of kanamycin in 

LB. Growth media for chloramphenicol resistant (CmpR) Bacillus subtilis (B. subtilis) 

was created by adding 5 mg of chloramphenicol to each liter of LB solution, resulting in 

a 15 μM  solution of chloramphenicol in LB.   These growth mediums are referred to as 

 

156 

the selective media (LB) or selective agar and are used to ensure only growth of the 

antibiotic-resistant bacteria of interest. 

 

4.2.4 Bacterial Strains 

The kanamycin-resistant (KanR) strain of Escherichia coli (E. coli) and the 

chloramphenicol-resistant (CmpR) strain of Bacillus subtilis (B. subtilis) were provided 

by Dr. David P. Weliky and Dr. Lee. R. Kroos, respectively.25-26 The same glycerol 

stocks used in Chapter 2 were used in this chapter.  A 20% glycerol stock solution of 

the bacterial culture was made for each culture and stored at -80 ºC.  The glycerol stock 

solution was created by making an 80% (v/v) glycerol solution in DDW using glyercol 

(spectrophotometric grade, 99.5+%) purchased from Aldrich (Milwaukee, WI) and then 

incubated for 45 minutes at 121 ºC.  The bacterial cultures were plated on selective 

agar plates and grown overnight to create colonies. A single colony was transferred 

from the agar plate to selective media and grown for five hours to mid-logarthmic phase 

in 10 mL of selective LB. A bacterial stock solution was created in a sterile tube by 

adding 200 µL of the 80% glycerol solution and 800 µL of the mid-logarithmic phase 

bacteria.  The bacterial stock was then mixed by vortexing and stored at -80 ºC. 

 

4.2.5 Sample Preparation 

 

This procedure was adapted from the susceptibility test procedure in 

Chapter 2 in order to be used with the 3D-printed, two-compartment model described in 

Chapter 3. These procedures are modification of the research performed by Memptin et 

al.27 Bacteria from a glycerol stock were transferred to an agar plate containing the 

 

157 

antibiotic that the bacteria are resistant via an inoculating loop. The plate is incubated 

overnight for E. coli or 48 hours for  B. subtilis. A single colony was transferred using a 

pipette tip to a culture tube containing 2 mL of selective LB so only the bacteria of 

interest grow.  This tube was then incubated at 350 rpm and 37 ºC overnight. The next 

morning, the OD600 of 150 µL of overnight culture was measured. A 5 mL diluted 

culture with an OD600 ~ 0.005 was prepared in the Veroclear petri dish and 

preincubated for 1 -2 hours depending on the strain bacteria at 210 rpm and 37 ºC.  The 

purpose of preincubating the bacteria in the Veroclear material was to provide selective 

pressure on the growing bacteria so that the bacteria that replicate can survive in the 

presence of the Veroclear material. The dimensions for the petri dish are shown in 

Figure 4.2. The OD600 and extracellular ATP of the preincubated culture are measured 

at this time. 

Prior to use, the two-compartment devices are soaked in distilled and deionized 

water (DDW) with a pH = 7 – 8. Devices are air dried 10 – 15 minutes before use. The 

devices are hooked up to the syringe pump and placed into the shaker adapter, which is 

bracketed to the platform of the modified orbital shaker.  Levofloxacin (125 µM) in 

selective LB was primed vertically through the channel of three of the devices while 

selective LB with no levofloxacin was primed vertically through the channel of the other 

three devices.  The priming rate is 500 µL/min.  After the devices are primed, the flow 

rate is set at 5 µL/min. The preincubated bacterial culture (300 µL) is transferred to the 

secondary compartment of each device.  Selective LB without levofloxacin (300 µL) is 

transferred to the primary compartment of each device.  The cap is then placed over the 

shaker adaptor and the modified cover is closed on the orbital shaker.  The devices are 

 

158 

then shaken for one hour at 350 rpm and 37 ºC to produce the absorption part of the PK 

curve.  After one hour, the channels of the devices that contain selective LB with 

levofloxacin are primed with selective LB without levofloxacin to replicate the excretion 

part of the PK curve.  About 500 µL of liquid needs to be primed through the channel in 

order to ensure a sufficient change in the concentration of levofloxacin in the channel.  

The devices then resume incubation at 37 ºC and 350 rpm for 1, 2, 3, or 4 hours after 

incubation.  After the predetermined amount of time, all contents of the secondary 

compartment are collected and the OD600 and extracellular ATP are measured.  There 

is only enough liquid in the secondary compartment for one or two measurements, so 

each time point was a different experiment on a different day.   

 

4.2.6 Determination of Extracellular ATP and OD600 

The procedure for the determination of extracellular ATP and the OD600 is the 

same as in Chapter 2.  The OD600 was measured to determine the relative amount of 

bacteria in solution. An OD600 is an absorbance measurement at a wavelength of 600 

nm.  A volume of 150 µL was transferred from the samples from the preincubation and 

the endpoint of the experiment to a clear bottom 96 well plate.  The remaining volumes 

of the samples were centrifuged for 30 seconds at 30,000 g.  The supernatant was 

transferred to a new tube and stored at 4 ºC until ATP measurements were performed.  

A luciferin-luciferase assay was used to measure the ATP concentrations of each 

aliquot. The luciferin-luciferase assay is a chemiluminescence measurement where the 

concentration of ATP is the limiting reagent in a reaction creating an excited oxyluciferin 

molecule that produces light.28  The amount of light produced by the reaction is directly 

 

159 

correlated with the concentration of ATP allowing for quantitative determination. The 

luciferin-luciferase solution was made by combining 2 mg of potassium luciferin (Gold 

Biotechnology, St. Louis, MO) and 10 mg of firefly lantern extract (Sigma) into 5 mL of 

DDW and vortexing.  The luciferin-luciferase reagent was aliquoted into 1 mL storage 

vials and stored at -20 ºC.  To measure the ATP concentration, 150 µL of sample were 

added to a black bottom 96 well plate and 20 µL of the luciferin-luciferase solution was 

added.  The samples were mixed directly before measuring the chemiluminescense.  

The concentration of ATP in the sample was quantified by comparing with the 

chemiluminescense signal of ATP standards.  An ATP working solution was prepared 

using standard ATP (Adenosine 5’-triphosphatte disodium salt hydrate (Grade I, ≥ 99%), 

Sigma) in DDW with a final concentration of 1 µM.  Serial dilutions were performed in 

selective LB to create the samples used for the standard curve.  The concentration of 

extracellular ATP is dependent on the number of bacteria cells present in the culture. 

Therefore, the ATP/OD600 ratio was used to indicate if the bacteria were alive.  The 

ATP/OD600 ratio can vary due to the age of the samples such as the bacteria or the 

growth mediums (agar plates and LB growth media).  The ATP/OD600 ratios were 

normalized against the ATP/OD600 ratio of the sample after the preincubation but 

before the drug of interest was added to account for the differences in the samples.  

