molecular engineering of selective recognition elements as coatings for sensor platforms by
TRANSCRIPT
Molecular Engineering of Selective Recognition Elements as Coatings for Sensor
Platforms
by
Justyn Wayne Jaworski
B.S. (Boston University) 2004
A dissertation submitted in partial satisfaction of the
requirements for the degree of
Joint Doctor of Philosophy
with the University of California, San Francisco
in
Bioengineering
in the
GRADUATE DIVISION
of the
UNIVERSITY OF CALIFORNIA, BERKELEY
Committee in charge:
Professor Seung-Wuk Lee, co-Chair
Professor Arun Majumdar, co-Chair
Professor Tejal Desai
Professor Ting Xu
Spring 2009
The dissertation of Justyn Wayne Jaworski is approved:
Prof. Seung-Wuk Lee, co-Chair Date
Prof. Arun Majumdar, co-Chair Date
Prof. Tejal Desai Date
Prof. Ting Xu Date
University of California, Berkeley
Spring 2009
Molecular Engineering of Selective Recognition Elements as Coatings for Sensor
Platforms
Copyright 2009
by
Justyn Wayne Jaworski
1
Abstract
Molecular Engineering of Selective Recognition Elements as Coatings for Sensor
Platforms
by
Justyn Wayne Jaworski
Joint Doctor of Philosophy
with the University of California, San Francisco
in
Bioengineering
in the
GRADUATE DIVISION
of the
UNIVERSITY OF CALIFORNIA, BERKELEY
Professor Seung-Wuk Lee, co-Chair
Professor Arun Majumdar, co-Chair
This dissertation focuses on the aspects of selectivity in chemical sensing systems. While
a number of sensing platforms exist that are capable of highly sensitive detection, the
common factor of poor selectivity continues to limit their widespread use. In this work,
we explore the use of sequence specific biopolymers identified through combinatorial
screening approaches for the creation of molecular recognition elements for chemical
sensor coatings. Particularly, a library of bacteriophage was screened to identify which
2
of the unique peptide sequences present on their protein coat could provide the highest
affinity binding to a target chemical. We specifically targeted small molecules including
trinitrotoluene (TNT) and dinitrotoluene (DNT). From phage display experiments, we
identify consensus peptide motifs, and we analyzed their binding efficacy based on
affinity and specificity. Additionally, we demonstrate that the standalone receptor for
TNT could be incorporated into a polymeric coating while retaining its functionality. In
doing so, a peptide based sensor coating was developed and implemented onto a common
Quartz Crystal Microbalance sensing platform. Liquid phase experiments demonstrated
the sensing ability of this system selectivity respond to TNT while remaining relatively
inert to the analogue DNT molecule. Furthermore, a polymeric based sensing system
was developed with the TNT receptive motif to create a widely deployable sensing
system. Integration was simply a matter of coupling a chromic responsive polymer at the
final step of receptor synthesis. In doing so, a modular sensing system was created which
demonstrated target binding to small molecules, such as TNT, or large cells, such as
fibroblasts, depending on the surface receptor motif. Finally, we show that the
fabrication approach could be optimized to enhance the sensitivity of the system to small
molecule targets.
Our results demonstrate that short amino acid sequences can be identified through phage
screening for small molecule binding and further developed into a sensor coating. The
receptors may be implemented onto a common QCM based sensor or onto a newly
develop chromic responsive system, thus demonstrating the broad sensor integration
capabilities of these receptive motifs. We anticipate this approach may lead to furthering
3
the development of molecular recognition elements by utilizing the biological toolkit of
evolutionary screening for selective receptors. In the future, we hope such approaches
will be used to gain a mechanistic understanding of molecular recognition which would
have a profound impact on the chemical sensing community.
_________________________
Co-Chair, Dissertation Committee
_________________________
Co-Chair, Dissertation Committee
i
Dedicated to Sonja Jaworski and Hilda Todd
ii
Table of Contents
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Chapter 1: Introduction
1.1 Challenges and Current Approaches in Chemical Sensing . . . . . . . . . . . . . . . . . . . . . 1
1.2 Molecular Recognition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Approaches to Achieving Molecular Recognition. . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Challenges and Future Trends in Chemical Sensor Coatings. . . . . . . . . . . . . . . . . . .19
Chapter 2: Phage Screening of Small Molecule Targets for Selectivity Motifs
2.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20
2.2 Phage Display for Selection of TNT and DNT Binding Peptide Motifs . . . . . . . . . .21
2.3 Phage Display for Selection of Methyl Parathion Binding Peptide Motifs. . . . . . . . 27
2.4 Phage Display for Selection of Eugenol Binding Peptide Motifs. . . . . . . . . . . . . . . 30
2.5 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Chapter 3: Analysis Techniques for Identifying Selective Binding Motifs
3.1 Selective Binding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Synthesis of Standalone Receptors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45
3.3 Mutational Analysis Binding Assay. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .46
3.4 Isothermal Titration Calorimetry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.5 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
iii
Chapter 4: Selective Coatings for Chemical Sensing
4.1 Introduction to Chemical Sensing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .53
4.2 Overview of a Peptide Based Sensor Coatings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3 Coating Design for Surface Stress Based Sensors. . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4 Coating Design for Quartz Crystal Microbalance Sensing Platform. . . . . . . . . . . . . 62
4.5 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Chapter 5: Development and Optimization of a Polymeric Sensing Vesicle
5.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .85
5.2 Overview of Polydiacetylene as a Sensing Platform. . . . . . . . . . . . . . . . . . . . . . . . . 86
5.3 Experimental Section. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.4 Results and Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .96
5.5 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Chapter 6: Summary and Outlook
6.1 Molecular Recognition Elements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.2 Sensor Coatings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6.3 Sensing Platforms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .114
Bibliography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
iv
List of Figures
Figure-1: a) Typical synthetic polymers consist of a monomer repeated multiple times.
Target analyte molecules generally have one binding site with such polymers. b)
Sequence-specific polymers have different residues strung together in a single chain.
Depending on the sequence, target molecules can form multiple binding sites with the
polymer, which leads to higher free energy of binding and thereby higher selectivity
against background interfering molecules.
Figure-2: Schematic of molecular imprinting approach to creating target selective
recognition motifs.
Figure-3: Overview of phage display screening process against molecular crystal TNT.
Figure-4: Schematic of SELEX screening process for identifying RNA or DNA based
aptamers for target specific binding.
Figure-5: Schematic diagram showing our biomimetic approach to develop selective
coatings for gas-phase explosive molecules. Identified molecular recognition elements
from the directed evolution process of phage display are used for their multivalent
recognition of explosive targets in liquid-phase screening. Through chemical
modification, the peptide receptors are linked with oligo(ethylene glycol) and
immobilized as coatings capable of binding explosive targets in air.
v
Figure-6: Phage display sequence results after the third and fourth rounds of phage
display screening against a TNT substrate.
Figure-7: Phage display sequence results after the third, fourth, and fifth rounds of phage
display screening against a DNT substrate.
Figure-8: Phage display screening results: a) converged amino acid sequences from the
fourth round of phage display screening with a noted percentage abundance obtained
from sequencing results, b) selectivity screening of the DNT receptor and TNT receptor
against DNT substrates and TNT substrates with the level of binding quantified from
phage titration. Nonspecific binding levels are identified by PS binding phages against
TNT and DNT substrates. All data are presented as the mean ± standard deviation.
Figure-9: Chemical structure of the pesticide methyl parathion, a small molecule
compound target used for phage display screening.
Figure-10: Phage display sequence results from experiment 1 after the third, fourth, and
fifth rounds of phage display screening against a methyl parathion substrate.
Figure-11: Phage display sequence results from experiment 2 after the third and fourth
rounds of phage display screening against a methyl parathion substrate.
vi
Figure-12: Overview of the method for immobilizing liquid target (eugenol) onto a solid
support (TentaGel beads) to allow a method for phage display screening of liquid targets.
Figure-13: Overview of the phage display screening process developed against a liquid
target immobilized onto a solid support. An initial screening against the solid support
alone is performed to remove phage that may bind to the support material.
Figure-14: Phage display sequence results from the fourth round of phage display
screening against eugenol, an immobilized liquid target.
Figure-15: Phage display sequence results from the fifth round of phage display
screening against eugenol.
Figure-16: Comparison of non-selective conventional polymer coatings to target specific
biopolymers.
Figure-17: The pIII receptors present on phage that were used for competitive screening
analysis. The number of appearances for a given receptor after competitive screening
(last column) were found from the DNA sequencing results.
Figure-18: a) Receptor sequences from phage primarily from TNT panning used in the
following specificity screening, and b) specificity screening results of phage against TNT
vii
and DNT substrate. Levels of binding represented as the ratio of phage remaining bound
after interaction and washing as compared to the initially exposed number of phage.
Figure-19: a) Receptor sequences from phage primarily from DNT panning used in the
following specificity screening, and b) specificity screening results of phage against TNT
and DNT substrate. Levels of binding represented as the ratio of phage remaining bound
after interaction and washing as compared to the initially exposed number of phage.
Figure-20: Phage display screening results of converged amino acid sequences from the
fourth round of phage display screening.
Figure-21: Selectivity screening of the DNT receptor and TNT receptor against DNT
substrates and TNT substrates with the level of binding quantified from phage titration.
Figure-22: Mutational analysis for identified TNT and DNT binding peptides. (a)
Sequence of receptors synthesized with C-terminal biotinylated lysine to allow
functionalization with the fluorescent probe Atto-425. TNT-BP and DNT-BP represent
the identified TNT and DNT binding receptor sequences, respectively, while H-Sub
(histidine substituted by alanine) and W-Sub (tryptophan substituted by alanine) are
mutations at amino acids 1 and 2 of TNT-BP. Scram-Ctrl is a nonspecific sequence
derived from scrambling the TNT-BP sequence to demonstrate the sequence importance
for encoding target selectivity and hence a negative control. (b) Fluorescence binding
assay against TNT substrate and DNT substrate, revealing the importance of tryptophan
viii
and histidine residues in TNT-BP as well as demonstrating the ability of the lone peptides
TNT-BP and DNT-BP to bind selectively to their targets in liquid when not associated
with the other phage body proteins. Fluorescence levels are normalized to the BSA
fluorescence background signal (P<0.0001, n=4). All data are presented as the mean ±
standard deviation.
Figure-23: Measurement of the dissociation constant of the complex between TNT and
the peptide TNT-BP determined from isothermal titration calorimetry. Data points
(values for the integrated heat change during each injection, normalized per mole of
TNT) are represented by filled squares. A one-site model was used to fit the data. The
solid red line is the calculated curve using the best-fit parameters.
Figure-24: Operation mechanism of micromembrane surface stress sensor.
Figure-25: Gas-phase screening for partition coefficients of various coatings on Si
exposed to DNT gas. The values are normalized to the DNT gas partition coefficient of
blank Si substrates to observe the contribution attributed solely to the coating layer.
Partition coefficients are calculated as the ratio of the concentration of analyte bound to
the coating (identified through thermal desorption GCMS) to the concentration of analyte
in the exposed gas headspace.
ix
Figure-26: Analysis of the level of TNT substrate binding for the designated TNT
binding peptide (TNT-BP) as well as a scramble TNT-BP sequence as control and a
trinitrobenzene binding sequence (TNB-Ctrl).
Figure-27: Selective gas-phase binding assay for DNT-specific coating: (a) schematic
diagram of the DNT binding peptide conjugated to oligo(ethylene glycol) and their
coating onto a gold surface for gas-phase selective binding; (b) partition coefficient of
DNT receptor coatings exposed to TNT gas and DNT gas. The values are normalized to
the target gas partition coefficient of OEG coating on a Au substrate to isolate the
contribution attributed to the DNT receptor element. Partition coefficients are identified
as the ratio of the concentration of analyte bound to the coating to the concentration of
analyte in the exposed gas headspace. The results are obtained through thermal
desorption GC-MS experiments on exposed coating surfaces (P < 0.001, n = 4). All data
are presented as the mean ± standard deviation.
Figure-28: Analysis of a hydrophilic and hydrophobic polymeric coatings using Quartz
Crystal Microbalance to identify the humidity response.
Figure-29: Chemical structure of the poly(ethylene-co-glycidyl methacrylate), called
'PEGM', which was found to have the lowest humidity response of the polymers tested.
Hence, it was chosen as the polymeric support for attachment of selective receptors and
coating onto the QCM.
x
Figure-30: Schematic of attachment mechanism for amine on receptor with epoxy grou
of poly(ethylene-co-glycidyl methacrylate).
Figure-31: XPS spectra of C 1s (first row), N 1s (second row) and S 2p (third row) peaks
of TNT receptor deposited on a bare Au slide before (column A) and after (column B)
washing in water overnight. Experimental spectra are plotted in solid line, fitted spectra
in dashed line, and fits for each peak component in dash-dotted line. The inset in the C
peak, column B, represents the C 1s spectrum recorded on a bare Au substrate.
Figure-32: XPS spectra of C 1s (first row), N 1s (second row) and S 2p (third row) peaks
of PEGM deposited on Au (sample 'PEGM/Au') (column A), and TNT receptor bound to
PEGM on Au (sample 'TNTrec/PEGM/Au') before (column B), and after (column C)
washing in water overnight. Experimental spectra are plotted in solid line, fitted spectra
in dashed line, and fits for each peak component in dash-dotted line.
Figure-33: XPS spectra of N 1s region for samples prepared by deposition of 660 nmol
of TNT on a) PEGM/Au and b) PEGM with TNT receptor. Experimental spectra are
plotted in solid line, fitted spectra in dashed line, and fits for each peak component in
dash-dotted line.
Figure-34: Percentage of TNT remaining on samples containing TNT receptor, DNT
receptor, and just PEGM after overnight washing in water. The values were calculated by
xi
dividing the normalized area underlying the peak at ∼409 eV relative to TNT after
washing the sample in water to the normalized area of the same peak measured before
washing the sample, and multiplying the result by 100, to obtain a percentage. Details
about the normalization of the peak relative to TNT and p values calculated according to
Student’s t test are reported in the text
Figure-35: GC/MS measurements of a) the amount of TNT remaining on samples
containing TNT receptor, DNT receptor, and just PEGM after overnight washing in
water and b) the amount of TNT and DNT remaining on samples containing the receptor
specific for TNT, after overnight washing in water.
Figure-36: Schematics showing the two modes of operation of the QCM setup.
Figure-37: Change in QCM resonance frequency measured on a crystal coated with
PEGM/TNT receptor, after exposure of a solution containing TNT a) and DNT b),
respectively.
Figure-38: Schematic diagram of synthesis, composition, and assembly parameters
which must be optimized to achieve an effective colorimetric PDA vesicle based sensor.
Figure-39: Dynamic light scattering measurements analysis was used to characterize the
diameter of the range of assembled PCDA-Trp-His-Trp particles to be on average 162 nm.
xii
Figure-40: Effect of PDA polymerization times were analyzed by a) varying exposure of
PDA vesicles to identify the conditions for achieving maximal chromic response (trend
line added as guide), and b) identifying the length of UV exposure required for full PDA
vesicle polymerization as indicated by irreversibility of the system.
Figure-41: Effect of PDA functional end-group on initial blue percentage for 100%
surface functionality of PCDA (carboxyl), PCDA-Trp, and PCDA-Trp-His-Trp
Figure-42: Dependence of initial blue percentage on the surface density of vesicles
comprising various concentrations of peptide-conjugate PCDA-Trp-His-Trp. The dashed
line represents the minimum blue percentage (42%) required to have a visually detectable
chromic response of 15% change relative to the blue percentage provided at 100%
surface density.
Figure-43: Visible-absorption spectra of polymerized vesicles containing: a) 4 mol %
Trp-His-Trp surface receptor; b) 2 mol % Trp and 2 mol % Trp-His-Trp surface
receptor; c) 4 mol % Trp and 4 mol % Trp-His-Trp surface receptor; d) 4 mol % Gly-
Arg-Gly-Asp-Ser surface receptor. Solid lines represent the non-exposed spectra while
dashed lines represent spectra attained after target exposure. Exposed spectra are
normalized to the corresponding non-exposed spectra and y-axes scaled to clarify the
change in chromic response.
xiii
Figure-44: Dependence of alkane chain length on PDA-Trp-His-Trp sensitivity to TNT
target. Decreasing PDA lengths facilitate a higher chromic response over a range of TNT
concentrations.
xiv
Acknowledgements
I have been extremely fortunate to have had the opportunity to work with two advisors.
Both Professor Arun Majumdar and Professor Seung-Wuk Lee have been extremely
encouraging and supportive, and I am highly grateful for their eagerness to take on
challenging projects in the exciting field of chemical sensing. I am also deeply indebted
to them for giving me the freedom to work on areas of my interest during my graduate
study. Their scientific advice and insight into future areas of importance have taught me
that science is not merely performed for its own sake, rather science and engineering
should be done on behalf of the public. This has left a lasting impression that I will bring
with me in my upcoming research work.
I also would like to thank the members of Professor Majumdar’s and Professor Lee’s
groups. I gained a lot from Dr. Woo-Jae Chung’s extensive experience in solid-phase
synthesis, and I highly appreciate his advice, technical expertise, and friendship. Dr. Ki-
Young Kwon has also offered terrific advice, not only in terms of research but also
advice on life and careers. I would especially like to thank Eddie Wang for his friendship,
keeping the laboratory a fun place to work, and always having a useful answer to my
many questions. Additionally, I would like to thank Jin Huh, with whom this research
began. I cannot thank him enough for all those long hours spent working on sequencing
and phage screening experiments. I would also like to thank Yang Guo for his time and
efforts in developing different phage display screening strategies for various targets.
Anna Merzlyak has been a great friend, and I really enjoyed her support and unity to
finish our PhD simultaneously. My professional interaction with other fellow lab
xv
members has been intellectually very satisfying and has had a significant influence on my
research work. A partial list of those members I’d like to thank includes: Masae
Kobayashi, Jonathan Phillips, Rohit Karnik, Pramod Reddy, Woochul Kim, Srinath
Satyanarayana, Si-Hyung Lim, Robert Wang, Tony Tong, Yang Zhao, Renkun Chen,
Suzanne Singer, Chuanhua Duan, Kaal Baheti, and Digvijay Raorane. I would especially
like to thank Keisuke Yokoyama and Chris Zueger for their collaboration on the chromic
responsive sensor work. Besides being great friends and members of the Sensing Team,
they have always provided terrific advice, and I appreciated their hard work in helping
create this exciting sensor platform. Additionally, Dr. Marta Cerruti has been a great
friend and motivator, and her expertise in surface chemistry has been very helpful in
making this project a success. My friendships with some of the greatest people in
Berkeley and San Francisco are unforgettable, and I am highly appreciative. Our
solidarity has helped ensure a healthy level of laboratory life. I also gratefully
acknowledge the funding agencies that made this research possible, namely, the Bill and
Melinda Gates Foundation, the National Science Foundation, the Office of Naval
Research, the Center of Integrated Nanomechanical Systems, and the Department of
Energy.
I would also like to thank Dr. Ron Zuckermann for providing me with project advice,
training, and access to equipment in the Molecular Foundry for my experiments.
Additionally, I would like to thank Professor Kevin Healy for his guidance in helping
form my project aims as my qualifying examination chairman. Additionally, I thank
Professor Song Li for his help in structuring my coursework as my curriculum advisor.
xvi
My research experience at Berkeley would have been quite different without the
wonderful research rotations I had with Professor Jay Keasling and Professor Luke Lee.
I highly appreciate their advice and guidance during those rotations and beyond. I also
want to thank Professor Tejal Desai and Professor Ting Xu for serving on my dissertation
committee and for their wisdom in helping me to prepare for this work as well as my
qualifying examination. Also, I highly appreciated their suggestions on project directions
and materials usage. Finally, I would like to thank my family, though words cannot
express the gratitude I have for their motivation and support of my education.
1
Chapter 1: Introduction
1.1 Challenges and Current Approaches in Chemical Sensing
The ability to detect and analyze volatile organic compounds (VOCs) such as
explosives,1 pesticides,
2 disease markers,
3 and food aroma
4 has significant bearing on our
security, health, and general well being. As opposed to mobility-based5 and optical
sensors,6 those based on ligand-receptor binding that emulate the olfactory system enable
miniaturization. While there are several highly sensitive ways to convert ligand-receptor
binding to electrical or optical signals,7-13
the lack of selectivity has remained a key
challenge, making such sensors inadequate in many applications and preventing their
widespread use. At issue are functional coatings, which in the past have relied on arrays
of oxide layers,14
readily available polymers, 15
or even de novo designed receptors.9, 16-18
Although such arrays provide a pattern of binding that can be mathematically processed
to distinguish molecular species, the affinity differences are often insufficient for highly
selective and sensitive chemical analysis in realistic conditions.
1.1.1 Introduction to Current Sensing Platforms
While the focus of this research lies in the selective coating layer, it is important to first
look at the various chemical sensing platforms available. This is necessary in order to
effectively create a sensor coating amenable to the various signal transduction approaches
used by these sensors. The sensitivity of the transducer to ligand-receptor binding events
2
depends heavily on which sensing platform is to be used. The transduction can occur in
many ways: (i) change in electrical resistance of a chemoresistor or a chemo-field effect
transistor (chemFET);19, 20
(ii) frequency shift in mechanical resonance devices such as
quartz crystal microbalances, surface acoustic wave (SAW) devices, cantilever beams;7,
21-23 (iii) frequency shift in optical resonances such as surface plasmon resonance, cavity
resonance,24, 25
(iv) structural deflections such as in cantilever beams and membranes.26, 27
The above chemical sensing systems are effective provided they have an adequate sensor
coating which will provide a physical change attributed to the binding event. For a
ligand-receptor interaction event (as in the case of our sensor coating), binding may result
in three fundamental changes: (i) changes in local dielectric constant; (ii) addition of
mass; (iii) generation of intermolecular forces.28
A device capable of detecting each of
these physical changes simultaneously could be broadly used in detection of the ligand-
receptor binding events. The micromembrane and microcantilever systems may provide
a means of identifying all three of these changes. A change in mass addition or even
surface stress (attributed to intermolecular forces) could produce changes in the resonant
frequency detectable by capacitance measurements via a micromembrane or
microcantilever based sensor. A capacitance change could also directly be a result of
dielectric changes attributed to target binding attributed to analyte replacement of water
in the capacitive sensor.
