Page 1
PEPTIDE SELF-ASSEMBLY FROM THE
MOLECULAR TO THE MACROSCOPIC SCALE AT
STANDARD CONDITIONS
Ahmad I. Athamneh
Dissertation submitted to the faculty of the Virginia Polytechnic Institute and State University in
partial fulfillment of the requirements for the degree of
Doctor of Philosophy In
Biological Systems Engineering
Justin R. Barone, Chair Chenming (Mike) Zhang
Ryan S. Senger Abby W. Morgan
November 16, 2010 Blacksburg, VA
Keywords: peptide materials, plasticized biopolymers, nanomaterials, multiscale self-assembly
Page 2
Peptide Self-Assembly from the Molecular to the Macroscopic
Scale at Standard Conditions
ABSTRACT
This dissertation attempts to address the problem of how to prepare protein-based materials with
the same level of order and precision at the molecular level similar to the structures we find in
nature. It is divided into two parts focusing on feedstock and processing. Part one is devoted to
discussing the use of agricultural proteins as a feedstock for material production. Particularly, it
focuses on the effect of hydrogen bonding, or lack thereof, between proteins as mediated by
hydration or plasticization. The effect of varying plasticizer (glycerol) levels on mechanical
properties of a series of proteins is discussed in the context of primary and secondary structure of
these proteins. We have found that the extent to which a protein can be plasticized is dependent
on its molecular and higher order structure and not simply molecular weight, as it was often
assumed in previous studies.
The second part of the dissertation focuses on the study of self-assembly as a way to make useful
peptide-based materials. There are major efforts underway to study protein self-assembly for
various medical and industrial reasons. Despite huge progress, most studies have focused on
nanoscale self-assembly but the crossover to the macroscopic scale remains a challenge. We
show that peptide self-assembly into macroscopic fibers is possible in vitro under physiological
conditions. We characterize the fibers and propose a mechanism by which they form. The
macroscopic fibers self-assemble from a combination of β- and α-peptides and are similar to
other naturally-occurring systems in which templated self-assembly is used to create functional
Page 3
iii
peptide materials. Finally, the ability to control macroscopic properties of the fiber by varying
the ratio of constituent peptides is demonstrated.
Owing to the richness of the amino acid building blocks, peptides are highly versatile structural
and functional building blocks. The ability to extend and control peptide self-assembly over
multiple length scales is a significant leap toward incorporating peptide materials into dynamic
systems of higher complexity and functionality.
Page 4
iv
Acknowledgments
I would like to thank my advisor Dr. Justin R. Barone and PhD committee members Dr. Chenming
(Mike) Zhang, Dr. Ryan Senger and Dr. Abby Morgan for their guidance and support through out
the course of this research.
.
Page 5
v
Dedication
My father, Ibrahim M. Athamneh, and mother, Ameena M. Al-Yaseen
My beloved wife Ruba
My brothers Safwan, Khaled, and Mohammad
My sister Hana
… To whom I owe everything I have achieved in my life.
.
Page 6
vi
TABLE OF CONTENT
Abstract...................................................................................................................................................ii
Table of Content.................................................................................................................................... vi
List of figures....................................................................................................................................... viii
List of tables ........................................................................................................................................xiii
CHAPTER ONE: Introduction ............................................................................................................. 1
Motivation..................................................................................................................................................................1
The Cross-β Structural Motif: A Model For Studying Self-Assembly .....................................................................3
Green Feedstock For a Green Process .......................................................................................................................6
Self-Assembly From the Nanometer to the Micrometer Scale..................................................................................7
Hypotheses.................................................................................................................................................................8
Dissertation outline ....................................................................................................................................................8
References..................................................................................................................................................................9
CHAPTER TWO: Conformational Changes and Molecular Mobility in Plasticized Proteins ...........16
Abstract....................................................................................................................................................................16
Introduction..............................................................................................................................................................16
Experimental section ...............................................................................................................................................20
Results and Discussion ............................................................................................................................................22
Conclusion ...............................................................................................................................................................36
References................................................................................................................................................................37
CHAPTER THREE: Self-Assembling Peptide From Non-Structural Proteins................................. 43
Abstract....................................................................................................................................................................43
Introduction..............................................................................................................................................................44
Materials and Methods ............................................................................................................................................46
Results and Discussion ............................................................................................................................................48
Conclusion ...............................................................................................................................................................58
Page 7
vii
References................................................................................................................................................................58
CHAPTER FOUR: Peptide Self-Assembly into Macroscopic Fibers at Physiological Conditions ... 64
Abstract....................................................................................................................................................................64
Introduction..............................................................................................................................................................65
Materials and Methods ............................................................................................................................................67
Results and Discussion ............................................................................................................................................71
Conclusion ...............................................................................................................................................................86
References................................................................................................................................................................88
CHAPTER FIVE: Templated Self-Assembly of Large Fibers Using a Model System....................... 96
Abstract....................................................................................................................................................................96
Introduction..............................................................................................................................................................97
Materials and Methods ............................................................................................................................................98
Results and Discussion ..........................................................................................................................................101
Conclusion .............................................................................................................................................................116
References..............................................................................................................................................................117
CHAPTER SIX: Conclusions .............................................................................................................123
Appendix..............................................................................................................................................125
A.1 Model fitting and optimization .......................................................................................................................126
A.2 Simulation: formation of fibrillar aggregates as a function of Gliadin level. .................................................132
Page 8
viii
LIST OF FIGURES
Chapter One: Introduction
Figure 1. Schematic drawing of a cross-β fibril............................................................................. 5
Figure 2. a, Transmission electron micrograph of negatively stained TWG nanofibrils. Scale bar:
100 nm. b, Scanning electron micrograph of TWG fiber. Scale bar: 5 µm. c, Optical image
showing a single TWG fiber glued to a glass fiber. Scale bar: 0.1 mm. ................................ 7
Chapter Two: Conformational Changes and Molecular Mobility in Plasticized Proteins
Figure 1. Elastic modulus of plasticized egg albumin (EA) (¡), lactalbumin (LA) (o), feather
keratin (FK) (¯), and wheat gluten (WG) (Í) films as a function of glycerol content. 22
Figure 2. Amide I peak position in the FT-IR spectra of plasticized egg albumin (¡),
lactalbumin (o), feather keratin (¯), wheat gluten (Í), and corn zein (+) films as a
function of glycerol content. ............................................................................................. 23
Figure 3. Amide II peak position in the FT-IR spectra of plasticized egg albumin (¡),
lactalbumin (o), feather keratin (¯), wheat gluten (Í), and corn zein (+) films as a
function of glycerol content. ............................................................................................. 24
Figure 4. Critical plasticization concentration, c*, obtained from modulus (¡), FT-IR amide I
(o), and amide II (¯) data as function of the number of polar amino acids per protein. 26
Figure 5. Absolute change in the FT-IR amide I peak position for each protein as a function of
the number of polar amino acids per protein (o), and critical plasticization concentration,
c*, as determined form FT-IR amide I data (¡)............................................................... 27
Page 9
ix
Figure 6. Absolute change in the amide I peak position (¡), and modulus, E, (o) at c* for each
protein as a function of the percent cysteine in egg albumin (EA), lactalbumin (LA),
feather keratin (FK), wheat gluten (WG), and corn zein (CZ). ........................................ 28
Figure 7. XRD patterns for WG made with variable glycerol concentrations (a) full pattern and
(b) peak around 0.46 nm (∼23 °2θ). ................................................................................. 29
Figure 8. XRD patterns for FK made with variable glycerol concentrations (a) full pattern and
(b) peak around 0.46 nm (∼23 °2θ). ................................................................................. 30
Figure 9. Molecular weight between intermolecular interactions, Mc, at c* as function of the
average distance between inter-molecular cross-links in the proteins.............................. 33
Figure 10. Molecular distance between peptide bonds and side groups on two different
molecules as a function of glycerol content. * denotes concentrations where new Bragg
peak was observed on WG................................................................................................ 35
Chapter Three: Self-Assembling Peptide From Non-Structural Proteins
Figure 1. Schematic drawing of a cross-β fibril........................................................................... 45
Figure 2. SDS-page profile of wheat gluten and hydrolysis products: (1) native wheat gluten, (2)
molecular weight marker, (3) Tryp-pH5.7, and (4) Tryp-pH8. ........................................ 49
Figure 3. SEM micrographs of (a) Control and (b-d) Tryp-pH8. ................................................ 50
Figure 4. FT-IR spectra of (1) Control, (2) Tryp-pH5.7, and (3) Tryp-pH8. .............................. 51
Figure 5. X-ray powder diffraction patterns of of (1) native (Control) and trypsin-hydrolyzed
wheat gluten at (2) pH 5.7 and (3) pH 8 ........................................................................... 52
Page 10
x
Figure 6. Effect of trypsin treatment on gliadin (top) and glutenin (bottom) with regard to
formation of cross-β structures as indicated by (a) FT-IR amide I band, and (b) XRD.
Control samples are shown in blue. .................................................................................. 56
Figure 7. Schematic illustration of cross-β fibril formation. ....................................................... 57
Chapter Four: Peptide Self-Assembly into Macroscopic Fibers at Physiological Conditions
Figure 1. a, Overlaid FPLC chromatograms. b, SDS-PAGE profile of native (lane 2) and
hydrolyzed (lane 3) wheat gluten showing reduction of most components to <10 kDa
fragments. The band at the line between stacking and running gels in lane 3 was due to
aggregates. Only two bands <10 kDa were present in the supernatant of partially
dissolved fibers (lane 4). ................................................................................................... 71
Figure 2. a, Transmission electron micrograph of negatively stained TWG nanofibrils. Scale bar:
100 nm. b, Scanning electron micrograph of TWG fiber. Scale bar: 5 µm. c, Optical
image showing a single TWG fiber glued to a glass fiber. Scale bar: 0.1 mm................. 72
Figure 3. Amide I and II of the Fourier transform-infrared spectra of native (dashed) and tryptic
wheat gluten showing decreased α content and increased β content. .............................. 73
Figure 4. Raman spectra of native (bottom) and tryptic wheat gluten (top) showing α to β
transformation. .................................................................................................................. 74
Figure 5. X-ray diffraction patterns of native (dotted) and tryptic wheat gluten......................... 75
Figure 6. Fluorescent microscopy images of the Th-T stained tryptic wheat gluten fiber. ......... 76
Page 11
xi
Figure 7. Tendency for β-aggregation and secondary structure prediction at each amino acid
position for Gd20, Gd46 and GtL75. Aggregation scores, percentage per residue, were
estimated using the TANGO algorithm. Secondary structure predicted using the GOR4
algorithm. .......................................................................................................................... 79
Figure 8. Amide I and II regions of the Raman spectra of TWG fibers oriented perpendicular
(top), 45° (middle) and perpendicular (bottom) to the laser’s plane of polarization. ....... 80
Figure 9. a, Shift to lower wavenumber of νs(CH3) and νs(CH2) bands with time indicated an
increasing hydrophobic environment. b, Estimated secondary structure content as a
function of time, determined by fitting of the FT-IR amide I band. c, The intensity ratios
νs(CH3) to νas(CH3) and νs(CH3) to δas(CH3) indicated amount of hydrophobic packing.82
Figure 10. a, TWG Fibers embedded in epoxy resin matrix. b, Nanoindentation image of
longitudinal and transverse sections showing indentation imprints. c, Comparison of
nanoindentation results in longitudinal and transverse sections. Error bars represent 95%
confidence interval; n=9. .................................................................................................. 83
Figure 11. Schematic illustration of the templated self-assembly process leading to formation of
macroscopic cross-β fibers................................................................................................ 84
Chapter Five: Templated Self-Assembly of Large Fibers Using a Model System
Figure 1. Size-exclusion chromatograms of native and tryptic myoglobin and gliadin. ........... 102
Figure 2. FE-SEM micrographs of macroscopic fibers formed after adding myoglobin to trypsin-
hydrolyzed gliadin. ......................................................................................................... 102
Page 12
xii
Figure 3. Dependence of the number of self-assembled macroscopic fibers found in the solution
on the ratio of gliadin to myoglobin. .............................................................................. 104
Figure 4. Time-dependent change in secondary structure in 50% Trypsin-gliadin-myoglobin
solution calculated from FT-IR spectroscopy data. ........................................................ 104
Figure 5. a) Time dependent α-helix and (b) β-sheet content in trypsin-myoglobin-gliadin
solutions estimated using FT-IR spectroscopy. c) Dependence of the rate of α-helix
unraveling and β-sheet increase on gliadin content. d) Dependence of the number of
fibers found in the solution on the rate of β-sheet increase. ........................................... 107
Figure 6. The intensity ratios (b) νs(CH3) to δas(CH3) indicated amount of hydrophobic packing.
........................................................................................................................................ 108
Figure 7. a, Scanning electron micrographs of macroscopic fibers formed in trypsin-myoglobin-
gliadin solution (TMG) using different gliadin to myoglobin ratios. Scale bars: 100 µm.
b, Comparison of macroscopic properties of different TMG fibers. Aspect ratio values are
averages of up to 16 independent measurements. Reduced modulus values are averages
of up to 24 independent measurements. Error bars represent standard error of the mean.
........................................................................................................................................ 110
Figure 8. Schematic illustration of the templated self-assembly process leading to formation of
macroscopic cross-β fibers.............................................................................................. 111
Figure 10. Comparison of experimental and simulated data. .................................................... 116
Page 13
xiii
LIST OF TABLES
Chapter Two: Conformational Changes and Molecular Mobility in Plasticized Proteins
Table 1. Amino acid composition of feather keratin (FK), egg albumin (EA), lactalbumin (LA),
wheat gluten (WG) and corn zein (CZ). ........................................................................... 18
Table 2. Summary of FT-IR amide I deconvolution. .................................................................. 25
Table 3. The critical plasticization point, c*, for each protein as defined from amide I, amide II,
and modulus data. ............................................................................................................ 27
Chapter Three: Self-Assembling Peptide From Non-Structural Proteins
Table 1. Amide I band deconvolution results. ............................................................................ 52
Table 2. Possible tryptic gliadin fragments with glutamine (Q), asparagine (N), proline (P) and
glycine (G) composition. .................................................................................................. 55
Chapter Four: Peptide Self-Assembly into Macroscopic Fibers at Physiological Conditions
Table 1. Possible tryptic wheat gluten fragments with tendency for β-aggregation predicted
using the TANGO algorithm and glutamine (Q), asparagine (N), proline (P) and glycine
(G) composition. ............................................................................................................... 77
Table 2. Properties of templating peptides: Gd20 and the templating peptides in curli and
barnacle cement proteins. ................................................................................................. 85
Page 14
xiv
Table 3. Properties of templated peptides: GtL75 and templated peptides in curli and barnacle
cement proteins. Conformational switches in these peptides are triggered upon
interaction with hydrophobic templating peptides, thus promoting β-aggregation.......... 86
Chapter Five: Templated Self-Assembly of Large Fibers Using a Model System
Table 1. Properties of templating peptides from gliadin and myoglobin................................... 102
Table 2. Net percent change in secondary structure content calculated from FT-IR spectroscopy
data for trypsin-gliadin-myoglobin solutions with variable gliadin content. Statistically
insignificant (α=0.05) values are not shown. ................................................................. 105
Page 15
1
CHAPTER ONE
INTRODUCTION
MOTIVATION
The problem this dissertation attempted to address was how to prepare protein-based materials
with the same level of order and precision at the molecular scale similar to the structures we find
in nature.
Nature relies on self-assembly to create sophisticated functional materials from simple
nanostructures. Natural proteinaceous materials are tough and lightweight. Silk, collagen,
cartilage and feather keratin are but a few examples of natural materials self-assembled from
protein molecules with physical properties hard to match in the synthetic world. Self-assembly
is the thermodynamically favored association of molecules. It is, therefore, inherently precise
and requires minimum energy input. Additionally, nature synthesizes proteins using the
‘greenest’ chemistry, i.e., at room to human body temperature, one atmosphere pressure and
close to neutral pH. If the mechanism of protein self-assembly could be understood, materials of
superior mechanical properties could be produced in a simple, green process using a biological,
green feedstock. The implications of self-assembly of proteins are actually wider than just
structural materials. It is know that a variety of serious, incurable disorders such as Alzheimer’s
disease, Type II diabetes, and Creutzfeldt–Jakob disease are caused by insoluble, self-assembled
protein fibrils generally known as amyloids [1].