The ATP/OD600 ratio of the antibiotic sample at a given time point was compared 

against the ATP/OD600 ratio of the control at the same time point to determine if there 

was a significant difference (α = 0.05)  A Flexstation 3 (Molecular Devices, Sunnyvale, 

CA) was used for all ATP and OD600 measurements. A Sorvall ST 8R Centrifuge 

(Thermo Scientific, Waltham, MA) was used for centrifugation.   

 

160 

4.3 Results 

4.3.1 Oral Pharmacokinetic Profile of Levofloxacin Using Fluorescein 

 

The purpose of running a similar procedure to that performed in Chapter 3 was to 

ensure that the 3D-printed modifications for the shaker/incubator did not result in drastic 

changes to the PK curve.  The results of these experiments are shown in Figure 4.3. 

The results of three devices run across three different days show that the devices under 

the same conditions produce relatively similar PK curves.  The resulting oral PK model 

resulted in a Cmax of 12.4 ± 3.0 µM, a tmax of 1 hour, and a half-life of 5.2 ± 0.5 hours in 

the primary compartment.  The concentration of fluorescein is representative of the 

concentration of levofloxacin.24, 29 The oral PK profile created by the device is similar to 

a human PK profile of an oral administration of levofloxacin.  In humans, the Cmax of 

levofloxacin is 7.8 – 14.4 µM, has a tmax of 1 – 2 hours, and has a half-life of 6 – 8 hours 

in healthy patients.30 In the secondary compartment, the concentration of fluorescein 

rises to 6.5 µM after about 7 hours. This procedure will be used for the dosing of 

bacteria with antibiotics under dynamic conditions, but the fluorescein solution in DDW 

will be replaced with a levofloxacin solution in selective LB.   

 

161 

 

 

Figure 4.3: The PK curves for both the primary and secondary compartment for 
an oral model of levofloxacin using fluorescein.  In the primary compartment (upper 
plot), the fluorescein concentration reached a Cmax of 12.4 ± 3.0 µM at a tmax of 1 hour 
with a half-life of 5.2 ± 0.5 hours.  In the secondary compartment (lower plot), the 
fluorescein concentration continues to increase to 6.5 µM after 7 hours. (n = 9, error = 
SEM) 

 

162 

Time (hours)02468Fluorescein Concentration (0246810121416Time (hours)02468Fluorescein Concentration (024684.3.2 Bacterial Strains Challenged by an Oral PK Curve of Levofloxacin 

 

Both Gram-negative E. coli and Gram-positive B. subtilis were exposed to a 

levofloxacin PK curve representative of an oral administration similar to that shown in 

Figure 4.3.  The purpose of this experiment was to determine a statistical difference in 

the normalized ATP/OD600 of a bacterial culture while under a dynamic PK curve of the 

antibiotic levofloxacin.  The results for kanamycin-resistant E. coli exposed to a 

levofloxacin PK curve are shown in Figure 4.4. The ATP/OD600 trend after the 

maximum peak, as seen in Chapter 2, is similar.  The ATP/OD600 of the controls 

continues to decrease while the culture challenged with the antibiotic increases.  In the 

first hour of the curve, the ATP/OD600 of the culture challenged with levofloxacin 

decreases in a similar way to the controls before the ATP/OD600 started to increase.  

This is most likely due to the concentration in the secondary compartment not being 

high enough to effectively start killing the bacteria (below the minimum effective 

concentration).  The earliest significant difference (α = 0.05) was detected 2 hours from 

the start of the PK curve, or 1 hours after Cmax. The concentration of the levofloxacin in 

the secondary compartment, as determined from the fluorescein curve in Figure 4.3, is 

equal to 3.5 ± 1.3 µM.  Clinical minimum inhibitory concentrations for E. coli against 

levofloxacin have been determined to be 0.69 – 5.5 µM, which confirms the results seen 

here.31 

 

163 

 

Figure 4.4: Normalized ATP/OD600 signals of a kanamycin-resistant E. coli strain 
exposed to a simulated levofloxacin oral administration PK curve. After 
preincubating an E. coli culture for 1 hour, the culture was exposed to a levofloxacin PK 
curve representative of an oral administration.  The normalized ATP/OD600 trend was 
the same under dynamic conditions as static conditions like in Chapter 2.  The 
ATP/OD600 of healthy growing bacteria (black bars) continues to decrease as the 
bacteria grow.  The ATP/OD600 of bacteria challenged with the antibiotic increased 
after reaching an effective concentration of the antibiotic.  The earliest statistical 
difference was detected 2 hours after the start of the PK curve or 1 hour after Cmax.  The 
corresponding secondary compartment concentration at this time point was 3.5 ± 1.3 
µM. (n = 3; n = 12 at 0 time point, error = standard deviation, * p < 0.05) 

 

164 

The results for chloramphenicol-resistant B. subtilis exposed to a levofloxacin PK 

curve is shown in Figure 4.5. The ATP/OD600 trend for both the controls and the culture 

exposed to levofloxacin follow similar trends to that seen with the E.coli, but the 

changes were not as drastic.  In the first hour of the curve, the ATP/OD600 of the 

culture challenged with levofloxacin decreases in a similar way to the controls before 

the ATP/OD600 started to increase.  This is most likely due to the concentration in the 

secondary compartment not being high enough to effectively start killing the bacteria.  

The earliest significant difference (α = 0.05) was detected 3 hours from the start of the 

PK curve, or 2 hours after Cmax. The concentration of the levofloxacin in the secondary 

compartment, as determined from the fluorescein curve in Figure 4.3, is equal to 4.8 ± 

1.8 µM.  A minimum inhibitory concentration for B. subtilis against levofloxacin was 

determined to be 2.76 µM,32 which is slightly lower than the results seen here and may 

be due to there being a time delay in measuring the difference in the ATP/OD600 once 

the MIC has been reached. The closely related strain Bacillus cereus showed a 

minimum inhibitory concentration range of 0.3 – 88.6 µM of levofloxacin in 26 clinical 

isolates.33  The reason that the normalized ATP/OD600 are so low and not increasing 

as fast as those for the E. coli may be due to B. subtilis being strictly aerobic and 

requiring oxygen to grow.  The current design of the device may be inadequate to 

provide sufficient oxygenation of the bacterial culture leading to slow growth, which may 

explain the low ratios. 