3
1.1.2 Introduction to Current Sensor Coatings
The creation of a chemical sensing device capable of selective detection of a target
chemical is a deeply sought after technology. The accurate detection of various
chemicals related to explosive threats would significantly enhance homeland security
measures through protection of civilians and military forces. Example target molecules
of interest include volatile and semi-volatile small molecules such as TNT
(trinitrotoluene) which typically have a molecular weight in the 50-500 g/mol range.
Current sensor technologies have the opportunity for highly sensitive detection;7
however, the lack of proper selectivity continues to be the main challenge in creating a
chemical detection system which is successful against a background of various
interfering agents.29
System-level performance of chemical sensing systems is often quantified in terms of
receiver operating characteristic curves in which an optimally sensitive sensor would
want to maximize selectivity (or minimize the occurrence of false positives). The
current gold standard for sensitive and selective systems is based on gas chromatography-
mass spectroscopy (GC/MS). Unfortunately, these machines are very large, expensive,
and require complex data analysis. Aside from this, many successful chemical sensors
for di-atomic and tri-atomic molecules, such as NOx and CO2, have demonstrated
selectivity.30-35
Though these chemicals are important, there remains a void for selective
sensors of volatile organic compounds in the 50g/mol – 500g/mol range due to the lack of
diversity available in sensor coatings.
4
The most promising candidates for portable systems are based on transduction of ligand-
receptor binding to an electrical signal.36
This style of device typically uses an array of
polymers as coatings to generate a binding signature pattern.37, 38
While an interesting
approach, these polymers are unfortunately non-specific in their binding,39
which is
detrimental in two ways: (i) the affinity ratios between target analytes are not enough to
distinguish molecules of sufficiently low concentrations; (ii) the binding of background
chemicals is too high causing creating uncertainty in the acquired data. As a result, such
polymer-based systems have poor selectivity. To address the selectivity challenge we
looked to biology, which relies on selective ligand-receptor interaction for complex
molecular networks and interactions. Particularly, we are interested in biology’s use of
sequence-specific heteropolymers which are rich in structure and chemical functionality
as compared to bulk homopolymer coatings.40
1.1.3 Sensor Coating Formation
Many applications of polymers and biopolymers as sensor coating utilize the formation of
monolayers on a sensing surface. Typically these consist of either drop-cast polymers or
the more stable self-assembled monolayer. Self-assembled monolayers, or SAMs, are
stable, highly ordered, single layers of molecules that organize spontaneously during
adsorption to a solid substrate.41-45
Molecules involved in SAM formation have a distinct
“head” with affinity for the solid substrate and a “tail” usually consisting of alkyl or
ethylene glycol chains.40
Covalent chemical bonding between the head group and the
5
surface often plays a major role to stabilize the monolayer structures.44
Systems forming
SAMs include the adsorption of alkylthiols onto gold, alkylamines onto platinum, and
alkylchlorosilanes onto silica.43, 44
The development of such a coating material that contains a high density of specific
receptors with intact binding efficacy is a critical step for the fabrication of specific and
sensitive sensing devices. Coating receptor immobilization and SAM formation are
typically formed by solution preparation methods, where a solid substrate is introduced
into a solution of solvent and molecules with a head group having affinity for the
substrate.41
Ellipsometry is generally used to determine layer thickness and surface
plasmon resonance is used to study their in situ formation.43, 46
Although more
information can be gained from the technique of x-ray photoelectron spectroscopy which
is employed to determine uniformity of coverage, the orientation and tilt of the
molecules, and extent of their order.43
Chemically and biologically selective coatings
have gained attention in a wide number of fields with application ranging from
microelectronic structures to biocompatible solid surface environments. SAMs in
particular are of general interest in surface science as model systems, as well ordered
monolayers may be engineered with different head-tail combinations to provide insight
into interfacial phenomena and the relationships between structure and function.43, 45
6
1.2 Molecular Recognition
The ability to recognize an individual molecule is a daunting task from a design
perspective, while Nature has proven to be an effective source for such interactions.
Biology is filled with examples of selective molecular recognition. Recognition is in fact
vital for most cellular level processes (protein-protein docking, protein-DNA interactions,
enzyme-substrate catalysis, and receptor-ligand binding to name a few). While the
importance of these binding events is vast, little has been developed in terms of
predictive mechanisms for biomolecular interactions. One case in which we have
achieved a working knowledge of biomolecular interactions lies in the formulaic Watson-
Crick base pairing of DNA. The advent of this discovery undoubtedly changed biology
forever. Unfortunately for protein-protein and protein-small molecule interactions, we
are not given a set of generic patterns such as hydrophobicity, charge pairing, and shape
that we can use directly to predict which atoms will be involved in a binding event. To
achieve such goals, researchers utilize spectroscopy and crystallization approaches to
resolve Angstrom level intramolecular interactions as well as intermolecular host/guest
interactions. By looking at these examples from nature, we one day hope to achieve a
mechanistic understanding of all molecular interactions.
The concept of molecular recognition requires a receptor (host) molecule to participate in
non-covalent bonding with a specific ligand (guest). Several coupling events can occur
between a given receptor-ligand pair including hydrogen bonding, electrostatic
interactions, aromatic interactions, or Van der Waals interactions. In a molecular
7
recognition event, it is typical for several combinations of these interactions to occur
simultaneously. This multi-valence binding is critical for receptors to discriminate
between different potential guest target molecules. This differentiation for a specific
guest is deemed the selectivity of a receptor. Several factors have been found to
contribute to a selective molecular recognition event including: complementary
molecular shape, appropriate structural rigidity, and appropriate available binding sites.
The importance of molecular shape rests in the need to maximize the number of potential
interaction sites for a given binding event. In having a complimentary shape, a proper
distance between functional groups can be achieved for effective hydrogen bonding and
aromatic interactions. In contrast, a steric hindrance may force a receptor-ligand pair to
rely solely on long-range electrostatic interactions. From such steric consideration, it is
has been made evident that selective molecular recognition is enhanced by the intrinsic
inaccessibility of larger, non-target molecules.47
Another critical attribute for a receptor is for it to maintain an appropriate level of
plasticity or structural rigidity. Host receptors are generally not loosely swinging
molecular ropes; rather receptors possess an adequately rigid structure which can
minimize entropic losses that will arise due to reorganization or freezing of rotating bond
which occurs upon binding with the designated target guest molecule. As such, many
supramolecular chemists have utilized cyclic structures as receptor templates including
crown ethers, cryptands, spherands, cavitands, calixarenes, and porphyrins.48-52
As an
exception to this, researchers have found a protein-DNA interaction which was deemed
8
the “fly-casting mechanism”.53
In this instance, long-range weak forces are used to bring
the protein in proximity to the DNA at which point folding occurs in concert with the
binding event. As such, this protein folding event allows distant folding interactions to
contribute to an increased stability of the overall final recognition event. With that said,
typical conformation changes of a peptide backbone are generally less that 1A,54
while
functional side-chain dynamics can be significantly altered and restricted upon binding
supporting the notion that complex formations are generally enthalpic in nature.55
Finally, selective recognition of target molecule requires a receptor to have the
appropriate functional groups available for binding to occur. Binding sites containing
acid functionalities are capable of interacting ionically with amine and amide functions
on a target molecule or through hydrogen bonding with a variety of polar functionalities
including carboxylic acids, carbamates and carboxylic esters.56
Aromatic groups offer
another set of interaction mechanisms including cation- π, anion- π , and π–π interactions,
These interactions are quite ubiquitous through biological molecular recognition. Other
important bonding mechanisms, including van der Waals forces, charge-transfer
interactions, and dipole-dipole interactions, may also be present in the binding site if
beneficial for selective recognition of the corresponding target molecule. The following
sections will present some current approaches being used by researchers to achieve
molecular recognition. In addition, we will take a closer look at some attempts to
incorporate molecular recognition into chemical sensing platforms and the associated
challenges.
9
1.3 Approaches to Achieving Molecular Recognition
1.3.1 Overview
Molecular recognition stems from specific binding attributed to complementary
intermolecular interactions defined by structure and chemical functionality.9, 50, 57, 58
In
biology, three-dimensional shapes and chemistry diversity arises from the massive
combinatorial possibility of nucleic acids and peptide sequences.59, 60
Our goal is to use
this diversity and emulate molecular recognition in the non-living world.
Figure 1: a) Typical synthetic polymers consist of a monomer repeated multiple
times. Target analyte molecules generally have one binding site with such polymers.
b) Sequence-specific polymers have different residues strung together in a single
chain. Depending on the sequence, target molecules can form multiple binding sites
with the polymer, which leads to higher free energy of binding and thereby higher
selectivity against background interfering molecules.
Sequence-specific heteropolymers are rich in structure and chemical functionality as
compared to bulk polymers.40
Natural biopolymers, like nucleic acids and polypeptides,
are classic examples of materials that can adopt complex, defined structures that present a
variety of different functional groups in a particular geometry (Figure-1). These folded
structures are capable of highly-selective molecular recognition and efficient catalysis of
a wide variety of chemical transformations.61, 62
While bulk polymers have many
10
sophisticated and useful properties (structural, mechanical, optical, electronic…), they are
not capable of the degree of molecular recognition and transformation as that of proteins.
The ability of proteins to bind a wide variety of substances with high-affinities (e.g. small
molecules, nucleic acids, peptides, other proteins, etc) has long been exploited to discover
novel receptors for ligands of interest.63-65
Large combinatorial libraries of peptides,
antibodies, RNA or DNA can be readily prepared and screened.66
Peptide libraries have
been a particularly fruitful source of molecular diversity, as they have provided numerous
functional peptides.67
Such libraries can be readily prepared by both chemical and
biological methods. Using a relatively short peptide libraries, where a sequence of <20
residues is randomized, offers a significant advantage, in that that the discovered
receptors can be readily synthesized on a large scale for coupling with other materials or
incorporation into devices. In order to improve the specificity of the sensory molecules,
our approach is to exploit established chemical and biological toolkits to develop specific
recognition elements.
It is widely accepted that molecular recognition is essential to biological signal
transduction, enzymatic processes, gene expression, and a number of other cellular level
processes which are vital for sustaining life. Using structural information of existing
receptor-ligand docking events, researchers have been able to identify steric and
electronic conformations of selective binding events. In addition, replacement of
functional groups through mutational analysis has also allowed key bonding components
of an interaction to be identified along with their contribution to binding affinity. In
11
general, thermodynamic and kinetic studies can provide a measure of substrate affinity
from association constants with which specificity can be determined. In order to better
control the selectivity of biological systems and to exploit these mechanisms for other
purposes (such as chemical sensing), researchers have attempted to devise ways of
creating receptors capable of selective molecular recognition. Herein we will discuss the
approaches to achieving molecular recognition through molecular imprinting,
supramolecular chemistry, protein engineering, and finally screening technologies
including phage display. Phage display will be discussed in depth, as this technology was
utilized in this study for identification of small molecule receptors.
1.3.2 Molecular Imprinting
The concept of molecular imprinting (depicted in Figure-2) utilizes a target molecule of
interest to create a template within a polymeric system which once removed will leave a
cavity (or binding site) available for selectively binding identical target molecules. To
create such binding sites, monomers possessing particular functional groups (usually
methacrylates) are mixed in solutions containing the target “template” molecules of
interest. A cross-linking agent is also added to the mixture in order to “cast” the
functional monomers around the template molecules. The target molecule is then
disrupted or washed from the polymeric film to leave an empty template site available for
future binding to occur.
12
Figure 2: Schematic of molecular imprinting approach to creating target selective
recognition motifs.
One such case of molecular recognition via molecular imprinting utilized L-tryptophan as
a template target molecule to demonstrate enantiomer selectivity against D-tryptophan.68
In this study, L-tryptophan was cast into a polymer film using acrylamide functional
monomers and various concentrations of cross-linking agent (1,1,1-Trimethylolpropane
trimethacrylate). The researchers implement the polymeric film onto a quartz crystal
microbalance based sensor to demonstrate preferential interaction for L-Tryptophan over
D-tryptophan. In fact, the sensor was capable of a detection limit of 80uM. It should be
noted, that D-tryptophan did produce a signal, though it was considerably less than that
for the target L-tryptophan. Interestingly, the researchers found that increasing the cross-
linking density resulted in enhanced selectivity to a certain point, which is believe to be
due to the stabilization of the polymer binding pocket.
While the concept of molecular imprinting is very alluring, one should be aware of its
limitations. Primarily, the incidence of non-specific interactions is still a concern as the
13
site formation is polyclonal. Since the polymer film may have various different template
sites which possess various confirmations and hence affinities, the polymer film cannot
achieve purely selective molecular recognition. In addition, cross-linked polymer
systems can easily undergo swelling or shrinkage depending on the environment. In
some instance, temperature and humidity can drastically deform imprints thereby
rendering them impractical. The extent of cross-linking can easily impact the percentage
of functioning imprints on the polymeric film. Extensive cross-linking makes many
imprint sites inaccessible, while insufficient cross-linking results in unconstrained
templates which can result in altered template morphology. Despite the short-comings of
the approach, it has provided a significant advance toward selectivity in chemical
sensing.
1.3.3 Supramolecular chemistry
Using computational and theoretical means, supramolecular chemistry has been used to
design receptors to be tailor made recognition elements for specific target molecules. By
engineering the correct curvature and functionality to fill the space complementary to the
given target molecule, research have demonstrated high affinity target binding. One such
receptor was designed to have a concave binding site with the right size and shape for
choline.52
It is expected that strong cation-π interaction are occurring, as the researchers
have demonstrated the ability to preferentially bind choline (Kd of 83 uM) over
acetylcholine (Kd of 250 uM).
14
Cavitands, similar to the one mentioned above for choline binding, have the ability to
bind to aliphatic and aromatic guests in both gas and liquid.69, 70
As such, several
research groups have implemented these supramolecular based receptors, including
cavitands, crown-ethers, and metalloporphyrins, onto chemical sensing platforms.16, 18, 52
A particular example of a sensor that utilizes supramolecular chemistry takes the form of
a quartz-crystal microbalance coated with cavitands which are responsive to organic
molecules in both liquid and vapor phase.71
The sensor, which was intended to be
selective, actually showed a response to multiple different targets. While a preference for
aromatic and chlorinated hydrocarbons was demonstrated, this demonstrates the difficulty
in achieving highly specific molecular recognition. The research did successfully
demonstrate that different cavity organizations and shapes elicit different response
behavior. This indicates that it should be possible to specifically control the selectivity
based on the supramolecular design.
1.3.4 Protein engineering
Perhaps, the most successful approaches to achieving molecular recognition have been
through the use of protein engineering. In protein engineering, molecular recognition is
achieved by either borrowing existing molecular recognition domains from nature or
modifying binding/catalytic sites for altered specificity. To facilitate this, researchers
have utilized a combination of a computational approach, in which receptive domains are
analyzed in silico with extensive modeling of interacting domains,17, 72
and a rapid
evolutionary approach, generally DNA shuffling and site directed mutagenesis.73
While
15
many mutagenesis strategies exist, it has become a consensus that the optimal mutation
rate is dependent on the methodology of the mutagenesis as well as the organism and
protein being modified.74
Using random mutagenesis, researchers group have found it possible to engineer proteins
for selective molecular recognition. A particularly interesting example of this involved
the directed evolution of G protein coupled receptors (GPCR) to be activated by a pre-
designated target molecule.75
Initially, a library of chimeric GPCRs were created, which
were made insensitive to acetylcholine by replacement and random mutagenesis of the
target recognition site. The GPCR library was cloned into yeast and designed such that
activation would result in histidine production. This was important for screening as the
yeast were then grown in His deficient media to screen for activation. Screening was
performed in the presence of clozapine-N-oxide which is a pharmacologically inert small
molecule. By identifying the surviving yeast clones, the researchers were able to isolate
GPCR which were activated by clozapine-N-oxide. The sequences were recovered and
underwent additional mutagenesis and screening at lower clozapine-N-oxide
concentration to enrich high affinity target binders. Through this strategy the group was
able to achieve selective molecular recognition via GPCRs with activation at less than
5nM of clozapine-N-oxide while remaining insensitive to acetylcholine. This reflects a
significant advance for molecular recognition and helps solidify protein engineering as a
viable approach to re-engineering receptor protein selectivity.
16
1.3.5 Phage Display and SELEX Screening technologies
Genetic engineering of phage provides a rapid tool to identify specific binding peptide
motifs against various substrates including proteins, cell surfaces, and even inorganic
semiconductor or metallic surfaces. By mimicking the evolution process in nature, the
phage is used as an information mining tool to identify specific peptide sequences
capable of recognizing desired materials at the molecular level. Phage display is a
combinatorial process to identify specific binding peptides which utilizes phage and
bacterial biology through a fast, directed-evolution method.76
M13 bacteriophage, a
bacterial virus, is comprised of a single stranded DNA encapsulated by various major and
minor coat proteins. It has a long-rod filament shape that is approximately 880 nm in
length and 6.6 nm in width. The viral peptide capsid is composed of 2,700 copies of
helically arranged major coat protein, known as pVIII, and 5-7 copies of other proteins
(pIII, pVI, pIX and pVII) located at either end of the phage. All of the above proteins can
be genetically modified to display short (8-12 amino acids) peptide sequences at various
locations on the phage. By inserting randomized DNA sequences into specific locations
of the phage genome, a highly diverse library of peptides (up to 109 random sequences)
can be displayed on the viral particles. To select the best binding peptide sequence for a
given target material, the engineered phage library pool goes through several rounds of
selection processes. This is depicted in the following figure with the case of a TNT
(trinitrotoluene) crystal target.
17
Figure 3: Overview of phage display screening process against a molecular crystal
target of TNT.
Initially, the phages are allowed to bind to the target. The non-bound phages are then
washed away, while the remaining (stronger binding) phages are eluted, captured, and
amplified through infection of an E. coli bacterial host. This evolutionary approach to
select the fittest binding peptide sequence is repeated several times with consecutively
more rigorous binding conditions to enrich the phage with the best affinity peptide for the
target. Finally the dominant binding peptides are identified by DNA analysis of the phage
genome.
Traditionally this technique has been used to identify small antibodies and study protein
interaction.77
Though, recently phage display has been utilized to identify peptide
sequences with affinity for a variety of inorganic substances, such as semiconducting,
18
magnetic, and metallic materials.78, 79
Other screening technologies include SELEX,
Systematic Evolution of Ligands by Exponential Enrichment, which utilizes RNA or
DNA libraries. Commonly, the libraries consist of RNA in a stem-loop structure with the
loop region comprising the variable region.80
Similar to the phage display screening
approach described above, a series of target binding, washing, and enrichments steps are
carried out until a consensus sequence is attained which binds well to the target of
interest. Generally, the binding mechanisms for these recognition elements take one of
the following forms: 1) a previously unstructured aptamer folds to conform to a stable
structure around the target; 2) a large stable secondary RNA structure acts as a docking
site for target binding.
Figure 4: Schematic of SELEX screening process for identifying RNA or DNA
based aptamers for target specific binding.
One such RNA aptamer, screened to bind to the target Vitamin B-12, was found to have a
dissociation constant of approximately 90nM.81
The strong aptamer-target recognition
19
was analyzed by NMR to show that indeed steric complementarities, hydrophobic
packing, and electrostatic interactions were essential for binding. Even though aptamer
screening technology can provide high-affinity binders, the major limitation of such
nucleic acid based recognition elements for chemical sensing platforms remains in their
instability as a coating material. Primarily, RNA and DNA are easily degraded which
unfortunately prevents their use as a practical sensor coating.
1.4 Challenges and Future Trends in Chemical Sensor Coatings
Molecular detection is an important research topic in such areas as environmental
analysis (detection of pollutants),82
food industry (food spoilage and quality tests),83
and
security (detection of explosives).1 While most sensors presented in the literature are
highly sensitive, often their selectivity is insufficient for performance in the field. In
practical applications selectivity is critical for successful detection, where the molecule of
interest exists in a complex environment together with many other species that can create
false positive signals. This is primarily attributed to the lack of diversity available for
sensor coatings specific for these targets,84
as most current coating technologies rely on
weak or non-specific molecular interaction.10, 11, 14, 15, 85-88
20
Chapter 2: Phage Screening of Small Molecule Targets for Selectivity Motifs
2.1 Introduction
Molecular biology is replete with examples where weak interactions based on hydrogen
bonding or van der Waals interactions can lead to highly specific molecular recognition
through multivalent or cooperative binding. The diversity of chemistry and structure
produced by sequence-specific biopolymers such as nucleic acids and peptides allows
specific multivalent receptors to be created for a wide range of target ligands. The
immune system, for example, utilizes this to create specific protein-based receptors
against a large variety of foreign antigens. The use of sequence-specific heteropolymers
and multivalent binding has, however, not been thoroughly explored for the detection of
VOCs. While oligomers of DNA and RNA have been used recently,89
the multivalency
of their binding against VOCs has remained undetermined. Moreover, the diverse
chemistry of amino acids suggests that peptides are better suited as receptors against a
wide range of target molecules. The challenge, however, is to identify the specific N-mer
amino acid sequence that provides the strongest multivalent binding among 20N possible
sequences. We utilized the combinatorial screening power of phage display, an
information mining tool, for the identification of peptides that can specifically recognize
a desired target material at the molecular level (Figure-5). Evolutionary screening
processes have been previously used in liquid environments for diverse material targets
including semiconductors,78, 79
metals,90-92
and proteins.77, 93
21
Figure 5: Schematic diagram showing our biomimetic approach to develop selective
coatings for gas-phase explosive molecules. Identified molecular recognition
elements from the directed evolution process of phage display are used for their
multivalent recognition of explosive targets in liquid-phase screening. Through
chemical modification, the peptide receptors are linked with oligo(ethylene glycol)
and immobilized as coatings capable of binding explosive targets in air.
2.2 Phage Display for Selection of TNT and DNT Binding Peptide Motifs
Utilizing phage display we pursued the selection of peptides which could selectively bind
to the explosive trinitrotoluene (TNT). Additional phage display experiments were
performed to identify selective molecular recognition elements for dinitrotoluene.
Receptor screening utilized a phage library which possessed 3.9x109 different peptides,
composed of both linear 12mer (Ph.D.™-12) and constrained 7mer (Ph.D.™-C7C)
random amino acids, fused to the pIII coat proteins of M13 phages. The input phage
library solution was prepared by adding 10 µL of each library to 1mL of 0.1% TBST
buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.1% [v/v] Tween-20). In order to
eliminate substrate heterogeneity, we screened the library against the molecular crystal
22
form of TNT and DNT using standard phage display for peptide selection.78, 79
Phage
display was performed with the initial binding condition of 0.1% TBST for 30 minutes at
room temperature on a rocking platform with 5 mg of target, TNT or DNT crystals.