Page 16
2
Green processing and green feedstocks are keys to mitigating the environmental impact of
petroleum usage. It is also an effective way to deal with the troubled economics of an unstable
and increasingly politicized petroleum supply. The polymer industry has been particularly
vulnerable to increasing petroleum prices because, unlike in the case of fuels, rising cost cannot
be easily transferred to consumers. Accordingly, there has been increased interest in the
development of products from renewable resources [2, 3] including biological feedstocks that are
the by-products of the food and agricultural industries. Insoluble proteins like corn zein (CZ),
wheat gluten (WG), feather keratin (FK), egg albumin (EA), and lactalbumin (LA) are some
examples of biological feedstocks that are normally disposed, cannot be converted to fuel very
easily, have limited nutritional value, but have been generally shown to be easily processed into
useful biodegradable polymers [4-8].
To be utilized, some of the protein structure must first be broken down into smaller units so that
a new product can be formed. Unfortunately, most protein materials lose their useful properties
when removed from their native assembly and reprocessed for use in other applications; often
showing increased density and brittle behavior [4]. Two factors with major impact on the
properties of proteinaceous materials are hydration and native hierarchical organization from the
atomic to the macroscopic scale. Hydration prevents excess intermolecular hydrogen bonding
thus retaining structural flexibility [9]. Dehydrated proteins form an extensive hydrogen-bonding
network leading to high modulus, stress to break, and density and low strain to break. In
addition, reprocessed proteins lose their native hierarchical organization, which impart shape and
properties to the material [10]. Individual protein molecules separated from their natural
assembly do not usually possess the properties of the assemblage of components. It is the
Page 17
3
interaction of protein molecules at different length scales that come together to form a
macroscopic material of high properties.
Past efforts to develop processing techniques to reprocess proteins into materials have essentially
addressed the dehydration problem and gave little to no attention to how much the native
structure was compromised [10]. Understanding how proteins interact with themselves and with
water is important as it determines the final structure and properties of the protein material. The
first goal of this research was to understand protein-protein and protein-solvent interactions as a
function of solvent concentration. We are interested in the case where the protein is not fully
dissolved in solvent but the solvent is the minority phase, similar to bound water in naturally
occurring structural materials. Studying this case in vitro is sometimes referred to as
“plasticization”. Plasticization is a popular method to utilize proteins for structural materials [10-
16]. It involves the use of small polar molecules such as water, glycerol, ethylene glycol, or
propylene glycol. Plasticizers decrease protein-protein hydrogen bonding interactions and
increase free volume and molecular mobility resulting in greater structural flexibility. This
mimics natural hydration and, therefore, is a step toward restoring the pre-processed, original
protein structure. So, while there will be a significant loss of properties due to disruption of the
native structure, plasticization can still be used to prepare useful polymers from proteins but
more importantly understand the role of protein-protein and protein-plasticizer interactions [13-
16]. Unfortunately, protein plasticization is poorly understood and assumed to be dependent on
the protein’s molecular weight without considering the diversity of structure and composition.
THE CROSS-β STRUCTURAL MOTIF: A MODEL FOR STUDYING SELF-ASSEMBLY
Page 18
4
In the first part of the research we focused simply on the amount of protein-protein interactions
as plasticizer molecules were added without concern for the type of protein-protein interaction.
Next we focused on a specific type of protein interaction that leads to self-assembly of well-
defined larger structures. Self-assembly of cross-β structure is a process driven by backbone
(main chain) interactions. It is independent of the amino acid sequence and occurs in unfolded
and/or partially folded protein molecules [17]. This type of interaction proceeds to form fibrils
with an exceptionally stable core structure known as the cross-β structure. It was termed cross-β
because extended β-strands run perpendicular to the fibril axis as illustrated in Figure 1 [18].
Cross-β fibrils are essentially made of stacked β-sheets that grow in the lateral direction into
structures of high aspect ratio. It is the same structural motif implicated in amyloidosis. Recently,
there has been growing interest in cross-β fibrils in materials science because of their mechanical
properties combined with their flexibility, versatility and ability to self-assemble [19]. It has been
suggested that cross-β fibrils could serve as a structural motif for nanomaterials applications or
in the fabrication of nanodevices [19-23]. Cross-β fibrils combine the versatility of protein
sequences with self-assembly, adjustable properties and topographies, outstanding strength, and
an ability to withstand varied and extreme environmental conditions. These properties made
cross-β structures an attractive structural motif for medical, bionanotechnology, and
nanotechnology applications. Cross-β fibrils can potentially be used as a platform for
development of biomaterials, advanced catalysis systems, biosensors and diagnostic devices.
One of the most interesting features of an amyloid fibril is that the formation of its basic cross-β
structure does not require the presence of specific side chain interactions or sequence patterns
and is primarily the result of main chain interactions [17]. Thus, it is a step above simply the
amount of hydrogen bonding but not at the level of highly specific amino acid interactions.
Page 19
5
Accumulating evidence suggests that virtually any polypeptide can fibrillate provided that the
right conditions are met [24, 25]. These are usually extreme denaturing conditions such as low
pH and high temperature that disrupt the protein’s ability to fold based on side chain interactions
allowing main chain interactions to dominate. Cross-β fibrils are a generic class of material
despite the diversity of peptide building blocks, which is a very interesting feature from a
material science point of view.
Although the amino acid sequence does not dictate the fundamental organization of the cross-β
structure, the nature of, and interactions between, the side chains have significant influence on
the rate and conditions under which it can form [17]. Certain features of the polypeptide,
stemming from the peculiarity of its amino acid sequence such as hydrophobicity, charge, and
secondary structure propensity, have strong influence on the kinetics of self-assembly [26]. This
perhaps explains the occurrence of cross-β structures in nature under physiological conditions, as
is the case in amyloidosis. Some living organisms, such as green Lacewing, bacteria and
barnacles, utilize the inherent ability of proteins to form cross-β structures to generate novel and
diverse structures that can endure harsh environments [27-31]. Although most of the functional
cross-β structures in nature are on the nanometer scale, micron-sized structures have also been
fibril axis
Figure 1. Schematic drawing of a cross-β fibril.
Page 20
6
identified [29]. For this and other reasons mentioned earlier, the cross-β motif represents a good
model for studying protein self-assembly.
GREEN FEEDSTOCK FOR A GREEN PROCESS
There are major efforts underway to develop cross-β-based materials for various applications
[32]. The current state of the art is to start with de novo peptides that are known, or designed, to
self-assemble into cross-β structures [33-35]. However, the high cost of de novo synthesis, as it
exists today, represents a major hurdle to large-scale production of cross-β-based materials [20].
We propose an alternative method for production of self-assembling peptides through distilling
peptide behavior to generic self-assembly features. A great example is the self-assembling “Q-
block” of polyglutamine [26]. Glutamine (Q) rich short peptides or Q-blocks can be obtained
from the proteolysis of plant proteins such as wheat glutens. This suggests that peptide self-
assembling materials can be produced through proteolysis of existing proteins that can be
obtained cheaply. Glutens contain repeated sequences of Q with frequent proline (P) at the end of
the repeat sequence [36]. This mimics the cross-β forming Q-blocks. Wheat gluten is
increasingly being removed from bread and cereal products because of its connection to Celiac
disease [37]. It contains 58% glutenin and 42% gliadin [36]. Gliadin has a molecular weight of
about 36,000 g/mol. Glutenin contains 28% high molecular weight fraction of about 68,000 -
100,000 g/mol and 72% low molecular weight fraction of about 36,000 - 44,000 g/mol. Glutens
are examples of plant storage proteins that have no structural function. They are not known to
fibrillate unless subjected to severe denaturing conditions [38] and/or external mechanical stress
through processes like extrusion [39] and spinning [40].
Page 21
7
To prove the concept of utilizing agricultural proteins to derive self-assembling peptides, we
chose to work with wheat gluten and trypsin as a model system. The choice to hydrolyze gluten
with trypsin was not fortuitous. Alzheimer’s disease is believed to occur following the
proteolysis of beta amyloid precursor protein (βAPP) to Aβ1-40 peptides that can then self-
assemble [41]. Serine proteases like trypsin are implicated in this proteolysis [42]. Additionally,
trypsin’s high specificity enables better-controlled experiments and informed analysis.
SELF-ASSEMBLY FROM THE NANOMETER TO THE MICROMETER SCALE
Upon incubating wheat gluten with trypsin, we observed the formation of macroscopic fibers.
Macroscopic fibers assembled in tryptic wheat gluten and were hierarchically structured showing
organization from the nanometer to the micrometer scale. With a diameter raging from 10 - 15
µm, the large fibers were bundles of 10 - 20 nm diameter cross-b fibrils (Figure 2). Cross-β
structural units formed in other enzyme-protein systems, including tryptic gliadin, but did not
Figure 2. a, Transmission electron micrograph of negatively stained TWG nanofibrils. Scale bar:
100 nm. b, Scanning electron micrograph of TWG fiber. Scale bar: 5 µm. c, Optical image
showing a single TWG fiber glued to a glass fiber. Scale bar: 0.1 mm.
Page 22
8
assemble into large fibers. Individually, neither tryptic glutenin nor tryptic gliadin formed fibers
at physiological conditions. The structure of tryptic glutenin showed little change and remained
rich in disordered and α-helix conformations. By contrast, tryptic gliadin formed elementary
cross-β structural units, but these did not assemble into large fibers. These observations
suggested a cooperative process involving more than one peptide required to form macroscopic
fibers.
Our results clearly demonstrated that peptide self-assembly beyond the nanoscale was possible in
vitro under physiological conditions. The question then was: what is the correct combination of
properties that makes it possible?
HYPOTHESIS
Macroscopic protein fibers self-assemble when α-helix-rich peptides undergo α to β transition in
the presence of a hydrophobic peptide with high β-aggregation potential.
DISSERTATION OUTLINE
The first part of this dissertation, chapter two, is devoted to discussing the effect of hydrogen
bonding, or lack thereof, between proteins as mediated by hydration or plasticization. The effect
of varying plasticizer (glycerol) levels on mechanical properties of a series of proteins (WG, FK,
LA, EA, CZ) is discussed in the context of primary and secondary structure of these proteins. We
found that the extent to which a protein can be plasticized was dependent on its primary structure
and not simply molecular weight, as it was often assumed in previous studies. The effect of
plasticizer is complex. It increased order (relative β-sheet content) at low concentration by
allowing chains to align but the addition of low molecular weight solvent counteracts the
Page 23
9
ordering to reduce modulus. At higher concentrations, an overall drop in modulus is observed
because hydrogen bonding between proteins is progressively taken away in favor of protein-
solvent interactions. The effect is mitigated by having less polar groups or more cysteine, i.e.,
protein-protein interactions that are not hydrogen bonding.
The second part, chapters three, four and five, focus on peptide self-assembly. Chapter three
demonstrates a new concept of using a complex protein such as wheat gluten to derive self-
assembling peptides for advanced nanomaterials applications. Chapter four concentrates on
characterizing self-assembled WG fibers and the process leading to their formation. The chapter
illustrates a hierarchical self-organization process form the nanometer to the macroscopic scale.
The chapter also discusses a possible templated self-assembly process leading to the formation of
micrometer scale fibers. Building on the conclusions of chapter four, chapter five is devoted to
discussing the kinetics of the process of fiber self-assembly.
REFERENCES
[1] Dobson, C. M., 2003, "Protein folding and misfolding," Nature, 426(6968), pp. 884-890.
[2] Wool, R. P., 2005, Bio-based polymers and composites Elsevier Academic Press,
Amsterdam; Boston.
[3] Mohanty, A. K., Misra, M., and Drzal, L. T., 2005, Natural fibers, biopolymers, and
biocomposites,, CRC Press, Boca Raton, FL.
[4] Barone, J. R., Schmidt, W. F., and Gregoire, N. T., 2006, "Extrusion of feather keratin,"
J. Appl. Polym. Sci., 100(2), pp. 1432-1442.
Page 24
10
[5] Galietta, G., Di Gioia, L., Guilbert, S., and Cuq, B., 1998, "Mechanical and
Thermomechanical Properties of Films Based on Whey Proteins as Affected by
Plasticizer and Crosslinking Agents," J. Dairy Sci., 81(12), pp. 3123-3130.
[6] Redl, A., Morel, M. H., Bonicel, J., Vergnes, B., and Guilbert, S., 1999, "Extrusion of
Wheat Gluten Plasticized with Glycerol: Influence of Process Conditions on Flow
Behavior, Rheological Properties, and Molecular Size Distribution," Cereal Chem.,
76(3), pp. 361-370.
[7] Di Gioia, L., and Guilbert, S., 1999, "Corn Protein-Based Thermoplastic Resins: Effect of
Some Polar and Amphiphilic Plasticizers," J. Agric. Food Chem., 47(3), pp. 1254-1261.
[8] Gennadios, A., Weller, C. L., Hanna, M. A., and Froning, G. W., 1996, "Mechanical and
Barrier Properties of Egg Albumen Films," J. Food Sci., 61(3), pp. 585-589.
[9] Zhang, X., Burgar, I., Do, M. D., and Lourbakos, E., 2005, "Intermolecular Interactions
and Phase Structures of Plasticized Wheat Proteins Materials," Biomacromolecules, 6(3),
pp. 1661-1671.
[10] Barone, J. R., Schmidt, W. F., and Liebner, C. F. E., 2005, "Thermally processed keratin
films," J. Appl. Polym. Sci., 97(4), pp. 1644-1651.
[10] Zhang, X., Hoobin, P., Burgar, I., and Do, M. D., 2006, "Chemical Modification of
Wheat Protein-Based Natural Polymers: Cross-Linking Effect on Mechanical Properties
and Phase Structures," J. Agric. Food Chem., 54(26), pp. 9858-9865.
[12] Hernandez-Munoz, P., Kanavouras, A., Ng, P. K. W., and Gavara, R., 2003,
Page 25
11
"Development and Characterization of Biodegradable Films Made from Wheat Gluten
Protein Fractions," J. Agric. Food Chem., 51(26), pp. 7647-7654.
[13] Gao, C., Stading, M., Wellner, N., Parker, M. L., Noel, T. R., Mills, E. N. C., and Belton,
P. S., 2006, "Plasticization of a Protein-Based Film by Glycerol: A Spectroscopic,
Mechanical, and Thermal Study," J. Agric. Food Chem., 54(13), pp. 4611-4616.
[14] Gueguen, J., Viroben, G., Noireaux, P., and Subirade, M., 1998, "Influence of plasticizers
and treatments on the properties of films from pea proteins," Ind. Crop. Prod., 7(2-3), pp.
149-157.
[15] Mangavel, C., Barbot, J., Gueguen, J., and Popineau, Y., 2003, "Molecular Determinants
of the Influence of Hydrophilic Plasticizers on the Mechanical Properties of Cast Wheat
Gluten Films," J. Agric. Food Chem., 51(5), pp. 1447-1452.
[16] Choi, S. G., Kim, K. M., Hanna, M. A., Weller, C. L., and Kerr, W. L., 2003, "Molecular
Dynamics of Soy-Protein Isolate Films Plasticized by Water and Glycerol," J. Food Sci.,
68(8), pp. 2516-2522.
[17] Fandrich, M., and Dobson, C. M., 2002, "The behaviour of polyamino acids reveals an
inverse side chain effect in amyloid structure formation," EMBO J., 21, pp. 5682-5690.
[18] Sunde, M., 1997, "Common core structure of amyloid fibrils by synchrotron X-ray
diffraction," J. Mol. Biol., 273, pp. 729-739.
[19] Cherny, I., and Gazit, E., 2008, "Amyloids: Not Only Pathological Agents but Also
Ordered Nanomaterials13," Angew. Chem. Int. Ed., 47(22), pp. 4062-4069.
Page 26
12
[20] Gazit, E., 2007, "Self-assembled peptide nanostructures: the design of molecular building
blocks and their technological utilization," Chem. Soc. Rev., 36(8), pp. 1263-1269.
[21] Scheibel, T., Parthasarathy, R., Sawicki, G., Lin, X.-M., Jaeger, H., and Lindquist, S. L.,
2003, "Conducting nanowires built by controlled self-assembly of amyloid fibers and
selective metal deposition," Proc. Nat. Acad. Sci., 100(8), pp. 4527-4532.
[22] Lashuel, H. A., LaBrenz, S. R., Woo, L., Serpell, L. C., and Kelly, J. W., 2000,
"Protofilaments, Filaments, Ribbons, and Fibrils from Peptidomimetic Self-Assembly:
Implications for Amyloid Fibril Formation and Materials Science," J. Am. Chem. Soc.,
122(22), pp. 5262-5277.
[23] Channon, K., and MacPhee, C. E., 2008, "Possibilities for 'smart' materials exploiting the
self-assembly of polypeptides into fibrils," Soft Matter, 4(4), pp. 647-652.
[24] Dobson, C. M., 1999, "Protein misfolding, evolution and disease," Trends Biochem. Sci.,
24(9), pp. 329-332.
[25] Uversky, V. N., and Fink, A. L., 2004, "Conformational constraints for amyloid
fibrillation: the importance of being unfolded," Biochim. Biophys. Acta, Proteins
Proteomics, 1698(2), pp. 131-153.