 

165 

 

Figure 4.5: Normalized ATP/OD600 signals of a chloramphenicol-resistant B. 
subtilis strain exposed to a simulated levofloxacin oral administration PK curve. 
After preincubating a B. subtilis culture for 2 hours, the culture was exposed to a 
levofloxacin PK curve representative of an oral administration. The ATP/OD600 of 
healthy growing bacteria (black bars) continues to decrease as the bacteria grow.  The 
ATP/OD600 of bacteria challenged with the antibiotic remains steady or increasing 
slightly after reaching an effective concentration of the antibiotic.  The earliest statistical 
difference was detected 3 hours after the start of the PK curve or 2 hours after Cmax.  
The corresponding secondary compartment concentration at this time point was 4.8 ± 
1.8 µM. (n = 3; n = 12 at 0 time point, error = standard deviation, * p < 0.05) 

 

166 

4.4 Discussion 
 
 

The devices described in Chapter 3 were used successfully to model the PK/PD 

characteristics of an antibiotic and its interaction with a strain of bacteria. The antibiotics 

effect on the bacteria was assessed by measuring a significant difference in the 

ATP/OD600, in a similar fashion to the static measurements performed in Chapter 2.  

Both a Gram-positive and Gram-negative bacterial strain were able to show a significant 

difference in the normalized ATP/OD600 after being exposed to an oral PK curve of 

levofloxacin.  The PK curve for an oral administration of levofloxacin was created 

utilizing fluorescein, which has a similar molecular weight and diffusion coefficient in 

water.24, 29  Fluorescein was flowed through the channel with DDW in the other two 

compartments.  After one hour, the fluorescein in the channel was replaced with just 

DDW.  The resulting concentration of fluorescein in the compartment resulted in a curve 

similar to that of levofloxacin in regards to the Cmax, tmax, and half-life.30  

 

Either a culture of kanamycin-resistant E. coli or chloramphenicol-resistant B. 

subtilis was preincubated in a 10 mL petri made of Veroclear material for one or two 

hours, respectively.  The preincubation in a Veroclear petri dish was to ensure that the 

bacteria could grow in the presence of the 3D printed material.  Some of these Polyjet 

printed materials are toxic to cells.34  After the preincubation, the bacteria were 

transferred to the secondary compartment and exposed to the levofloxacin oral PK 

curve.  For E. coli, a statistical difference in the normalized ATP/OD600 was calculated 

at 2 hours after the start of the PK curve, or 1 hour after Cmax.  The corresponding 

secondary compartment concentration was 3.5 ± 1.3 µM levofloxacin.  For B. subtilis, a 

statistical difference in the normalized ATP/OD600 was calculated at 3 hours after the 

 

167 

start of the PK curve, or 2 hour after Cmax.  The corresponding secondary compartment 

concentration was 4.8 ± 1.8 µM levofloxacin. Both of these concentrations are within 

reasonable range of experimental minimum inhibitory concentrations.31-33 

 

The limitations of this procedure centers around the parameters of PK that are 

missing.  These devices are capable of mimicking absorption, distribution, and excretion 

through the use of porous membranes.  However, in the body these diffusion barriers 

are made up of various types of cells.  Polyethersulfone membranes can be modified to 

allow for the adhesion of various cell types so these diffusion barriers could be more 

representative to those seen in vivo.35-37  These diffusion barriers also limit the size of 

molecules that can diffuse through, which in vivo means that molecules bound to 

plasma proteins are not able to diffuse and elicit their mechanism of action. The addition 

of plasma proteins such as albumin, globulins, and glycoproteins are required to provide 

a more in vivo like equilibrium of the molecule, which can change the PK curve.38  

 

Another limitation is that the current design of the device does not appear to 

provide adequate oxygenation, which is required for aerobic bacteria.  Low oxygen 

could be the reason that the ATP/OD600 are so much lower for the B. subtilis 

experiment than the E. coli.  Changing the shape of the secondary compartment from a 

cube to a cylindrical shape, could provide better oxygenation.  Different types of 

bacteria need to be tested with this device provided the necessary changes to address 

oxygenation to ensure that the device itself is robust in use with other bacterial types. 

 

In this procedure, only a bactericidal antibiotic was used.  More antibiotics need 

to be tested to ensure that the device can be used robustly with other antibiotics as well 

as other drug administrations such as intravenous administrations. In Chapter 2, one of 

 

168 

the problems with the static procedure was that the pharmacological concentrations of 

bacteriostatic antibiotics were too high leading to bactericidal effects.  Bacteriostatic 

antibiotics need to be tested using this device to provide more adequate interstitial fluid 

concentrations of antibiotic in the secondary compartment while recreating a 

pharmacological concentration in the primary compartment.  This could address the 

problems in Chapter 2 regarding the bactericidal endpoint.   

 

Overall, the 3D printed, two compartment devices were capable of dosing both a 

Gram-positive and Gram-negative strain of bacteria against an oral PK curve of 

levofloxacin.  The data obtainable from this device include PK information including 

concentration in both the primary and secondary compartment, the Cmax, tmax, and half-

life of the molecule, as well as PD information regarding minimum inhibitory 

concentrations.  Compared to other susceptibility methods, this procedure exposes 

bacteria to an in vivo-like PK curve of the antibiotic of interest, instead of just a constant 

concentration of antibiotic. Compared to other dynamic methods, like the hollow fiber 

chamber reactor, this procedure requires much less antibiotic solution and the cells in 

the secondary compartment can be monitored at various time points during the PK 

curve. 

 

 

 

 

 

169 

REFERENCES 

 

170 

DiMasi, J. A.; Grabowski, H. G.; Hansen, R. W., Innovation in the pharmaceutical 

Hay, M.; Thomas, D. W.; Craighead, J. L.; Economides, C.; Rosenthal, J., 

Scannell, J. W.; Blanckley, A.; Boldon, H.; Warrington, B., Diagnosing the decline 

 

REFERENCES 

 

Culyba, M. J.; Mo, C. Y.; Kohli, R. M., Targets for Combating the Evolution of 

Falagas, M. E.; Karageorgopoulos, D. E., Pandrug Resistance (PDR), Extensive 

Waring, M. J.; Arrowsmith, J.; Leach, A. R.; Leeson, P. D.; Mandrell, S.; Owen, 

Munos, B., Lessons from 60 years of pharmaceutical innovation. Nat Rev Drug 

1. 
Acquired Antibiotic Resistance. Biochemistry 2015, 54 (23), 3573-3582. 
 
2. 
Drug Resistance (XDR), and Multidrug Resistance (MDR) among Gram-Negative 
Bacilli: Need for International Harmonization in Terminology. Clinical Infectious 
Diseases 2008, 46 (7), 1121-1122. 
 
3. 
industry: New estimates of R&D costs. Journal of Health Economics 2016, 47, 20-33. 
 
4. 
Discov 2009, 8 (12), 959-68. 
 
5. 
R. M.; Pairaudeau, G.; Pennie, W. D.; Pickett, S. D.; Wang, J., An analysis of the 
attrition of drug candidates from four major pharmaceutical companies. Nature reviews 
Drug discovery 2015, 14 (7), 475. 
 