Following binding, a series of 10 wash steps were performed using the same binding
buffer to remove non-specific binders. Specific binding phages were eluted with 1mL of
0.2 M Glycine-HCl [pH 2.2], 1 mg/ml BSA. The progressive screening rounds utilized
increasing surfactant concentration to increase stringency of binding to TNT or DNT
targets. Screening results were obtained through sequence analysis of the receptor region
following each screening round.
23
Figure 6: Phage display sequence results after the third and fourth rounds of phage
display screening against a TNT substrate.
24
The sequences seen in Figure-6 represent the phages sequenced after the third and fourth
rounds of phage display screening against a TNT substrate under a binding condition of
pH 7.5. The sequence ID represents “Substrate - Binding Conditions - Round/Experiment
- Phage Sample”. The abundance listed on the right reflects the number of times a given
receptor sequence appeared from the sequencing results. After each round, approximately
20 phage samples were submitted for sequencing.
25
Figure 7: Phage display sequence results after the third, fourth, and fifth rounds of
phage display screening against a DNT substrate.
The sequences seen in Figure-7 represent the phages sequenced after the third, fourth,
and fifth rounds of phage display screening against a DNT substrate under a binding
condition of pH 7.5. The sequence ID represents “Substrate - Binding Conditions -
Round/Experiment - Phage Sample”. The abundance listed on the right reflects the
26
number of times a given receptor sequence appeared from the sequencing results. After
each round, approximately 20 phage samples were submitted for sequencing.
The phage display results from third and fourth round screenings against molecular TNT
crystals are represented in the Figure-6. After four rounds of TNT screening, we
identified TNT binding sequences with the consensus motif of Trp-His-Trp-X (X:
hydroxylated, amine, or positively-charged side chain) at the N terminus of the receptor.
Similarly, the resulting phage display sequences from third, fourth, and fifth round
screenings against the molecular DNT crystal target can be seen in Supplementary
Figure-7. Through these experiments, we arrived at the consensus DNT binding motifs
and TNT binding motifs depicted in Figure 8a.
27
Figure 8: Phage display screening results: a) converged amino acid sequences
from the fourth round of phage display screening with a noted percentage
abundance obtained from sequencing results, b) selectivity screening of the DNT
receptor and TNT receptor against DNT substrates and TNT substrates with the
level of binding quantified from phage titration. Nonspecific binding levels are
identified by PS binding phages against TNT and DNT substrates. All data are
presented as the mean ± standard deviation.
2.3 Phage Display for Selection of Methyl Parathion Binding Peptide Motifs
In addition to explosive molecules, we aimed to utilize this phage display based selection
strategy for other small molecule targets of interest. Particularly, we were interested in
using this system for identifying receptors to organophosphates. Organophosphate
28
detection holds significant interest in terms of environmental safety, as this class of
chemicals constitutes pesticides and chemical warfare agents.94
To identify a receptor for
a particular organophosphate, we pursued the screening of methyl parathion, seen in
Figure-9.
Figure 9: Chemical structure of the pesticide methyl parathion, a small molecule
compound target used for phage display screening.
Methyl parathion is a small molecule compound used extensively as an agricultural
chemical to control the presence of insects. Methyl parathion has show to be extremely
hazardous to human health. Exposure has shown to cause disorders in the reproductive
and cardiovascular system, and its role as an acetylcholinesterase inhibitor leads to
neurological degeneration.95
Unfortunately, this highly toxic chemical often becomes
present in nearby streams and water sources as a result of run-off.96
As such, this
molecule is an attractive candidate for chemical sensing applications. Detection of
methyl parathion presence, may aid in preventing its inadvertent consumption, thereby
preventing the occurrence of a number of associated physiological disorders. Previously,
researchers have attempted to use quartz crystal resonator based sensing devices in
combination with antibodies for methyl parathion.97
A number of other research groups
have utilize the enzyme, Organophosphorus Hydrolase, as the selectivity element on
potentiometric sensing platform,98
optical based biosensors,99
and amperometric
biosensors.100
Here in, we pursued an improved selective coating to the large protein or
29
antibody for methyl parathion by using our evolutionary screening approach. By
identifying a shorter peptide sequence capable of binding to methyl parathion, we hoped
to maintain the selectivity of the antibody or enzyme while incorporating the high
stability of small peptides onto a sensor coating.
Figure 10: Phage display sequence results from experiment 1 after the third, fourth,
and fifth rounds of phage display screening against a methyl parathion substrate.
Using the same screening approach described in the previous section for determining
explosive binding receptors, we performed phage display to identify homologous leads as
potential receptors for methyl parathion.
30
Figure 11: Phage display sequence results from experiment 2 after the third and
fourth rounds of phage display screening against a methyl parathion substrate.
Notice that the Asn-Ile-Ile-Thr-Thr-Gln-Asn-Trp-Trp-Lys-Gln-Thr sequence appeared as
the most abundant in both experiments. Researchers have identified important amino
acid residues involved in the binding of methyl parathion, by identifying the crystal
structure of methyl parathion hydrolase.96
Looking at the binding site for methyl
parathion in this enzymes, the researchers found that the binding pockets are lined by
hydrophobic residues including Leu65, Leu67, Phe119, Trp179, Phe196, Leu273 and
Leu258. After modeling of the methyl parathion substrate into this cavity, the
researchers data suggests that the the Phe119 and Phe196 are ideally placed to anchor the
phenyl group of methyl parathion.96
Interestingly, our results from evolutionary
screening also show an aromatic rich domain of Trp-Trp which may also participate in
non-covalent interaction with phenyl groups.
2.4 Phage Display for Selection of Eugenol Binding Peptide Motifs
The final target we pursued to identify a binding motif for was eugenol, a liquid target
which has a characteristic aromatic scent of cloves. This scent was chosen in particular
to provide a chemical which has high vapor pressure. To utilize phage display for this
31
liquid target, a new screening strategy was developed. Initially TentaGel beads
(TentaGel HL-Br, Sigma Aldrich, St. Louis, MO) were utilized to immobilize the
eugenol to a solid support.
Figure 12: Overview of the method for immobilizing liquid target (eugenol) onto a
solid support (TentaGel beads) to allow a method for phage display screening of
liquid targets.
Considering that the eugenol target is a liquid, immobilization onto a solid support
facilitates a method for separating the bound phage via centrifugation. Immobilization of
eugenol consisted of incubating the TentaGel beads for 2 hours at room temperature at
equimolar concentration of NaOH. Several washing step in TBS were performed to
remove all trace amounts of un-bound eugenol.
32
Figure 13: Overview of the phage display screening process developed against a
liquid target immobilized onto a solid support. An initial screening against the solid
support alone is performed to remove phage that may bind to the support material.
The phage display approach, seen in Figure-13, utilizes an initial round of library
screening against uncoated TentaGel beads. The purpose of this initial screening round is
to remove phage from the M13 library which would bind to the TentaGel bead material
(polystyrene). As such, the beads were discarded, and the supernatant was retained as the
input for eugenol screening. Hence, after 30 minutes of binding to TentaGel beads, the
supernatant was removed and added to 5mg of eugenol coated TentaGel beads to
facilitate eugenol binding. After this, the screening was performed in the same fashion as
that of the TNT, DNT, and methyl parathion.
33
Figure 14: Phage display sequence results from the fourth round of phage display
screening against eugenol, an immobilized liquid target.
From fourth round screening results, two major sequences were found to be dominant.
These sequences were Gly-Ser-Arg-Met-Ser-Gln-Ser-Ser-Lys-Arg-Asn-Leu and Phe-Ser-
Leu-Pro-Ser-Lys-Ala-Leu-Pro-Trp-Gln-Leu as seen in Figure-14. Interestingly, this
latter sequence was also seen to dominate the fifth round screening results. By
examining the sequences in more detail, we found that From this, the we examined the
potential for the sequence Phe-Ser-Leu-Pro-Ser-Lys-Ala-Leu-Pro-Trp-Gln-Leu to be a
good binder for eugenol. We found that the initial screening used to remove TentaGel
binding phage, may not have been sufficient to remove non-specific binders. The theory
behind the screening of the liquid target while logical was not as effective in practice.
Particularly, the presence of the solid support offers too many sites for non-specific
binding to occur in comparison the available eugenol targets. As such, we found that
these results may not be effective representation of eugenol binding peptides.
Figure 15: Phage display sequence results from the fifth round of phage display
screening against eugenol.
34
2.5 Conclusion
Here in we have demonstrated the use of a diverse chemistry of amino acids present on
M13 bacteriophage to identify receptors for small molecule targets. Two distinct
strategies were utilized by performing phage display screening against the chemical
target. In one case, a molecular crystal form of the small molecule was used which works
well in the case of the target having low solubility in aqueous solution. The other
approach utilized prior immobilization of the target onto a polymeric bead. Utilizing the
combinatorial screening approach of phage display, we were able to mine information
from the available library for peptides which have preferential binding to the desired
targets. The first target screened was trinitrotoluene in the molecular crystal form. From
phage display results we identified a consensus motif comprising or aromatics and amino
acids capable of hydrogen bonding. In addition to this explosive molecule, we pursued
the screening of an analogue target dinitrotoluene from which we also found a consensus
binding sequences from phage display which was different from that of the TNT binding
motif. The last small molecule that was screened with a molecular crystal target strategy
was the organophosphate, methyl parathion. Since organophosphate
(pesticide/neurotoxin) detection holds interest to the public in terms of environmental
safety, we pursued the screening of methyl parathion to identify a potential receptor for
its detection. Our results show an aromatic rich domain of Trp-Trp which may
participate in non-covalent interaction with the phenyl group of methyl parathion.
Finally, we screened a liquid target, eugenol, by immobilizing it onto a solid support.
Screening for the eugenol sequence, while theoretically sound, did not yield effective
phage display results. We believe this is a result of the initial screening that was required
35
against the TentaGel bead solid support. To summarize, we have found several
consensus sequences from phage display for our targets of interest. We have found it is
highly necessary to have a molecular crystal target or a target with sufficiently identical
binding sites available throughout in order to have effective phage display results. In the
following chapters, we will discuss the steps used to identify the efficacy of these
sequences in binding to there respective targets with selectivity.
36
Chapter 3: Analysis Techniques for Identifying Selective Binding Motifs
3.1 Selective Binding
Achieving selective binding has been severely limited in sensor development by the use
of “off-the-shelf” polymers. While polymers such as polystyrene and PVC may offer
good structural characteristics, they are not well suited for discriminating between target
molecules due to the use of a repeating functional motif. Conversely, biopolymers such
as the specific binding pockets found in antibodies and GPCR (G-Protein Coupled
Receptors) elicit multivalent binding to offer a mode of achieving selective binding as
depicted in Figure-16.
Figure 16: Comparison of non-selective conventional polymer coatings to target
specific biopolymers.
37
3.1.1 Competitive Affinity Screening
In order to determine the strongest TNT binding receptors from the large subset of phage
display results, the phages bearing TNT binding peptides were collected. These 33
individual phage samples from TNT phage display screening results were picked and
separately amplified and diluted into a single mini-library of TNT binding phage having
106 pfu/µL of each phage. A single round of phage screening was performed against 5
mg of TNT in 0.2% TBST. Here, the TNT binding phages were simultaneously exposed
to the TNT target substrate in solution. After this competitive binding, the weak binding
phages were washed from the target, while the remaining strong binding phages were
captured. These remaining phage were titrated on LB Xgal/IPTG agar plates.79
Phage
titration was used to select phage plaques with receptor inserts.79
Twenty plaques, which
appeared blue, were picked and sequenced to reveal the strongest binding sequence.
Similarly, we identified the strongest DNT binding sequences from DNT target
screening.
38
Figure 17: The pIII receptors present on phage that were used for competitive
screening analysis. The number of appearances for a given receptor after
competitive screening (last column) were found from the DNA sequencing results.
The 33 sequences in Figure-17 represent the pIII receptors expressed on the M13
bacteriophage which were amplified and mixed to equal concentrations of 106 pfu/uL of
each phage for competitive screening. After a single round of phage screening with 0.2%
TBST, the remaining bound phage were eluted, captured, and titrated onto LB Xgal/IPTG
39
agar plates. From the blue plaques which appeared after overnight incubation at 37°C,
twenty of these were picked for sequence analysis. The last column of Figure-17
represents the number of appearances of a given receptor amino acid sequence from the
DNA sequencing results. The most abundant of these receptor sequences indicates the
highest level of binding. Importantly, 19 of the 20 phages sequenced possessed a Trp-
His-Trp motif at the N terminus.
Within the competitive screening experiments, 33 phage samples were screened against
TNT, and their corresponding pIII receptor sequences can be seen in Figure-17. The
remaining bound phages were captured and 20 of the phage were randomly picked for
sequence analysis. The last column of Figure-17 represents the number of appearances of
a given receptor amino acid sequence from the 20 random phages. The most abundant of
these receptor sequences indicates the highest level of TNT binding. Importantly, 95% of
these strongest TNT binding phage exhibited the same N terminal tetra-peptide motif:
Trp-His-Trp-X. Through this rigorous competitive screening, we assigned the most
abundant binding sequence as the strongest TNT binding peptide candidate (Trp-His-Trp-
Gln-Arg-Pro-Leu-Met-Pro-Val-Ser-Ile: TNT-BP). Similarly from the identified DNT
binding sequences, His-Pro-Asn-Phe-Ser-Lys-Tyr-Ile-Leu-His-Gln-Arg, was found to be
the strongest DNT binding sequences and was denoted DNT-BP.
40
3.1.2 Substrate Specificity Screening
To determine the most selective receptors from phage display screening results, TNT
binding phage and DNT binding phage were separately screened against both TNT and
DNT substrates. To evaluate the extent of receptor binding activity to these target
molecules, we measured the ratio of the amount of phage present after one round of
screening with stringent washing steps as compared to the amount of phage initially
introduced (Output/Input). Specifically, the phages were amplified to a concentration of
106 pfu/µL (Input), and each phage sample underwent one round of the screening process
with 0.2% TBST against 5 mg TNT and DNT separately. The amount of phage was
identified via UV spectrometer as well as titration. After titration, blue plaques were
counted to determine the concentration of bound phage (Output). This ratio of
Output/Input was calculated for each phage sample and then related to that of polystyrene
(PS-BP) specific phage which has no particular binding preference to either TNT or
DNT. PS-BP was, therefore, used as a negative ‘non-specific’ control. The receptor
with largest ratio difference between TNT and DNT substrate binding was designated
most specific for its target substrate.
41
Figure 18: a) Receptor sequences from phage primarily from TNT screening results
used in the following specificity screening, and b) specificity screening results of
phage against TNT and DNT substrate. Levels of binding represented as the ratio
of phage remaining bound after interaction and washing as compared to the initially
exposed number of phage.
42
The phages bearing receptors identified in Figure-18a were amplified to a concentration
of 106 pfu/uL. This number of phages was identified as the “Input” and was verified via
UV spectrometer as well as titration. 1mL of the phage samples were then individually
screened against 5mg of TNT, washed, and the remaining bound phage were eluted. This
elution, known as the “Output” number of phages, was quantified using phage titration.
The ratio of Output / Input was calculated and the values can be seen in Figure-18b.
Similarly, phage samples indicated * were also screened against 5mg of DNT and the
corresponding ratio of Output / Input can be seen in Figure-18b. The results reflect the
preferential binding of a given phage for the TNT substrate and hence the selectivity of
the receptor sequence.
43
Figure 19: a) Receptor sequences from phage primarily identified from DNT
screening used in the following specificity screening, and b) specificity screening
results of phage against TNT and DNT substrate. Levels of binding represented as
the ratio of phage remaining bound after interaction and washing as compared to
the initially exposed number of phage.
Similarly, the phages bearing receptors identified in Figure-19a were amplified to a
concentration of 106 pfu/uL. This number of phages was identified as the “Input” and was
verified via a UV spectrometer and also titration of phage. 1mL of the phage samples
were then individually screened against 5mg of DNT, washed, and the remaining bound
44
phage were eluted. This elution, known as the “Output” number of phages, was
quantified using phage titration. The ratio of Output / Input was calculated and the values
can be seen in Figure-19b. Similarly, phage samples indicated * were also screened
against 5mg of TNT and the corresponding ratio of Output / Input can be seen in Figure-
19b. The results reflect the preferential binding of DNT_ph7.5-Va-11 for the DNT
substrate and hence the selectivity of the receptor sequence His-Pro-Asn-Phe-Ser-Lys-
Tyr-Ile-Leu-His-Gln-Arg.
Figure 20: Phage display screening results of converged amino acid sequences from
the fourth round of phage display screening.
From the phage display results, we assessed the best binding phage by greatest target
selectivity. This target selectivity was determined based on the relative level of phage
binding to TNT and DNT target substrates. Among the subset of TNT binding
sequences, several exhibited varying affinity for DNT. We selected the phage with the
largest binding difference between TNT and DNT targets (those with strong binding to
TNT and low affinity for DNT), which was determined to have the TNT binding peptide,
TNT-BP. Similarly from DNT phage display results, we screened various DNT binding
phage against TNT and DNT and identified the most selective phage to contain the
45
sequence from DNT-BP (Figure-18). From our selectivity screening, the peptide TNT-
BP was found to have only non-specific interactions with DNT as demonstrated by
comparison to the negative control phage (Figure-21). Interestingly, DNT-BP binding to
TNT also remained on the order of non-specificity. This important result demonstrates
that M13 linked receptor sequences identified from phage display can selectively bind a
pre-determined target.
Figure 21: Selectivity screening of the DNT receptor and TNT receptor against
DNT substrates and TNT substrates with the level of binding quantified from phage
titration.
3.2 Synthesis of Standalone Receptors
All receptors were synthesized using standard Fmoc chemistry based solid-phase peptide
synthesis101
with amino acids and pre-loaded (cysteine or biotinylated-lysine) Wang
46
resins. OEG conjugated DNT receptors were synthesized by coupling three subunits of
Fmoc-amino-diethoxy-acetic acid to cysteine resin prior to addition of the DNT-BP
sequence. Cleavage reactions were performed for 2 hours with the a cocktail of 82.5%
trifluoroacetic acid, 5% thioanisole, 2.5% water, 5% phenol, 2.5% ethanedithiol, and
2.5% tri-isopropyl silane. Samples were purified by HPLC to >95% purity.
3.3 Mutational Analysis Binding Assay
Utilizing alanine substituted control peptides; we could identify the influence of
individual substitutions on the receptor’s binding capability and substrate specificity, and
thus identify the role of multivalent binding. By using a tetrapeptide biotinylated linker,
the receptors could be functionalized with Atto 425-Streptavidin probes for fluorescence
binding assays. In this assay, 10 µL of 100 µg/mL TNT or DNT in acetonitrile was
placed in 96 well TCPS plates. The TNT or DNT target was coated to the surface by
introduction of 300 µL TBS (50 mM Tris-HCl [pH 7.5], 150 mM NaCl) and set overnight
at room temperature. Solutions were removed and rinsed with TBS to remove non-
adsorbed TNT or DNT target. 1mM of peptides were introduced to the TNT or DNT
coated wells for 30 minutes. After binding, the substrate was washed with 0.1% TBST.
A bovine serum albumin (BSA) blocking buffer was then used for 30 minutes to prevent
non-specific adsorption of the fluorophore. An avidin labeled fluorophore, Atto 425-
Streptavidin, was then introduced while allowing 20 minutes for binding to the attached
biotinylated peptides. The substrate was washed with 0.1% TBST in order to remove any
non-specifically bound fluorophore. Finally, the bound receptor peptides, now
47
conjugated with the fluorophore, were eluted from the TNT substrate with vigorous
washing with 0.5% TBST. An Electromax Gemini EM plate reader was used to obtain
emission intensity at λ = 476 nm with excitation at λ = 436 nm in order to characterize
the amount of eluted peptide and the relative fluorescence signal was compared to the
BSA background.
48
Figure 22: Mutational analysis for identified TNT and DNT binding peptides. (a)
Sequence of receptors synthesized with C-terminal biotinylated lysine to allow
functionalization with the fluorescent probe Atto-425. TNT-BP and DNT-BP
represent the identified TNT and DNT binding receptor sequences, respectively,
while H-Sub (histidine substituted by alanine) and W-Sub (tryptophan substituted
by alanine) are mutations at amino acids 1 and 2 of TNT-BP. Scram-Ctrl is a
nonspecific sequence derived from scrambling the TNT-BP sequence to demonstrate
the sequence importance for encoding target selectivity and hence a negative control.
(b) Fluorescence binding assay against TNT substrate and DNT substrate, revealing
the importance of tryptophan and histidine residues in TNT-BP as well as
demonstrating the ability of the lone peptides TNT-BP and DNT-BP to bind
selectively to their targets in liquid when not associated with the other phage body
proteins. Fluorescence levels are normalized to the BSA fluorescence background
signal (P<0.0001, n=4). All data are presented as the mean ± standard deviation.
While target specific phage binding was a critical first step, the fluorescence binding
studies indicated that the receptors retained high substrate specificity despite no longer
being attached to the M13 phage (Figure-22a,b). The selectivity of the TNT receptor is
demonstrated by the significantly higher fluorescence levels of TNT-BP to the TNT
49
substrate as compared to the low levels of binding against the DNT substrate.
Substitutions in the N terminal region with alanine residues demonstrate the importance
of the conserved tryptophan and histidine residues in the binding event. This supports the
importance of multi-site interactions, as the tryptophan replacement resulted in a 58%
decrease in binding while the histidine replacement decreased binding by 48%. By
comparing the TNT binding sequence, scrambled TNT binding sequence, histidine
substitute, and tryptophan substitute for different substrates, we provide evidence of
substrate selectivity through multivalent binding. Direct comparison of the TNT binding
sequence for TNT and DNT substrates demonstrates the ability of the isolated TNT
receptor to bind selectively (over 3 fold increase) to the TNT substrate while remaining
relatively inert to DNT substrate (on the order of the non-specific scrambled sequence
and background (Figure-22b). Similarly, our isolated DNT receptor preferentially bound
the DNT substrate (over 2 fold increase compared to response against the TNT substrate
(Figure-22b)).