[26] Perutz, M. F., Johnson, T., Suzuki, M., and Finch, J. T., 1994, "Glutamine repeats as
polar zippers: their possible role in inherited neurodegenerative diseases," Proc. Nat.
Acad. Sci., 91, pp. 5355-5358.
[27] Chiti, F., and Dobson, C. M., 2006, "Protein Misfolding, Functional Amyloid, and
Page 27
13
Human Disease," Annu. Rev. Biochem., 75(1), pp. 333-366.
[28] Barlow, D. E., Dickinson, G. H., Orihuela, B., Kulp, J. L., Rittschof, D., and Wahl, K. J.,
2010, "Characterization of the Adhesive Plaque of the Barnacle Balanus amphitrite:
Amyloid-Like Nanofibrils Are a Major Component," Langmuir, 26(9), pp. 6549-6556.
[29] Sullan, R. M. A., Gunari, N., Tanur, A. E., Chan, Y., Dickinson, G. H., Orihuela, B.,
Rittschof, D., and Walker, G. C., 2009, "Nanoscale structures and mechanics of barnacle
cement," Biofouling: The Journal of Bioadhesion and Biofilm Research, 25(3), pp. 263 -
275.
[30] Wang, X., and Chapman, M. R., 2008, "Curli provide the template for understanding
controlled amyloid propagation," Prion, 2(2), pp. 57-60.
[31] Claessen, D., Rink, R., de Jong, W., Siebring, J., de Vreugd, P., Boersma, F. G. H.,
Dijkhuizen, L., and Wösten, H. A. B., 2003, "A novel class of secreted hydrophobic
proteins is involved in aerial hyphae formation in Streptomyces coelicolor by forming
amyloid-like fibrils," Gene Dev, 17(14), pp. 1714-1726.
[32] Arnold, C., 2008, "From Diseases To Devices," Chemical & Engineering News,
American Chemical Society, pp. 48-50.
[33] Rubin, N., Perugia, E., Goldschmidt, M., Fridkin, M., and Addadi, L., 2008, "Chirality of
Amyloid Suprastructures," J. Am. Chem. Soc., 130(14), pp. 4602-4603.
[34] Oppenheim, T., Knowles, T. P. J., Lacour, S. P., and Welland, M. E., 2010, "Fabrication
and characterisation of protein fibril-elastomer composites," Acta Biomaterialia, 6(4), pp.
Page 28
14
1337-1341.
[35] Kopecek, J., and Yang, J., 2009, "Peptide-directed self-assembly of hydrogels," Acta
Biomaterialia, 5(3), pp. 805-816.
[36] Fox, P. F., and Condon, J. J., 1982, Food Proteins, Applied Science Publishers, London,
UK.
[37] Kasarda, D. D., and D'Ovidio, R., 1999, "Deduced Amino Acid Sequence of an α-Gliadin
Gene from Spelt Wheat (Spelta) Includes Sequences Active in Celiac Disease," Cereal
Chem., 76(4), pp. 548-551.
[38] Kasarda, D. D., Bernardin, J. E., and Thomas, R. S., 1967, "Reversible Aggregation of α-
Gliadin to Fibrils," Science, 155(3759), pp. 203-205.
[39] Reddy, N., and Yang, Y., 2007, "Novel Protein Fibers from Wheat Gluten,"
Biomacromolecules, 8(2), pp. 638-643.
[40] Woerdeman, D. L., Ye, P., Shenoy, S., Parnas, R. S., Wnek, G. E., and Trofimova, O.,
2005, "Electrospun Fibers from Wheat Protein: Investigation of the Interplay between
Molecular Structure and the Fluid Dynamics of the Electrospinning Process,"
Biomacromolecules, 6(2), pp. 707-712.
[41] Benjannet, S., Elagoz, A., Wickham, L., Mamarbachi, M., Munzer, J. S., Basak, A.,
Lazure, C., Cromlish, J. A., Sisodia, S., Checler, F., Chretien, M., and Seidah, N. G.,
2001, "Post-translational Processing of β-Secretase (β-Amyloid-converting Enzyme) and
Its Ectodomain Shedding.," J. Biol. Chem., 276(14), pp. 10879-10887.
Page 29
15
[42] Katona, G., Berglund, G. I., Hajdu, J., Gráf, L., and Szilágyi, L., 2002, "Crystal structure
reveals basis for the inhibitor resistance of human brain trypsin," J. Mol. Biol., 315(5),
pp. 1209-1218.
Page 30
CHAPTER TWO
CONFORMATIONAL
CHANGES AND MOLECULAR
MOBILITY IN PLASTICIZED
PROTEINS
Reprinted with permission from Ahmad I. Athamneh, et al., Conformational Changes and
Molecular Mobility in Plasticized Proteins. Biomacromolecules, 2008. 9(11): p. 3181-3187.
Copyright 2009 American Chemical Society.
ABSTRACT
Most biopolymers exist in a plasticized state whether it is naturally with water or unnaturally
with glycerol or other suitable polyol to make a flexible material. We have found that the extent
to which a biopolymer can be plasticized is dependent on its molecular and higher order
structures outside of simply molecular weight. Lactalbumin, ovalbumin, corn zein, wheat gluten,
and feather keratin were plasticized with glycerol from very low to very high amounts. The
conformation of the proteins was monitored with Fourier transform-infrared (FT-IR)
spectroscopy and x-ray powder diffraction (XRD) and correlated with the tensile modulus.
Page 31
17
Protein conformational changes were pronounced for polar proteins with a low amount of
cysteine. FT-IR showed that the conformational changes resulted in ordering of the protein at
low to moderate plasticization levels. For proteins with little resistance to conformational
changes, additional small-scale ordering occurred around the glass transition as observed in
XRD. Accurate comparison of plasticized proteins was dependent on knowing whether or not the
protein was glassy or rubbery at room temperature as no differences arose in the glassy state. The
transition from glassy to rubbery behavior with plasticization level can be found from modulus,
FT-IR, and XRD data.
INTRODUCTION
Growing concern with the long-term availability, political implications, and environmental
impact of petroleum usage has led to increased interest in the development of products from
renewable resources [1, 2]. The petroleum dilemma is not only important for the production of
fuels but for the production of pharmaceuticals, chemicals, and plastics. Approximately 200
billion pounds of plastic are produced worldwide each year, of which over 60 billion tons are
thrown away in the United States alone [1]. The production of this plastic requires between 6 and
8 percent of the world’s total oil production [2].
One way to improve this situation is through the production of polymers from renewable
resources [3, 4]. However, attempts to make fuels, chemicals, and plastics from plant materials
such as corn starch or soybean oil have had the perverse effect of driving up agricultural
commodities and food prices to all-time highs as well as reducing exportation of needed food to
other countries. While the use of renewable resources may be noble in producing fuels,
chemicals, and plastics, it does not appear as if there is enough for all of the industries that would
Page 32
18
rely on them. One alternative may be to use biological feedstocks that are the by-products of the
food and agricultural industries such as insoluble proteins. Some examples include corn zein
(CZ), wheat gluten (WG), feather keratin (FK), egg albumin (EA), and lactalbumin (LA). The
primary structure information of these proteins is listed in Table 1. These proteins cannot be
converted to fuel very easily, have limited nutritional value in the food or feed markets, but have
been generally shown to be easily processed into useful biodegradable polymers [5-9].
Table 1. Amino acid composition of feather keratin (FK), egg albumin (EA), lactalbumin (LA),
wheat gluten (WG) and corn zein (CZ).
FKa EAb LAc WGd CZe Ala 4 34 3 11 29 Arg 5 15 1 13 2 Asn 3 17 12 17 10 Asp 2 14 9 16 Cys 7 6 8 12 2 Glu 2 33 9 1 Gln 5 16 4 147 41 Gly 11 19 6 14 5 His 7 3 8 2 Ile 5 25 8 13 9 Leu 6 32 13 28 43 Lys 20 12 8 Met 16 1 6 Phe 4 20 4 20 13 Pro 12 14 2 50 23 Ser 16 38 7 19 15 Thr 4 15 7 10 5 Try 3 4 6 Tyr 1 10 4 12 8 Val 9 31 6 14 5 total 96 385 123 424 213 molecular weight, g/mol 11455 49670 16364 57519 27114 number of polar residues 38 185 68 250 84
a Arai et al [10]. b Nisbet et al [11]. c Whitaker and Tannenbaum [12]. d Kinsella [13]; data
calculated based on amino acid composition of glutenin and gliadin. Wheat gluten is 58%
glutenin and 42% gliadin. e Di Gioia et al [14].
Page 33
19
Proteins perform structural (i.e., skin, hair, muscle) and biological (i.e., enzymes for digestion,
nitrogen storage) functions in nature. While CZ, FK, WG, LA, and EA comprise both classes,
the focus here is on using them as starting materials for structural polymers. To that end, the
primary, secondary, tertiary, and quarternary structures of the proteins determine physical
properties such as strength, stiffness, and flexibility [15]. A protein such as WG appears to serve
only as a means for the plant to store nitrogen and amino acids. The lack of secondary structure
in wheat gluten is not surprising because that would be an obstacle to the plant accessing the
needed nitrogen and amino acids for biosynthesis. Conversely, FK serves an important structural
function for the bird. Feather keratin has a high amount of β-sheet content and inter-molecular
cystine bonding to achieve its structural function [16].
Upon isolation from the native state, all the proteins show increased density and brittle behavior.
Solubilized and regenerated proteins will hydrogen bond strongly with one another. Modulus,
stress to break, and density for these proteins are high and strain to break is low. One popular
method to utilize these proteins for structural polymers is to plasticize them with small polar
molecules such as water, glycerol, ethylene glycol, or propylene glycol [17-23]. While the
intrinsic protein molecular arrangement is not regained, some level of hydration mimicking
natural hydration is so plasticization is one small step toward the protein behaving more
naturally. Glycerol is the most preferred because of its low price, bio-based nature, ability to
survive typical thermal processing because of a high boiling point, and ability to be retained by
the protein in the solid state. It is believed that plasticizers decrease protein-protein hydrogen
bonding interactions and increase free volume and molecular mobility resulting in greater
polymer flexibility. While there is a cost to modulus and stress at break, useful polymers are still
Page 34
20
possible because many applications do not require the very high values found in unplasticized
proteins [20-23].
There are many reports in the literature comparing the properties of plasticized proteins [24-27].
However, it is dubious that valid comparisons can be made based on equal weight or molar
amounts of plasticizer alone. Each protein not only has a different molecular weight, but varying
polar group amounts, secondary structure, and cysteine contents. If the proteins were in a mostly
glassy state, it would be difficult to find large differences in their physical properties. It is the
intent of this study to examine the effects of varying plasticizer amount on a series of proteins of
varying primary and secondary structure and separate out the effects of each on polymer physical
properties. Then a consistent model to compare the physical properties of different plasticized
proteins can emerge.
EXPERIMENTAL SECTION
Materials. Feather keratin was obtained from Featherfiber Corporation (Nixa, MO). The feathers
were ground according to a previously used procedure [28]. Wheat gluten and corn zein were
obtained from MP Biomedicals LLC (Solon, OH). Type II egg albumin and lactalbumin were
obtained from Sigma-Aldrich (Saint Louis, MO). Glycerol was obtained from VWR
International (West Chester, PA).
Preparation of Films. Samples of 40 g total were prepared at 0-50 wt% glycerol. The samples
were mixed on a Waring blender on high speed for approximately 4 minutes. After mixing the
temperature of the samples was 35-40 °C. Films were prepared by pressing 5 g of each sample
between Teflon coated aluminum on a Carver Press Autofour/30 Model 4394 for two minutes at
Page 35
21
160 °C and pressing stress ranging from 8.6 to 46.0 MPa depending on the protein and
plasticization level. It has been shown that this type of processing can change all cystine bonding
in the protein to inter-molecular and this was the intent [29].
Fourier Transform Infrared (FT-IR) Spectroscopy. FT-IR spectra were obtained using a
Thermo Electron Avatar 370DTGS Spectrometer with a Pike Miracle ATR diamond cell. The
spectra were collected using 64 scans at 4 cm-1 resolution from 4000-525 cm-1. A blank was run
between each sample to ensure that the cell was completely clean and a background was
collected prior to each run. Deconvolution of the amide I band into individual components was
accomplished with OMNIC v 7.3 software. The spectral region 1725-1575 cm-1 of the original
spectra was fitted with Gaussian/Lorentzian peaks. The number of peaks and their positions were
determined by the automatic peak finding feature of the program at low sensitivity and full width
at half height of 1.928. All spectra were fitted using no baseline and a target noise of 10.0. The
same procedure was followed for all samples.
X-Ray Diffraction (XRD). Film samples were scanned on a PANalytical X'Pert PRO X-ray
diffractometer using Co radiation (wavelength = 1.78901 Å) generated at 40 kV and 40 mA.
Scanning proceeded with a Theta/Theta goniometer from 2-70 °2θ with a step size of 0.0668545
°2θ at a time of 600 s.
Mechanical Properties. Test samples (2.54 cm wide x 5.08 cm gage length) were prepared
according to ASTM D882-91. The average thickness of each sample was obtained from three
measurements using a micrometer. Force and elongation data were collected using a TA-XT2i
mechanical tester at 1.2 mm/s and converted into stress-strain data based on original sample
dimensions using a spreadsheet program. Modulus was determined from the initial low strain
Page 36
22
linear range of the stress-strain curve. Eight films were tested for each glycerol concentration of
every protein and the results are presented as averages with 95% confidence intervals.
RESULTS AND DISCUSSION
Physical Properties. Figure 1 shows the modulus, E, behavior for each protein as a function of
plasticization. CZ data was not obtained because of the difficulty in molding large enough
samples to tensile test. Plotting the data as a function of the weight percentage of glycerol
showed a split between all of the curves. Plotting the data as a function of the number of glycerol
molecules per protein molecule (mol glycerol/mol protein) showed that the modulus values were
very similar at low levels of plasticization. As more glycerol was added to the proteins, WG and
EA showed a marked decrease in E while LA and FK showed much smaller deviations. This
behavior was reminiscent of the glassy to rubbery transition in polymers as a function of
Figure 1. Elastic modulus of plasticized egg albumin (EA) (¡), lactalbumin (LA) (o), feather
keratin (FK) (¯), and wheat gluten (WG) (Í) films as a function of glycerol content.
Page 37
23
temperature. “At temperatures below the glass transition, polymers have approximately the same
value of modulus”[30] and the observed behavior suggested that at 298 K all of the proteins were
in a glassy state at low levels of glycerol plasticization [31]. Mechanical studies of whole
feathers have produced modulus values of about 1.5-2.5 GPa [32, 33]. Extrapolation of the
modulus data back to 0 wt% glycerol yielded a similar value indicating that modulus was
maintained upon reconstitution of the proteins from their native state.
Figure 2. Amide I peak position in the FT-IR spectra of plasticized egg albumin (¡),
lactalbumin (o), feather keratin (¯), wheat gluten (Í), and corn zein (+) films as a function of
glycerol content.
Conformational Changes. FT-IR analysis can be used to assess secondary structure changes in
proteins [34, 35]. Specifically, the amide I and II bands at ca. 1625 cm-1 and 1550 cm-1,
respectively, can be used to note changes in protein structure [20, 36]. In this case, changes in the
protein originating from plasticization were observed. Amide I originated from C=O stretching
Page 38
24
vibrations and has been shown to be a sensitive indicator of different types of protein secondary
structure [34]. Figure 2 shows the change in the amide I peak position as a function of added
glycerol. In general, the amide I peak shifted to lower wavenumber until reaching an inflection
and then saturating. Shifts in amide I to lower values were indicative of more ordering,
specifically β-sheet formation and the extent of the ordering was protein-dependent [20, 34].
This method of analyzing the amide I data was more straightforward than the method used by
Gao et al [20] that relied on plotting the ratio of the peak intensity at 1625 cm-1 (β-sheet) to those
at 1650 cm-1 and 1660 cm-1 (other structures and disordered portions). The inflection in the
amide I vs. glycerol curve defined a critical point in the plasticization. Table 2 summarizes
results of the deconvolution of the amide I band into individual components representing various
structural conformations. Peaks were assigned according to Goormaghtigh et al [37] and
Figure 3. Amide II peak position in the FT-IR spectra of plasticized egg albumin (¡),
lactalbumin (o), feather keratin (¯), wheat gluten (Í), and corn zein (+) films as a function of
glycerol content.
Page 39
25
references therein. Plotting the ratio of β-sheet and α-helix (ordered) to random and β-turn
(disordered), as determined from the deconvolution, at each glycerol concentration (not shown)
was consistent with Figure 2.
Table 2. Summary of FT-IR amide I deconvolution.