6. 
Clinical development success rates for investigational drugs. Nature biotechnology 
2014, 32 (1), 40. 
 
7. 
in pharmaceutical R&D efficiency. Nature reviews Drug discovery 2012, 11 (3), 191. 
 
8. 
Berndt, E. R.; Nass, D.; Kleinrock, M.; Aitken, M., Decline in economic returns 
from new drugs raises questions about sustaining innovations. Health affairs (Project 
Hope) 2015, 34 (2), 245-52. 
 
9. 
pipeline? The Pharmaceutical Journal 2013, 291, 520. 
 
10.  Blaskovich, M. A.; Butler, M. S.; Cooper, M. A., Polishing the tarnished silver 
bullet: the quest for new antibiotics. Essays in biochemistry 2017, 61 (1), 103-114. 
 
11.  National Action Plan for Combating Antibiotic-Resistant Bacteria. The White 
House: Washington D.C., 2015. 
 
12.  National Action Plan For Combating Antibiotic-Resistant Bacteria: Progress 
Reprot for Years 1 and 2. The White House: Washington D.C., 2017. 
 

Sukkar, E., Why are there so few antibiotics in the reserach and development 

 

171 

13.  Obama, B., Executive Order - Combating Antibiotic-Resistant Bacteria. The 
White House: Washington D.C., 2014. 
 
14. 
Jorgensen, J. H.; Ferraro, M. J., Antimicrobial susceptibility testing: a review of 
general principles and contemporary practices. Clinical infectious diseases : an official 
publication of the Infectious Diseases Society of America 2009, 49 (11), 1749-55. 
 
15.  Balouiri, M.; Sadiki, M.; Ibnsouda, S. K., Methods for in vitro evaluating 
antimicrobial activity: A review. Journal of Pharmaceutical Analysis 2016, 6 (2), 71-79. 
 
16.  Vaddady, P. K.; Lee, R. E.; Meibohm, B., In vitro 
pharmacokinetic/pharmacodynamic models in anti-infective drug development: focus on 
TB. Future Medicinal Chemistry 2010, 2 (8), 1355-1369. 
 
17.  Heller, A. A.; Spence, D. M., A rapid method for post-antibiotic bacterial 
susceptibility testing. PLOS ONE 2019, 14 (1), e0210534. 
 
18.  Heller, A. A.; Lockwood, S. Y.; Janes, T. M.; Spence, D. M., Technologies for 
Measuring Pharmacokinetic Profiles. Annual Review of Analytical Chemistry 2018, 11 
(1), 79-100. 
 
19.  Heller, A. A. In Silico Pharmacokinetics and the Potential Effects on 
Pharmacodynamics: The Missing Method in Antibiotic Development. Michigan State 
Unviersity, East Lansing, MI, 2018. 
 
20.  Dedrick, R. L., Animal scale-up. Journal of Pharmacokinetics and 
Biopharmaceutics 1973, 1 (5), 435-461. 
 
21.  Drusano, G. L., Role of pharmacokinetics in the outcome of infections. 
Antimicrobial Agents and Chemotherapy 1988, 32 (3), 289-297. 
 
22.  Boxenbaum, H., Interspecies scaling, allometry, physiological time, and the 
ground plan of pharmacokinetics. Journal of Pharmacokinetics and Biopharmaceutics 
1982, 10 (2), 201-227. 
 
23.  Pankey, G. A.; Sabath, L. D., Clinical Relevance of Bacteriostatic versus 
Bactericidal Mechanisms of Action in the Treatment of Gram-Positive Bacterial 
Infections. Clinical Infectious Diseases 2004, 38 (6), 864-870. 
 
24.  Vrany, J. D.; Stewart, P. S.; Suci, P. A., Comparison of recalcitrance to 
ciprofloxacin and levofloxacin exhibited by Pseudomonas aeruginosa bofilms displaying 
rapid-transport characteristics. Antimicrobial agents and chemotherapy 1997, 41 (6), 
1352-1358. 
 
 

 

172 

Fish, D. N.; Chow, A. T., The Clinical Pharmacokinetics of Levofloxacin. Clinical 

25.  Ratnayake, P. U.; Ekanayaka, E. A. P.; Komanduru, S. S.; Weliky, D. P., Full-
Length Trimeric Influenza Virus Hemagglutinin II Membrane Fusion Protein and Shorter 
Constructs Lacking the Fusion Peptide or Transmembrane Domain: 
Hyperthermostability of the Full-Length Protein and the Soluble Ectodomain and Fusion 
Peptide Make Significant Contributions to Fusion of Membrane Vesicles. Protein 
expression and purification 2016, 117, 6-16. 
 
26.  Halder, S.; Parrell, D.; Whitten, D.; Feig, M.; Kroos, L., Interaction of 
intramembrane metalloprotease SpoIVFB with substrate Pro-σK. Proceedings of the 
National Academy of Sciences 2017, 114 (50), E10677-E10686. 
 
27.  Mempin, R.; Tran, H.; Chen, C.; Gong, H.; Kim Ho, K.; Lu, S., Release of 
extracellular ATP by bacteria during growth. BMC Microbiology 2013, 13, 301-301. 
 
28.  Hirano, T.; Hasumi, Y.; Ohtsuka, K.; Maki, S.; Niwa, H.; Yamaji, M.; Hashizume, 
D., Spectroscopic Studies of the Light-Color Modulation Mechanism of Firefly (Beetle) 
Bioluminescence. Journal of the American Chemical Society 2009, 131 (6), 2385-2396. 
 
Lockwood, S. Y.; Meisel, J. E.; Monsma, F. J.; Spence, D. M., A Diffusion-Based 
29. 
and Dynamic 3D-Printed Device That Enables Parallel in Vitro Pharmacokinetic Profiling 
of Molecules. Analytical Chemistry 2016, 88 (3), 1864-1870. 
 
30. 
Pharmacokinetics 1997, 32 (2), 101-119. 
 
31.  DeFife, R.; Scheetz, M. H.; Feinglass, J. M.; Postelnick, M. J.; Scarsi, K. K., 
Effect of Differences in MIC Values on Clinical Outcomes in Patients with Bloodstream 
Infections Caused by Gram-Negative Organisms Treated with Levofloxacin. 
Antimicrobial Agents and Chemotherapy 2009, 53 (3), 1074-1079. 
 
32. 
H.; Shafiee, A., Synthesis and antibacterial activity of levofloxacin derivatives with 
certain bulky residues on piperazine ring. European Journal of Medicinal Chemistry 
2007, 42 (7), 985-992. 
 