3.4 Isothermal Titration Calorimetry
Isothermal titration calorimetry (ITC) was used to obtain binding isotherms of the
complex between TNT and the peptide TNT-BP, from which the dissociation constant
was determined. The ITC reservoir cell (kept at 25°C under constant mixing) contained a
10 µM solution of TNT-BP in acetonitrile to which a stock solution of TNT (100 µM in
acetonitrile) was added with an initial 5.0µL equilibration injection volume followed by
15µL injections every 5 minutes. The heat released per ITC injection of TNT was
50
measured, and the integrated heat plotted against the molar ratio of TNT added to TNT-
BP was obtained to give a complete binding isotherm for the interaction. To determine
the dissociation constant, the data was fit to a one-site binding model. The binding
isotherm obtained from the titration of TNT into a solution of TNT-BP is provided in
Figure-23. This isotherm, calculated from the integrated heat change per mole of TNT
injected and fit to a one-site binding model, provided the best-fit parameters values of the
binding constant and the enthalpy of K = 1.4 x 107 ±3 x 10
6 M
-1 and ∆H = -49 ±1
kcal/mol respectively. The inverse of the calculated K value gives us the dissociation
constant of 71 nM for TNT-BP interaction with TNT. These results support our ability
to use evolutionary screening to identify a receptor for a pre-determined target.
Figure 23: Measurement of the dissociation constant of the complex between TNT
and the peptide TNT-BP determined from isothermal titration calorimetry. Data
points (values for the integrated heat change during each injection, normalized per
mole of TNT) are represented by filled squares. A one-site model was used to fit the
data. The solid red line is the calculated curve using the best-fit parameters.
51
3.5 Conclusion
Using directed evolutionary screening processes, we identified specific TNT and DNT
binding peptides. Compared with the resulting TNT binding peptide, (Trp-His-Trp-X:
where X represents Gln, Ser, Asn, or Lys), the active site of nature’s known TNT binding
protein Enterobacter cloacae’s pentaerythritol tetranitrate reductase (Trp102, His181,
Tyr186, Thr26) contains many compositional similarities.102
PETN-reductase and other
TNT binding proteins display a highly conserved tryptophan residue involved in the
binding event.102, 103
Furthermore, these previous studies show that changes in the
tryptophan’s neighboring amino acid, histidine, can drastically modulate the TNT binding
ability of these proteins. Similarly, through mutational analysis of our TNT binding
motif (Figure-22b), we have demonstrated the role of multivalent binding involved with
neighboring tryptophan and histidine residues. Various mechanisms exist by which
tryptophan may attribute its strong role in the binding motif for TNT. The PETN-
reductase utilizes the aromatic stacking between tryptophan and the ring structures of
nitroaromatics.104
Through computational approaches, researchers have identified similar
dual aromatic residues as part of a high affinity TNT binding motif which parallels the
tryptophan arrangement of our own TNT binding sequence.17
Tryptophan’s interaction
with TNT may take on a donor-acceptor character due to the electron deficiency of the
ring in TNT while those of tryptophan are electron rich.105
Histidine can also contribute
to π-π interactions.106
In addition, imidazole side-chains can coordinate with the nitro
group in the TNT molecules through partial charge-charge interactions or hydrogen
bonding.102, 107
52
In summary, we have shown the successful evolutionary screening of highly selective
peptide receptors for explosive targets, such as TNT and DNT. We discovered a peptide
motif which coincides with the TNT binding site in PETN-reductase that has evolved in
nature. Using mutational analysis, we demonstrated that multivalent binding is the key to
selectivity of the TNT binding motif
53
Chapter 4: Selective Coatings for Chemical Sensing
4.1 Introduction to Chemical Sensing
Susceptibility of chemical sensors to false positive signals remains a common drawback
due to insufficient sensor coating selectivity. By mimicking biology, we have
demonstrated the use of sequence-specific biopolymers to generate highly selective
receptors for trinitrotoluene (TNT) and 2, 4-dinitrotoluene (DNT). Using mutational
analysis, we show that the identified binding peptides recognize the target substrate
through multivalent binding with key side-chain amino acids elements. Additionally, our
peptide-based receptors embedded in a hydrogel show selective binding to target
molecules in the gas phase. These experiments demonstrate the technique of receptor
screening in liquid to be translated to selective gas phase target binding, potentially
impacting the design of a new class of sensor coatings.
4.2 Overview of a Peptide Based Sensor Coatings
Utilizing the evolutionary screening approach discussed in detail in previous chapters, we
discovered molecular recognition motifs capable of selectively binding explosive
molecules TNT and DNT. Furthermore, to make these relevant for VOC detection, we
have translated liquid-phase screening into gas-phase selective binding through the
formation of receptor-laden hydrogels that attempts to emulate the olfactory system. As
demonstrated in the previous chapters, phage display can be used to obtain highly
54
selective receptors for small volatile molecules. In particular, by exploiting the
combinatorial screening power of this technique, it is shown that oligopeptides are
capable of selectively binding the explosive TNT while remaining inactive towards DNT
and vice versa.
4.3 Coating Design for Surface Stress Based Sensors
Chemo-mechanical transducers such as a cantilever beams and micro-membranes can
detect surface stresses created by ligand-receptor binding at very low concentration with
sufficiently high signal-to-noise ratio.8, 108
Such high sensitivity opens many applications
such as detection of explosives and chemical warfare agents. Figure-24 demonstrates the
operation principle of micromembrane surface stress sensors.
Figure 24: Operation mechanism of micromembrane surface stress sensor.
The molecular interaction between probe molecules and target molecules generate a
surface stress on the thin gold layer. This surface stress causes the structural deflection of
the membrane, which generates the capacitance change in electrical sensing. In order to
55
design a coating for such a surface stress based sensor, we implemented our DNT
receptor onto a gold sensor surface as a mock sensor layer.
4.3.1 Gas Phase Binding Assay Methods
A (100) silicon wafer was cleaned with heated Piranha solution. A 5 nm chrome layer
was thermally evaporated onto the wafer as an adhesion layer between the gold and
silicon. A 25 nm thick layer of gold was then thermally evaporated onto the wafer. The
wafer was then protected using 2 µm of G-line photoresist prior to dicing into 3mm by
3mm chips. The photoresist was then stripped using heated PRS-3000 solution, and
chips were cleaned and dried.
As a novel extension to standard liquid phase receptor identification and to make it
relevant for gas-phase chemical sensing, we embedded the selective DNT receptors in a
hygroscopic oligo (ethylene glycol), OEG, coating to test gas phase binding. Given
DNT’s higher vapor pressure compared to TNT, we found it more applicable to focus our
gas phase experiments solely on the identified DNT binding peptide.109, 110
Multiple
coating conditions were analyzed including: (i) DNT Receptor-OEG-Cys on gold chips
(ii) OEG-Cys on gold chips (iii) Blank gold surface (iv) DNT Receptor-OEG-Cys on
blank silicon surface (v) OEG-Cys on blank silicon surface (vi) Blank silicon surface.
Immobilization of coating layers was carried out by immersing the different chips in 1
mM solution of either DNT-OEG-Cys or OEG-Cys solutions for 24 hours utilizing the
available gold-thiol bond chemistry. The coatings were then exposed to target gas in
56
ambient air by placing the chips inside a scintillation tube containing crystalline DNT or
TNT which was heated uniformly to 600C for 2 hours using a custom designed aluminum
heat block with an NIST certified temperature controller (VWR Inc.) to generate DNT or
TNT gas. All experiments were performed with chips exposed to 18ppm of DNT gas.
The chips were immediately analyzed for the amount of bound DNT and TNT by
placement in the thermal desorption tube of a Unity Thermal Desorption System, which
heated the chips to 3000C and passed the desorbed particles directly to an Agilent GC-MS
system (Santa Clara, CA). Partition coefficients are identified as the ratio of
concentration of analyte bound to the coating compared to the concentration of analyte in
exposed gas headspace and normalized to the appropriate control condition (ie. blank Si
and OEG-Cys coatings).
4.3.2 Gas Phase Binding Assays Results
The results of Figure-25 represent the various control experiments performed to identify
the extent to which DNT would interact with the various components of the Au-DNT BP
coating. Silicon chips were exposed to DNT gas and used as the background signal for
DNT partition coefficient measurements for the various coatings of the Si chips. Six
chips conditions were utilized for DNT gas experiments: (i) DNT Receptor-OEG-Cys on
gold chips (ii) OEG-Cys on gold chips (iii) blank gold surface (iv) DNT Receptor-OEG-
Cys on blank silicon surface (v) OEG-Cys on blank silicon surface (vi) blank silicon
surface (control). Importantly, the amount of DNT bound to conditions (iv) and (v) were
relatively the same as that for the blank Au control (iii). This indicates the OEG-Cys or
57
DNT Receptor-OEG-Cys coating attachment is inhibited on Si substrate as compared to
there attachment to Au coated substrates under the same conditions. Furthermore,
Figure-25 identifies the highest DNT partition coefficient for condition (i) in which DNT
Receptor-OEG-Cys was used as the coating for the Au chip. By displaying this
comparatively large DNT partition coefficient using the DNT receptor, we demonstrate
the ability to translate from liquid phase screened receptors into gas phase target binding.
Figure 25: Gas-phase screening for partition coefficients of various coatings on Si
exposed to DNT gas. The values are normalized to the DNT gas partition coefficient
of blank Si substrates to observe the contribution attributed solely to the coating
layer. Partition coefficients are calculated as the ratio of the concentration of
analyte bound to the coating (identified through thermal desorption GCMS) to the
concentration of analyte in the exposed gas headspace.
The sequence identified as the best TNT binder was designated TNT-BP and possessed
the sequence Trp-His-Trp-Gln-Arg-Pro-Leu-Met-Pro-Val-Ser-Ile. Scram-Ctrl represents
58
the same amino acids as TNT-BP but in a different configuration. TNB-Ctrl represents
the positive control as it is known to bind TNT in liquid. These sequences were
synthesized with C terminal biotinylated lysine followed by a tri-Glycine linker to allow
functionalization with a fluorescent probe Atto-425 for the comparative analysis of the
loan peptide receptors (no longer attached to the phage body). Here, the sequences were
exposed to TNT coated polystyrene wells, fluorescently labeled, washed, captured, and
measured of their fluorescence relative to the BSA background signal. Figure-26
represents the level of binding exhibited by the given sequence for the TNT substrate.
The sequence TNT-BP exhibited much greater binding to the sequence Scram-Ctrl which
possessed the same amino acids in a different configuration. This result demonstrates the
importance of the configuration of the amino acids in the binding event, thereby
identifying that it is not solely the apparent charge or the functional groups present on the
receptor but instead their particular arrangement. TNT-BP exhibit higher level of TNT
binding as compared to the positive control, TNB-BP. (P < 0.0001, n = 4). All data
presented as mean ± standard deviation.
59
Figure 26: Analysis of the level of TNT substrate binding for the designated TNT
binding peptide (TNT-BP) as well as a scramble TNT-BP sequence as control and a
trinitrobenzene binding sequence (TNB-Ctrl).
The gas phase binding results for the DNT binding peptide (Figure-27b) show a 4 fold
increase in the partition coefficient for DNT over TNT as a result of the DNT receptor.
The preferred coating condition for selective binding of DNT gas was that of the
identified DNT-BP. Additionally, we found the DNT-BP partition coefficient for TNT
gas is on the same range as that of the OEG coated chip without a receptor indicating that
the selectivity of the DNT receptor remains when implemented in gas phase.
60
Figure 27: Selective gas-phase binding assay for DNT-specific coating: (a)
schematic diagram of the DNT binding peptide conjugated to oligo(ethylene glycol)
and their coating onto a gold surface for gas-phase selective binding; (b) partition
coefficient of DNT receptor coatings exposed to TNT gas and DNT gas. The values
are normalized to the target gas partition coefficient of OEG coating on a Au
substrate to isolate the contribution attributed to the DNT receptor element.
Partition coefficients are identified as the ratio of the concentration of analyte
bound to the coating to the concentration of analyte in the exposed gas headspace.
The results are obtained through thermal desorption GC-MS experiments on
exposed coating surfaces (P < 0.001, n = 4). All data are presented as the mean ±
standard deviation.
While successful biosensors have previously been developed for DNT detection in
aqueous solution, this demonstration of a selective coating for DNT in gas phase is of
particular importance as these short OEG embedded receptors are capable of retaining
efficacy outside of the liquid environment. The success of gas phase binding may be
attributed to the properties of OEG indicated by various research groups including: i) the
ability of OEG to retain the conformation of biomolecules111
and ii) the selectivity of
peptides remaining unaffected by OEG conjugation.101
Additionally, PEG is often used
for its non-fouling properties which may be beneficial in terms of minimizing false
positives in sensing applications.112
61
Unfortunately, the high hygroscopic nature of PEG was actually a drawback in terms of
implementation onto a real world sensor. This is because the PEG accumulates multi-
layers of water with increasing humidity inst of saturating at low humidity levels as
previous hoped. Looking to several potentials polymer for coating onto a sensing
platform, the humidity response was analyzed for the polymeric coatings using Quartz
Crystal Microbalance. Some consideration taken into account included a low humidity
response and the ability to immobilize the receptor.
Figure 28: Analysis of a hydrophilic and hydrophobic polymeric coatings using
Quartz Crystal Microbalance to identify the humidity response.
After analyzing both hygroscopic polymers (ie. poly (acrylic acid) and PEG) as well as
hydrophobic polymers (ie. PCGEF and PEGM), it was determined that the PEGM, poly
(ethylene-co-glycidyl methacrylate), had the lowest response to humidity. Additionally,
62
the epoxy group offered a point of receptor attachment via amino groups as is discussed
in the next section.
4.4 Coating Design for Quartz Crystal Microbalance Sensing Platform
For many sensing applications a high density of receptor is required to yield a measurable
signal when the target concentration is low. This is the case for TNT, which has a vapor
pressure of ~2x10-4
Torr at 25° C,113
and water solubility of 100 mg/L.114
Because of
this, we have examined suitable polymers to use as matrices to embed the receptors in,
with the goal of increasing the number of bound receptors available for interactions with
the target molecules. The polymeric matrices were selected to contain reactive groups
readily available for coupling with the functional groups present on the receptors. In this
paper we report on a particular co-polymer meeting such requirements: poly(ethylene-co-
glycidyl methacrylate), called 'PEGM'.
Figure 29: Chemical structure of the poly(ethylene-co-glycidyl methacrylate), called
'PEGM', which was found to have the lowest humidity response of the polymers
tested. Hence, it was chosen as the polymeric support for attachment of selective
receptors and coating onto the QCM.
63
This co-polymer, seen in Figure-29, contains an epoxy group in the glycidyl methacrylate
monomer, which reacts with amino groups present in the receptor structure to form stable
covalent linkages. Such bonds are not hydrolized in water, thus making this coating
particularly suited not only for gas sensing but also for detection in aqueous
environments. This is of relevance for DNT and TNT detection, because these explosives
are commonly stored in ammunition depots or weapon testing sites, and can seep into the
surrounding ground waters.115
As they are known to be harmful to health,116
their
detection in liquid is especially important to monitor and regulate the contamination in
ground water. In what follows, we show the successful attachment of an oligopeptide
TNT receptor (developed through phage display) to PEGM as verified using X-ray
photoelectron spectroscopy (XPS). The ability of the oligopeptide-PEGM matrix to retain
its functional activity as a receptive coating for TNT is demonstrated both in liquid phase
and in vapor phase using GC/MS. Using this strategy, the use of PEGM-receptor
conjugates as a sensor coating material to facilitate specific interactions with target
molecules is shown. Finally, we demonstrate the real-time selective detection of TNT in
the liquid phase using a QCM sensing platform.
4.4.1 Theory of QCM Operations
The quartz crystal is driven by an internal phase lock voltage controlled oscillator. For
measurements, the crystal current is monitored where the magnitude of the current is
proportional to the crystal conductance. From the voltage, a frequency and amplitude is
obtained. The frequency part of the signal is sent to a buffer, amplified, sent to a counter
64
and then to the acquisition software. Here the crystal current and voltage are monitored.
If there is no different in phase between the crystal current and voltage, then the system is
at the crystal resonant frequency. Addition of mass to the sensor will result in a change
in the measured frequency which is recorded as the signal. Using this, significant binding
events can be identified by a change in the mass and hence a decrease in the resonant
frequency.
4.4.2 Synthesis of PEGM / Receptor
TNT and DNT receptors found through phage display were created using solid-phase
peptide synthesis to provide pure forms of the receptor for analysis and for attachment to
the polymer matrix. Using standard Fmoc chemistry,101
pre-loaded cysteine-Wang resin
was used for 0.1mmol scale synthesis of the TNT and DNT binding peptides. Resins
were pre-swelled for 30 minutes in NMP prior to deprotection. Deprotection steps using
3mL of 3% DBU in NMP were carried out for 20 minutes on a rocking platform.
Washing steps involved rinses with 4mL NMP, 4mL methanol followed 4mL
dichloromethane, and with 4mL NMP. Coupling steps of 3mL 0.2M amino acid, Hobt,
and DIC were carried out on a rocking platform for 2 hours. Kaiser tests were performed
to identify presence of primary amines in order to monitor extent of reaction.117
Cleavage reactions were performed for 2 h with a cocktail of 82.5% trifluoroacetic acid,
5% thioanisole, 2.5% water, 5% phenol, 2.5% ethanedithiol, and 2.5% triisopropylsilane.
Samples were purified by HPLC to >95% purity.
65
To prepare the receptor coating, gold substrates were used to deposit the polymeric and
receptor coatings. Gold substrates were created by sputtering a 5 nm layer of Cr on a
(100) Si wafer, followed by 25 nm of Au. They were diced to the desired dimensions (5 x
5 mm2 for XPS analysis, 3 x 3 mm
2 for GC/MS analysis), sonicated in acetone for 15
min, and further cleaned using UV-Ozeonolysis (UVO-Cleaner, model 42, Jelight
Company, Inc. (Irvine, CA)) for 5 min. A droplet of PEGM solution was immediately
placed onto the Au substrate and allowed to dry, leaving a fairly uniform coating on the
substrate. A volume of 10 uL was used for XPS experiments and 5 uL for GC/MS
experiments. The TNT or DNT receptor attachment was carried out by exposing the
polymer coated chips to solutions containing 2.5 mg/ml of receptor, in 90% acetonitrile
(ACN, Fischer) and 10% ultra-pure 18 MOhm water. Volumes of 15 uL for XPS
experiments, and 5 uL for GC/MS were exposed to the polymer, followed by 15 and 5 uL
of tri-ethyl amine (TEA, 0.8 mg/ml in Acetonitrile) for XPS experiments and GC/MS
substrates, respectively. TEA was used as a catalyst to favor the reaction between the
polymer and the receptor sketched in Figure-30.
Figure 30: Schematic of attachment mechanism for amine on receptor with epoxy
grou of poly(ethylene-co-glycidyl methacrylate).
66
4.4.3 Qualitative Analysis of Coating with XPS
Sample preparation used in XPS analysis of receptor attachment and activity
Samples used for XPS analysis were prepared by drying 30 uL of a 22 mM solution of
TNT onto gold chips coated with PEGM and receptive peptides. The samples were then
evacuated in the loading-lock chamber for thirty minutes, and then transferred to the
measurement chamber, with a base pressure of 10-9
Torr, at which point the spectra were
acquired to quantify the amount of TNT on the samples. To identify the extent of receptor
activity, samples were then washed overnight in water to remove non-specifically bound
TNT, followed by nitrogen drying. XPS measurements were carried out after this wash to
identify the extent of TNT that remained bound to the surface. All XPS measurements
were performed with an Omicron EA125 electron energy analyzer and an Omicron
DAR400 source with Al Kα X-rays at an energy of 1486.6 eV. The detector angle was 0°
to the surface normal. Spectral deconvolution was performed after background
subtraction with the Shirley method. The spectra were analyzed using Origin 6.0
software, and peaks were fit to Gaussians or Lorentzians to determine peak positions. For
energy calibration purposes, the Au 4f7/2 peak was used from the substrate as a reference
at 84 eV.118
If the Au from the substrate was too low in intensity to read, the
hydrocarbon contamination peak was used as a reference at 285 eV, as explained by
Clark and Thomas.119
67
XPS analysis of TNT receptor attachment to PEGM
XPS spectra of TNT receptor deposited on a bare Au slide before and after rinsing in
water overnight are shown in Figure 31.
Figure 31: XPS spectra of C 1s (first row), N 1s (second row) and S 2p (third row)
peaks of TNT receptor deposited on a bare Au slide before (column A) and after
(column B) washing in water overnight. Experimental spectra are plotted in solid
line, fitted spectra in dashed line, and fits for each peak component in dash-dotted
line. The inset in the C peak, column B, represents the C 1s spectrum recorded on a
bare Au substrate.
68
The C 1s spectrum of the TNT receptor deposited on Au shows different components,
representing the many types of carbon environments found in an oligopeptide. A first
broad peak is centered at 284.5 eV, and is related to aliphatic C atoms. A second peak
located at 286.5 eV corresponds to carbon bonded to an amino group (C-NH2).120
A
second, intense and broad peak is centered at 288.1 eV, and may be assigned to carbon in
amide groups.121
This peak is typically indicative of the presence of a peptide on the
surface.122
The last peak located at 290.4 eV may be assigned to a ππ* shake-up band
due to the presence of aromatic groups within the peptide.123
The N 1s spectrum shows
two main peaks. The lowest peak in intensity is centered at ~400 eV and is related to non-
protonated amines and amides. The peak at higher energy is indicative of the presence of
a large amount of protonated N atoms,124, 125
thus indicating that most amines or imines
in the lateral chains of the amino acids composing the peptide are in fact charged. The S
2p spectrum shows a peak at ~168 eV, indicative of the presence of oxidized S in the
terminal cysteine present in the receptor.126
After washing in water overnight, most of the peptide is washed away from the Au
surface. The C 1s spectrum no longer displays peaks at ~288 and ~291 eV; rather it
becomes similar to that measured on a bare gold substrate (shown in the inset, where only
peaks relative to surface contamination are observed ). The peaks that most compellingly
indicate the removal of the peptide from the surface are the N and S peaks: the latter is
completely disappeared, and in the N spectrum, the most intense peak is now at ~400 eV
with only a shoulder observed at ~403 eV. This could be indicative of the presence of a
thin layer of peptide on the surface, as a similar effect was observed in the past for
69
histidine films on Au.127
This is confirmed by the drastic decrease in total N measured on
the sample soaked in water. The ratio of the total area underlying the N peaks to that
underlying the Au peaks was 23.5 before soaking in water and 0.07 afterwards, which
indicates that virtually all the peptide is washed away by overnight soaking in water, and
the Au surface is left mostly bare.