% Glycerol
0 10 20 30 40 50 FK
β-sheet 42 47 42 44 39 36
Random 17 12 14 21 25 22 α-helix 16 18 22 18 21 19
β-turn 25 23 23 17 15 23
EA β-sheet 41 42 44 53 47 41
Random 23 12 12 13 17 21 α-helix 20 21 19 18 19 17
β-turn 16 25 25 16 17 21
LA β-sheet 48 46 55 62 61 53
Random 11 8 10 10 12 12 α-helix 17 22 18 14 12 12
β-turn 24 24 17 15 14 22
WG β-sheet 34 37 43 45 33 34
Random 30 17 15 9 13 11 α-helix 20 22 24 23 36 29
β-turn 16 25 19 23 18 26
The amide II band originated in C-N stretching and N-H deformation vibrations [34]. As
glycerol was added, it was observed that the amide II peak position shifted to a higher frequency
for all of the proteins as shown in Figure 3. Eventually, the shifting of the amide II peak
Page 40
26
saturated. The saturation point was defined as a critical point in amide II behavior with
plasticization.
Figure 4. Critical plasticization concentration, c*, obtained from modulus (¡), FT-IR amide I
(o), and amide II (¯) data as function of the number of polar amino acids per protein.
Shown in Figure 1 is an arrow pointing to the critical point found for each protein from the
amide I data in Figure 2. For WG and EA, the rapid decrease in modulus occurred right after the
critical point while the modulus of LA and FK had a much less severe drop, or deviation from
the solid line, after the critical point. A critical point was also defined in the modulus as the point
at which the modulus deviated from the line in Figure 1. The critical point from amide I, amide
II, and modulus data was defined as the critical plasticization, c*. These values are listed in
Table 3. Figure 4 shows a plot of c* obtained from each data set versus the number of polar
amino acids per protein. Moles of glycerol/moles of protein was a number indicating how many
Page 41
27
glycerol molecules were associated with the protein and the amide I results showed a nearly 1:1
correlation. This indicated that all polar groups were associated with a glycerol molecule at c*.
Table 3. The critical plasticization point, c*, for each protein as defined from amide I, amide II,
and modulus data.
c*, (mol glycerol/mol protein) Amide I Amide II Modulus Wheat gluten 133 69 119 Lactalbumin 75 129 76 Egg Albumin 232 254 190 Feather Keratin 13 124 56 Corn zein 33 69
100
101
102
101
102
103
! A
mid
e I
Number of polar amino acids/protein or c*
0.74, r2=0.95
0.42, r2=0.71
WG
FK
EA
LA
CZ
WG
EA
FKCZ
LA
Figure 5. Absolute change in the FT-IR amide I peak position for each protein as a function of
the number of polar amino acids per protein (o), and critical plasticization concentration, c*, as
determined form FT-IR amide I data (¡).
Page 42
28
The absolute change in the amide I peak position from no glycerol to c* for each protein is
plotted in Figure 5. A strong correlation with c* and the number of polar amino acids per protein
was again found. Plotting the change in the amide II peak position again revealed a constant
number around 14 cm-1. Although both amide I and amide II changed as a function of
plasticization, amide I was strongly correlated with the level of plasticization and the protein
polarity whereas amide II was not.
Each protein also had different levels of cysteine. Figure 6 plots the change in the amide I peak
as a function of the percent cysteine in the protein. The change was less pronounced for more
cysteine indicating that disulfide bonds resisted conformational change with plasticization.
Figure 6. Absolute change in the amide I peak position (¡), and modulus, E, (o) at c* for each
protein as a function of the percent cysteine in egg albumin (EA), lactalbumin (LA), feather
keratin (FK), wheat gluten (WG), and corn zein (CZ).
Page 43
29
Figures 7 and 8 show the XRD spectra as plasticization level increased. For brevity, only WG
and FK are shown, which showed the largest and smallest change in the amide I peak position in
Figure 5. As more glycerol was added to WG, the peak around 1.00 nm decreased until it was
nearly flat at 50 wt% glycerol, indicative of amorphous behavior, as shown in Figure 7a. This
was a concentration effect from more glycerol and less ordered protein per unit volume. Also
with an increase in plasticization a new Bragg peak appeared on top of the original peak around
0.46 nm beginning at 20 wt% glycerol and disappearing at 50 wt% glycerol as shown in Figure
7b. The new Bragg peak was maximum at 20 wt% glycerol, which was consistent with Figure 2
and the ratio of ordered to disordered secondary structure obtained from the deconvolution of
amide I data. However, the new Bragg peak persisted after c*. Interestingly, the intensity of the
peak around 1.00 nm did not steadily decrease and remained steady over this plasticization
range.
1 104
2 104
3 104
4 104
5 104
5 10 15 20 25 30 35 40
0 wt%10 wt%20 wt%30 wt%40 wt%50 wt%
Inte
nsity
2!
2.5 104
3 104
3.5 104
4 104
4.5 104
18 20 22 24 26 28
0 wt%10 wt%20 wt%30 wt%40 wt%50 wt%
Inte
nsity
2!
(a) (b)
Figure 7. XRD patterns for WG made with variable glycerol concentrations (a) full pattern and
(b) peak around 0.46 nm (∼23 °2θ).
Page 44
30
The same did not happen for FK as shown in Figure 8a. The peak around 1.00 nm decreased in
intensity but saturated at 40-50 wt% glycerol and did not become flat like in WG. The original
peak around 0.46 nm did not show evidence of a new superimposed peak indicating that the
original β-sheet structure was intact as shown in Figure 8b. FT-IR amide I deconvolution also
indicated more resistance to conformational changes in FK than in WG as evidenced by the less
severe change in the order/disorder ratio for FK as a function of plasticization. In both FK and
WG, the peak around 1.00 nm showed a more marked intensity decrease than did the peak
around 0.46 nm.
1 104
2 104
3 104
4 104
5 104
5 10 15 20 25 30 35 40
10 wt%20 wt%30 wt%40 wt%50 wt%
Inte
nsity
2!
2.5 104
3 104
3.5 104
4 104
4.5 104
18 20 22 24 26 28
0 wt%10 wt%20 wt%30 wt%40 wt%50 wt%
Inte
nsity
2!
(a) (b)
Figure 8. XRD patterns for FK made with variable glycerol concentrations (a) full pattern and
(b) peak around 0.46 nm (∼23 °2θ).
Molecular Mobility and Effect on Modulus. At 298 K, c* was more than likely the
plasticization concentration where the glass transition temperature, Tg was equal to 298 K. The
Page 45
31
Couchman-Karasz relationship (eq 1) has been shown to be an effective way to predict the Tg of
biopolymers as a function of plasticization level [30, 38, 39].
!
lnTgTg,gly
"
# $ $
%
& ' ' =
wp lnTg,pTg,gly
"
# $ $
%
& ' '
wgly
Tg,pTg,gly
"
# $ $
%
& ' ' + wp
(1)
where the glass transition temperature of glycerol, Tg,gly was 180 K, wgly and wp were the mass
fraction of glycerol and protein, respectively, and Tg,p was the glass transition temperature of
protein.[40] The weak glass transition temperatures of the individual proteins have been shown to
be difficult to find experimentally using traditional techniques such as differential scanning
calorimetry [41]. Indeed, we have been able to find only the Tg of WG to be 453 K using DSC,
which was in good agreement with literature values [42, 43]. Using Tg,p values for the WG, CZ,
and EA proteins found in Katayama et al [41] at 448 K, 437 K, and 481 K, respectively, and
setting Tg=298 K, it was possible to find values for wgly through wgly+wp=1 and the Couchman-
Karasz relationship. Transforming wgly to c* values yields 198, 190, and 91 mol glycerol/mol
protein for WG, EA, and CZ, respectively. These were values that occurred near the inflection in
the amide I data in Figure 2 indeed suggesting that the conformational changes at the inflection
point in FT-IR occurred because of Tg=298 K being reached at c*. Although the proteins
possessed differences in secondary structure and cystine bonding, these differences did not
manifest in the glassy state. In comparing the properties of the different proteins, differences did
not occur until after reaching c*.
In the glassy state, addition of plasticizer increased order and the increase in ordering was more
pronounced for proteins with more polar groups and less cysteine. However, this ordering was
Page 46
32
not enough to differentiate the modulus of the proteins because the glassy behavior continued to
dominate. Proteins were able to hydrogen bond multiple times per amino acid and this was the
origin of the glassy behavior. Small amounts of plasticization did not allow the protein to
become rubbery. The large drop in modulus at c* was a transition from a glassy to a rubbery
state at high levels of plasticization. This transition was pronounced for proteins with a large
number of polar groups and little cysteine. However, for less polar proteins with large amounts
of cysteine, there was little deviation from the original glassy behavior, even after a transition in
the amide I peak in FT-IR. As plasticization level was increased, the Tg=298 K of the proteins
was met at c*. WG and EA showed more typical polymeric behavior of a drop in modulus before
reaching the rubbery plateau. FK and LA showed no transition from the glassy state to the
rubbery plateau because of the large cysteine content forming a highly cross-linked network. For
the rubbery plateau, the modulus was described by
!
E ="RTMc (2)
where Mc was the molecular weight between inter-molecular interactions, ρ was density, R was
the ideal gas constant, and T was absolute temperature so the modulus was directly proportional
to the number of molecules per unit volume [31]. The density of each protein was found from the
method of Fischer et al [44]. Adding glycerol decreased the density of the material slightly, by
about 3 % at the highest plasticization levels. The molecular weight between intermolecular
interactions, Mc, was calculated from Equation 2 using the experimental data for E. It was found
that Mc increased with plasticization, which was indicative of protein molecules being pushed
apart by glycerol molecules so that protein-glycerol interactions replaced protein-protein
intermolecular interactions. Figure 9 plots Mc at c* as a function of the distance between inter-
Page 47
33
molecular cross-links in the proteins, obtained by dividing the protein molecular weight by the
number of cysteine amino acids. A very strong correlation was found and the dashed line in
Figure 9 shows a slope of 1. Plotting E at c* versus the cysteine percentage in each protein as
shown in Figure 6 yielded a good correlation but not as strong as that in Figure 9. The transition
from glassy to rubbery behavior can be found using FT-IR by observing changes in the amide I
peak. However, for highly cross-linked networks, modulus will not show traditional changes
associated with a glass to rubber transition. Cystine bonding not only resisted conformational
changes induced by plasticization but maintained modulus at temperatures higher than Tg. Many
reports of plasticized biopolymers, especially proteins, exist to show the potential of these
materials as replacements for synthetic polymers [17-27]. To properly compare proteins to note
effects of secondary structure and cystine bonding on properties, the proteins needed to be in a
state of differentiation, which was not the glassy state. At room temperature, 298 K, the most
Figure 9. Molecular weight between intermolecular interactions, Mc, at c* as function of the
average distance between inter-molecular cross-links in the proteins.
Page 48
34
likely use temperature for plasticized proteins, comparing proteins at equal weight or mole
fraction of plasticizer did not ensure that all proteins were outside the glassy state and any
modulus differences may occur simply because one protein was in the glassy state and one was
not. At Tg=298 K, proteins above c* were in a state where modulus was differentiated by
differences in secondary structure (for instance, β-sheets) or high amounts of cystine bonding as
shown in Figure 9.
Primary Sequence Features Sensitive to Plasticization. Carbonyls and amines originated in
the peptide bond as well as the amino acid side groups in the protein. As glycerol entered the
protein structure, the hydrogen on the glycerol hydroxyl hydrogen-bonded with at least the
oxygen on a protein carbonyl group as suggested by the FT-IR data. In some cases the oxygen
on the glycerol hydroxyl hydrogen-bonded with the hydrogen on a protein amine. Clearly, the
number of polar side groups plays a role in the extent of plasticization of the molecule. The more
polar side groups, the more the protein can be plasticized. Figure 10 shows the changes in the
XRD peak positions at 0.46 and 1.00 nm, which corresponded to two distances between protein
molecules: a main chain carbonyl on one molecule hydrogen bonding to a main chain amine on
another (0.46 nm) and the total distance between two amino acid side groups interacting (1.00
nm), which was normal to the 0.46 nm distance [45]. As glycerol was added to FK and WG, the
peptide bonds on two different protein molecules were brought slightly closer together up to c*
and the saturation point more closely resembled that obtained from amide II data. For WG, the
hydrogen-bonding distance between two adjacent main chains stayed roughly constant over the
range of glycerol concentration where the new Bragg peak was observed. The side groups on two
different molecules were also brought closer together for FK up to c*. For WG, the side groups
were pushed away from each other up to c* then showed little change until the highest
Page 49
35
plasticization. This region of plasticization where there is little change in molecular distances but
the appearance of a new Bragg peak may indicate that an increase in possible conformational
states upon reaching a rubbery state resulted in small-scale ordering of some of the amino acids
in WG, which was only possible because of its large polarity and lack of cross-linking. In a
highly plasticized state, both proteins ended up with approximately the same molecular
dimensions in both directions but with different amounts of order.
0.44
0.445
0.45
0.455
0.46
0.465
0.95
0.96
0.97
0.98
0.99
1
1.01
0 100 200 300 400 500 600 700
WG
FK
WG
FK
Mole
cula
r D
ista
nce (
C=
O--
H-N
), n
m
Mole
cula
r D
ista
nce (
C-R
1--
R2-C
), n
m
mol glycerol/mol protein
* *
*
* *
*
Figure 10. Molecular distance between peptide bonds and side groups on two different
molecules as a function of glycerol content. * denotes concentrations where new Bragg peak was
observed on WG.
Taking the FT-IR and XRD data together, the amide II region was sensitive to changes in the
distance between two adjacent main chains and this did not change much while the amide I
Page 50
36
region was sensitive to overall changes with plasticization especially changes in interactions
between side groups. Glycerol ordered the carbonyl side groups found in asparagine, glutamine,
aspartic acid, and glutamic acid. Since these amino acids make up 37% of WG and only 18% of
FK, WG showed the largest change in conformation as evidenced by amide I and in fact showed
the creation of an entirely new ordered phase upon reaching a rubbery state as evidenced in XRD
data (Figure 7b). Alternately, the presence of cysteine made the protein less sensitive to
plasticization. Cystine bonding increased the resistance of proteins to conformational changes
induced by glycerol and allowed proteins of high cysteine content to maintain modulus even at
high levels of plasticization above Tg.
CONCLUSION
To properly compare proteins to note effects of primary and secondary structural features on
mechanical properties, the proteins need to be in a state of differentiation, which is not the glassy
state, i.e., Tg greater than 298 K. Plasticization did not differentiate the proteins in the glassy
state. In the glassy state, addition of plasticizer increased order in the protein. The extent of the
ordering was protein-dependent and was more pronounced for proteins with more polar groups
and less cysteine.
At low levels of glycerol plasticization all of the proteins remained in a glassy state at 298 K. As
more glycerol was added a critical plasticization point, c*, was reached when all polar groups in
the protein were associated with glycerol molecules. At c*, Tg dropped to 298 K, and differences
in secondary structure and cystine bonding were reflected in the modulus. Sensitivity to
plasticization increased with increased amount of polar side groups in the primary structure,
especially those with carbonyl groups, i.e, glutamine, asparagine, glutamic acid, and aspartic
Page 51
37
acid. Conversely, the presence of cysteine made the protein less sensitive to plasticization due to
formation of a robust cystine cross-linked network.
REFERENCES
[1] Barone, J. R., and Schmidt, W. F., 2006, "Nonfood Applications of Proteinaceous
Renewable Materials," J. Chem. Educ., 83(7), pp. 1003-1009.
[2] Stevens, E. S., 2002, Green Plastics: An Introduction to the New Science of
Biodegradable Plastics, Princeton University Press, Princeton, NJ.
[3] Wool, R. P., 2005, Bio-based polymers and composites Elsevier Academic Press,
Amsterdam; Boston.
[4] Mohanty, A. K., Misra, M., and Drzal, L. T., 2005, Natural fibers, biopolymers, and
biocomposites,, CRC Press, Boca Raton, FL.
[5] Barone, J. R., Schmidt, W. F., and Gregoire, N. T., 2006, "Extrusion of feather keratin,"
J. Appl. Polym. Sci., 100(2), pp. 1432-1442.
[6] Galietta, G., Di Gioia, L., Guilbert, S., and Cuq, B., 1998, "Mechanical and
Thermomechanical Properties of Films Based on Whey Proteins as Affected by
Plasticizer and Crosslinking Agents," J. Dairy Sci., 81(12), pp. 3123-3130.
[7] Redl, A., Morel, M. H., Bonicel, J., Vergnes, B., and Guilbert, S., 1999, "Extrusion of
Wheat Gluten Plasticized with Glycerol: Influence of Process Conditions on Flow
Behavior, Rheological Properties, and Molecular Size Distribution," Cereal Chem.,
76(3), pp. 361-370.