33.  Horii, T.; Notake, S.; Tamai, K.; Yanagisawa, H., Bacillus cereus from blood 
cultures: virulence genes, antimicrobial susceptibility and risk factors for blood stream 
infection. FEMS Immunology & Medical Microbiology 2011, 63 (2), 202-209. 
 
34.  Walsh, M. E.; Ostrinskaya, A.; Sorensen, M. T.; Kong, D. S.; Carr, P. A., 3D-
Printable Materials for Microbial Liquid Culture. 3D Printing and Additive Manufacturing 
2016, 3 (2), 113-118. 
 
35.  Azadbakht, M.; Madaeni, S. S.; Sahebjamee, F., Biocompatibility of 
polyethersulfone membranes for cell culture systems. Engineering in Life Sciences 
2011, 11 (6), 629-635. 

Foroumadi, A.; Emami, S.; Mansouri, S.; Javidnia, A.; Saeid-Adeli, N.; Shirazi, F. 

 

173 

36.  Unger, R. E.; Huang, Q.; Peters, K.; Protzer, D.; Paul, D.; Kirkpatrick, C. J., 
Growth of human cells on polyethersulfone (PES) hollow fiber membranes. Biomaterials 
2005, 26 (14), 1877-1884. 
 
37.  Unger, R. E.; Peters, K.; Huang, Q.; Funk, A.; Paul, D.; Kirkpatrick, C. J., 
Vascularization and gene regulation of human endothelial cells growing on porous 
polyethersulfone (PES) hollow fiber membranes. Biomaterials 2005, 26 (17), 3461-
3469. 
 
38.  Busher, J. T., Serum Albumin and Globulin. In Clinical Methods: The History, 
Physical, and Laboratory Examinations, rd; Walker, H. K.; Hall, W. D.; Hurst, J. W., Eds. 
Butterworths 
Butterworth Publishers, a division of Reed Publishing.: Boston, 1990. 
 

 
 

 

 

174 

CHAPTER 5 

 

OVERALL CONCLUSIONS & FUTURE WORK 

 

5.1 Overall Conclusions 

 

Antibiotics are an important pharmaceutical asset that have made many modern 

medical procedures possible due to their various mechanisms of action to fight bacterial 

infections.1-11  However, the over prescription and misuse of antibiotics has influenced 

the evolution of bacteria that are resistant to multiple and, in extreme circumstances, all 

available antibiotics due to bacteria’s ability to develop and transfer antibiotic resistance 

genes.12-16 In the United States alone, antibiotic resistant bacteria are responsible for 

over 2 million illnesses and 23,000 deaths. Novel antibiotics are needed to treat 

antibiotic resistance bacteria, but the nature of bacteria will eventually lead the bacteria 

to become resistant to the new antibiotics.17  Currently, many pharmaceutical 

companies are not pursuing research into new antibiotics because antibiotic 

development does not provide an adequate return on investment. The threat of 

resistance, the market’s unwillingness to pay high price for novel antibiotics, and the 

increasing cost of drug development all correlate with the decreasing numbers of novel 

antibiotics approved by the U.S. Food & Drug Administration (FDA).18-19      

 

Currently, the average investment for a pharmaceutical company to develop a 

drug, like antibiotics, through approval by the FDA is over $2.5 billion dollars and can 

take greater than 10 years to achieve.  The investment for preclinical trials is 

responsible for $1.098 billion and 31.2 months and the investment for clinical trials is 

 

175 

responsible for $1.460 billion and 80.8 months.20 The high cost of drug development 

has been associated with the high failure rate of potential drugs in clinical trials due to 

safety or efficacy concerns that were not determined in in vitro or in vivo animal trials 

during preclinical trials.21-22  This dissertation described the research performed in order 

to provide more clinically-relevant pharmacokinetic and pharmacodynamic (PK/PD) data 

for antibiotics from in vitro susceptibility assays.   

 

The most popular methods for in vitro antibiotic susceptibility assays provide PD 

data, but the bacteria are not subjected to the PK of the antibiotic in the body, only a 

constant concentration.  These assays also tend to take an overnight incubation to 

obtain a result.23 In Chapter 2, the research of Mempin et al. was used to create a 

susceptibility assay that determined if an antibiotic would be efficient in treating a 

particular strain of bacteria. Mempin et al. showed that ATP/OD600 of a growing 

bacteria culture increased during logarithmic growth and then decreased once the 

bacteria entered stationary phase.  This trend in the ATP/OD600 only occurred when 

the bacteria was alive.24 This trend lead to the hypothesis in Chapter 2 that the addition 

of a functional antibiotic after the peak ATP/OD600 curve would lead to a difference in 

the ATP/OD600 when compared with a culture of healthy growing bacteria.  Monitoring 

the ATP/OD600 of a growing culture showed that removing the absorbance caused by 

the media and the 96 well plate resulted in the ATP/OD600 increasing during the lag 

and early logarithmic phase and then decreasing throughout the logarithmic phase.  

Adding the antibiotic after the peak ATP/OD600 allowed for a statistical difference (α = 

0.05) to be determined within one hour of adding an antibiotic for both a Gram-positive 

strain of bacteria, a Gram-negative strain of bacteria, and a mixture of two bacteria.25  

 

176 

The shortcomings of this method are that the static concentrations used to challenge 

the bacteria were maximum plasma concentrations (Cmax), which would be higher than 

the minimum inhibitory or minimum bactericidal concentrations of antibiotics. This 

means that any bacteriostatic antibiotics that were tested resulted in bactericidal results.  

To combat this problem, a two compartment, 3D printed device was required in order to 

create a plasma compartment, where Cmax occurs, and an interstitial compartment, 

where bacteria reside, in order to create the PK curve of the antibiotic in the plasma 

compartment and expose the bacteria to more in vivo like concentrations.   

 

The design and characterization of the two compartment, 3D printed device were 

described in Chapter 3.  The Spence Lab has created many different 3D models that 

replicate different PK parameters.26-28  One of these devices was even used to mimic 

the PK curve of antibiotic and used it to challenge bacteria in order to determine the 

effect of the antibiotic.29 However, the device was designed with the bacteria in the 

plasma compartment resulting in antibiotic concentrations, like in Chapter 2, that are 

higher than what is experienced in the interstitial fluid.  A Statasys J750 polyjet-style 3D 

printer was capable of printing the two compartment device due to its ability to print 

multiple materials and the ability to integrate porous membranes into designs via the 

print-pause-print method or the further modified Z-axis drop method.26-27  In Chapter 3, 

the 3D printed device and inserts were tested for leakage with fluorescein to ensure that 

the joint between the device and insert as well as the integration of the membranes did 

not lead to unwanted movement of molecules.  No significant difference (α = 0.05) was 

detected in any test signifying leakage. The device was equipped with an insert 

 

177 

containing porous membranes and PK curves were able to be replicated for an oral, 

intermittent intravenous (IV), and continuous IV administration. 