A very different result is obtained if the Au surface is covered with PEGM, and the
receptor is bound to it, as shown in Figure-32.
Figure 32: XPS spectra of C 1s (first row), N 1s (second row) and S 2p (third row)
peaks of PEGM deposited on Au (sample 'PEGM/Au') (column A), and TNT
receptor bound to PEGM on Au (sample 'TNTrec/PEGM/Au') before (column B),
70
and after (column C) washing in water overnight. Experimental spectra are plotted
in solid line, fitted spectra in dashed line, and fits for each peak component in dash-
dotted line.
The C 1s spectrum of just PEGM on Au displays two main peaks, related to the presence
of aliphatic carbon (peak at 285.2 eV) and carboxyl groups (peak at 289.4 eV), as
expected from the molecular structure of the polymer. No N and S were detected on the
polymer film. The spectra recorded on the sample containing the TNT receptor bound to
PEGM are shown in inset B. Two new peaks appear in the C 1s spectrum, located at
287.8 eV and 290.6 eV. These peaks are in similar positions to those related to amide and
carboxylic groups present in the TNT receptor. In particular, the peak at ~288 eV is
usually considered to be a very specific indication of the presence of peptides. Also the N
and S spectra resemble those analyzed for the receptor on Au. The spectra of the sample
TNTrec/PEGM/Au soaked in water overnight are shown in Figure-32. In the C 1s region,
the peak at ~288 eV is still present, which is an indication that the receptor is still present
on the substrate. An increase in the peak at 285 eV may be related to the adsorption of
impurities on the sample surface, as well as the appearance of some of the underlying
PEGM substrate, which has a carbon component at a similar position. The N 1s spectrum
splits in two peaks, which is indicative of a modification in the charge of some of the
amines and imines in the peptide (the lower energy component increasing in intensity is
related to the presence of a larger amount of neutral species). This split may also reflect
the removal of multilayers of peptide which had physisorbed onto the surface. The S 2p
peak remains visible on the sample, and maintains the same position it had before
soaking in water. All these observations indicate that most of the TNT receptor remains
bound to the surface of the PEGM/Au sample. This is a different observation from the
71
sample of TNT receptor drop-casted directly onto gold. The ratio of the total area
underlying the N peaks to that underlying the Au peaks remains virtually unchanged after
soaking in water, going from 2.4 measured before soaking in water to 2.8 measured after
soaking, which is within sample-to-sample variability.
Qualitative analysis of PEGM-TNT receptor activity using XPS
The experiments discussed above show that the TNT receptor can be stably bound to
PEGM. It is now crucial to determine if the attachment of the peptide based receptor to
the polymer compromises its activity. Previous work showed that the amino acid
sequence triplet, tryptophan-histidine-tryptophan, present at the N terminus of the TNT
receptor is critical to the target specificity (Trp-His-Trp). The chemistry used to attach the
receptor to PEGM involves the reaction of primary amino groups from the peptide with
epoxy groups from the polymer. To analyze the receptor activity, a series of XPS and
GC/MS experiments were performed to assess and quantify the presence of TNT bound
to the receptor after overnight washing in water. This amount was then compared with
the results obtained for non-specific adsorption of TNT on PEGM and for adsorption on a
DNT-receptor with no specific affinity to TNT, also bound to PEGM.
72
Figure 33: XPS spectra of N 1s region for samples prepared by deposition of 660
nmol of TNT on a) PEGM/Au and b) PEGM with TNT receptor. Experimental
spectra are plotted in solid line, fitted spectra in dashed line, and fits for each peak
component in dash-dotted line.
XPS was used to determine whether the TNT receptor linked to PEGM was still active
towards TNT binding. The N spectrum of a droplet containing 660 nmol of TNT
deposited on PEGM/Au is shown in Figure-33. As PEGM does not contain nitrogen in its
structure, the appearance of a peak at 408.7 eV after TNT deposition is readily assigned
to the NO2 groups of TNT. This is in agreement with the position observed for nitric
groups in other inorganic and organic structures. The position of NO2 at such high
binding energies allows for an effective identification of the presence of TNT also when
other N-containing species are present on the sample. As an example, the spectrum of a
similar TNT droplet deposited on a sample of TNTrec/PEGM/Au is shown in Figure-32
above The position of this peak is distant from other peaks relative to the peptide, thus
implying that the amount of TNT can be determined without the need of laborious and
somewhat uncertain spectral deconvolution.
73
In order to test the activity of the TNT receptor bound to PEGM, the area of the N peak at
~409 eV right after the deposition of the TNT droplet is compared with the area of the
same peak after soaking the TNT-containing sample in water overnight. Washing in
water overnight removes only the physisorbed TNT, whereas the TNT bound to the
receptor remains on the sample. As previously mentioned, the effect of TNT removal by
washing overnight in water is analyzed for samples containing the TNT receptor bound to
PEGM (TNTrec/PEGM/Au), samples of just PEGM/Au and samples containing the
peptide that is a receptor for DNT instead of TNT (samples DNTrec/PEGM/Au). This
allows one to confirm if the specific binding of TNT to the TNT receptor is retained after
washing. In order to be able to compare the area of the N peak at ~409 eV before and
after washing, one needs to decide which components of the spectrum to be used for
normalization. The Au peaks arising from the substrate are a good choice for the sample
PEGM/Au. However, the intensity of these peaks is too low for the DNTrec/PEGM/Au
and TNTrec/PEGM/Au samples, due to the extreme surface sensitivity of XPS, and
cannot be used for the normalization of the NO2 peak. Hence, the normalization for these
samples is performed using the area underlying all the components of the N 1s spectrum,
which includes the peak relative to TNT and the peaks relative to the amines, imines and
amides from the peptide. As previously shown, the peptide bound to PEGM is stable to
soaking in water overnight, and hence the only difference in the components of the N
spectrum before and after overnight soaking in water is due to the removal of TNT. The
results obtained from the percentage ratio of the normalized area underlying the N 1s
peak relative to TNT before and after soaking in water overnight, multiplied by 100 in
order to obtain a percentage, are shown in the Figure-34 for all the samples analyzed.
74
Figure 34: Percentage of TNT remaining on samples containing TNT receptor,
DNT receptor, and just PEGM after overnight washing in water. The values were
calculated by dividing the normalized area underlying the peak at ∼∼∼∼409 eV relative
to TNT after washing the sample in water to the normalized area of the same peak
measured before washing the sample, and multiplying the result by 100, to obtain a
percentage. Details about the normalization of the peak relative to TNT and p
values calculated according to Student’s t test are reported in the text.
Almost no TNT is removed after overnight washing of the sample containing the TNT
receptor bound to PEGM. On the contrary, when a non-specific receptor, such as the
DNT receptor, is used, about 50% of the TNT is washed away by overnight soaking in
water. This is only slightly higher than the non-specific adsorption of TNT on the
polymer PEGM. These results are not quantitative due to the assumption that none of the
peptide was removed during overnight soaking, required for the normalization of the area
underlying the TNT peak for samples containing receptors. In fact, the possible removal
of some of the peptide during overnight washing may be the reason for the value of
greater than 100% for the TNT on the sample containing the TNT receptor. Also, the
measurements are performed in vacuum, which implies that some of the TNT might have
75
desorbed. Still, these results clearly show a high level of selective binding of TNT
molecules to the TNT receptor, compared to the binding to DNT receptor or non-specific
adsorption on the polymeric matrix.
4.4.4 Quantitative Analysis of PEGM Receptor Conjugate Coating using GC/MS
Quantitative Validation of PEGM-TNT receptor activity in solution using GC/MS
Samples used for GC/MS were prepared by drying 5 uL of a 1 mM ethanolic solution of
TNT on gold chips coated with PEGM and receptors. The samples were then placed
inside the desorption tube of a thermal desorber (Markes Intl. Inc.) and heated at 300°C
using a Unity Thermal Desorption System. The desorbed material was passed directly
into an Agilent Gas Chromatograph-Mass Spectrometer calibrated for TNT
quantification. Control experiments were performed to ensure that a consistent amount of
TNT was initially exposed to each substrate. Analogous GC/MS measurements were
carried out on samples exposed to TNT after soaking in water overnight. More than one
chip was loaded in the thermal desorber to provide a sufficiently readable signal for
samples washed overnight. This was necessary because the TNT bound to only one
sample of PEGM or DNT receptor functionalized surfaces was too low to be detected.
Thus, at least three chips of samples containing the DNT receptor and at least 15 chips of
samples coated with just PEGM were loaded in the thermal desorber, and the amount of
TNT measured was then divided by the number of samples used in order to get the
average amount of TNT left on each sample. A final set of experiments was carried out
76
with GC/MS, by drying 5 uL of a 1 mM ethanolic solution of DNT. GC/MS
measurements were carried out on these samples before and after overnight washing in
water. More than one chip was placed in the thermal desorber to obtain a detectable
signal, as explained above. The amount of TNT and DNT remaining on the samples
washed overnight was evaluated by comparing the GC/MS response obtained for such
samples with that obtained for the samples prepared containing 5 nmol of TNT or DNT,
prior to washing.
A quantitative analysis of the amount of TNT remaining on samples after washing in
water overnight can be performed using GC/MS. The same set of experiments described
in the XPS section above was conducted, using a lower amount of TNT because of the
very high sensitivity of GC/MS. In particular, a droplet of a solution containing 5 nmol of
TNT was exposed to each sample. The samples were then washed in water overnight, and
the amount of TNT left on each sample was evaluated with GC/MS
Figure 35: GC/MS measurements of a) the amount of TNT remaining on samples
containing TNT receptor, DNT receptor, and just PEGM after overnight washing
in water and b) the amount of TNT and DNT remaining on samples containing the
receptor specific for TNT, after overnight washing in water.
77
The results shown in Figure-35 confirm those obtained with XPS, namely the amount of
TNT retained on samples containing the receptor specific to TNT was much higher than
that measured on samples containing a non-specific DNT receptor. The non-specific
adsorption of TNT on PEGM was very low. Quantitatively, the ratio of TNT remaining
on samples TNTrec/PEGM/Au to that left on samples DNTrec/PEGM/Au to that
measured on PEGM/Au samples was 1:0.33:0.06 for GC/MS, and 1:0.46:0.29 for XPS.
This shows that the trend measured with the two techniques was the same, although a
higher non-specific adsorption was measured with XPS. Also, it must be noted that the
absolute amounts of TNT remaining on the samples after overnight soaking measured by
XPS was much higher than that measured by GC/MS. Virtually the same amount of TNT
was measured by XPS before and after soaking in water overnight, on the sample
containing the receptor specific for TNT. Instead, less than 1% of it was found with
GC/MS, for the same type of samples. This observation is attributed to the following
factors. XPS was performed in ultra-high vacuum, thus only the amount of TNT that
resisted evacuation was detected; as such, most of the physisorbed TNT molecules were
most likely removed before the spectra were collected. Hence, soaking in water overnight
removed a further portion of TNT only on the samples that did not contain the receptor
specific for TNT, whereas almost no changes were observed in the amount of TNT
measured on samples containing the receptor specific for TNT. Instead, no evacuation
was performed before GC/MS measurements, which implies that the physisorbed TNT
was removed only during the overnight soaking. Hence, the absolute amount of TNT
resisting the overnight soaking was much lower than that initially evaporated on the
samples. This explains why the percentages of TNT left on the samples after overnight
78
washing measured by GC/MS are much lower than those measured by XPS. Moreover,
only a thin, superficial layer of the solid TNT dried on the samples was in direct contact
with the receptors bound to the PEGM/Au surface. This would lead to only a small
amount of TNT remaining on the samples after overnight washing, as measured by
GC/MS, since only a small fraction of the initial TNT could interact with the receptors,
while most of the TNT was dissolved in solution. Lastly, the reason for the proportionally
higher non-specific adsorption on PEGM measured by XPS compared to GC/MS is that a
much higher absolute amount of TNT was dried on the samples used for XPS compared
to those used for GC/MS (660 nmol vs. 5 nmol), due to the very different surface
sensitivity of the two techniques, which may imply that the overnight washing may not
have been sufficient to completely remove all the physisorbed TNT on the samples used
for XPS.
These sets of experiments demonstrate the selectivity of TNT binding on the receptor
specific for TNT compared to the binding to a non-specific DNT receptor or the lone
PEGM polymeric matrix. Another set of experiments was performed with GC/MS in
order to analyze the selectivity for TNT when the receptors were embedded in the PEGM
matrix. To identify this selectivity, the same concentration of DNT solution was exposed
to samples containing TNT receptors bound to PEGM, followed by washing in water
overnight. GC/MS spectra were collected before and after the wash to measure the DNT
signal and compare it to that obtained with TNT in the experiments described above.
After soaking in water overnight, virtually no DNT could be detected on the samples, as
shown in Figure-35, where this result is compared to that obtained when samples with
79
TNT receptors were exposed to TNT. This analysis shows that only TNT was strictly
bound to the TNT receptors and resisted overnight soaking, whereas DNT was not, and
the amount of DNT remaining on the samples was below the detection limit of GC/MS.
4.4.5 Integration with a Quartz Crystal Microbalance Sensing Platform
These experiments were performed using a Research Quartz Crystal Microbalance
(Maxtek, Inc. (Cypress, CA)). The quartz crystal was placed in a teflon crystal holder
which also acted as a reaction chamber. The crystal holder equipped for liquid flow was
connected by teflon tubings to two syringe pumps, containing DI water and a solution of
either DNT or TNT, respectively. A schematic of the system used for this set of
experiments is shown in Figure-36.
Figure 36: Schematics showing the two modes of operation of the QCM setup.
A 1” diameter, gold-coated quartz crystal was coated with 5 uL of PEGM solution
prepared as described above. TNT receptor was then bound to PEGM following the same
procedure. As shown in the figure above, the flow of solutions coming out of the syringes
is controlled using a set of three-way valves, adjusted in such a way that only one
solution flows through the sensor chamber (quartz crystal holder) at any given time.
80
Therefore, the setup can be used in either “solution mode” or “purge mode”, thus
allowing the chamber to be flushed with a solution containing the molecule of interest
(TNT or DNT), or DI water, respectively. Prior to every experiment, the frequency of
quartz crystal was measured and a stable baseline was obtained in the “purge mode”.
Then, the system was switched to “solution mode”, and the crystal coated with TNT
receptor/PEGM was exposed to either TNT or DNT solutions. The change in resonance
frequency of the quartz crystal was measured in real time. The system was maintained in
this mode till the equilibrium was reached. Finally, the system was switched back to
“purge mode”, and the crystal was rinsed with DI water, until a stable baseline was
obtained again.
Here we show the application of the receptor/polymeric coating on a well-known realistic
sensing platform, using QCM in liquid phase. In typical QCM experiments, the resonance
frequency of a quartz crystal is measured, which decreases if the mass adsorbed on the
crystal increases.128
PEGM was deposited on the gold-coated QCM crystal, and TNT
receptor was bound to PEGM as described previously. After reaching a stable baseline by
flowing DI water, the crystal was exposed to either TNT or DNT solution.
81
Figure 37: Change in QCM resonance frequency measured on a crystal coated with
PEGM/TNT receptor, after exposure of a solution containing TNT a) and DNT b),
respectively.
As seen in Figure-37, the resonance frequency measured with QCM decreased noticeably
when the crystal was exposed to TNT solution. Equilibrium was achieved in less than 30
minutes. After switching back to flowing DI water, the resonance frequency returned to
its previous value before the exposure. The decrease in frequency of ~6 Hz observed after
82
exposure to TNT solution is related to an increase in mass adsorbed on the crystal, due to
the interaction between TNT and TNT receptor. The same coating exposed to a solution
of DNT shows no appreciable changes in the resonance frequency of the crystal, thus
indicating that DNT molecules did not bind appreciably to the TNT receptor. These
results illustrate that the PEGM/TNT receptor coating maintained its high selectivity for
TNT towards DNT also in liquid phase, and that it can be used on a real-time sensing
platform.
4.5 Conclusion
Using the phage display identified receptors for gas phase chemical sensing, we created a
biomimetic coating for highly selective detection of DNT in ambient conditions. We
believe that this approach of evolutionary peptide screening followed by the creation of
biomimetic coatings, when generalized for other target molecules, reflects a significant
advance to enable highly selective and sensitive miniaturized chemical sensors.
In summary, a strategy for the preparation of a selective coating for TNT sensing is
demonstrated. An oligopeptide was identified as a specific receptor for TNT using phage
display, and was stably bound to a co-polymer, PEGM. The attachment was easily
performed by reacting the epoxy groups present in one of the monomers of PEGM to
primary amino groups present in the receptor. The attachment was stable to overnight
washing with water, as shown by XPS. The oligopeptide bound to PEGM maintained its
activity as a receptor for TNT, as shown by experiments performed by exposure to
83
solutions of TNT on samples containing TNT receptors bound to PEGM, and washing
such samples in water overnight to remove the unbound TNT. Both XPS and GC/MS
showed that the amount of TNT left on samples containing the specific TNT receptor was
much higher than that remaining on samples containing a non-specific receptor or the
lone PEGM matrix. In particular, the quantitative GC/MS results proved that the specific
adsorption of TNT in solution on TNT receptor was approximately 3 times larger than
that of a non-specific receptor and about 10 times larger than the physisorbed amount on
PEGM. GC/MS experiments also showed the selectivity of TNT over DNT binding in
solution for the TNT receptors was retained when the receptors were conjugated to the
PEGM matrix. A real-time highly selective explosive detection in the liquid phase was
also demonstrated, using QCM as a sensing platform, and coating the QCM crystal with
PEGM/TNT receptor. A decrease in resonance frequency of the QCM was observed only
in the presence of TNT solution, whereas no change in resonance frequency was noted
when the crystal was exposed to DNT solution. The potential of using this sensitive and
selective liquid detection of TNT could be particularly appealing for analysis of
contamination of ground waters near ammunition depots, for example.
In conclusion, these results show that it is possible to make a very selective sensing
coating by embedding specific receptors in a polymer, and that this can be used in a
standard sensor platform such as QCM. Ease of incorporation of these receptors in a
GC/MS sensor highlights versatility of the coating technique. This is of great importance,
because with a similar strategy large concentrations of receptors can be bound to the
sensor surface, if a porous and high surface area polymeric matrix is used. Additionally,
84
its possible to control the surface properties of the sensor by changing the polymeric
matrix used. For example, the hydrophobicity of the polymer used in the present work
makes this coating an optimal candidate for most real-life gas sensing applications, where
it is highly desirable to avoid interactions of the sensor with the humidity present in the
air.
85
Chapter 5: Development and Optimization of a Polymeric Sensing Vesicle
5.1 Introduction
In this chapter, we discuss the development of a polymeric sensing vesicle. Due
to their unique optical and stimuli-response properties, many sensor groups are utilizing
polydiacetylene vesicles as an alternative to complex mechanical and electrical sensing
systems. Utilizing a modular synthesis strategy to couple peptide based recognition
elements to chromic responsive polydiacetylene (PDA), we fabricated vesicles and
explored the essential design parameters necessary for an effective sensor from the
aspects of synthesis, composition, and assembly. Investigating the potential for these
systems, we explored the detection of small molecules with the specific example of the
explosive molecule trinitrotoluene (TNT). Furthermore, we investigated large targets by
incorporating cell binding domains for fibroblast detection. By observing shifts in the
absorption spectra of these sensor vesicle systems, we were able to identify the important
role of side-chain composition and surface density on transduction of chromic transition
from surface interactions. Additionally, we identified a tradeoff in surface density
considerations between proper monomer alignment and sufficiently high receptor density
to elicit a visually detectable chromic response. Finally, we identified a dependence of
polydiacetylene alkane chain length on the sensitivity of the system to chromic response
by the small molecule target, TNT. Overall, these systems offer the advantages of easy
and low cost manufacturing, simple analysis, and modular incorporation of target
recognition elements. These benefits make the PDA sensor a viable candidate for widely
86
deployable chemical sensing given the proper design parameters examined here are taken
into consideration.
5.2 Overview of Polydiacetylene as a Sensing Platform
Our environment contains a vast assortment of chemical and biological information,
some of which we can perceive through our senses. For instance, our olfaction and
gustation systems provide us with a means of molecular identification through smells and
tastes. Unfortunately, many surrounding harmful biological and chemical species, which
may threaten our health, remain undetectable. As such, we rely on the ability of sensor
technologies to identify the hidden information of our environment. As an alternative to
complex electromechanical sensing platforms, researchers have explored the use of
molecular target interactions and subsequent transduction into a detectable color change
as a means of gaining environmental information. One such colorimetric based sensor
framework utilizes the ability of inorganic nanoparticles to possess different absorption
spectra depending on their inter-particle spacing (or aggregation). Several research
groups have indeed demonstrated this with noble metal nanoparticles129-135
and other
inorganic nanostructures including quantum dots.136-138
When incorporated with
molecular recognition elements, these nanoparticles are effective sensing systems capable
of a visible color change. Other colorimetric based sensing approaches have minimized
the need for signal transduction hardware by using receptors designed for specific
molecular recognition which may elicit large spectral shifts upon ligand binding. Such
systems make use of engineering the curvature and functionality of the receptor to
87
provide the shape, size, and binding complementary for a given target molecule. These
include cavitands, crown-ether, and metalloporphyrin dyes which can provide a visually
detectable color change depending on the particular target-receptor pair.16, 18, 52
Researchers have also utilized chromic responsive polymers functionalized with
recognition sites for sensing purposes.139-145
In this work, we examine the creation of
these chemical sensors which utilize peptide based recognition elements functionalized to
such chromic responsive vesicles comprising conjugated polymer systems.
Several groups have exploited π-conjugated polymers, particularly polydiacetylenes
(PDAs), for chemical sensing applications due to the ability of their absorption spectra to
change in response to target binding.142-144, 146-150
PDA is an amphiphilic polymer
comprised of a carboxylic acid head-group and alkyl tail which facilitates its formation
into supramolecular assemblies such as vesicles.151, 152
When these assemblies are
exposed to UV irradiation, a conjugated PDA (ene-yne) carbon backbone is formed
which exhibits strong absorbance peaks at 640 nm or 540 nm resulting in a blue or red
color, respectively.153
The existence of a blue or red state is highly dependent on extent
of planarity or π orbital overlap of the conjugated PDA backbone.153-155
PDA has been
used extensively due to this susceptibility of the absorption spectrum to be altered by the
shortening of the effective conjugation length resulting from external stimuli.156-160
External stimuli, including interfacial perturbations or binding, can induce strain and
distortions within side-chains of PDA resulting in conformational transitions in the
backbone.161
This transition breaks the conjugated backbone network planarity resulting
in increased absorption at lower wavelengths and hence a red chromatic transition.162, 163
88
Molecular interactions resulting in disturbance of interfacial hydrogen bonds have been
proven to create a stress on the polymer backbone sufficient to effect the polymers
conjugations length.164-167
This change in the effective conjugation length of the π
conjugated backbone can be observed directly through the PDA absorption spectra
thereby making it ideal for sensing purposes.160, 168
Only a few degrees of (ene-yne) C-C
bond rotation can cause the irreversible chromic shift to the red state.153
The side-chain
arrangement is therefore critical to the chromic response of the system as we investigate
below. In this work, we study the design parameters which affect the functionality of
these conjugated polymer systems in colorimetric sensing applications for detection of
targets, such as the explosive TNT.