Page 52
38
[8] Di Gioia, L., and Guilbert, S., 1999, "Corn Protein-Based Thermoplastic Resins: Effect of
Some Polar and Amphiphilic Plasticizers," J. Agric. Food Chem., 47(3), pp. 1254-1261.
[9] Gennadios, A., Weller, C. L., Hanna, M. A., and Froning, G. W., 1996, "Mechanical and
Barrier Properties of Egg Albumen Films," J. Food Sci., 61(3), pp. 585-589.
[10] Arai, K. M., Takahashi, R., Yokote, Y., and Akahane, K., 1983, "Amino-Acid Sequence
of Feather Keratin from Fowl," European Journal of Biochemistry, 132(3), pp. 501-507.
[11] Nisbet, A. D., Saundry, R. H., Moir, A. J. G., Fothergill, L. A., and Fothergill, J. E.,
1981, "The Complete Amino-Acid Sequence of Hen Ovalbumin," European Journal of
Biochemistry, 115(2), pp. 335-345.
[12] Whitaker, J. R., and Tannenbaum, S. R., 1977, Food proteins, AVI Publishing, Westport,
CT.
[13] Kinsella, J. E., 1982, "Relationships between Structures and Functional Properties of
Food Proteins," Food proteins, P. F. Fox, and J. J. Condon, eds., Applied Science
Publishers, New York, p. 87.
[14] Di Gioia, L., Cuq, B., and Guilbert, S., 1999, "Thermal properties of corn gluten meal
and its proteic components," Int. J. Biol. Macromol., 24(4), pp. 341-350.
[15] Barone, J. R., Dangaran, K., and Schmidt, W. F., 2006, "Blends of Cysteine-Containing
Proteins," J. Agric. Food Chem., 54(15), pp. 5393-5399.
[16] Fraser, R. D. B., MacRae, T. P., Parry, D. A. D., and Suzuki, E., 1971, "The structure of
feather keratin," Polymer, 12(1), pp. 35-56.
Page 53
39
[17] Zhang, X., Hoobin, P., Burgar, I., and Do, M. D., 2006, "Chemical Modification of
Wheat Protein-Based Natural Polymers: Cross-Linking Effect on Mechanical Properties
and Phase Structures," J. Agric. Food Chem., 54(26), pp. 9858-9865.
[18] Zhang, X., Burgar, I., Do, M. D., and Lourbakos, E., 2005, "Intermolecular Interactions
and Phase Structures of Plasticized Wheat Proteins Materials," Biomacromolecules, 6(3),
pp. 1661-1671.
[19] Hernandez-Munoz, P., Kanavouras, A., Ng, P. K. W., and Gavara, R., 2003,
"Development and Characterization of Biodegradable Films Made from Wheat Gluten
Protein Fractions," J. Agric. Food Chem., 51(26), pp. 7647-7654.
[20] Gao, C., Stading, M., Wellner, N., Parker, M. L., Noel, T. R., Mills, E. N. C., and Belton,
P. S., 2006, "Plasticization of a Protein-Based Film by Glycerol: A Spectroscopic,
Mechanical, and Thermal Study," J. Agric. Food Chem., 54(13), pp. 4611-4616.
[21] Gueguen, J., Viroben, G., Noireaux, P., and Subirade, M., 1998, "Influence of plasticizers
and treatments on the properties of films from pea proteins," Ind. Crop. Prod., 7(2-3), pp.
149-157.
[22] Mangavel, C., Barbot, J., Gueguen, J., and Popineau, Y., 2003, "Molecular Determinants
of the Influence of Hydrophilic Plasticizers on the Mechanical Properties of Cast Wheat
Gluten Films," J. Agric. Food Chem., 51(5), pp. 1447-1452.
[23] Choi, S. G., Kim, K. M., Hanna, M. A., Weller, C. L., and Kerr, W. L., 2003, "Molecular
Dynamics of Soy-Protein Isolate Films Plasticized by Water and Glycerol," J. Food Sci.,
68(8), pp. 2516-2522.
Page 54
40
[24] Choi, W.-S., and Han, J. H., 2001, "Physical and Mechanical Properties of Pea-Protein-
based Edible Films," J. Food Sci., 66(2), pp. 319-322.
[25] Gennadios, A., Brandenburg, A. H., Weller, C. L., and Testin, R. F., 1993, "Effect of pH
on properties of wheat gluten and soy protein isolate films," J. Agric. Food Chem.,
41(11), pp. 1835-1839.
[26] Lim, L. T., Mine, Y., and Tung, M. A., 1999, "Barrier and Tensile Properties of
Transglutaminase Cross-linked Gelatin Films as Affected by Relative Humidity,
Temperature, and Glycerol Content," J. Food Sci., 64(4), pp. 616-622.
[27] McHugh, T. H., and Krochta, J. M., 1994, "Sorbitol- vs Glycerol-Plasticized Whey
Protein Edible Films: Integrated Oxygen Permeability and Tensile Property Evaluation,"
J. Agric. Food Chem., 42(4), pp. 841-845.
[28] Barone, J. R., Schmidt, W. F., and Liebner, C. F. E., 2005, "Thermally processed keratin
films," J. Appl. Polym. Sci., 97(4), pp. 1644-1651.
[29] Cuq, B., Boutrot, F., Redl, A., and Lullien-Pellerin, V., 2000, "Study of the Temperature
Effect on the Formation of Wheat Gluten Network: Influence on Mechanical Properties
and Protein Solubility," J. Agric. Food Chem., 48(7), pp. 2954-2959.
[30] Fried, J. R., 1995, Polymer Science and Technology, Prentice Hall PTR, Englewood
Cliffs.
[31] Ferry, J., 1980, Viscoelastic Properties of Polymers, John Wiley and Sons, Inc., New
York.
Page 55
41
[32] Bonser, R., and Purslow, P., 1995, "The Young's modulus of feather keratin," J Exp Biol,
198(4), pp. 1029-1033.
[33] Taylor, A. M., Bonser, R. H. C., and Farrent, J. W., 2004, "The influence of hydration on
the tensile and compressive properties of avian keratinous tissues," J. Mater. Sci., 39(3),
pp. 939-942.
[34] Pelton, J. T., and McLean, L. R., 2000, "Spectroscopic Methods for Analysis of Protein
Secondary Structure," Anal. Biochem., 277(2), pp. 167-176.
[35] Mangavel, C., Barbot, J., Popineau, Y., and Gueguen, J., 2001, "Evolution of Wheat
Gliadins Conformation during Film Formation: A Fourier Transform Infrared Study," J.
Agric. Food Chem., 49(2), pp. 867-872.
[36] Wellner, N., Mills, E. N. C., Brownsey, G., Wilson, R. H., Brown, N., Freeman, J.,
Halford, N. G., Shewry, P. R., and Belton, P. S., 2005, "Changes in Protein Secondary
Structure during Gluten Deformation Studied by Dynamic Fourier Transform Infrared
Spectroscopy," Biomacromolecules, 6(1), pp. 255-261.
[37] Goormaghtigh, E., Cabiaux, V., and Ruysschaert, J.-M., 1990, "Secondary structure and
dosage of soluble and membrane proteins by attenuated total reflection Fourier-transform
infrared spectroscopy on hydrated films," European Journal of Biochemistry, 193(2), pp.
409-420.
[38] Matveev, Y. I., Grinberg, V. Y., and Tolstoguzov, V. B., 2000, "The plasticizing effect of
water on proteins, polysaccharides and their mixtures. Glassy state of biopolymers, food
and seeds," Food Hydrocolloids, 14(5), pp. 425-437.
Page 56
42
[39] Barone, J. R., and Medynets, M., 2007, "Thermally processed levan polymers,"
Carbohydr. Polym., 69(3), pp. 554-561.
[40] Win, K. Z., and Menon, N., 2006, "Glass transition of glycerol in the volume-temperature
plane," Phys. Rev. E, 73(4), p. 040501(R).
[41] Katayama, D. S., Carpenter, J. F., Manning, M. C., Randolph, T. W., Setlow, P., and
Menard, K. P., 2008, "Characterization of amorphous solids with weak glass transitions
using high ramp rate differential scanning calorimetry," J. Pharm. Sci., 97(2), pp. 1013-
1024.
[42] Micard, V., Morel, M. H., Bonicel, J., and Guilbert, S., 2001, "Thermal properties of raw
and processed wheat gluten in relation with protein aggregation," Polymer, 42(2), pp.
477-485.
[43] Pouplin, M., Redl, A., and Gontard, N., 1999, "Glass Transition of Wheat Gluten
Plasticized with Water, Glycerol, or Sorbitol," J. Agric. Food Chem., 47(2), pp. 538-543.
[44] Fischer, H., Polikarpov, I., and Craievich, A. F., 2004, "Average protein density is a
molecular-weight-dependent function," Protein Sci, 13(10), pp. 2825-2828.
[45] Springall, H. D., 1954, The Structural Chemistry of Proteins, Academic Press, Inc., New
York.
Page 57
CHAPTER THREE
SELF-ASSEMBLING PEPTIDE
FROM NON-STRUCTURAL
PROTEINS
ABSTRACT
Wheat gluten is an amorphous storage protein. Trypsin hydrolysis of wheat gluten produced
glutamine-rich peptides. Some peptides were able to self-assemble into fibrous structures
extrinsic to native wheat gluten. The final material was an in situ formed peptide composite of
highly ordered nanometer-sized fibrils and micron-sized fibers embedded in an unassembled
peptide matrix. Fourier transform infrared spectroscopic and X-ray diffraction data suggested
that the new structures resembled cross-β fibrils found in some insect silk and implicated in prion
diseases. The largest self-assembled fibers were about 10 µm in diameter with right-handed
helicity and appeared to be bundles of smaller nanometer-sized fibrils. Results demonstrated the
potential for utilizing natural mechanisms of protein self-assembly to design advanced materials
that can provide a wide range of structural and chemical functionality.
Page 58
44
INTRODUCTION
Naturally occurring structural components, such as the materials comprising bone, muscle, joint,
skin, feather, hoof, horn, nail and hair, are structures of self-assembled protein molecules.
Proteins removed from structural materials do not retain the properties of the original material
demonstrating that it is the native assembly of proteins responsible for the properties [1]. Past
research has focused on developing processing techniques to reform proteins into materials for
various applications without regard to how much the native structure was compromised [1].
Nature routinely arranges proteins into larger structures via one of two mechanisms. First,
through folding into a highly selected and specific conformation based on side chain interactions.
Hence, the amino acid sequence of the protein determines the secondary and tertiary structures,
which in turn determine the arrangement of multiple folded protein molecules in the quaternary
structure. Folding is the most common mechanism of building proteinaceous structures in nature.
Second, unfolded and/or partially folded protein molecules can assemble through main chain
interactions independent of the amino acid sequence [2]. The resulting fibrous structures have a
highly stable, robust common core structure known as the cross-β in which peptide strands run
perpendicular to the fibril axis as illustrated in Figure 1 [3]. Cross-β fibrils are made of stacked
β-sheets that grow in the lateral direction into structures of high aspect ratio. Understanding
mechanisms of protein self-assembly would be key to designing and constructing useful
materials for various applications. The implications of self-assembly of proteins are actually
wider than just structural materials. It is known that prion diseases such as Alzheimer’s disease,
Type II Diabetes, and Creutzfeldt-Jakob disease are caused by self-assembled cross-β fibrils
generally known as amyloids [4, 5]. Accumulating evidence suggests that cross-β fibrils form a
Page 59
45
generic class of material despite the diversity of peptide building blocks [6, 7]. Although the
amino acid sequence does not dictate the fundamental organization of the cross-β structure [2],
studies have shown that certain features such as enrichment of glutamine residues facilitate self-
assembly into cross-β structures [5].
fibril axis
Figure 1. Schematic drawing of a cross-β fibril.
The superior mechanical properties, flexibility, versatility and ability to self-assemble have made
cross-β fibrils an attractive structural motif for nanomaterials applications and for the fabrication
of nanodevices [8]. The current state of the art for cross-β-based materials is to start with
peptides that are known cross-β formers such as polyglutamine, Aβ1-40, hen lysozyme, and
SAA1-12 [5, 9]. Large molecular weight peptides like the known cross-β formers are best
produced through de novo methods [10]. However, the high cost of de novo materials, as it exists
today, represents a major hurdle to large-scale production of peptide-based materials [10].
Here, we show that peptide-starting materials for cross-β structures can be produced through
proteolysis of an abundant plant protein that can be obtained cheaply. Proteolysis captures the
simplicity and potential large-scale production of short chain peptides. Wheat gluten (WG) is the
term for a nonstructural, highly disordered mixture of nitrogen storage proteins with an
abundance of glutamine (Q) in the amino acid composition. It is easily hydrated and gives dough
Page 60
46
its elasticity [11]. Wheat gluten contains repeated sequences of Q with proline (P) at the end of
the repeat sequence [12]. This mimics the cross-β forming so-called “Q-blocks” studied by
Perutz and others [5]. Gluten is increasingly being removed from bread and cereal products
because of its connection to Celiac disease [13]. It contains 58% glutenin (Gt) and 42% gliadin
(Gd) [12]. Gd has a molecular weight of ~30 kDa and a compact globular conformation with a
low content of β-structures [14]. Gt contains 28% high molecular weight (HMW) fraction of
~68-100 kDa and 72% low molecular weight (LMW) fraction of ~36-44 kDa [14]. The structure
of LMW Gt is similar to Gd [14]. HMW Gt subunits are disordered with little regular structure
but their mobility increases and β-sheets form at intermediate levels of hydration [11, 15]. Gluten
proteins are not known to fibrillate unless subjected to severe denaturing conditions [16-18]
and/or external mechanical stress through processes like extrusion [19] and spinning [20]. Here,
tryptic treatment of WG produced glutamine-rich peptides or “Q-blocks” that self-assembled into
the cross-β structure forming building blocks for highly ordered, fibrous structures from the
nanometer to the micrometer scale. Findings demonstrated the potential for utilizing natural
mechanisms of self-assembly and cheap protein sources to develop advanced high-performance
materials.
MATERIALS AND METHODS
Proteolysis. Twenty g WG (protein 75-80%; starch 9-19%; moisture 5-8%; fat (ether extract 0.5-
1.5%; ash .8-1.2%) (VWR International, West Chester, PA) was suspended in 800 ml de-ionized
H2O at 37°C. The reaction was initiated by the addition of trypsin (Type I from bovine pancreas,
Sigma-Aldrich, Saint Louis, MO) to give a final enzyme-to-substrate ratio of 1:1000 (w/w). In
the case of Gd and Gt (TCI America, Portland, OR), 5 g were suspended in 200 ml de-ionized
Page 61
47
H2O and the same enzyme-to-substrate ratio was used. Two enzymatic reactions were carried
out. The first with pH maintained in the range 8.0–8.3 by manual addition of 1.0 M sodium
hydroxide (Tryp-pH8). The second reaction was conducted with no pH adjustment (Tryp-pH5.7),
which started at pH 5.7 and ended at pH 5.2. Control samples were prepared with no enzyme
addition at pH 5.7 and pH 8 and showed the same results so the pH 5.7 data is presented as
“Control”. Mixtures were incubated at 37°C with continuous gentle stirring for 48 h. At the end
of the reaction, mixtures were heated to 95°C for 30 min to deactivate the enzyme, poured into
dishes made of Teflon-coated aluminum sheets, and dried under the fume hood at room
temperature.
SDS-polyacrylamide gel electrophoresis (SDS-PAGE). A Mini-PROTEAN Tetra
Electrophoresis System with a precast Tris-HCl gel consisting of 4% stacking gel and 15%
separating gel was used for SDS-PAGE. All materials and equipment for SDS-PAGE were
purchased from Bio-Rad Laboratories (Hercules, CA) unless otherwise indicated. Reducing
sample buffer was prepared by adding 2-mercaptoethanol (Sigma-Aldrich, Saint Louis, MO) to
Laemmli sample buffer (62.5 mM Tris-HCl, pH 6.8, 25% glycerol, 2% SDS, 0.01 Bromophenol
Blue) for a final concentration of 5%. Dried protein samples (1 mg) were suspended in 300 µl
sample buffer, centrifuged at 4000 rpm for 5 min, and the supernatants (10 µl) were used to load
the gel. Electrophoresis was done at a constant current of 20 mA for 100 min. The gel was
stained by Coomassie Brilliant Blue R-250.