 

In Chapter 4, the research from the previous two chapters were combined to 

create a susceptibility test that replicated the human PK of the antibiotic levofloxacin 

administered orally in the plasma compartment and challenged bacteria in the lateral 

interstitial compartment.  The bacteria were either a Gram-positive or Gram-negative 

strain of bacteria.  The results of the susceptibility tests determine both the efficacy of 

the antibiotic as well as PD data such as minimum bactericidal concentrations for the 

given PK curve.  Fluorescein was used to replicate an oral PK curve of levofloxacin due 

to their similar molecular weights and diffusion coefficients in water.28, 30 The PK curve 

resulted in a Cmax of 12.4 ± 3.0 µM and a half-life of 5.2 ± 0.5 hours. For a kanamycin-

resistant strain of Escherichia coli (E. coli) that was challenged to this PK curve, the 

earliest significant difference (α = 0.05) in the normalized ATP/OD600 compared to a 

control was detected at 2 hours (1 hour after Cmax).  At this time point, the minimum 

bactericidal concentration was determined to be 3.5 ± 1.3 µM. For a chloramphenicol 

resistant strain of Bacillus subtilis (B. subtilis), the earliest significant difference (α = 

0.05) in the normalized ATP/OD600 compared to a control was detected at 3 hours (2 

hours after Cmax). At this time point, the minimum bactericidal concentration was 

determined to be 4.8 ± 1.8 µM.  

 

 

 

 

 

178 

5.2 Future Work 

5.2.1 More Antibiotics and Bacterial Strains 

 

In the experiments of this dissertation, only a few antibiotics and bacterial strains 

were used.  In Chapter 2, the bacteriostatic antibiotics that were used were causing 

bactericidal effects in that the ATP/OD600 was increasing, which can only occur by 

increasing extracellular ATP or decreasing OD600.  Both of these are signs of dying 

bacteria cells. The bactericidal effects of the bacteriostatic antibiotics could possibly be 

due to the bacteria being exposed to a concentration of the antibiotic that was too 

high.31 Bacteriostatic antibiotics should be used in the procedure for the static 

concentration susceptibility test from Chapter 2, but at minimum inhibitory 

concentrations to determine if a difference can be measured in the ATP/OD600 after 

adding the antibiotic.   

Additional bacteriostatic and bactericidal antibiotics should be tested in the 

procedure from Chapter 4 involving the dynamic susceptibility test using the 3D printed 

two compartment devices and membrane inserts.  As shown in Chapter 3, the two 

compartment devices are capable of mimicking different types of drug administration.  

The different types of antibiotics should be loaded into the device to create the desired 

administration resulting in the desired Cmax.  For example, gentamicin is an 

aminoglycoside antibiotic that is delivered to the bloodstream by intermittent intravenous 

administration to a Cmax of 31.4 – 41.9 µM.32 In the two compartment model this would 

be replicated by adding a 31.4 – 41.9 µM gentamicin in LB solution into the primary 

compartment, a growing bacterial culture into the second, and flowing an LB solution 

without gentamicin through the channel. This will lead to the bacteria cells being 

 

179 

exposed to more similar concentrations of the antibiotic to what are found in the 

interstitial fluid. 

More bacterial strains should undergo both procedures from Chapter 2 and 

Chapter 4.  The purpose of using a Gram-positive B. subtilis and a Gram-negative E. 

coli strain of bacteria was to show that the procedures function for both types of 

bacteria.  These two particular strains were used due to them being non-pathogenic.  

The above procedure should be tested on more bacterial strains to ensure that the 

procedures will function for all bacterial strains.  Mempin et al. showed that the 

ATP/OD600 trend was also seen in Salmonella enterica, Acinetobacter junii, 

Pseudomonas aeruginosa, Klebsiella pneumoniae, Klebsiella oxytoca, and 

Staphylococus aureus.24 The procedures should also be performed with the 18 bacterial 

strains that the Centers for Disease Control & Prevention considers a threat or of 

concern to determine if the ATP/OD600 trend is seen in these strains and if the data 

collected from these procedures is similar to the data collected from current procedure 

for antibiotic susceptibility.17  

 

5.2.2 More Human-Like Pharmacokinetics 

 

Currently, there is not a lot of information about the concentration of antibiotics in 

the interstitial fluids of humans.33  This lack of information makes it difficult to confirm 

the concentration of antibiotic in the secondary compartment. In Chapter 3, it was noted 

that diffusion rate into the primary compartment was different than the diffusion rate into 

the secondary compartment caused by differences in the surface area of the membrane 

insert that was available for diffusion.  As more information about interstitial fluid 

 

180 

concentration of antibiotics is determined, the surface areas or the percent porosity of 

the different membranes can be adjusted to better match the PK of the molecule of 

interest.  An alternative to mimicking the diffusion of molecules into the different 

compartments is to adhere similar cell types to the membranes to replicate in vivo 

barriers. 

 

Currently the design of the two compartment devices allows for absorption, 

excretion, and distribution to be mimicked.  While these portions of PK have led to a 

procedure that exposes bacteria to PK curve similar to that seen in a human, there are 

still many missing PK parameters that could be added to the device through the use of 

cells and proteins.  The polyethersulfone (PES) membranes used to create the porous 

membrane insert of the two compartment device have been reported to be able to 

sustain viable, confluent monolayers of various types of human cells including 

endothelial, epithelial, keratinocytes, hepatocytes, fibroblasts, osteoblasts, and glial 

cells.34-36 The adhesion of human cells such as endothelial cells could provide a more 

lifelike barrier to the secondary compartment than the current porous membranes 

leading to more clinically relevant PK and subsequent PD data. This could be achieved 

by adhering endothelial cells to the PES membrane with much larger pores on the 

inside of the membrane insert, allowing for the cell layer to primarily control the diffusion 

instead of the pores. This would create a more realistic model of a plasma compartment 

than what is being used currently in the device.  The adhesion of other cell types, like 

hepatocytes, could introduce metabolism to the device, which is currently missing.  This 

would be beneficial since a drug that is in circulation will undergo metabolism, primarily 

in the liver, which will result in a metabolite that may or may not be an inactive form of 

 

181 

the drug.  Prodrugs actually require activation by metabolism in order to elicit their 

effect.37 The adhesion of cells could broaden the types of pharmacological experiments 

that could be performed with these devices as well as fix issues related to diffusion such 

as the shorter half-life of drugs in the 3D printed devices. 