Figure 38: Schematic diagram of synthesis, composition, and assembly parameters
which must be optimized to achieve an effective colorimetric PDA vesicle based
sensor.
89
Using a modular synthesis strategy, we created sensors comprising TNT recognition
elements, identified previously through phage display,169
coupled to chromic responsive
PDA elements. With this system, we investigated the role of side-chain composition and
the role of surface density on the transduction of surface interactions to the conjugated
PDA backbone. Additionally, we identified the effect of these parameters on the self-
assembly by observing their absorption spectra. By investigating these assembly
parameters, we have demonstrated the importance of polymerization timescales in
achieving maximal effective conjugation. The irreversible nature of the chromatic
transition, from a meta-stable blue phase into a more thermodynamically stable red phase,
was also explored through polymerization and heating experiments.163
Finally, we
investigated the potential of these systems for small molecule target (e.g., TNT) as well
as whole cell target (e.g., fibroblast) detection. We believe these parameters are critically
useful to researchers considering fabrication of an effective PDA based sensing system.
5.3 Experimental Section
5.3.1 PDA-Peptide Conjugate Synthesis
To produce PDA-peptide conjugates, standard solid-phase peptide synthesis was carried
out using Fmoc chemistry.101
Fmoc protected amino acids and rink amide resins were
obtained from EMD Biosciences (San Diego, CA), while 10, 12-pentacosadiynoic acid
(PCDA) was obtained from Sigma Aldrich (St. Louis, MO). Resins were pre-swelled for
30 minutes in NMP prior to deprotection. Deprotection steps using 3 mL of 3% DBU in
90
NMP were carried out for 20 minutes on a rocking platform. The washing steps
proceeded as follows: 3 washes with 4 mL NMP, 6 washes with the series of 4mL
Methanol followed by 4 mL Dichloromethane, and 3 washes with 4 mL NMP. Coupling
steps of 3 mL and 0.2 M amino acid, Hobt, and DIC were carried out on a rocking
platform for 2 hours. A final coupling step with 3 mL of 0.2 M 10, 12 pentacosadiynoic
acid (PCDA), Hobt, and DIC was performed with the same incubation period of 2 hours
to create the PCDA-peptide conjugate. Kaiser tests were performed at each step to
identify the presence of primary amines in order to monitor the extent of reaction.117
Resins were washed with methylene chloride, dried, and underwent cleavage. Cleavage
reactions were performed for 2 hours while shaking with a cocktail of 82.5%
trifluoroacetic acid, 5% water, 5% phenol, 5% ethanedithiol, and 2.5% triisopropylsilane.
Rotary evaporation followed by precipitation in diethyl ether provided the removal of
cleavage solvents and protecting groups. Samples were then suspended and mixed in
water followed by centrifugation for 10 minutes at 10,000 rpm to pellet the product at
which point the supernatant was discarded to remove any trace contaminants. The
suspension and centrifugation steps were repeated until all cleavage contaminants were
removed. Lyophilization was then performed and samples were stored at 4°C. The
PDA-peptide conjugates which were synthesized are as follows: PCDA-Trp (promoter),
PCDA-Trp-His-Trp (TNT binding motif), and PCDA-Gly-Arg-Gly-Asp-Ser (cell binding
motif). There structures can be found in Figure-38.
91
5.3.2 Controlling Surface Composition and Density in Supramolecular Assemblies
PCDA and PCDA-peptide conjugates were suspended to 10 mM in water. Solutions
were sonicated for 20 minutes and heated at 85°C for 20 minutes to melt any existing
super-molecular structure. Additional sonication for 20 minutes was performed prior to
mixing of PCDA with PCDA-peptide conjugates. Volumetric mixing was utilized to
provide the exact molar amount of PCDA and peptide conjugate. For example, 4%
PCDA-Trp-His-Trp and 96% PCDA utilized a 4:96 volumetric ratio of PCDA-Trp-His-
Trp to PCDA. After mixing, the sample was heated for another 5 minutes at 85°C and
sonicated to ensure ample distribution of PCDA-peptide conjugates and PCDA. To allow
self-assembly into vesicle structures, the solution was allowed to cool to room
temperature then incubated at 4°C for 24 hours. The solution is then brought to room
temperature and the self-assembled vesicles are polymerized using a 4 W UV lamp at 254
nm wavelength. Polymerization is carried out for 25 minutes of exposure. After filtering
through a 0.8 µm cellulose acetate filter, vesicle size was measured using a dynamic light
scattering (DLS) particle sizer (Malvern Instruments, Southborough, MA).
5.3.3 UV Exposure, Extent of Polymerization, and Irreversibility Measurements
Extent of PCDA polymerization was characterized from the visible absorption spectra
(400 nm-800 nm wavelength scan with Beckman Coulter UV-Visible
Spectrophotometer). Particularly, the blue percentage, [Abs640 / (Abs640 + Abs540)] *
100%, and the chromic response, [(initial blue percentage – exposed blue percentage) /
92
initial blue percentage] * 100%, were used to characterize the parameter of
polymerization of the conjugated π backbone. To identify the irreversibility of the
system as well as the timescale at which UV polymerization is complete, a 10 mM
solution of PCDA, prepared as described above, was initially polymerized by exposure to
5 minutes of UV light (4W lamp at 254 nm). The resulting spectrum was obtained and
the blue percentage determined. The solution was then exposed to 70°C for 5 minutes to
thermally induce a chromatic transition to the red phase at which point the blue
percentage color was again obtained through spectral analysis. This process was repeated
until the blue percentage remained constant indicating complete polymerization and
irreversibility of the system.
To confirm the polymerization time necessary to achieve the greatest potential chromic
response, freshly prepared PCDA samples were exposed to UV light for periods of 5 s,
10 s, 30 s, 1 min, 2.5 min, 5 min, 7.5 min, 10 min, 15 min, 20 min, or 25 min. The blue
percentage was calculated from obtained spectra, and the solutions were then exposed to
thermal stimulus of 70°C overnight to induce complete chromatic transition. From the
resulting spectra, the exposed blue percentage was obtained to calculate chromic response.
The necessary polymerization conditions for maximal chromic response could then be
determined.
93
5.3.4 Examination of Surface Density
PCDA vesicle mixtures containing 0, 4, 8, 14, 25, 30, 50, and 100% PCDA-Trp-His-Trp
surface density were prepared as described above. Importantly, all samples were
prepared from the same initial stock solutions in order to minimize variation in spectral
response. The samples were UV polymerized for 25 minutes and the resulting blue
percentage was measured to identify the effect of surface density on assembly into planar
conjugated system and hence the propensity for steric induced red phase transitions.
5.3.5 TNT Target Exposure
Vesicle mixtures containing compositions of PCDA-Trp-His-Trp:PCDA-Trp:PCDA of
4:0:96, 2:2:96, and 4:4:92 were prepared and polymerized by 25 minutes of UV exposure.
Individual samples were separated into No TNT Exposure and TNT Exposure aliquots of
200 µL. TNT Exposure aliquots were given 4 µL of 572 µM TNT in water for a final
concentration of 11 µM TNT. No TNT Exposure aliquots were given 4 µL of water.
Samples were allowed to incubate at room temperature for 1 hour prior to visible
spectrum analysis. The resulting chromic response to TNT exposure was calculated from
the blue percentage of TNT Exposure aliquots in relation to No TNT Exposure aliquots.
94
5.3.6 Changing the Alkane Chain Length
In addition to the PCDA-Trp-His-Trp, several other TNT binding motif bearing PDA
constructs with varying alkane chain lengths were produced. This was carried out by
using the same synthesis protocol as listed above but replacing the PCDA component
with one of the following purchased from GFS Chemicals, Inc. (Powell, OH): 10, 12
tricosadiynoic acid (TCDA) and 10, 12 heptacosadiynoic acid (HCDA). The resulting
structures can be seen in Figure-38. To identify the effect of chain length TCDA-peptide
and HCDA-peptide and PCDA-peptide vesicles were created using the same protocol as
listed above but with compositions containing a 10% surface density of peptide. That is
to say, vesicle mixtures were prepared with 10% molar ratio of PDA-Trp-His-Trp in PDA.
Samples were prepared in 200 uL aliquots within a 96 well clear bottom plate for analysis
of the absorbance spectra using a Safire2 plate reader (TECAN, Männedorf, Switzerland).
The plate was shaken for 2 seconds prior to sample measurement, and each sample was
subjected to 10 readings at which point the absorbance was averaged. Samples were then
subjected to TNT at concentrations of 5.2, 9.5, 20.3, 33.6 µM. Again the absorbance
spectra were obtained after TNT exposure using the Safire2 plate reader. The resulting
chromic response to TNT exposure was calculated from the blue percentage of TNT
Exposure aliquots in relation to No TNT Exposure aliquots as described above.
95
5.3.7 Fibroblast Culture
IMR-90 human lung fibroblasts (obtained from UC Berkeley Cell Culture Facility) were
grown in Dulbecco's Minimum Essential Media (ATCC, Manassas, VA) containing fetal
bovine serum (10%) and penicillin/streptomycin (1%). Cells were incubated at 37°C in
the presence of 5% CO2 in a humidified incubator. Media was changed every 48 hours,
and cells were split as they approached 80% confluence. Before the experiment, cells
were placed in 1 mL of PBS and a cell scraper was used for detachment from the tissue
culture polystyrene (TCPS) plates. The suspended cells were place in a microcentrifuge
tube, briefly pelleted, and re-suspended at a concentration of 106 cells per 100 µL of
water.
5.3.8 Fibroblast Target Response
Vesicle containing 4% PCDA-Gly-Arg-Gly-Asp-Ser were polymerized by 25 minutes of
UV exposure. Individual samples were separated into “No Cell Exposure” and “Cell
Exposure” aliquots of 200 µL. “Cell Exposure” aliquots were given 2 µL cell stock (104
fibroblast/µL) to give a final concentration of 100 cells/µL of human fibroblasts in water.
“No Cell Exposure” aliquots were given 2 µL of water. Samples were allowed to
incubate at room temperature for 10min prior to visible spectrum analysis. The resulting
chromic response to fibroblast cell exposure was calculated from the blue percentage of
“Cell Exposure” aliquots in relation to “No Cell Exposure” aliquots.
96
5.4 Results and Discussion
5.4.1 Vesicle Characterization
Confirming the formation of PCDA-Trp-His-Trp vesicles, we characterized the formation
of PCDA based sensors using dynamic light scatter. Analysis of PCDA-Trp-His-Trp
particle hydrodynamic diameter was identified to be 162 nm on average from dynamic
light scattering measurements. The size distribution of vesicles was identified to be
mono-modal indicating that a single size of vesicle is preferred (Figure-39). These values
of vesicle diameter coincide well with previously reported PCDA vesicles diameters.170
Figure 39: Dynamic light scattering measurements analysis was used to
characterize the diameter of the range of assembled PCDA-Trp-His-Trp particles to
be on average 162 nm.
5.4.2 Maximal Chromic Response Dependence on the Extent of Polymerization
To identify the polymerization time necessary to achieve an optimal initial blue
percentage for sensing purposes, we investigated the relation between chromic response
and UV polymerization of PDA vesicles. Achieving complete polymerization of
97
assembled PDA vesicles is vital in the fabrication of consistent and effective PDA based
sensors. It is this parameter which controls the maximal chromic response that can be
achieved for a given vesicle composition. Additionally, we show that insufficient UV
polymerization can give a false impression of a chromic reversible system. Researchers
have extensively explored the irreversibility of polydiacetylene polymers and the energy
barriers associated with these thermal induced bond rotations have shown strong
dependence on the bulkiness and interactions between side-chain groups.171-173
Modification of these polydiacetylene side groups has shown it possible to create chromic
reversible supramolecular assemblies and to control the polymers thermal response
properties.172, 174, 175
Figure 40: Effect of PDA polymerization times were analyzed by a) varying
exposure of PDA vesicles to identify the conditions for achieving maximal chromic
response (trend line added as guide), and b) identifying the length of UV exposure
required for full PDA vesicle polymerization as indicated by irreversibility of the
system.
We demonstrate the irreversibility of our system as well as identification of the UV
exposure necessary for full PDA vesicle polymerization. By monitoring the visible
absorption spectra at various time intervals after heating and UV exposure, we were able
98
to qualitatively observe changes in the effective conjugation length. The blue percentage
was used to characterize the extent of π conjugated backbone planarity, where a smaller
blue percentage indicates that a larger red phase had been triggered (decreased
conjugation). The results seen in Figure-40 demonstrate the pseudo-reversibility of the
system after heat exposure if the system is insufficiently polymerized. We see that
additional UV exposure will create π conjugated PDA backbones from any previously
un-polymerized PDA. The result is a mixture of absorption spectra between previously
thermally triggered and newly polymerized PDA. The irreversible nature of the
chromatic transition allows a meta-stable blue phase to enter a more thermodynamically
stable red phase upon heating.163
After several heat and UV cycles as seen in Figure-40b,
no change in the π conjugated backbone can be identified, as the blue percentage remains
constant after 20 minutes of total UV exposure. UV polymerization for longer periods of
time does not indicate any return to the blue phase. This demonstrates that all available
polymerization sites have been reacted. We qualify the polymerization as having reached
completion after this 20 minute exposure, since after this time scale the system exhibits
its true irreversible nature in terms of chromic response.
To confirm the conditions of complete polymerization and to identify the conditions for
maximal chromic response, we analyzed the thermally triggered chromic response of
PDA samples after various levels of UV exposure. From Figure-40a, we observe that the
maximal chromic response of 60.2 ± 1.5% is achieved after 20min (~150 J/cm2) of 254
nm UV exposure. From this data, we have identified the minimum polymerization
99
conditions necessary to elicit an optimal chromic response for the given vesicle
composition.
5.4.3 Effects of side-chain composition
In order to have an effective sensor, it is necessary to incorporate recognition
motifs capable of transducing a detectable signal. We demonstrate the ability to
incorporate recognition motifs onto a PDA vesicle while identifying the important
consideration of receptor length. The ability to tune side-chain composition is essential
for controlling the surface interactions experienced by the sensor vesicle as well as the
extent of monomer alignment in self-assembled vesicles.176
We functionalized the side-
chains with different peptide moieties referred to as receptor and promoter elements.
Receptor side-chains, such as PCDA-Trp-His-Trp, are generally bulky components used
to interact with particular target molecules of interest in the surrounding environment,
which is TNT in this case. Transduction of receptor-target binding to a break in
backbone planarity is dependent on side-chain rearrangement.177, 178
Because receptor
elements can be large relative to small molecule targets, it has proven advantageous to
incorporate promoter elements, such as PCDA-Trp, to assist in cascading a side-chain
conformation change from the bound receptor-target complexes.179
This parameter, as
demonstrated in subsequent sections, improves target recognition; therefore, it is an
essential part of designing an effective PDA based sensor. Additionally, we show that
side-chain composition dictates whether self-assembled sensor vesicles will have
sufficient alignment upon polymerization to elicit a significant chromic response.
100
Figure 41: Effect of PDA functional end-group on initial blue percentage for 100%
surface functionality of PCDA (carboxyl), PCDA-Trp, and PCDA-Trp-His-Trp
Using 100% surface density of PCDA, PCDA-Trp, and PCDA-Trp-His-Trp, we
identified the steric effect of side chain size on the ability of the vesicles to retain a planar
conjugated backbone. Here we demonstrate the importance of packing ability on
controlling the extent of blue percentage. From Figure-41, we see that increasing the
steric hindrance through bulkier PDA-peptide conjugates affects the ability for effective
packing of the PDA side-chains causing a noticeable decrease in blue percentage. The
observed color of PDA is a direct result of the effective conjugation length of the
delocalized π-conjugated polymer backbones.160
Properly aligned monomer side-chain
units which undergo UV irradiation will polymerize by means of the reactive diacetylene
groups to form blue PDA arrangements which retain their original mesostructure.180
Aside from stimulus driven red shifts, misalignment within the PDA vesicle can results in
a chromatic transition due to increased HOMO-LUMO spacing of the delocalized π
101
electrons along the PDA backbone. This is directly attributed to the decrease in the
effective conjugate length from reduction in π-π stacking between adjacent chains.181, 182
With 100% Trp-His-Trp receptor surface density, we see that even prior to the
introduction of an external stimulus the vesicle based sensor has already reached its red
phase indicating that the conjugation length is insufficient to render any further chromic
response. From this, we conclude that high concentrations of bulky receptor side-chains
sterically restrict the proper alignment of monomers, thereby making such vesicles
unusable for sensing purposes. Depending on the target sensing application, it may be
essential to use bulky receptors. Therefore, it is important to consider the use of
promoter elements and the appropriate surface density to achieve an effective
composition of the desired receptor as we explore in the following sections.
5.4.4 Effects of surface density
For the sensor to be practical, the surface must be incorporated with receptive motifs at
the appropriate density to ensure a sufficiently high initial blue percentage as well as
sufficient density to elicit a change in the absorption spectra. To investigate this in
relation to steric effects, we analyzed the effect of receptor surface density on the
chromic response of the sensor. Because of steric hindrance effects, 100% surface
receptor density will prevent proper initial monomer alignment during assembly. This
results in insufficient conjugation lengths to elicit a detectable chromic response. A
visually detectable chromic response is on the order of a 15% change.183
To design an
102
effective sensor vesicle, it is important to consider the parameter of PDA-peptide
conjugate surface density which is able to allow such a chromic response. Therefore, the
surface density must be sufficiently high to provide enough transduction sites to disrupt
the planar backbone such that the chromic response is greater than 15%. Conversely, the
surface density must be low enough to ensure that the pre-exposed vesicles do not exist in
the triggered red phase due to steric hindrance, as discussed above. To analyze this
parameter, we observed the blue percentage from vesicle spectra as a function of Trp-
His-Trp receptor surface density.
Figure 42: Dependence of initial blue percentage on the surface density of vesicles
comprising various concentrations of peptide-conjugate PCDA-Trp-His-Trp. The
dashed line represents the minimum blue percentage (42%) required to have a
visually detectable chromic response of 15% change relative to the blue percentage
provided at 100% surface density.
From Figure-42, we can easily identify the extent to which steric hindrance plays a role in
breaking planarity of the conjugated π-backbone of PDA. We are particularly interested
in identifying this failure point in optical quality in which sterics prevent extended
backbone planarity after polymerization. The blue percentage decreases rapidly with
103
increasing surface density such that only vesicles with Trp-His-Trp surface receptor
densities less than 14% may be considered practical from a sensor perspective. This is
because an initial blue percentage of at least 42% is required to possibly attain the 15%
chromic response necessary for visual differentiation. Investigation of this parameter is
dependent on the sterics associated with the side-chain functionality. To design an
effective sensor vesicle, researchers must consider the surface density effect on blue
percentage for their particular surface receptor. Ideally, this parameter would be the
largest possible surface density that also allows a chromic response of 15% to be realized.
5.4.5 Evaluating the Effect of Target Size
Target size can have a significant effect on the extent of backbone rearrangement that
occurs, thereby affecting the observed color change. We evaluate the effects of target
size on the chromic response of PDA as well as the effect of utilizing promoter elements
to enhance the chromic response. Using our modular approach, we incorporated different
recognition elements onto the vesicle surface to facilitate binding of a given target. The
versatile nature of this design allowed us to detect a variety of different targets ranging
from small molecules to large cells. First, we explored the use of the peptide conjugate
PCDA-Trp-His-Trp for the purpose of TNT detection. Exposure to 11 µM of TNT
resulted in a red shift in wavelength absorption as identified from spectra analyses,
Figure-43a. Specifically, a change in blue percentage was calculated with a resulting
chromic response of 2.7%. While this demonstrated that the system can detect small
molecules, the level of chromic response is below that discernable by the naked eye. In
104
an attempt to enhance the chromic response, a promoter element (PCDA-Trp), was
incorporated at equimolar surface density with the PCDA-Trp-His-Trp receptor elements.
The use of analogous promoter elements have been reported to facilitate chromic
transition by relaying the receptor-target binding to transduction of side-chain
conformation changes resulting in breaking of backbone planarity.179
Figure-43b
demonstrates the use of the receptor/promoter system with 2% PCDA-Trp-His-Trp and
2% PCDA-Trp in the presence of 11 µM of TNT. Analysis of the spectra reveals a
chromic response of 3.5% which is only a small increase that remains below that chromic
response necessary for visual detection. The last sensor vesicle system tested in 11 µM
of TNT was comprised of 4% PCDA-Trp-His-Trp and 4% PCDA-Trp. Again the
chromic response was not visually detectable, but the positive effect of the promoter was
easily revealed in the spectrophotometer measurements showing a slight increase of
chromic response to 4.1%, see Figure-43c. Control experiments with vesicles comprised
of only 4% PCDA-Trp promoter show a chromic response of 1.1% indicating the
promoter is not actively binding.
105
Figure 43: Visible-absorption spectra of polymerized vesicles containing: a) 4
mol % Trp-His-Trp surface receptor; b) 2 mol % Trp and 2 mol % Trp-His-Trp
surface receptor; c) 4 mol % Trp and 4 mol % Trp-His-Trp surface receptor; d) 4
mol % Gly-Arg-Gly-Asp-Ser surface receptor. Solid lines represent the non-
exposed spectra while dashed lines represent spectra attained after target exposure.
Exposed spectra are normalized to the corresponding non-exposed spectra and y-
axes scaled to clarify the change in chromic response.
To test the system for larger targets, we pursued the detection of human fibroblasts.
Using our modular sensor approach we could easily incorporate the known sequence
derived from fibronectin protein for binding to cell surface receptors known as integrins.
The sequence was synthesized as the peptide conjugate, PCDA-Gly-Arg-Gly-Asp-Ser.