Fourier transform infrared (FT-IR) spectroscopy. FT-IR spectra of dried material were
obtained using a Thermo Electron Nicolet 6700 Spectrometer with a Smart Orbit ATR diamond
cell. The spectra were collected using 64 scans at 4 cm-1 resolution from 4000-525 cm-1. A blank
Page 62
48
was run between each sample to ensure that the cell was completely clean and a background was
collected prior to each run. Deconvolution of the amide I band into individual components was
performed using OMNIC v 7.3 software. The range 1725−1575 cm-1 of the original spectra was
fitted with Gaussian/Lorentzian peaks [21]. The number and position of peaks were determined
by the automatic peak finding feature of the OMNIC program using low sensitivity and full
width at half-height of 3.857. All spectra were fitted using no baseline and a target noise of 10.0.
Scanning electron microscopy (SEM). SEM micrographs were obtained using a FEI Quanta
600 FEG environmental SEM instrument in high vacuum mode with a 10-14 mm working
distance, accelerating voltage of 5 kV, and Everhardt-Thornley SE-detector.
X-ray diffraction (XRD). XRD patterns were recorded on a PANalytical X'Pert PRO X-ray
diffractometer using Co radiation (wavelength = 1.78901 Å) generated at 40 kV and 40 mA, and
scanned with a Theta/Theta goniometer from 2-70 °2θ with a step size of 0.0668545 °2θ at a
time of 600 s.
RESULTS AND DISCUSSION
The SDS-PAGE profile of native and hydrolyzed WG indicated that the peptides resulting from
hydrolysis at pH 8, which was the optimum for trypsin activity, were smaller and more
uniformly sized (Figure 2). Only three bands were observed in the Tryp-pH8 hydrolysate: two
sharp around 37 and 29 kDa and one broad below 10 kDa. In comparison, seven bands were
observed in Tryp-pH5.7 hydrolysate: six sharp around 37, 29, 25, 18, 15 and 10 kDa and one
broad below 10 kDa. Tryp-pH5.7 hydrolysate was completely water-soluble, whereas Tryp-pH8
hydrolysate was only partially soluble. No fibrils or fibers were observed in the Control or the
Page 63
49
hydrolysate at pH 5.7 (Figure 3a). However, fibrous structures were evident in the SEM
micrographs of trypsin-hydrolyzed WG at pH 8 (Figures 3b-d). Small fibrils with diameter,
D~100-500 nm and length, L~2.5-10 mm were observed throughout the sample as shown in
Figure 3b. Larger fibers had more consistent diameters and lengths of Df~10 µm and Lf~100 mm
and appeared to be bundles of smaller fibrils with a right-handed helical twist (Figure 3c and d).
The large fibers were sporadic throughout the sample. Material between fibrils and fibers was
unassembled peptides.
Figure 2. SDS-PAGE profile of wheat gluten and hydrolysis products: (1) native wheat gluten,
(2) molecular weight marker, (3) Tryp-pH5.7, and (4) Tryp-pH8.
The shape and maxima of the amide I peak in the FT-IR spectra of Tryp-pH8 (Figure 4)
suggested that the new structure resembled that of cross-β fibrils with a sharp maximum at 1624
cm-1 [22]. Cleavage of WG at bulky side groups arginine and lysine produced shorter glutamine-
rich peptides that favored assembly into cross-β structures. Hydrolysis at pH 5.7 did not produce
as sharp an amide I maximum but the data indicated some cross-β formation albeit weaker than
Page 64
50
at pH 8. The modicum of cross-β structures produced in Tryp-pH5.7 was not sufficient to
assemble into mature fibrils that could be detected in SEM. Compared with Tryp-pH5.7, the
sharper peak at 1624 cm-1 in Tryp-pH8 indicated a higher amount of cross-β structure and less
native β-sheets [22]. Within the 48 hour time frame of the experiment, the smaller glutamine rich
Q-blocks of Tryp-pH8 could self-assemble into fibrils and fibers whereas the larger peptides of
Tryp-pH5.7 could not. Allowing the experiments to proceed for longer time showed the
emergence of fibrils in Tryp-pH5.7. Hydrolyzing wheat gluten at the optimal pH of 8 and then
dropping pH to 5.7 did not interfere with assembly kinetics. This suggested a conformational
dependence on self-assembly as shorter peptides could extend their backbone faster and allow
them to hydrogen bond together independently of pH. Given literature reports of cross-β
formation under high or low pH, it is highly likely that a severe decrease or increase in the pH of
Tryp-pH5.7 would allow it to self-assemble [6]. Tryp-pH5.7 peptides were longer and would
require stronger denaturing conditions to uncoil the molecules and allow them to self-assemble
side-by-side.
Figure 3. SEM micrographs of (a) Control and (b-d) Tryp-pH8.
Page 65
51
1500 1550 1600 1650 1700
1624
Ab
so
rba
nc
e
Wavenumber, cm-1
2
3
1
Figure 4. FT-IR spectra of (1) Control, (2) Tryp-pH5.7, and (3) Tryp-pH8.
The deconvolution of the amide I band (1600–1700 cm-1) in the spectra of Control, Tryp-pH5.7
and Tryp-pH8 revealed additional peaks listed in Table 1. The characteristic peak at 1684 cm-1 in
Tryp-pH8 indicated anti-parallel β-sheet formation likely originating from the non-fibril peptides
surrounding cross-β fibrils [23]. The peaks at 1675 cm-1 in Tryp-pH5.7 and 1671 cm-1 in Tryp-
pH8 were contributions from the asymmetrical CN3H5 stretching vibration of arginine side
groups [24]. Arginine was abundant on peptide ends in the trypsin-hydrolyzed protein unlike the
Control where a similar peak was not detected. The arginine values were similar for Tryp-pH5.7
and Tryp-pH8 but the SDS-PAGE showed a much more thorough hydrolysis in Tryp-pH8.
Therefore, the deconvolution may reveal that hydrolysis at arginine was not pH dependent within
the pH range studied but hydrolysis at lysine was, with lysine being inaccessible to trypsin at pH
5.7. Contributions from glutamine residues were evident in the spectra of both native and
hydrolyzed WG. The characteristic peaks at 1615 cm-1 in Control, 1608 cm-1 in Tryp-pH5.7, and
Page 66
52
1615 cm-1 in Tryp-pH8 originated from a glutamine side chain NH2 bending vibration [24]. The
glutamine contribution to the amide I band was reduced after enzymatic hydrolysis from 21% in
Control to 15% in Tryp-pH5.7 and 2% in Tryp-pH8. This reduction correlated positively with the
amount of cross-β structures in the hydrolyzed protein and was likely due to the involvement of
glutamine side chains in rigorous hydrogen bonding regimes ascribed to cross-β structures [25].
It was more difficult to infer structural information from the amide II band (1500–1600 cm-1)
over the amide I band despite the clear shift in amide II from lower to higher wavenumber.
Generally, amide II has been shown to be less sensitive to variation in secondary structure [21].
Additionally, this spectral region was sensitive to individual amino acid side chains such as those
of arginine and lysine [26]. Significant contributions from individual amino acids would mask
any secondary structure information especially for shorter peptides.
Table 1. Amide I band deconvolution results.
Control Tryp-pH5.7 Tryp-pH8 assignment frequency, cm-1 % frequency, cm-1 % frequency, cm-1 % β-sheet 1629 21 1623 27 1626 44 Random coil 1642 18 1640 19 1644 2 31-helix 1653 13 1652 14 1656 28 β-turn 1666 16 1664 14 1697 5 1690 1694 Antiparallel β-sheet 1680 8 1684 5 1684 10 Glutamine side group 1615 21 1608 15 1615 2 Arginine side group - - 1675 7 1671 8
XRD patterns confirmed the formation of cross-β structures in the trypsin-hydrolyzed WG
(Figure 5). The characteristic reflection of cross-β structures at ~0.48 nm was evident in Tryp-
pH5.7 and Tryp-pH8 [3]. The Control displayed a diffraction pattern indicative of a
predominantly amorphous material with very little order. The 0.46 nm length represented the
distance between peptide main chains in a native β-sheet but was very broad [27]. Upon
Page 67
53
treatment with trypsin, a sharp peak appeared superimposed on the Tryp-pH5.7 at 0.46 nm. The
sharp peak at 0.46 nm in Tryp-pH5.7 indicated an increase in traditional β-sheet content. Tryp-
pH5.7 had more unhydrolyzed, long peptides that formed traditional β-sheets between certain
peptide segments but those segments would not further self-assemble into distinct fibrils and
fibers. Tryp-pH5.7 did show a very small peak at 0.48 nm indicative of some cross-β formation.
The size of the peak at 0.48 nm correlated with the sharpness of the FT-IR peak at 1624 cm-1.
Although a small amount of cross-β building blocks were available in Tryp-pH5.7 these building
blocks were not enough to initiate fibril and fiber formation because the longer peptide segments
could not extend enough to continue assembly. Conversely, with more efficient hydrolysis at pH
8, there were enough short peptides to extend and form mature cross-β fibrils and fibers. The
cross-β structure has been shown to be insoluble in many solvents and this may explain why
2
3
1
0.5 1 1.5 2 2.5 3 3.5
0.48 nm
0.46 nm
1.85 nm
1.00 nm
1.54 nm
1/d , nm-1
Inte
nsit
y
Figure 5. X-ray powder diffraction patterns of of (1) native (Control) and trypsin-hydrolyzed
wheat gluten at (2) pH 5.7 and (3) pH 8
Page 68
54
Tryp-pH5.7 was more soluble than Tryp-pH8 [28]. In fact, the insoluble Tryp-pH8 fraction gave
a strong FT-IR peak at 1624 cm-1 similar to that in Figure 4 and was rich with fibers when
examined under SEM.
The larger lengths around and greater than 1.00 nm represented the distances between β-sheets
and were a function of peptide side groups [25, 27]. As shown in Figure 5, the peak at 1.00 nm
shifted and sharpened after trypsin treatment. The inter-sheet distance shifted from 1.0 nm in the
Control to 1.54 nm in Tryp-pH5.7. For Tryp-pH8, two peaks were superimposed on one another,
one at 1.54 nm and one at 1.85 nm. The reported inter-sheet distance of the cross-β structure of
de novo synthesized glutamine-rich short peptides was 1.6 nm [25], and the 1.85 nm length in
Tryp-pH8 likely originated from arginine or lysine [29]. So it appeared that at pH 5.7 the
majority of β-sheets were of the traditional type with glutamine mediating the inter-sheet
distance, which was shown in reference 14. In contrast, at pH 8, β-sheets of the cross-type
formed most likely at peptide ends where arginine and lysine were more abundant due to a
higher degree of hydrolysis.
Wheat gluten was a crude protein mixture so to discern the role of each component, glutenin and
gliadin were hydrolyzed separately with trypsin at pH 8. FT-IR and XRD data showed that the
trypsin-hydrolyzed gliadin peptides formed cross-β structures and trypsin-hydrolyzed glutenin
did not (Figure 6). Interestingly, FT-IR and XRD data for glutenin control indicated that the
sample may have already contained some cross-β structures using the peak maxima in both data
sets as an indicator. This may have been a result of the separation procedures used because
glutenin has been shown to form the cross-β structure under strong denaturing conditions [17].
However, no significant cross-β formation was observed in trypsin-hydrolyzed glutenin in 48
Page 69
55
hours. Concentrating on the observed stronger cross-β former gliadin, possible tryptic fragments
were evaluated in terms of their ability to form cross-β structures based on literature information.
Cross-β forming ability has been linked to high Q and asparagine (N) and low P and glycine (G)
content [30, 31]. Table 2 shows possible Q-rich tryptic gliadin fragments along with Q, N, P and
G composition. Among the Q-rich fragments, the 5.4 kDa fragment had less P and G and more
N. Therefore, it was most likely the 5.4 kDa tryptic gliadin fragment that self-assembled into
cross-β fibrils. Although very strong cross-β formers, no large fibers were observed in trypsin-
hydrolyzed gliadin peptides. This suggested a cooperative assembly process to build large cross-
β fibers where a larger molecular weight component was required.
Table 2. Possible tryptic gliadin fragments with glutamine (Q), asparagine (N), proline (P) and
glycine (G) composition.
sequence [32] molecular weight, KDa Q, % N, % P, % G, %
VPVPQLQPQNPSQQQPQEQVPLVQ
QQQFLGQQQPFPPQQPYPQPQPF
PSQQPYLQLQPFLQPQLPYSQPQP
FRPQQPYPQPQPQYSQPQQPISQQ
QQQQQQQQQQQQQQQQQIIQQIL
QQQLIPCMDVVLQQHNIVHGK
16.4 43.9 1.4 20.1 1.4
SQVLQQSTYQLLQELCCQHLWQIPE
QSQCQAIHNVVHAIILHQQQK
5.4 28.3 2.2 2.2 -
QQQQPSSQVSFQQPLQQYPLGQG
SFRPSQQNPQAQGSVQPQQLPQFE
EIR
5.7 36.0 2.0 14.0 6.0
Page 70
56
Figure 6. Effect of trypsin treatment on gliadin (top) and glutenin (bottom) with regard to
formation of cross-β structures as indicated by (a) FT-IR amide I band, and (b) XRD. Control
samples are shown in blue.
The observed cross-β fibrils suggested the following self-assembly mechanism, depicted
schematically in Figure 7: proteolysis followed by a heating step produced 1) short, extended
peptides with the ability to 2) self-assemble into ordered building blocks (cross-β sheets) through
main–chain hydrogen bonding. 3) An “elementary unit” of two cross-β sheets stacked face-to-
face was formed with inter-sheet distance mediated by the size of the glutamine, arginine, and
lysine side chains, ~1.6-1.9 nm, and inter-peptide spacing of 0.48 nm (Figure 5.3). When enough
self-assembling peptides were present in the solution, more elementary units formed and stacked
Page 71
57
into 4) cross-β superstructures. These superstructures further assembled into 5) fibrils of D~100-
500 nm or 67-333 stacked cross-β sheets and L~2.5-10 µm or 5,000-20,000 peptides side-by-side
along the fiber length with extended peptides oriented perpendicular to the fibril axis (Figure 3b).
Control and trypsin-hydrolyzed Gt never reached stage 1. Tryp-pH5.7 made it to stage 3.
Trypsin-hydrolyzed Gd reached stage 4. Combined Gt and Gd peptides formed fibrils that
twisted together into a right-handed conformation to form fibers of ultimate Df~10 µm and
Lf~100 µm (Figure 3c and 3d). Tryp-pH8 assembled beyond stage 5 to micron-sized fibers. The
fibrils were pitched 60o to the fiber axis with pitch distance of hf~1 µm. Fiber pitch was
determined by measuring the angle of the fibril twist relative to the fiber axis on SEM
micrographs (Figure 3). Eventually, the fiber cannot accommodate any more fibrils because the
energetic cost to keep twisting is too high so the final fiber size is thermodynamically limited [7,
33-35]. The chirality of naturally-occurring amino acids predicts a left-handed helical
conformation [36]. The observed fibrils and fibers were much larger than previously seen self-
assembled fibrous peptides and had right-handed helicity.
proteolysis,
95˚C
long chain,
disordered proteinshorter, extended
peptides
ordered building block
(!-sheet)cross-!
“elementary unit”
cross-! superstructure
(stable template)
cross-! fibril
main-chain
interaction
stacking
further assembly
(vertical and lateral)
more stacking
fibril axis,
direction of growth
~1.5 nm
! " #
$
%
Figure 7. Schematic illustration of cross-β fibril formation.
Page 72
58
CONCLUSION
Proteolysis of wheat gluten followed by brief thermal treatment initiated a self-assembly process
through aggregation of shorter extended peptides into cross-β sheets, which in turn provided the
basic building blocks for further assembly. Stacking of cross-β sheets, at an inter-sheet distance
of ~1.6-1.9 nm, then provided a stable template for further peptide assembly in the lateral
direction leading to the formation of a mature cross-β fibril of high aspect ratio. Concentration of
the solution allowed cross-β fibrils to further assemble into ~10 µm diameter fibers. The fact that
these highly ordered cross-β structures originated from wheat gluten demonstrated the potential
for large-scale production of amyloid-forming peptides through proteolysis of abundant proteins.
These can then be utilized to construct advanced nanomaterials that can provide a wide range of
structural and chemical functionality.
REFERENCES
[1] Barone, J. R., Schmidt, W. F., and Liebner, C. F. E., 2005, "Thermally processed keratin
films," J. Appl. Polym. Sci., 97(4), pp. 1644-1651.
[2] Fandrich, M., and Dobson, C. M., 2002, "The behaviour of polyamino acids reveals an
inverse side chain effect in amyloid structure formation," EMBO J., 21, pp. 5682-5690.
[3] Sunde, M., and Blake, C. C. F., 2000, "From the globular to the fibrous state: protein
structure and structural conversion in amyloid formation," Q. Rev. Biophys., 31(01), pp.
1-39.
[4] Nelson, R., Sawaya, M. R., Balbirnie, M., Madsen, A. Ø., Riekel, C., Grothe, R., and
Eisenberg, D., 2005, "Structure of the cross-β spine of amyloid-like fibrils," Nature, 435,
Page 73
59
pp. 773-778.