 

Distribution of the drug in these devices is based on the equilibrium of the drug of 

interest. However, in the body distribution is more complex because there are proteins 

in the plasma that bind these drugs leading to a portion of drug that is bound to these 

plasma proteins and a portion that are not bound.  Drug that is bound to plasma 

proteins cannot diffuse and will remain in the plasma, which increases the half-life of the 

drug, but decreases the effectiveness of the drug.  These plasma proteins consist of 

albumin, glycoproteins, and globulins. Together, these proteins have a total level of 6 – 

8 g/dL in the serum with albumin composing 3.5 – 5.0 g/dL of that level.38 Albumin has 

been show to play an important role in delivery of various proteins and molecules, 

including antibiotics.39-42 In order to provide more human-like conditions, the proper 

concentration of plasma proteins should be added to the primary compartment in future 

experiments.  A combination of human cell adhesion and the introduction of plasma 

proteins would provide more clinically-relevant PD data because the PK would then be 

controlled by the diffusion of the molecules through cell monolayers, plasma protein 

binding would lead to a fraction of free drug as well as a longer half-life, and metabolism 

would occur.  These three aspects are currently missing from the current two 

compartment devices. 

 

 

 

 

182 

5.2.3 Device Alterations 

 

Alterations to the current design of the device may provide benefits to the current 

procedure.  As seen in Chapter 4, the B. subtilis challenged with an oral administration 

of levofloxacin for 4 hours did not produce the increase in the ATP/OD600 as the E. coli 

did. This may have been due to B. subtilis being strictly aerobic and there not being 

adequate oxygen due to inadequate shaking.  This would lead to poor growth, which 

would result in lower ATP/OD600 after similar exposure times. Better oxygenation of the 

bacterial culture may be achievable by designing the secondary compartment as a 

semi-cylinder instead of a cube.  Another design change that could benefit the device is 

the addition of a second secondary compartment.  In the body, drug molecules may be 

delivered to muscular, hepatic, and renal tissue at the same time.  Adding a second 

secondary compartment could allow for a second parameter to be measured or 

examined.  For example, a circulating antibiotic will undergo metabolism, possibly 

deactivating the antibiotic.  A liver compartment could be created in another secondary 

compartment by culturing hepatocytes in that compartment.  The hepatocytes would 

metabolize the antibiotic an then the metabolites would diffuse back into the primary 

compartment and be free to diffuse into the other compartments as the metabolites 

would in the body.  At the same time, hepatotoxicity could be monitored along with the 

efficacy on the bacteria and the plasma concentration.  The designs for both a circular 

secondary compartment and a second secondary compartment are shown in Figure 

5.1. These devices as they are now are complex, but still primitive and can be adapted 

in numerous ways in order to provide more human-like PK and more clinically-relevant 

 

183 

PD data, which may have the potential to reduce the reliance on animal models and 

decrease the cost of pre-clinical trials and drug development in general. 

 

184 

Figure 5.1: Designs of alternations that could benefit the PK/PD data. (A) A semi-
cylindrical secondary compartment could provide better mixing and oxygenation to the 
bacterial culture leading to healthier growth. (B) A second secondary compartment 
could be used to monitor another PK parameter such as metabolism or a PD parameter 
such as toxicity of another cell type.   

 

 

185 

REFERENCES

 

186 

REFERENCES 

 

Yocum, R. R.; Rasmussen, J. R.; Strominger, J. L., The mechanism of action of 

Donowitz, G. R.; Mandell, G. L., Beta-Lactam antibiotics (1). The New England 

Choi, S.-W.; Mason, J. B., Folate Status: Effects on Pathways of Colorectal 

Sköld, O., Sulfonamide resistance: mechanisms and trends. Drug Resistance 

Mingeot-Leclercq, M.-P.; Glupczynski, Y.; Tulkens, P. M., Aminoglycosides: 

Chopra, I.; Roberts, M., Tetracycline Antibiotics: Mode of Action, Applications, 

Crider, K. S.; Yang, T. P.; Berry, R. J.; Bailey, L. B., Folate and DNA Methylation: 

1. 
penicillin. Penicillin acylates the active site of Bacillus stearothermophilus D-alanine 
carboxypeptidase. Journal of Biological Chemistry 1980, 255 (9), 3977-86. 
 
2. 
journal of medicine 1988, 318 (7), 419-26. 
 
3. 
Carcinogenesis. The Journal of Nutrition 2002, 132 (8), 2413S-2418S. 
 
4. 
A Review of Molecular Mechanisms and the Evidence for Folate’s Role. Advances in 
Nutrition 2012, 3 (1), 21-38. 
 
5. 
Updates 2000, 3 (3), 155-160. 
 
6. 
Activity and Resistance. Antimicrobial Agents and Chemotherapy 1999, 43 (4), 727-737. 
 
7. 
Molecular Biology, and Epidemiology of Bacterial Resistance. Microbiology and 
Molecular Biology Reviews 2001, 65 (2), 232-260. 
 
8. 
729. 
 
9. 
Lincosamides and Streptogramin B Reveals the Nascent Peptide Exit Path in the 
Ribosome. Journal of Molecular Biology 2003, 330 (5), 1005-1014. 
 
10.  Marne, G.; Alexander, S. M., Macrolide Antibiotics: Binding Site, Mechanism of 
Action, Resistance. Current Topics in Medicinal Chemistry 2003, 3 (9), 949-960. 
 
11.  Aldred, K. J.; Kerns, R. J.; Osheroff, N., Mechanism of Quinolone Action and 
Resistance. Biochemistry 2014, 53 (10), 1565-1574. 
 
12. 
Drug Resistance (XDR), and Multidrug Resistance (MDR) among Gram-Negative 
Bacilli: Need for International Harmonization in Terminology. Clinical Infectious 
Diseases 2008, 46 (7), 1121-1122. 
 

Tenson, T.; Lovmar, M.; Ehrenberg, M., The Mechanism of Action of Macrolides, 

Falagas, M. E.; Karageorgopoulos, D. E., Pandrug Resistance (PDR), Extensive 

Smilack, J. D., The Tetracyclines. Mayo Clinic Proceedings 1999, 74 (7), 727-

 

187 

Taubes, G., The Bacteria Fight Back. Science 2008, 321 (5887), 356-361. 

Levy, S. B.; Marshall, B., Antibacterial resistance worldwide: causes, challenges 

13.  Culyba, M. J.; Mo, C. Y.; Kohli, R. M., Targets for Combating the Evolution of 
Acquired Antibiotic Resistance. Biochemistry 2015, 54 (23), 3573-3582. 
 
14.  Castanon, J. I. R., History of the Use of Antibiotic as Growth Promoters in 
European Poultry Feeds. Poultry Science 2007, 86 (11), 2466-2471. 
 
15. 
and responses. Nat Med 2004. 
 
16.  Butaye, P.; Devriese, L. A.; Haesebrouck, F., Antimicrobial Growth Promoters 
Used in Animal Feed: Effects of Less Well Known Antibiotics on Gram-Positive 
Bacteria. Clinical Microbiology Reviews 2003, 16 (2), 175-188. 
 