The spectrum shown in Figure-43d shows the effect of exposing the sensor vesicles to
100 cells/µL of fibroblasts. The average chromic response was calculated to be 36.7%,
106
which is significantly larger than that achieved in small molecule target detection. In fact,
the chromic response was easily visually detectable, as it was well above 15%. Negative
controls of 100% PCDA vesicles were subjected to 100 cells/µL of fibroblasts resulting
in a chromic response of 4.5%. This indicates the integrin binding sequence was critical
to initiating a larger spectral response. This more noticeable spectral shift seen for cell
targets compared to small molecule targets may potentially arise due to the large area
available for cell–vesicle interaction which could allow long-range structural changes in
the planar backbone to occur. With the proper consideration of composition and
assembly parameters outlined above, we have successfully demonstrated the versatility,
efficacy, and potential for this platform as a widely deployable sensor. Additionally, the
modular nature of the platform may allow the detection of a variety of targets ranging
from small molecules to large cells.
5.4.6 Effect of Alkane Chain Length on Sensitivity
By increasing the chromic response of the system over a range of concentrations, we may
effectively amplify the color change for enhanced signal detection. We showed that
alkane chain length can influence the sensitivity of the PDA based sensor thereby
providing a means of increasing chromic response. Using the same modular approach,
Trp-His-Trp was incorporated with PDA elements having shorter or longer alkane
elements as compared to PCDA. By monitoring the TNT response of HCDA, PCDA,
and TCDA conjugates of Trp-His-Trp, we were able to identify an effect of alkane chain
length on the sensitivity of the vesicle system to target binding. HCDA, PCDA, and
107
TCDA have alkanes of decreasing chain length as shown in Figure-38. As the alkane
chain length decreased, we identified an increase in sensitivity of the systems to the
presence of TNT over a range of concentrations, Figure-44. Additionally, our data
demonstrates a chromic response increase with increasing TNT concentration. This
effect recapitulates the concentration dependence of chromic response seen in other PDA
based sensors.142-144
From these experiments, the largest chromic response to TNT,
which was 5.1%, occurred at the TNT concentration of 33.5 µM using 10% TCDA-Trp-
His-Trp. Through this, we demonstrated that the alkane chain length is indeed a critical
parameter to be taken into consideration when designing a PDA based sensing system.
Figure 44: Dependence of alkane chain length on PDA-Trp-His-Trp sensitivity to
TNT target. Decreasing PDA lengths facilitate a higher chromic response over a
range of TNT concentrations.
108
5.5 Conclusion
Utilizing a modular synthesis strategy to couple recognition elements to chromic
responsive polydiacetylene elements, we created a vesicle based sensor to explore the
essential parameters that must be considered in the sensor design process. We
demonstrated that, using this modular approach, sensing systems can be created for a
variety of targets ranging from small molecules (TNT) to large biological systems
(human fibroblasts). In depth discussion of the system parameters is presented from the
aspects of synthesis (receptor and promoter characteristics), composition (surface density
control), and assembly (polymerization considerations). Specific examples of sensor
design for target response have been demonstrated. Choosing a PDA with an
appropriately short alkane chain length proved to be highly important for amplifying the
sensitivity of chromic response to the presence of the small molecule target TNT. With
the PCDA sensor vesicle systems, we were also able to identify the important role of
side-chain composition and surface density on the transduction of chromic transition
from surface interactions by observing shifts in the absorption spectra. Additionally, we
identified a tradeoff in surface density considerations between proper monomer
alignment and sufficiently high receptor density to elicit a visually detectable chromic
response. While we see that low surface receptor densities can present a potential
limitation to these systems, the advantages of easy and low cost manufacturing, simple
analysis, and modular incorporation of target recognition elements make it a viable
candidate for a widely deployable chemical sensing system.
109
Chapter 6: Summary and Outlook
6.1 Molecular Recognition Elements
By utilizing phage display, molecular recognition elements have been developed for TNT
and DNT. The TNT recognition element was found to be a peptide with a conserved
amino acid sequence with high similarity to the binding site of pentaerythritol tetranitrate
reductase, one of nature’s own TNT binding proteins. Such TNT binding proteins
generally exhibit a highly conserved tryptophan residue which is involved in the binding
event. Similarly, the studies outline in this work also demonstrated the importance of
tryptophan in binding to TNT. Through mutational analysis of our TNT binding motif as
well as through NMR spectroscopic methods, we have demonstrated the role of
multivalent binding involving neighboring tryptophan residues. There are several
mechanisms by which tryptophan can contribute to TNT binding. Similar to what is seen
for the PETN-reductase, we have found that our TNT binding motif utilizes the aromatic
stacking between tryptophan and the ring structures of nitroaromatics. Furthermore, this
work demonstrates the successful evolutionary screening against not only TNT but also
DNT. In doing so, phage display was identified as successful for developing selective
peptide based receptors for various small molecules. Additionally, the multivalent
binding was identified to be critical to achieving the necessary selectivity.
While biology is extremely effective at creating specific molecular interaction,
researchers still struggle to achieve molecular recognition. Generally, the occurrence of
110
such recognition has arisen out of necessity, as most cellular processes rely on highly
important binding events. To identify the mechanisms of such molecular recognition
events, we need to achieve a working knowledge of biomolecular interactions. While
this exists to DNA-DNA interaction, protein-protein and protein-small molecule
interactions do not have a formulaic answer. As such, research will continue to achieve
incremental gains in the knowledge base of biomolecular interactions via spectroscopy
and crystallization techniques. From this, the full understanding of model systems may
allow a more comprehensive level of intermolecular host/guest interaction mechanisms to
be achieved.
The concept of molecular recognition requires a receptor (host) molecule to participate in
non-covalent bonding with a specific ligand (guest). Several coupling events can occur
between a given receptor-ligand pair including hydrogen bonding, electrostatic
interactions, aromatic interactions, or Van der Waals interactions. In a molecular
recognition event, it is typical for several combinations of these interactions to occur
simultaneously. This multi-valence binding is critical for receptors to discriminate
between different potential guest target molecules. This differentiation for a specific
guest is deemed the selectivity of a receptor. Several factors have been found to
contribute to a selective molecular recognition event including: complementary
molecular shape, appropriate structural rigidity, and appropriate available binding sites.
By optimizing the number of potential interaction as well as the shape of the binding site
in relation to the target molecule, a proper distance between functional groups can be
achieved for effective hydrogen bonding and aromatic interactions. However, if size
111
exclusion becomes evident, a receptor-ligand binding affinity will be greatly reduced as it
may only have binding contributions from long-rand electrostatic interactions. By
looking at the effects of steric hindrance, one can see that size exclusion helps to
contribute to selectivity in a molecular recognition event.
Aside from the role of steric, an effective molecular recognition element must have an
optimized level of structural rigidity. Particularly, an adequate level of ridigity in the
binding site can minimize entropic loss that would occur upon rearrangement of a main
chain (ie. peptide backbone) in a recognition event. The locking of such rotating bonds
would add to this loss in binding energy. Typical conformation changes of a peptide
backbone are less that 0.1nm, though functional side-chain dynamics can be significantly
altered and restricted upon binding. Supramolecular chemists have taken this into account
and have utilized cyclic structure as receptor templates for a variety of target molecules.
In addition to the above consideration of shape and structural plasticity, the use of
appropriate functional groups continues to be a key component for the design of selective
receptors. The selective recognition of a target may necessitate a binding site which
contains acid functionalities for interacting with amine and amide functions on a target
molecule. Hydrogen bonding is generally a main contributor to recognition events given
as these interactions occur for a variety of polar functionalities including carboxylic
acids, carbamates and carboxylic esters. Aromatic functionalities offer another set of
binding mechanisms including cation-π, anion-π , and π–π interactions. Charge transfer
interaction, dipole-dipole interactions, and van der Waals interactions also may contribute
112
to a selective recognition event. By tuning the location of such functional groups to be
complimentary to those of the target molecule, researchers may design effective
recognition elements for a target of interest. It is the hope of the molecular recognition
research community to implement principles gained from research such as this to develop
a generalized set of mechanisms to predict selective target binding.
6.2 Sensor Coatings
Development of a chemical sensor coating, which holds the ability to selectively
discriminate between target and non-target chemicals, is a highly desirable technology.
Such technology would facilitate the detection of the surrounding chemical vapor
landscape. This is of importance to identify information about potential environmental
contaminants and health threats, including industrial carcinogen, explosive, or pesticides.
Such target molecules of interest include volatile small molecules which typically have a
molecular weight in the range of 50-500 g/mol. Current chemical sensors, while very
sensitive, have a lack of proper selectivity. As such, the main challenge in chemical
sensing continues to be creating a chemical detection system which is successful in the
midst of various potential interfering agents that may elicit a false positive.
Utilizing the peptides recognized through phage display, we implemented a strategy for
creating selective coatings for sensor platform technologies. Specifically, we
incorporated a receptor which binds to TNT onto a PEGM polymer via an easy coupling
mechanism. Using an epoxy group present on one of the monomers of PEGM, primary
113
amino groups present on the receptor were capable of covalently attaching, and this
attachment was found to be highly stable. Using both XPS and GC/MS, we identified the
receptor as stably attached and capable of retaining its target selectivity. To further
explore the use of the coating, the PEGM/TNT receptor conjugate was implemented on a
real-time sensing platform (QCM). We identified the selective detection of the explosive
TNT in water by a decrease in resonance frequency of the QCM, while there was no
change in resonance frequency when the coated QCM crystal was exposed to DNT
solution. The potential application of this sensitive and selective liquid detection of TNT
is particularly appealing for analysis of contamination of ground waters near ammunition
depots. We foresee the use of such coatings for various sensing applications which can
be tailored by changing the selectivity motif.
Several other interesting coating technologies have recently been implemented including
polymeric binding pockets from molecular imprinting, de novo designed receptors using
supramolecular chemistry, and engineering of existing recognition elements from
biological receptors. While molecular imprinting has offered impressive selectivity by
demonstrating the ability to discriminate between enantiomers, there are inherent
limitations in using a cross-linked polymer for a QCM sensor. Particularly, swelling of
the polymer in different environments can easily cause a false-positive signal.
Additionally, the non-uniformity of cavity formation results in “poly-clonal” binding
sited which provides mechanisms for non-specific interactions in a complex targeting
environment. The coating technologies developed using supramolecular chemistry do
not fall victim to this, as the de novo designed receptors are uniform. Using
114
computational and theoretical means, supramolecular chemists have been able to design
receptors specific to a target by implementing a complimentary curvature, size, and
functionality. This approach offers high affinity, though intensive design and trial-and-
error are required, as such only a small number of specialized receptors have been
accomplished via this coating approach. A more generalized approach, which has shown
similar efficacy, utilizes engineering of existing protein binding site. By modifying the
molecular recognition motif of natural occurring receptor proteins, researchers have
found ways of altering specificity toward a particular target of interest. Using
mutagenesis and computational modeling, highly selective receptive motifs have been
identified though there use on chemical sensors is not without drawback. Particularly,
there implementation onto sensing platforms has seen limited success due to stability
issue of large protein domains over long time scales and un-natural environments.
Perhaps, these technologies will soon provide improved approaches to chemical sensor
coatings, though progress in overcoming fundamental sensor coating problems must be
taken into account, such as overcoming humidity response issues, eliminating non-
specific interaction, and maintaining stability in a variety of sensing environments.
6.3 Sensing Platforms
The current standard for chemic sensing systems are those based on chromatography-
mass spectroscopy, as these systems are capable of both high sensitivity and high
selectivity. GC/MS sensing systems are currently employed at points of security
screening, although there large size, expensive price, and complex data analysis has
115
hindered there widespread use. To overcome this problem, researchers have explored the
creation of inexpensive and portable chemical sensing systems. Some of the most
promising candidates for widely deployable chemical sensing systems utilize a
technology based on the transduction of ligand-receptor binding into an electrical signal.
Many target binding based sensing platforms utilize an array of polymer coatings from
which a signature pattern may be generated for a give target molecule. This interesting
approach works very well for pure target exposure, though as these polymers are non-
specific they may not be effective in a real world sensing environment in which there are
a variety of background chemical signals. In particular, this non-specificity means the
affinity ratio among different analytes is not discernable and the binding of background
chemicals would create too much uncertainty in the signal. While we have focused
mainly on the creation of selective coating layers, it is important to note there are several
highly sensitive chemical sensing platforms available. It was necessary to look closely at
the sensing mechanisms of theses sensors in order to effectively create a coating that
works well in relation to the signal transduction mechanism used by these sensors. The
sensitivity of the transducer to ligand-receptor binding events depends heavily on which
sensing platform is to be used. Transduction can occur via changes in electrical
resistance of a chemoresistor, a frequency shift in mechanical resonance devices such as
the quartz crystal microbalance used in this work, frequency shift in the optical
resonances of platforms such a SPR (surface plasmon resonance), or even surface stress
based transduction for creating structural deflections such as in cantilever based sensors.
116
Each of the above chemical sensing systems are effective at providing highly sensitive
detection of chemicals. By incorporating such devices with a sensor coating, we can
provide a mechanism for selectivity of the binding event. Future work may utilize these
sensor coatings with work being develop to exploit several simultaneous transduction
mechanisms. For a ligand-receptor interaction event, binding may result in changing a
local dielectric constant, causing an addition of mass, and generation of intermolecular
forces. By creating a device capable of detecting each of these physical changes
simultaneously, one could gains a variety of binding information to overcome the
challenges associated with selective molecular recognition. The micromembrane and
microcantilever systems being developed in our laboratory may provide a means of
identifying several of these changes simultaneously. For instance, a change in mass
addition or surface stress may produce changes in the resonant frequency detectable by
capacitance measurements via a micromembrane sensor. Capacitance changes due to
dielectric changes from target analyte replace of water may also be used simultaneously
in such a sensor. Incorporation of multiple modes of target selective signal
discrimination will remain critical to achieving a robust sensing platform.
With this in mind, the knowledge gained from our research has provided a way to
overcome the poor selectivity of conventional polymer-based coating. By utilizing
sequence-specific heteropolymers akin to biological mechanisms of achieving selectivity,
we have utilized peptides rich in structure and chemical functionality to find a receptor
for the explosive TNT. Utilizing a variety of polymers, we identified an effective coating
which minimizes the humidity response while allowing the receptor to remain specific for
117
TNT. We believe this coating could be used on a variety of existing sensing platforms in
addition to QCM which has already been demonstrated. Additionally, we utilized a
modular synthesis strategy to create a colorimetric sensor via chromic responsive
polydiacetylene elements liked to a receptor of interest. We demonstrated that, using this
modular approach, sensing systems can be created for a variety of targets ranging from
small molecules (TNT) to large biological systems (human fibroblasts). By optimizing
various parameters of the sensor development, we were able to increase the sensitivity of
the sensor to TNT. We believe the advantages of easy and low cost manufacturing,
simple analysis, and modular incorporation of target recognition elements make this
strategy a viable approach for the development of widely deployable chemical sensing
systems.
118
Bibliography
1. Yang, Y.-S.; Swager, T. M. J. Am. Chem. Soc. 1998, 120,, 11864-11873.
2. Mulchandani, A.; Chen, W.; Mulchandani, P.; Wang, J.; Rogers, K. R. Biosensors
& Bioelectronics 2001, 16, (4-5), 225-230.
3. Turner, A. P. F.; Magan, N. Nature Reviews Microbiology 2004, 2, (2), 161-166.
4. Venugopal, V. Biosensors & Bioelectronics 2002, 17, (3), 147-157.
5. Berezkin, V. B.; Drugov, Y. S.; Drugov, Y., Gas Chromatography in Air
Pollution Analysis. Elsevier Science Ltd.: 1991.
6. Baldini, F.; Martellucci, S.; Chester, A. N.; Homola, J., Optical Chemical Sensors.
Springer Verlag 2006.
7. Burg, T. P.; Godin, M.; Knudsen, S. M.; Shen, W.; Carlson, G.; Foster, J. S.;
Babcock, K.; Manalis, S. R. Nature 2007, 446, (7139), 1066-1069.
8. Thundat, T.; Majumdar, A., Microcantilevers for Physical, Chemical, and
Biological Sensing. In Sensors and Sensing in Biology and Engineering, F. G. Barth, J. A.
C. H., T.W. Seecomb, Ed. Springer-Verlag: New York, 2003.
9. Grate, J. W.; Patrash, S. J.; Abraham, M. H.; Du, C. M. Analytical Chemistry
1996, 68, (5), 913-917.
10. Patel, S. V.; Mlsna, T. E.; Fruhberger, B.; Klaassen, E.; Cemalovic, S.; Baselt, D.
R. Sensors and Actuators B-Chemical 2003, 96, (3), 541-553.
11. Hagleitner, C.; Hierlemann, A.; Lange, D.; Kummer, A.; Kerness, N.; Brand, O.;
Baltes, H. Nature 2001, 414, (6861), 293-296.
12. Freund, M. S.; Lewis, N. S. Proceedings of the National Academy of Sciences of
the United States of America 1995, 92, (7), 2652-2656.
13. Shekhawat, G.; Tark, S. H.; Dravid, V. P. Science 2006, 311, (5767), 1592-1595.
14. Suehle, J. S.; Cavicchi, R. E.; Gaitan, M.; Semancik, S. Ieee Electron Device
Letters 1993, 14, (3), 118-120.
15. Hierlemann, A.; Ricco, A. J.; Bodenhofer, K.; Dominik, A.; Gopel, W. Analytical
Chemistry 2000, 72, (16), 3696-3708.
16. Rakow, N. A.; Suslick, K. S. Nature 2000, 406, (6797), 710-713.
17. Looger, L. L.; Dwyer, M. A.; Smith, J. J.; Hellinga, H. W. Nature 2003, 423,
(6936), 185-190.
18. Suman, M.; Freddi, M.; Massera, C.; Ugozzoli, F.; Dalcanale, E. Journal of the
American Chemical Society 2003, 125, (40), 12068-12069.
19. Cho, W. S.; Moon, S. I.; Paek, K. K.; Lee, Y. H.; Park, J. H.; Ju, B. K. Sensors
and Actuators B-Chemical 2006, 119, (1), 180-185.
20. Sant, W.; Pourciel, M. L.; Launay, J.; Do Conto, T.; Martinez, A.; Temple-Boyer,
P. Sensors and Actuators B-Chemical 2003, 95, (1-3), 309-314.
21. Yan, S. L.; Fang, Y. J.; Gao, Z. X. Biosensors & Bioelectronics 2007, 22, (6),
1087-1091.
22. Fechete, A. C.; Wlodarski, W.; Kalantar-Zadeh, K.; Holland, A. S.; Antoszewski,
J.; Kaciulis, S.; Pandolfi, L. Sensors and Actuators B-Chemical 2006, 118, (1-2), 362-367.
23. Sadek, A. Z.; Wlodarski, W.; Shin, K.; Kaner, R. B.; Kalantar-zadeh, K.
Nanotechnology 2006, 17, (17), 4488-4492.
24. Esteban, O.; Gonzalez-Cano, A.; Diaz-Herrera, N.; Navarrete, M. C. Optics
Letters 2006, 31, (21), 3089-3091.
119
25. Shen, J.; Mandelis, A.; Ashe, T. International Journal of Thermophysics 1998, 19,
(2), 579-593.
26. Satyanarayana, S.; McCormick, D. T.; Majumdar, A. Sensors and Actuators B-
Chemical 2006, 115, (1), 494-502.
27. Yue, M.; Lin, H.; Dedrick, D. E.; Satyanarayana, S.; Majumdar, A.; Bedekar, A.
S.; Jenkins, J. W.; Sundaram, S. Journal of Microelectromechanical Systems 2004, 13,
(2), 290-299.
28. Matz, M. V. Nature Biotechnology 1999, 17, (12), 1227-1227.
29. Grate, J. W.; Abraham, M. H. Sensors and Actuators B-Chemical 1991, 3, (2), 85-
111.
30. Li, M.; Tang, H. X.; Roukes, M. L. Nature Nanotechnology 2007, 2, (2), 114-120.
31. Ponzoni, A.; Comini, E.; Sberveglieri, G.; Zhou, J.; Deng, S. Z.; Xu, N. S.; Ding,
Y.; Wang, Z. L. Applied Physics Letters 2006, 88, (20).
32. Gouma, P. I.; Prasad, A. K.; Iyer, K. K. Nanotechnology 2006, 17, (4), S48-S53.
33. Wang, H. T.; Kang, B. S.; Ren, F.; Tien, L. C.; Sadik, P. W.; Norton, D. P.;
Pearton, S. J.; Lin, J. Applied Physics Letters 2005, 86, (24).
34. Pengfei, Q. F.; Vermesh, O.; Grecu, M.; Javey, A.; Wang, O.; Dai, H. J.; Peng, S.;
Cho, K. J. Nano Letters 2003, 3, (3), 347-351.
35. Kong, J.; Franklin, N. R.; Zhou, C. W.; Chapline, M. G.; Peng, S.; Cho, K. J.; Dai,
H. J. Science 2000, 287, (5453), 622-625.
36. Cosnier, S. Electroanalysis 2005, 17, (19), 1701-1715.
37. Ahmad, M. N.; Ismail, Z.; Chew, O. S.; Islam, A.; Shakaff, A. Y. M. Sensors
2006, 6, (10), 1333-1344.
38. Strike, D. J.; Meijerink, M. G. H.; Koudelka-Hep, M. Fresenius Journal of
Analytical Chemistry 1999, 364, (6), 499-505.
39. Maksymiuk, K. Electroanalysis 2006, 18, (16), 1537-1551.
40. Barron, A. E.; Zuckermann, R. N. Current Opinion in Chemical Biology 1999, 3,
(6), 681-687.
41. Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M.
Chemical Reviews 2005, 105, (4), 1103-1169.
42. Schreiber, F. Journal of Physics-Condensed Matter 2004, 16, (28), R881-R900.
43. Sandhyarani, N.; Pradeep, T. International Reviews in Physical Chemistry 2003,
22, (2), 221-262.
44. Adamson, A. W., Gast, A.P., Physical Chemistry of Surfaces. 6th ed.; John Wiley
& Sons: New York, 1997; Vol. 1.
45. Ulman, A. Chemical Reviews 1996, 96, (4), 1533-1554.
46. Yan, L.; Huck, W. T. S.; Whitesides, G. M. Journal of Macromolecular Science-
Polymer Reviews 2004, C44, (2), 175-206.