[5] Perutz, M. F., Johnson, T., Suzuki, M., and Finch, J. T., 1994, "Glutamine repeats as
polar zippers: their possible role in inherited neurodegenerative diseases," Proc. Nat.
Acad. Sci., 91, pp. 5355-5358.
[6] Knowles, T. P., Fitzpatrick, A. W., Meehan, S., Mott, H. R., Vendruscolo, M., Dobson,
C. M., and Welland, M. E., 2007, "Role of Intermolecular Forces in Defining Material
Properties of Protein Nanofibrils," Science, 318(5858), pp. 1900-1903.
[7] Chiti, F., and Dobson, C. M., 2006, "Protein Misfolding, Functional Amyloid, and
Human Disease," Annu. Rev. Biochem., 75(1), pp. 333-366.
[8] Cherny, I., and Gazit, E., 2008, "Amyloids: Not Only Pathological Agents but Also
Ordered Nanomaterials13," Angew. Chem. Int. Ed., 47(22), pp. 4062-4069.
[9] Rubin, N., Perugia, E., Goldschmidt, M., Fridkin, M., and Addadi, L., 2008, "Chirality of
Amyloid Suprastructures," J. Am. Chem. Soc., 130(14), pp. 4602-4603.
[10] Gazit, E., 2007, "Self-assembled peptide nanostructures: the design of molecular building
blocks and their technological utilization," Chem. Soc. Rev., 36(8), pp. 1263-1269.
[11] Shewry, P. R., Halford, N. G., Belton, P. S., and Tatham, A. S., 2002, "The Structure and
Properties of Gluten: An Elastic Protein from Wheat Grain," Philosophical Transactions:
Biological Sciences, 357(1418), pp. 133-142.
[12] Fox, P. F., and Condon, J. J., 1982, Food Proteins, Applied Science Publishers, London,
UK.
Page 74
60
[13] Kasarda, D. D., and D'Ovidio, R., 1999, "Deduced Amino Acid Sequence of an α-Gliadin
Gene from Spelt Wheat (Spelta) Includes Sequences Active in Celiac Disease," Cereal
Chem., 76(4), pp. 548-551.
[14] Shewry, P. R., and Tatham, A. S., 1990, "The prolamin storage proteins of cereal seeds:
structure and evolution," Biochem. J., 267(1), pp. 1-12.
[15] Anjum, F. M., Khan, M. R., Din, A., Saeed, M., Pasha, I., and Arshad, M. U., 2007,
"Wheat Gluten: High Molecular Weight Glutenin Subunits: Structure, Genetics, and
Relation to Dough Elasticity," J. Food Sci., 72(3), pp. R56-R63.
[16] Kasarda, D. D., Bernardin, J. E., and Thomas, R. S., 1967, "Reversible Aggregation of α-
Gliadin to Fibrils," Science, 155(3759), pp. 203-205.
[17] Mackintosh, S. H., Meade, S. J., Healy, J. P., Sutton, K. H., Larsen, N. G., Squires, A.
M., and Gerrard, J. A., 2009, "Wheat glutenin proteins assemble into a nanostructure with
unusual structural features," J. Cereal Sci., 49(1), pp. 157-162.
[18] McMaster, T. J., Miles, M. J., Kasarda, D. D., Shewry, P. R., and Tatham, A. S., 2000,
"Atomic Force Microscopy of A-Gliadin Fibrils and in situ Degradation," J. Cereal Sci.,
31(3), pp. 281-286.
[19] Reddy, N., and Yang, Y., 2007, "Novel Protein Fibers from Wheat Gluten,"
Biomacromolecules, 8(2), pp. 638-643.
[20] Woerdeman, D. L., Ye, P., Shenoy, S., Parnas, R. S., Wnek, G. E., and Trofimova, O.,
2005, "Electrospun Fibers from Wheat Protein: Investigation of the Interplay between
Page 75
61
Molecular Structure and the Fluid Dynamics of the Electrospinning Process,"
Biomacromolecules, 6(2), pp. 707-712.
[21] Vedantham, G., Sparks, H. G., Sane, S. U., Tzannis, S., and Przybycien, T. M., 2000, "A
Holistic Approach for Protein Secondary Structure Estimation from Infrared Spectra in
H2O Solutions," Anal. Biochem., 285(1), pp. 33-49.
[22] Zandomeneghi, G., Krebs, M. R. H., McCammon, M. G., and Fandrich, M., 2004, "FTIR
reveals structural differences between native β-sheet proteins and amyloid fibrils,"
Protein Sci, 13(12), pp. 3314-3321.
[23] Zurdo, J., Guijarro, J. I., and Dobson, C. M., 2001, "Preparation and Characterization of
Purified Amyloid Fibrils," J. Am. Chem. Soc., 123(33), pp. 8141-8142.
[24] Venyaminov, S. Y., and Kalnin, N. N., 1990, "Quantitative IR spectrophotometry of
peptide compounds in water (H2O) solutions. I. Spectral parameters of amino acid residue
absorption bands," Biopolymers, 30(13-14), pp. 1243-1257.
[25] Sikorski, P., and Atkins, E., 2005, "New Model for Crystalline Polyglutamine
Assemblies and Their Connection with Amyloid Fibrils," Biomacromolecules, 6(1), pp.
425-432.
[26] Barth, A., 2000, "The infrared absorption of amino acid side chains," Progress in
Biophysics and Molecular Biology, 74(3-5), pp. 141-173.
[27] Springall, H. D., 1954, The Structural Chemistry of Proteins, Academic Press, Inc., New
York.
Page 76
62
[28] Ippel, J. H., Olofsson, A., Schleucher, J. r., Lundgren, E., and Wijmenga, S. S., 2002,
"Probing solvent accessibility of amyloid fibrils by solution NMR spectroscopy," Proc.
Nat. Acad. Sci., 99(13), pp. 8648-8653.
[29] Fraser, R. D. B., and Mac Rae, T. P., 1973, Conformation in Fibrous Proteins and Related
Synthetic Polypeptides, Academic Press, New York.
[30] DePace, A. H., Santoso, A., Hillner, P., and Weissman, J. S., 1998, "A critical role for
amino-terminal glutamine/asparagine repeats in the formation and propagation of a yeast
prion," Cell, 93, pp. 1241-1252.
[31] Rauscher, S., Baud, S., Miao, M., Keeley, F. W., and PomËs, R., 2006, "Proline and
Glycine Control Protein Self-Organization into Elastomeric or Amyloid Fibrils,"
Structure, 14(11), pp. 1667-1676.
[32] Okita, T. W., Cheesbrough, V., and Reeves, C. D., 1985, "Evolution and heterogeneity of
the alpha-/beta-type and gamma-type gliadin DNA sequences," J. Biol. Chem., 260(13),
pp. 8203-8213.
[33] Davies, R. P. W., Aggeli, A., Beevers, A. J., Boden, N., Carrick, L. M., Fishwick, C. W.
G., McLeish, T. C. B., Nyrkova, I., and Semenov, A. N., 2006, "Self-assembling β-Sheet
Tape Forming Peptides," Supramol. Chem., 18(5), pp. 435-443.
[34] Nyrkova, I. A., Semenov, A. N., Aggeli, A., and Boden, N., 2000, "Fibril stability in
solutions of twisted β-sheet peptides: a new kind of micellization in chiral systems," Eur.
Phys. J. B, 17(3), pp. 481-497.
Page 77
63
[35] Nyrkova, I. A., Semenov, A. N., Aggeli, A., Bell, M., Boden, N., and McLeish, T. C. B.,
2000, "Self-assembly and structure transformations in living polymers forming fibrils,"
Eur. Phys. J. B, 17(3), pp. 499-513.
[36] Aggeli, A., Nyrkova, I. A., Bell, M., Harding, R., Carrick, L., McLeish, T. C. B.,
Semenov, A. N., and Boden, N., 2001, "Hierarchical self-assembly of chiral rod-like
molecules as a model for peptide β-sheet tapes, ribbons, fibrils, and fibers," Proc. Nat.
Acad. Sci., 98(21), pp. 11857-11862.
Page 78
CHAPTER FOUR
PEPTIDE SELF-ASSEMBLY
INTO MACROSCOPIC FIBERS
AT PHYSIOLOGICAL
CONDITIONS
ABSTRACT
There are major efforts underway to study protein self-assembly for various medical and
industrial reasons. Despite huge progress, most studies have focused on nanoscale self-assembly
but the crossover to the macroscopic scale remains a challenge. In this chapter, self-assembly of
macroscopic fibers from a combination of peptides is reported. The hierarchical multiscale self-
assembly was a cooperative process involving hydrophobic β- and α-peptides. Characterization
of the fibers is discussed and a mechanism by which they form is proposed. Finally, a
comparison with other natural systems is presented to show that the proposed self-assembly
mechanism may be a route utilized by several very different natural systems to create
sophisticated functional peptide materials. The ability to extend peptide self-assembly beyond
the nanoscale will have significant implications on design and fabrication of new functional
materials.
Page 79
65
INTRODUCTION
Nature relies on self-assembly from the molecular to the macroscopic scale to create
sophisticated functional materials. Naturally occurring protein-based structures have become the
subject of growing interest in materials science because of their superior mechanical properties,
versatility, and ability to self-assemble [1-3]. Of particular interest are cross-β fibrils which are
highly organized, self-assembled protein aggregates featuring an extensive β-sheet structure
stabilized by a dense network of backbone hydrogen bonds [4, 5]. Recent studies highlighted the
fact that cross-β fibrils are among the most robust biological materials and are comparable in
strength to steel [6, 7]. It has been suggested that cross-β structure could serve as a structural
motif for nanomaterials applications or in the fabrication of nanodevices [1-3, 8-10].
Furthermore, cross-β fibrils have been the subject of an explosive number of studies published in
recent years primarily because of their connection to human pathological conditions ranging
from neurodegenerative disorders to systemic amyloidoses [11]. There are major efforts
underway to understand peptide self-assembly in general and to develop cross-β-based materials
for various applications. Despite huge progress, studies have focused on nanoscale self-assembly
but the crossover to the macroscopic scale remains a challenge. Furthermore, studies have
mainly relied on self-assembly under non-natural conditions, i.e., low pH, high temperature or
high ionic strength, even though in nature most structures assemble under physiological or near
physiological conditions.
The formation of the cross-β structure does not require the presence of specific amino acid
sequence and is primarily the result of main chain interactions [12]. Accumulating evidence
suggests that virtually any polypeptide can fibrillate provided that the right conditions are met
Page 80
66
[13, 14]. These are usually extreme denaturing conditions that enable the conversion from a
soluble functional state into highly ordered fibrillar aggregates. In effect, cross-β fibrils are a
generic class of material despite the diversity of peptide building blocks, which is a very
interesting feature from a material science point of view. Although the amino acid sequence does
not dictate the fundamental organization of the cross-β structure, the nature of, and interactions
between, the side chains have significant influence on the rate and conditions under which it can
form [12]. Certain features of the polypeptide, such as hydrophobicity, charge, and secondary
structure propensity, have strong influence on the kinetics of self-assembly and these arise from
the sequence [15]. This perhaps explains the occurrence of cross-β structures in nature under
physiological conditions, as is the case in amyloidosis. Additionally, some living organisms
utilize the inherent ability of proteins to form cross-β structures to generate structures that can
endure harsh environments [16-20]. Although most of the functional cross-β structures in nature
are on the nanometer scale, macroscopic structures have also been identified [18].
It was possible to form macroscopic fibers in tryptic wheat gluten (TWG) [21]. In the previous
chapter, we presented preliminary data that suggested a cooperative assembly process was
required to form the fibers, but the process was not clearly understood. Here, we analyze the
hierarchical self-assembly of TWG peptides more in depth. Unlike in the previous study, TWG
fiber described here self-assembled under physiological conditions (pH 8 and 37°C) without
post-hydrolysis thermal treatment. The goal was to eliminate temperature as a variable, but
observations could yield insight into biological peptide aggregation as it relates to disease.
Results clearly demonstrated that peptide self-assembly beyond the nanoscale was possible in
vitro under physiological conditions. Our observations suggest that the macroscopic fibers self-
assembled as a result of the interaction between hydrophobic β-sheet-forming peptides and α-
Page 81
67
helix rich peptides. Interestingly, this kind of assembly, which involved templating and an α to β
transition, might be more prevalent in nature than what we previously thought. For example,
curli and chaplin proteins in bacteria and barnacle cement proteins self-assemble into cross-β
fibrils designed to perform various biological functions in adverse environments [17, 19, 20].
These proteins have similarities to the TWG self-assembling peptides in terms of
hydrophobicity, secondary structure and β-aggregation tendency. Barnacle cement proteins are
particularly interesting because they offer an example of self-assembly from the nanometer to the
micrometer scale.
MATERIALS AND METHODS
Enzymatic Hydrolysis. Wheat gluten (VWR International, West Chester, PA) was incubated in
a 2.5% w/w aqueous suspension with trypsin (Sigma-Aldrich, Saint Louis, MO) at 1:1000
enzyme-to-substrate ratio by weight. The solution was maintained at 37 °C and pH 8 by manual
addition of 1.0 M sodium hydroxide with continuous gentle stirring. Control reactions were
carried out under identical conditions without adding trypsin. Aliquots were frequently taken and
dried on Teflon-coated aluminum dishes under the fume hood at room temperature for further
analysis.
Size-exclusion chromatography (SEC). SEC was performed using the Superdex 200 HR
column on the ÄKTApurifier 10 FPLC system (GE Healthcare Biosciences, Pittsburgh, PA).
Samples were prepared by dissolving the protein in 0.1 M acetic acid, vortexing, centrifuging at
4000 rpm for 15 min, and injecting 0.5 ml of the supernatant onto the column. The column was
eluted with 0.1 acetic acid at 0.5 ml/min. Fraction were collected in 1 ml eppendorf tubes using
Frac-920 Fraction Collector (GE Healthcare Biosciences, Pittsburgh, PA).
Page 82
68
SDS-PAGE characterization of fibers. Fibers were manually isolated and washed 10 times
with pure H2O to remove any unassembled proteins. The washing step involved suspending in
H2O, vortexing for 2 min, then centrifuging and discarding the supernatant. Fibers were then
incubated overnight under vigorous agitation in SDS-PAGE sample buffer containing 2% SDS.
The mixture was then centrifuged and the supernatant run on a 4-15% Tris-HCl gradient gel.
SDS-PAGE was performed using a Mini-PROTEAN Tetra Electrophoresis System with a
precast Tris-HCl gel (Bio-Rad Laboratories, Hercules, CA). Reducing sample buffer was
prepared by adding 2-mercaptoethanol (Sigma-Aldrich, Saint Louis, MO) to Laemmli sample
buffer (62.5 mM Tris-HCl, pH 6.8, 25% glycerol, 2% SDS, 0.01 Bromophenol Blue) for a final
concentration of 5%. Electrophoresis was done at a constant current of 20 mA for 100 min. The
gel was stained with Biosafe Coomassie Stain (Bio-Rad Laboratories, Hercules, CA).
Thioflavin-T (Th-T) Binding Assay. Fluorophore Th-T (Sigma-Aldrich, Saint Louis, MO) was
used to confirm the core cross-β structure of the self-assembled TWG fibers. Twenty µl TWG
solution was dried on a glass slide, stained with 1% Th-T solution for 10 min and gently washed
with de-ionized H2O. Spatially resolved fluorescence images of the Th-T stained TWG fibers
were taken using a Zeiss Axio Imager M1 microscope. Th-T dye was excited at ~480 nm and
emission was collected at ~510 nm through a 10x objective.
Fourier Transform Infrared (FT-IR) Spectroscopy. Native and dried TWG (5 mg) were
mixed with 200 mg KBr and pressed into a disc using a die press. Transmission FT-IR spectra
were recorded on a Thermo Nicolet 6700 FT-IR Spectrometer (Thermo Fisher Scientific Inc.,
Waltham, MA) using 256 scans at 4 cm-1 resolution from 400-4000 cm-1. To study the reaction
kinetics, spectra of dried samples were obtained using the Smart Orbit ATR diamond cell. The
Page 83
69
spectra were collected using 256 scans at 4 cm-1 resolution from 4000-525 cm-1. A blank was run
between each sample to ensure that the cell was clean and a background was collected prior to
each run. Deconvolution and fitting of the amide I band was performed using OMNIC v 7.3
software. The spectral range 1700−1600 cm-1 was fitted with Gaussian/Lorentzian peaks. The
number and position of peaks were determined by the automatic peak finding feature of the
OMNIC program using low sensitivity and full width at half-height of 3.857. All spectra were
fitted using constant baseline and a target noise of 10.0.