17.  Antibiotic Resistance Threats in the United States, 2013. CDC, Ed. CDC: Atlanta, 
2013. 
18.  Sukkar, E., Why are there so few antibiotics in the reserach and development 
pipeline? The Pharmaceutical Journal 2013, 291, 520. 
 
19. 
 
20.  DiMasi, J. A.; Feldman, L.; Seckler, A.; Wilson, A., Trends in risks associated 
with new drug development: success rates for investigational drugs. Clinical 
pharmacology and therapeutics 2010, 87 (3), 272-7. 
 
21.  Heller, A. A.; Lockwood, S. Y.; Janes, T. M.; Spence, D. M., Technologies for 
Measuring Pharmacokinetic Profiles. Annual Review of Analytical Chemistry 2018, 11 
(1), 79-100. 
 
22.  Hay, M.; Thomas, D. W.; Craighead, J. L.; Economides, C.; Rosenthal, J., 
Clinical development success rates for investigational drugs. Nature biotechnology 
2014, 32 (1), 40. 
 
23. 
Jorgensen, J. H.; Ferraro, M. J., Antimicrobial susceptibility testing: a review of 
general principles and contemporary practices. Clinical infectious diseases : an official 
publication of the Infectious Diseases Society of America 2009, 49 (11), 1749-55. 
 
24.  Mempin, R.; Tran, H.; Chen, C.; Gong, H.; Kim Ho, K.; Lu, S., Release of 
extracellular ATP by bacteria during growth. BMC Microbiology 2013, 13, 301-301. 
 
25.  Heller, A. A.; Spence, D. M., A rapid method for post-antibiotic bacterial 
susceptibility testing. PLOS ONE 2019, 14 (1), e0210534. 
 
26.  Pinger, C. W.; Heller, A. A.; Spence, D. M., A Printed Equilibrium Dialysis Device 
with Integrated Membranes for Improved Binding Affinity Measurements. Analytical 
Chemistry 2017, 89 (14), 7302-7306. 
 

 

188 

Lockwood, S. Y. Utilizing Fluidic Platforms for the Development of In Vitro 

27.  Castiaux, A. D.; Pinger, C. W.; Spence, D. M., Ultrafiltration binding analyses of 
glycated albumin with a 3D-printed syringe attachment. Analytical and bioanalytical 
chemistry 2018, 410 (29), 7565-7573. 
 
28. 
Lockwood, S. Y.; Meisel, J. E.; Monsma, F. J.; Spence, D. M., A Diffusion-Based 
and Dynamic 3D-Printed Device That Enables Parallel in Vitro Pharmacokinetic Profiling 
of Molecules. Analytical Chemistry 2016, 88 (3), 1864-1870. 
 
29. 
Pharmacokinetic/Pharmacodynamic Models. Ph.D. Disseration, Michigan State 
University, East Lansing, MI, 2014. 
 
30.  Vrany, J. D.; Stewart, P. S.; Suci, P. A., Comparison of recalcitrance to 
ciprofloxacin and levofloxacin exhibited by Pseudomonas aeruginosa bofilms displaying 
rapid-transport characteristics. Antimicrobial agents and chemotherapy 1997, 41 (6), 
1352-1358. 
 
31.  Pankey, G. A.; Sabath, L. D., Clinical Relevance of Bacteriostatic versus 
Bactericidal Mechanisms of Action in the Treatment of Gram-Positive Bacterial 
Infections. Clinical Infectious Diseases 2004, 38 (6), 864-870. 
 
32.  Hodiamont, C. J.; Janssen, J. M.; de Jong, M. D.; Mathôt, R. A.; Juffermans, N. 
P.; van Hest, R. M., Therapeutic drug monitoring of gentamicin peak concentrations in 
critically ill patients. Therapeutic drug monitoring 2017, 39 (5), 522-530. 
 
33.  Kiang, T. K. L.; Häfeli, U. O.; Ensom, M. H. H., A Comprehensive Review on the 
Pharmacokinetics of Antibiotics in Interstitial Fluid Spaces in Humans: Implications on 
Dosing and Clinical Pharmacokinetic Monitoring. Clinical Pharmacokinetics 2014, 53 
(8), 695-730. 
 
34.  Unger, R. E.; Huang, Q.; Peters, K.; Protzer, D.; Paul, D.; Kirkpatrick, C. J., 
Growth of human cells on polyethersulfone (PES) hollow fiber membranes. Biomaterials 
2005, 26 (14), 1877-1884. 
 
35.  Unger, R. E.; Peters, K.; Huang, Q.; Funk, A.; Paul, D.; Kirkpatrick, C. J., 
Vascularization and gene regulation of human endothelial cells growing on porous 
polyethersulfone (PES) hollow fiber membranes. Biomaterials 2005, 26 (17), 3461-
3469. 
 
36.  Azadbakht, M.; Madaeni, S. S.; Sahebjamee, F., Biocompatibility of 
polyethersulfone membranes for cell culture systems. Engineering in Life Sciences 
2011, 11 (6), 629-635. 
 
37.  Rautio, J.; Kumpulainen, H.; Heimbach, T.; Oliyai, R.; Oh, D.; Järvinen, T.; 
Savolainen, J., Prodrugs: design and clinical applications. Nature Reviews Drug 
Discovery 2008, 7, 255. 

 

189 

Liu, Y.; Chen, C.; Summers, S.; Medawala, W.; Spence, D. M., C-peptide and 

38.  Busher, J. T., Serum Albumin and Globulin. In Clinical Methods: The History, 
Physical, and Laboratory Examinations, rd; Walker, H. K.; Hall, W. D.; Hurst, J. W., Eds. 
Butterworths 
Butterworth Publishers, a division of Reed Publishing.: Boston, 1990. 
 
39. 
zinc delivery to erythrocytes requires the presence of albumin: implications in diabetes 
explored with a 3D-printed fluidic device. Integrative biology : quantitative biosciences 
from nano to macro 2015, 7 (5), 534-43. 
 
40.  Seedher, N.; Agarwal, P., Complexation of fluoroquinolone antibiotics with 
human serum albumin: A fluorescence quenching study. Journal of Luminescence 
2010, 130 (10), 1841-1848. 
 
41.  Chen, J.; Zhou, X.; Zhang, Y.; Gao, H., Potential toxicity of sulfanilamide 
antibiotic: Binding of sulfamethazine to human serum albumin. Science of The Total 
Environment 2012, 432, 269-274. 
 
42.  Chi, Z.; Liu, R., Phenotypic Characterization of the Binding of Tetracycline to 
Human Serum Albumin. Biomacromolecules 2011, 12 (1), 203-209. 
 

 
 

 

190