47. Spivak, D. A.; Campbell, J. Analyst 2001, 126, (6), 793-797.
48. Atwood, J. L.; Gokel, G. W. Nature 1991, 354, (6352), 354-355.
49. Gutsche, C. D.; Alam, I.; Iqbal, M.; Mangiafico, T.; Nam, K. C.; Rogers, J.; See,
K. A. Journal of Inclusion Phenomena 1989, 7, (1), 61-72.
50. Lehn, J. M. Science 1985, 227, (4689), 849-856.
51. Cram, D. J. Angewandte Chemie-International Edition in English 1988, 27, (8),
1009-1020.
120
52. Ballester, P.; Shivanyuk, A.; Far, A. R.; Rebek, J. Journal of the American
Chemical Society 2002, 124, (47), 14014-14016.
53. Shoemaker, B. A.; Portman, J. J.; Wolynes, P. G. Proceedings of the National
Academy of Sciences of the United States of America 2000, 97, (16), 8868-+.
54. Cherfils, J.; Janin, J. Current Opinion in Structural Biology 1993, 3, (2), 265-269.
55. Lee, A. L.; Kinnear, S. A.; Wand, A. J. Nature Structural Biology 2000, 7, (1),
72-77.
56. Takeuchi, T.; Haginaka, J. Journal of Chromatography B-Analytical Technologies
in the Biomedical and Life Sciences 1999, 728, (1), 1-20.
57. Lehn, J. M. Angewandte Chemie-International Edition in English 1988, 27, (1),
89-112.
58. Rebek, J. Science 1987, 235, (4795), 1478-1484.
59. Sergeeva, A.; Kolonin, M. G.; Molldrem, J. J.; Pasqualini, R.; Arap, W. Advanced
Drug Delivery Reviews 2006, 58, (15), 1622-1654.
60. Degraaf, M. E.; Miceli, R. M.; Mott, J. E.; Fischer, H. D. Gene 1993, 128, (1), 13-
17.
61. Luchansky, S. J.; Lansbury, P. T.; Stein, R. L. Biochemistry 2006, 45, (49),
14717-14725.
62. Wennemers, H. Chimia 2003, 57, (5), 237-240.
63. Richman, S. A.; Healan, S. J.; Weber, K. S.; Donermeyer, D. L.; Dossett, M. L.;
Greenberg, P. D.; Allen, P. M.; Kranz, D. M. Protein Engineering Design & Selection
2006, 19, (6), 255-264.
64. Mathonet, P.; Fastrez, J. Current Opinion in Structural Biology 2004, 14, (4),
505-511.
65. Hartley, O. Journal of Receptor and Signal Transduction Research 2002, 22, (1-
4), 373-392.
66. Dekoster, H. S.; Amons, R.; Benckhuijsen, W. E.; Feijlbrief, M.; Schellekens, G.
A.; Drijfhout, J. W. Journal of Immunological Methods 1995, 187, (1), 179-188.
67. Falciani, C.; Lozzi, L.; Pini, A.; Bracci, L. Chem. Biol. 2005, 12, 417-426.
68. Liu, F.; Liu, X.; Ng, S. C.; Chan, H. S. O. Sensors and Actuators B-Chemical
2006, 113, (1), 234-240.
69. Soncini, P.; Bonsignore, S.; Dalcanale, E.; Ugozzoli, F. Journal of Organic
Chemistry 1992, 57, (17), 4608-4612.
70. Vincenti, M.; Pelizzetti, E.; Dalcanale, E.; Soncini, P. Pure and Applied
Chemistry 1993, 65, (7), 1507-1512.
71. Dalcanale, E.; Hartmann, J. Sensors and Actuators B-Chemical 1995, 24, (1-3),
39-42.
72. Rothlisberger, D.; Khersonsky, O.; Wollacott, A. M.; Jiang, L.; DeChancie, J.;
Betker, J.; Gallaher, J. L.; Althoff, E. A.; Zanghellini, A.; Dym, O.; Albeck, S.; Houk, K.
N.; Tawfik, D. S.; Baker, D. Nature 2008, 453, (7192), 190-U4.
73. Bloom, J. D.; Meyer, M. M.; Meinhold, P.; Otey, C. R.; MacMillan, D.; Arnold, F.
H. Current Opinion in Structural Biology 2005, 15, (4), 447-452.
74. Drummond, D. A.; Iverson, B. L.; Georgiou, G.; Arnold, F. H. Journal of
Molecular Biology 2005, 350, (4), 806-816.
121
75. Armbruster, B. N.; Li, X.; Pausch, M. H.; Herlitze, S.; Roth, B. L. Proceedings of
the National Academy of Sciences of the United States of America 2007, 104, (12), 5163-
5168.
76. Petrenko, V. A.; Smith, G. P.; Gong, X.; Quinn, T. Protein Engineering 1996, 9,
(9), 797-801.
77. Smith, G. P.; Petrenko, V. A. Chem. Rev. 1997, 97, 391-410.
78. Lee, S.-W.; Mao, C.; Flynn, C. E.; Belcher, A. M. Science 2002, 892-895.
79. Whaley, S. R.; English, D. S.; Hu, E. L.; Barbara, P. F.; Belcher, A. M. Nature
2000, 405, (6787), 665-668.
80. Ellington, A. D.; Szostak, J. W. Nature 1990, 346, (6287), 818-822.
81. Sussman, D.; Nix, J. C.; Wilson, C. Nature Structural Biology 2000, 7, (1), 53-57.
82. Roscoe, H. K.; Clemitshaw, K. C. Science 1997, 276, (5315), 1065-1072.
83. Di Natale, C.; Macagnano, A.; Davide, F.; D'Amico, A.; Paolesse, R.; Boschi, T.;
Faccio, M.; Ferri, G. Sensors and Actuators B-Chemical 1997, 44, (1-3), 521-526.
84. Betz, N.; Clochard, M. C.; Goncalves, M.; Bittencourt, C.; Pireaux, J. J.; Gionnet,
K.; Deleris, G.; Le Moel, A. Nuclear Instruments & Methods in Physics Research,
Section B (Beam Interactions with Materials and Atoms) 2005, 236, (1-4), 208-15.
85. Casalini, R.; Kilitziraki, M.; Wood, D.; Petty, M. C. Sensors and Actuators B-
Chemical 1999, 56, (1-2), 37-44.
86. Giordano, M.; Russo, M.; Cusano, A.; Cutolo, A.; Mensitieri, G.; Nicolais, L.
Applied Physics Letters 2004, 85, (22), 5349-5351.
87. Grate, J. W.; Nelson, D. A.; Skaggs, R. Analytical Chemistry 2003, 75, (8), 1868-
1879.
88. Janata, J.; Josowicz, M. Nature Materials 2003, 2, (1), 19-24.
89. Staii, C.; Johnson, A. T. Nano Letters 2005, 5, (9), 1774-1778.
90. Naik, R. R.; Stringer, S. J.; Agarwal, G.; Jones, S. E.; Stone, M. O. Nature
Materials 2002, 1, (3), 169-172.
91. Sarikaya, M.; Tamerler, C.; Jen, A. K. Y.; Schulten, K.; Baneyx, F. Nature
Materials 2003, 2, (9), 577-585.
92. Sano, K. I.; Shiba, K. Journal of the American Chemical Society 2003, 125, (47),
14234-14235.
93. McCafferty, J.; Griffiths, A. D.; Winter, G.; Chiswell, D. J. Nature 1990, 348,
(6301), 552-554.
94. Dumas, D. P.; Durst, H. D.; Landis, W. G.; Raushel, F. M.; Wild, J. R. Archives
of Biochemistry and Biophysics 1990, 277, (1), 155-159.
95. Edwards, F. L. T., Paul B. . International Journal of Environmental Research
and Public Health 2005, 2, (3), 430-441.
96. Dong, Y. J.; Bartlam, M.; Sun, L.; Zhou, Y. F.; Zhang, Z. P.; Zhang, C. G.; Rao,
Z. H.; Zhang, X. E. Journal of Molecular Biology 2005, 353, (3), 655-663.
97. Ngehngwainbi, J.; Foley, P. H.; Kuan, S. S.; Guilbault, G. G. Journal of the
American Chemical Society 1986, 108, (18), 5444-5447.
98. Mulchandani, P.; Mulchandani, A.; Kaneva, I.; Chen, W. Biosensors &
Bioelectronics 1999, 14, (1), 77-85.
99. Rogers, K. R.; Wang, Y.; Mulchandani, A.; Mulchandani, P.; Chen, W.
Biotechnology Progress 1999, 15, (3), 517-521.
122
100. Mulchandani, A.; Mulchandani, P.; Chen, W.; Wang, J.; Chen, L. Analytical
Chemistry 1999, 71, (11), 2246-2249.
101. Grun, J.; Revell, J. D.; Conza, M.; Wennemers, H. Bioorganic & Medicinal
Chemistry 2006, 14, (18), 6197-6201.
102. Khan, H.; Barna, T.; Harris, R. J.; Bruce, N. C.; Barsukov, I.; Munro, A. W.;
Moody, P. C. E.; Scrutton, N. S. Journal of Biological Chemistry 2004, 279, (29), 30563-
30572.
103. Little, J. R.; Eisen, H. N. Biochemistry 1967, 6, (10), 3119-&.
104. Dower, S. K.; Gettins, P.; Jackson, R.; Dwek, R. A.; Givol, D. Biochemical
Journal 1978, 169, (1), 179-188.
105. Guan, X. Y.; Gu, L. Q.; Cheley, S.; Braha, O.; Bayley, H. Chembiochem 2005, 6,
(10), 1875-1881.
106. McGaughey, G. B.; Gagne, M.; Rappe, A. K. Journal of Biological Chemistry
1998, 273, (25), 15458-15463.
107. Khan, H.; Harris, R. J.; Barna, T.; Craig, D. H.; Bruce, N. C.; Munro, A. W.;
Moody, P. C. E.; Scrutton, N. S. Journal of Biological Chemistry 2002, 277, (24), 21906-
21912.
108. Pinnaduwage, L. A.; Boiadjiev, V.; Hawk, J. E.; Thundat, T. Applied Physics
Letters 2003, 83, (7), 1471-1473.
109. Rittfeldt, L. Analytical Chemistry 2001, 73, (11), 2405-2411.
110. Lenchitz, C.; Velicky, R. W. Journal of Chemical and Engineering Data 1970, 15,
(3), 401-&.
111. Keeling-Tucker, T.; Rakic, M.; Spong, C.; Brennan, J. D. Chemistry of Materials
2000, 12, (12), 3695-3704.
112. Otsuka, H.; Nagasaki, Y.; Kataoka, K. Current Opinion in Colloid & Interface
Science 2001, 6, (1), 3-10.
113. Pushkarsky, M. B.; Dunayevskiy, I. G.; Prasanna, M.; Tsekoun, A. G.; Go, R.;
Patel, C. K. N. Proceedings of the National Academy of Sciences of the United States of
America 2006, 103, (52), 19630-19634.
114. Ro, K. S.; Venugopal, A.; Adrian, D. D.; Constant, D.; Qaisi, K.; Valsaraj, K. T.;
Thibodeaux, L. J.; Roy, D. Journal of Chemical and Engineering Data 1996, 41, (4),
758-761.
115. Pennington, J. C.; Brannon, J. M. Thermochimica Acta 2002, 384, (1-2), 163-172.
116. Hannink, N.; Rosser, S. J.; French, C. E.; Basran, A.; Murray, J. A. H.; Nicklin,
S.; Bruce, N. C. Nature Biotechnology 2001, 19, (12), 1168-1172.
117. Kaiser, E.; Colescot.Rl; Bossinge.Cd; Cook, P. I. Analytical Biochemistry 1970,
34, (2), 595-&.
118. Seah, M. P.; Gilmore, L. S.; Beamson, G. Surface and Interface Analysis 1998, 26,
(9), 642-649.
119. Clark, D. T.; Thomas, H. R. Journal of Polymer Science Part a-Polymer
Chemistry 1978, 16, (4), 791-820.
120. Xiao, S. J.; Textor, M.; Spencer, N. D.; Sigrist, H. Langmuir 1998, 14, (19), 5507-
5516.
121. Olivares, O.; Likhanova, N. V.; Gomez, B.; Navarrete, J.; Llanos-Serrano, M. E.;
Arce, E.; Hallen, J. M. Applied Surface Science 2006, 252, (8), 2894-2909.
123
122. Rezania, A.; Johnson, R.; Lefkow, A. R.; Healy, K. E. Langmuir 1999, 15, (20),
6931-6939.
123. Clochard, M. C.; Betz, N.; Goncalves, M.; Bittencourt, C.; Pireaux, J. J.; Gionnet,
K.; Deleris, G.; Le Moel, A. Nuclear Instruments & Methods in Physics Research Section
B-Beam Interactions with Materials and Atoms 2005, 236, 208-215.
124. Provatas, A.; Matisons, J. G.; Smart, R. S. Langmuir 1998, 14, (7), 1656-1663.
125. Tanaka, Y.; Doi, H.; Iwasaki, Y.; Hiromoto, S.; Yoneyama, T.; Asami, K.; Imai,
H.; Hanawa, T. Materials Science & Engineering C-Biomimetic and Supramolecular
Systems 2007, 27, (2), 206-212.
126. Willey, T. M.; Vance, A. L.; van Buuren, T.; Bostedt, C.; Terminello, L. J.;
Fadley, C. S. Surface Science 2005, 576, (1-3), 188-196.
127. Zubavichus, Y.; Zharnikov, M.; Yang, Y. J.; Fuchs, O.; Heske, C.; Umbach, E.;
Tzvetkov, G.; Netzer, F. P.; Grunze, M. Journal of Physical Chemistry B 2005, 109, (2),
884-891.
128. Sauerbrey, G. Zeitschrift Fur Physik 1959, 155, (2), 206-222.
129. Nath, N.; Chilkoti, A. Analytical Chemistry 2002, 74, (3), 504-509.
130. Liu, J. W.; Lu, Y. Journal of the American Chemical Society 2003, 125, (22),
6642-6643.
131. Shimada, T.; Ookubo, K.; Komuro, N.; Shimizu, T.; Uehara, N. Langmuir 2007,
23, (22), 11225-11232.
132. Chen, S. J.; Huang, Y. F.; Huang, C. C.; Lee, K. H.; Lin, Z. H.; Chang, H. T.
Biosensors & Bioelectronics 2008, 23, (11), 1749-1753.
133. Chah, S.; Hammond, M. R.; Zare, R. N. Chemistry & Biology 2005, 12, (3), 323-
328.
134. Sugunan, A.; Thanachayanont, C.; Dutta, J.; Hilborn, J. G. Science and
Technology of Advanced Materials 2005, 6, (3-4), 335-340.
135. Haes, A. J.; Zou, S. L.; Schatz, G. C.; Van Duyne, R. P. Journal of Physical
Chemistry B 2004, 108, (22), 6961-6968.
136. Lucas, L. J.; Chesler, J. N.; Yoon, J. Y. Biosensors & Bioelectronics 2007, 23, (5),
675-681.
137. Medintz, I. L.; Clapp, A. R.; Mattoussi, H.; Goldman, E. R.; Fisher, B.; Mauro, J.
M. Nature Materials 2003, 2, (9), 630-638.
138. Liu, J. W.; Lee, J. H.; Lu, Y. Analytical Chemistry 2007, 79, (11), 4120-4125.
139. Wang, C. G.; Ma, Z. F. Analytical and Bioanalytical Chemistry 2005, 382, (7),
1708-1710.
140. Kolusheva, S.; Kafri, R.; Katz, M.; Jelinek, R. Journal of the American Chemical
Society 2001, 123, (3), 417-422.
141. Su, Y. L.; Li, J. R.; Jiang, L. Colloids and Surfaces B-Biointerfaces 2004, 38, (1-
2), 29-33.
142. Biesalski, M.; Tu, R.; Tirrell, M. V. Langmuir 2005, 21, (13), 5663-5666.
143. Charych, D. H.; Nagy, J. O.; Spevak, W.; Bednarski, M. D. Science 1993, 261,
(5121), 585-588.
144. Kim, J. M.; Lee, Y. B.; Yang, D. H.; Lee, J. S.; Lee, G. S.; Ahn, D. J. Journal of
the American Chemical Society 2005, 127, (50), 17580-17581.
145. Ma, G. Y.; Cheng, Q. Langmuir 2005, 21, (14), 6123-6126.
146. Kolusheva, S.; Shahal, T.; Jelinek, R. Biochemistry 2000, 39, (51), 15851-15859.
124
147. Ahn, D. J.; Kim, J. M. Accounts of Chemical Research 2008, 41, (7), 805-816.
148. Lee, S. W.; Kang, C. D.; Yang, D. H.; Lee, J. S.; Kim, J. M.; Ahn, D. J.; Sim, S. J.
Advanced Functional Materials 2007, 17, (13), 2038-2044.
149. Reppy, M. A.; Pindzola, B. A. Chemical Communications 2007, (42), 4317-4338.
150. Song, J.; Cisar, J. S.; Bertozzi, C. R. Journal of the American Chemical Society
2004, 126, (27), 8459-8465.
151. Sukwattanasinitt, M.; Wang, X.; Li, L.; Jiang, X.; Kumar, J.; Tripathy, S. K.;
Sandman, D. J. Chemistry of Materials 1998, 10, (1), 27-29.
152. Okada, S.; Peng, S.; Spevak, W.; Charych, D. Accounts of Chemical Research
1998, 31, (5), 229-239.
153. Carpick, R. W.; Sasaki, D. Y.; Marcus, M. S.; Eriksson, M. A.; Burns, A. R.
Journal of Physics-Condensed Matter 2004, 16, (23), R679-R697.
154. Kobayashi, A.; Kobayashi, H.; Tokura, Y.; Kanetake, T.; Koda, T. Journal of
Chemical Physics 1987, 87, (8), 4962-4966.
155. Tanaka, H.; Gomez, M. A.; Tonelli, A. E.; Thakur, M. Macromolecules 1989, 22,
(3), 1208-1215.
156. Lu, Y. F.; Yang, Y.; Sellinger, A.; Lu, M. C.; Huang, J. M.; Fan, H. Y.; Haddad,
R.; Lopez, G.; Burns, A. R.; Sasaki, D. Y.; Shelnutt, J.; Brinker, C. J. Nature 2001, 410,
(6831), 913-917.
157. Jung, Y. K.; Park, H. G.; Kim, J. M. Biosensors & Bioelectronics 2006, 21, (8),
1536-1544.
158. Ma, G. Y.; Cheng, Q. Talanta 2005, 67, (3), 514-519.
159. Pang, J. B.; Yang, L.; McCaughey, B. F.; Peng, H. S.; Ashbaugh, H. S.; Brinker,
C. J.; Lu, Y. F. Journal of Physical Chemistry B 2006, 110, (14), 7221-7225.
160. Tang, Z. H.; Yang, B. A. Journal of Materials Science 2007, 42, (24), 10133-
10137.
161. Hoofman, R.; Gelinck, G. H.; Siebbeles, L. D. A.; de Haas, M. P.; Warman, J. M.;
Bloor, D. Macromolecules 2000, 33, (25), 9289-9297.
162. Burns, A. R.; Carpick, R. W.; Sasaki, D. Y.; Shelnutt, J. A.; Haddad, R. Tribology
Letters 2001, 10, (1-2), 89-96.
163. Valenta, C.; Steininger, A.; Auner, B. G. European Journal of Pharmaceutics and
Biopharmaceutics 2004, 57, (2), 329-336.
164. Chance, R. R. Macromolecules 1980, 13, (2), 396-398.
165. Huo, Q.; Russell, K. C.; Leblanc, R. M. Langmuir 1999, 15, (11), 3972-3980.
166. Jonas, U.; Shah, K.; Norvez, S.; Charych, D. H. Journal of the American
Chemical Society 1999, 121, (19), 4580-4588.
167. Kim, J. M.; Lee, J. S.; Lee, J. S.; Woo, S. Y.; Ahn, D. J. Macromolecular
Chemistry and Physics 2005, 206, (22), 2299-2306.
168. Cai, M.; Mowery, M. D.; Menzel, H.; Evans, C. E. Langmuir 1999, 15, (4), 1215-
1222.
169. Jaworski, J. W.; Raorane, D.; Huh, J. H.; Majumdar, A.; Lee, S. W. Langmuir
2008, 24, (9), 4938-4943.
170. Stanish, I.; Singh, A. Chemistry and Physics of Lipids 2001, 112, (2), 99-108.
171. Bredas, J. L.; Heeger, A. J. Macromolecules 1990, 23, (4), 1150-1156.
172. Kim, J. M.; Lee, J. S.; Choi, H.; Sohn, D.; Ahn, D. J. Macromolecules 2005, 38,
(22), 9366-9376.
125
173. Carpick, R. W.; Mayer, T. M.; Sasaki, D. Y.; Burns, A. R. Langmuir 2000, 16,
(10), 4639-4647.
174. Park, H.; Lee, J. S.; Choi, H.; Ahn, D. J.; Kim, J. M. Advanced Functional
Materials 2007, 17, (17), 3447-3455.
175. Ye, Q.; Zou, G.; You, X.; Yu, X. W.; Zhang, Q. J. Materials Letters 2008, 62,
(24), 4025-4027.
176. Cheng, Q.; Yamamoto, M.; Stevens, R. C. Langmuir 2000, 16, (12), 5333-5342.
177. Song, J.; Cheng, Q.; Zhu, S. M.; Stevens, R. C. Biomedical Microdevices 2002, 4,
(3), 213-221.
178. Kew, S. J.; Hall, E. A. H. Analytical Chemistry 2006, 78, (7), 2231-2238.
179. Charych, D.; Cheng, Q.; Reichert, A.; Kuziemko, G.; Stroh, M.; Nagy, J. O.;
Spevak, W.; Stevens, R. C. Chemistry & Biology 1996, 3, (2), 113-120.
180. Peng, H. S.; Tang, J.; Pang, J. B.; Chen, D. Y.; Yang, L.; Ashbaugh, H. S.;
Brinker, C. J.; Yang, Z. Z.; Lu, Y. F. Journal of the American Chemical Society 2005,
127, (37), 12782-12783.
181. Li, L. S.; Stupp, S. I. Macromolecules 1997, 30, (18), 5313-5320.
182. Lio, A.; Reichert, A.; Ahn, D. J.; Nagy, J. O.; Salmeron, M.; Charych, D. H.
Langmuir 1997, 13, (24), 6524-6532.
183. Gill, I.; Ballesteros, A. Angewandte Chemie-International Edition 2003, 42, (28),
3264-3267.