Raman Spectroscopy. Spectra were obtained using a Bruker Senterra dispersive Raman
spectrometer with confocal microscope (Bruker, Billerica, MA). Samples were excited with a
785 nm laser. Analyses were carried out with a spatial resolution of 9-15 cm-1, output laser
power of 100 mW, and 10 s acquisition time. Scanning sites were examined following each
measurement to make sure the laser beam did not cause any damage. Spectra were baseline
corrected in the OPUS software using the Concave Rubberband Method with 5 iterations and 64
baseline points.
Scanning Electron Microscopy (SEM). Twenty µl of protein-enzyme solutions were deposited
on a SEM stub and allowed to dry under the fume hood. SEM micrographs were obtained using a
FEI Quanta 600 FEG environmental SEM (FEI Company, Hillsboro, OR) in high vacuum mode
with a 10-14 mm working distance, accelerating voltage of 5 kV, and Everhardt-Thornley SE-
detector.
Transmission Electron Microscopy (TEM). Samples were prepared by suspending 5 mg
protein in 10 ml de-ionized H2O followed by sonication for 5 min. The solution was 10-fold
diluted before a 5 µl drop was applied to a 200 mesh carbon-coated support film on a copper grid
Page 84
70
and allowed to dry at room temperature. The samples were then negatively stained by applying a
5 µl drop of 2% uranyl acetate (Electron Microscopy Sciences, Hatfield, PA) solution to the grid
for 5 min and finally dried under the fume hood. All specimens were examined on a Philips EM
420 Scanning Transmission Electron Microscope at an accelerating voltage of 120 kV.
X-Ray Diffraction (XRD). XRD patterns were recorded on a PANalytical X'Pert PRO X-ray
diffractometer (Westborough, MA) using Co radiation (wavelength = 1.78901 Å) generated at 40
kV and 40 mA. Scanning was done with a Theta/Theta goniometer from 2-70 °2θ with a step
size of 0.0668545 °2θ at a time of 600 s.
Nanoindentation. Nanoindentation measurements were carried out on microtomed surfaces of
transverse and longitudinal cross-sections of fibers embedded in EMbed 812 resin (Electron
Microscopy Sciences, Hatfield, PA) using a Hysitron TriboIndenter (Minneapolis, MN) with a
Berkovich diamond 142.3 degree, 3-sided pyramidal tip. Indentations were carried out at
ambient conditions under displacement rate control of 100 nm s-1 and 1000 nm maximum
displacement. The reduced modulus, Er, was determined according to Oliver-Pharr [22], as
!
Er
="S
2 Ac
(1)
where Ac was the contact area and S was the unloading stiffness, determined as the initial slope
of a polynomial function fit over 95-20% of the unloading curve. The fiber elastic modulus, Ef,
was related to the reduced modulus by
!
1Er
=1"# f
2
E f
+1"# t
2
Et (2)
where νf and νt were Poisson’s ratio of the fiber and indenter tip, respectively.
Page 85
71
RESULTS AND DISCUSSION
Macroscopic fibers assembled from TWG peptides under physiological conditions
SEC and SDS-PAGE analyses of native and hydrolyzed WG confirmed that trypsin had reduced
WG into smaller fragments (Figure 1). There were several distinct bands in native WG between
25 and 100 kDa that disappeared 24 hrs following tryptic treatment. Only three bands remained
in THWG: two sharp around 37 and 25 kDa, and one very broad below 10 kDa. The apparent
band at around 250 kDa in THWG8 was most likely protein aggregates and occurred at the line
between stacking and running gels. Conservation of bands around 37 and 25 kDa indicated
protease-resistant peptides known to WG [23].
Figure 1. a, Overlaid FPLC chromatograms. b, SDS-PAGE profile of native (lane 2) and
hydrolyzed (lane 3) wheat gluten showing reduction of most components to <10 kDa fragments.
The band at the line between stacking and running gels in lane 3 was due to aggregates. Only
two bands <10 kDa were present in the supernatant of partially dissolved fibers (lane 4).
Page 86
72
Macroscopic fibers extrinsic to native WG were observed in TWG solution after 24 hrs. Dried
TWG solution contained numerous fibrils and fibers that appeared to be consistent in size and
morphology as evidenced in TEM and SEM micrographs (Figure 2). Larger fibers were helical,
about 10 - 15 µm in diameter, and appeared to be bundles of smaller fibrils of about 10 - 20 nm
in diameter. SDS-PAGE analysis of deconstructed fibers, showed the presence of two peptides of
molecular weight <10 kDa. This was a key finding as it provided an important clue into what
peptides were forming the fibers.
TWG fibers possessed cross-β core structure
FT-IR, Raman, and XRD data suggested significant structural rearrangement involving α to β
transformation and the formation of cross-β structures as requisites for large fiber formation.
Dried WG and TWG samples were examined by FT-IR and Raman spectroscopy and XRD to
assess the underlying secondary structure of the resulting fibrous material. Overall change in the
Figure 2. a, Transmission electron micrograph of negatively stained TWG nanofibrils. Scale bar:
100 nm. b, Scanning electron micrograph of TWG fiber. Scale bar: 5 µm. c, Optical image
showing a single TWG fiber glued to a glass fiber. Scale bar: 0.1 mm.
Page 87
73
shape and position of the FT-IR amide I and amide II bands suggested a significant structural
reorganization involving unfolding of a native helical structure and increasing β-aggregation
(Figure 3). Relative decrease in the absorbance of amide I 1659 cm-1 and amide II 1538 cm-1
peaks was consistent with reduction in α-helical content [24]. In general, the range of cross-β
structures extends roughly from 1611 cm–1 to 1630 cm–1, while native β-sheet proteins produce
peaks around 1630 cm–1 to 1643 cm–1 [25]. The spectrum of TWG had a maximum at 1628 cm-1
indicating predominantly cross-β conformation. The increase in β content was also indicated by
the shoulder at 1520 cm-1 [24, 26].
Raman spectroscopy of native and hydrolyzed WG corroborated FT-IR data and showed
significant α to β transformation. As shown in Figure 4, most of the peaks assignable to α-helix
1500 1550 1600 1650 1700
Wavenumber, cm-1
1628cross-
side chainCOOH
-sheet
-helix
-helix
TWG
WG
Figure 3. Amide I and II of the Fourier transform-infrared spectra of native (dashed) and tryptic
wheat gluten showing decreased α content and increased β content.
Page 88
74
in WG, such as 1656, 1375, 1340, 1300, 1107, and 935 cm-1, shifted, weakened or disappeared in
TWG. At the same time, peaks assignable to β-sheets, such as 1680, 1670, 1028 and 980 cm-1
were more evident in TWG [26-28]. The presence of cross-β structures in THWG was indicated
by the characteristic peaks at 1558 cm-1 and 1400 cm-1 [28].
The XRD pattern of WG (Figure 5) was typical of a protein with a “fringed micelle”
microstructure with two diffuse halos occurring at Bragg spacings of about 0.46 and 1.00 nm
[29, 30]. The 0.46 nm spacing represented the distance between two β-strands in a native
(globular) β-sheet independent of the protein’s primary structure [29]. The 1.0 nm spacing
represented the inter-sheet distance between two stacked β-sheets and was a function of the
shape and size of the amino acid side chains [30]. The 0.46 nm peak shifted to 0.47 nm in TWG
wavenumber cm-1
ram
an
in
ten
sity
cro
ss-!
glo
bula
r-!
cro
ss-!
!-s
heet
!-s
heet
!-s
heet
CH3 p
ackin
g
"-h
elix
"-h
elix
"-h
elix
"-h
elix
"-h
elix
"-h
elix
"-h
elix
900 1000 1100 1200 1300 1400 1500 1600 1700
Figure 4. Raman spectra of native (bottom) and tryptic wheat gluten (top) showing α to β
transformation. Most of the peaks associated with α-helix in wheat gluten shifted, weakened or
disappeared in in tryptic wheat gluten.
Page 89
75
consistent with the inter-peptide distance in cross-β structures [4, 30]. Reflections with spacings
of 0.28 to 0.29 nm occur frequently in fibrils with the cross-β structure [4]. The inter-sheet
distance at 1.00 nm in WG shifted and sharpened (a strong indicator of ordering in powder
diffraction) after trypsin treatment suggesting a new distinct structural arrangement. The position
of superimposed peaks in TWG was closer together with peak maxima appearing at about 1.86
nm and 1.43 nm, which suggested more uniformity to the structure. It has been well established
that the inter-sheet stacking distance in cross-β structures can be closely related to the van der
Waals volumes of the amino acid residues, suggesting the amino acid side groups maintain the
inter-sheet distance [12]. Inter-sheet spacings ranged from 0.35 to 2.09 nm and were greatest for
peptides rich in amino acids with bulky side groups like arginine (R), lysine (K), glutamine (Q),
leucine (L) and isoleucine (I) [12, 15, 30, 31]. It appeared that these amino acids, which were
abundant at chain ends and in bulk of chain, were key determinants of inter-sheet stacking
distance in the TWG cross-β fibrils.
1 104
1.5 104
2 104
2.5 104
3 104
3.5 104
4 104
0.5 1 1.5 2 2.5 3 3.5 4
Inte
nsity
1/d , nm-1
0.47 nm
0.29 nm
0.28 nm
1.86 nm
1.43 nm
1.0 nm 0.46 nm
Figure 5. X-ray diffraction patterns of native (dotted) and tryptic wheat gluten.
Page 90
76
The cross-β core structure of the self-assembled THWG fibers was further confirmed with
thioflavin-T (Th-T) binding assay. Th-T molecules bind specifically to cross-β proteins by
inserting themselves into channels made by the regular repeat of residues and run along the
length of the β-sheet [32]. Since native β-sheets are unlikely to have these binding channels,
Thioflavin-T has been extensively used to detect the presence of cross-β structures and their rates
of formation [32]. Figure 6 shows a representative fluorescence microscopy snapshots of the Th-
T stained THWG fibers. Strong fluorescence indicated cross β-sheet structural arrangement of
the self-assembled THWG fibers.
Figure 6. Fluorescent microscopy images of the Th-T stained tryptic wheat gluten fiber.
TWG Fiber assembly was a cooperative process involving a hydrophobic β-strand and α-
helix forming peptides.
Neither glutenin nor gliadin, the two components of WG, formed fibers when hydrolyzed at the
same conditions as TWG. Tryptic glutenin (TGt) remained rich in disordered and α-helix
conformations while tryptic gliadin (TGd) formed cross-β structural units but did not assemble
into large fibers [21]. This suggested a cooperative assembly process involving peptides from
both gliadin and glutenin to build large cross-β fibers. Accordingly, individual Gt and Gd
Page 91
77
peptides (Table 1) were examined with regard to their role in the fiber assembly process. Peptide
names are listed in Table 1 and will be used henceforth.
Table 1. Possible tryptic wheat gluten fragments with tendency for β-aggregation predicted
using the TANGO algorithm and glutamine (Q), asparagine (N), proline (P) and glycine (G)
composition.
possible tryptic peptide sequence name
MW, KDa TANGO score
P (%) G (%) Q (%) N (%)
gliadin [43] MK Gd2 0.3 0 0.0 0.0 0.0 0.0 TFLILALLAIVATTATTAVR Gd20 2.1 1477 0.0 0.0 0.0 0.0 VPVPQLQPQNPSQQQPQEQVPLVQQQQFLGQQQPFPPQQPYPQPQPFPSQQPYLQLQPFLQPQLPYSQPQPFRPQQPYPQPQPQYSQPQQPISQQQQQQQQQQQQQQQQQQQIIQQILQQQLIPCMDVVLQQHNIVHGK
Gd139 16.4 2 0.201 0.014 0.439 0.014
SQVLQQSTYQLLQELCCQHLWQIPEQSQCQAIHNVVHAIILHQQQK
Gd46 5.4 207 0.022 0.0 0.283 0.022
QQQQPSSQVSFQQPLQQYPLGQGSFRPSQQNPQAQGSVQPQQLPQFEEIR
Gd50 5.7 0 0.14 0.06 0.36 0.02
NLARK Gd5 0.6 0
glutenin HMW [44]
QWQQSGQGQQGHYPTSLQQPGQGQQGHYLASQQQPGQGQQGHYPASQQQPGQGQQGHYPASQQQPGQGQQGHYPASQQEPGQGQQGQIPASQQQPGQGQQGHYPASLQQPGQGQQGHYPTSLQQLGQGQQTGQPGQK
GtH137 14.6 0 0.109 0.204 0.387 0.0
QQPGQGQQTGQGQQPEQEQQPGQGQQGYYPTSLQQPGQGQQQGQGQQGYYPTSLQQPGQGQQGHYPASLQQPGQGQPGQR
GtH80 8.6 0 0.125 0.225 0.413 0.0
QPGQGQQGYYPTSPQQPGQGQQLGQGQQGYYPTSPQQPGQGQQPGQGQQGHCPTSPQQSGQAQQPGQGQQIGQVQQPGQGQQGYYPTSVQQPGQGQQSGQGQQSGQGHQPGQGQQSGQEQQGYDSPYHVSAEQQAASPMVAK
GtH142 14.8 0 0.12 0.211 0.366 0.0
glutenin LMW [45]
PWQQQPLPPQQTFPQQPLFSQQQQQQLFPQQPSFSQQQPPFWQQQPPFSQQQPILPQQPPFSQQQQLVLPQQPPFSQQQQPVLPPQQSPFPQQQQQHQQLVQQQIPVVQPSILQQLNPCK
GtL120 14.1 0 0.225 0.0 0.425 0.8
AIIYSIILQEQQQVQGSIQSQQQQPQQLGQCVSQPQQQSQQQLGQQPQQQQLAQGTFLQPHQIAQLEVMTSIALR
GtL75 8.5 685 0.053 0.053 0.400 0.0
Page 92
78
Cross-β fibril forming ability has also been linked, among other factors, to high Q and asparagine
(N) and low proline (P) and glycine (G) content [33, 34]. Qd46 had relatively low P and G and
high Q and N, and GtL75 had low P and G and high Q (Table 1). The β-aggregating propensity
of individual Gt and Gd peptides was evaluated using The TANGO algorithm [35]. The accuracy
of the TANGO algorithm has been established in the literature and was found to be more than
90% for a set of 176 experimentally validated peptides [36]. TANGO scores showed that only
Gd20, Gd46 and GtL75 had β-aggregation propensity at 37°C and pH 8. The whole length of the
Gd20 sequence had a strong β-aggregation tendency compared to weaker tendency of only small
portions of the Gd46 and GtL75 sequences (Figure 7). Other algorithms focusing on different
aspects of the amino acid sequence further corroborated TANGO predictions [37-41]. Gd20 was
mostly hydrophobic and as such had a strong tendency to form nonfibrillar β-aggregates,
consistent with XRD and FT-IR data on trypsin hydrolyzed Gd [21]; whereas fibrillar aggregates
generally involve peptides with polar sequences like that of Gd46 and GtL75 [42]. Gd46 and
Gd20 alone, without GtL, could not assemble into fibrillar structures as was evident from the
separate TGd and TGt experiments. These observations suggested that TWG fibers formed as a
result of the interaction between Gd20 and GtL75. This conclusion was confirmed with SDS-
PAGE analysis of deconstructed fibers, which showed the presence of 2 peptides of molecular
weight <10 kDa (Figure 1) consistent with the molecular weights of Gd20 and GtL75.
Ding et al [46] suggested a generic scenario of templated cross-β aggregation in which a
preformed β-sheet with a hydrophobic surface catalyzes the transition to β-conformation in α-
rich peptides. The interaction between the hydrophobic β-sheet surface and the hydrophobic
portions of the α-helix reduces the free energy barrier thus enabling an α to β transformation. α
to β transitions appear to be promoted on hydrophobic surfaces. Gd20 was mostly hydrophobic,
Page 93
79
rich with β-stands and had the highest β-aggregation score. Secondary structure prediction,
performed using the GOR4 algorithm [47] showed that 35% of the Gd20 sequence was likely to
form β-strands compared with 4.4% for Gd46 and 6.7% for GtL75 (Figure 7). About half of the
GtL75 sequence adopted the α-helical conformation. It was likely that Gd20 peptides formed the
template that enabled α to β transformation in the α-rich GtL75 peptides and the subsequent
formation of large fibers. The α to β transformation was evident in FT-IR and Raman data as
discussed earlier.
Figure 7. Tendency for β-aggregation and secondary structure prediction at each amino acid
position for Gd20, Gd46 and GtL75. Aggregation scores, percentage per residue, were estimated
using the TANGO algorithm. Secondary structure predicted using the GOR4 algorithm.
Page 94
80
Raman spectroscopy provided additional experimental evidence of the significant contribution of
Gd20 to TWG fiber assembly. Scanning of the TWG fibers while noting orientation revealed a
mutually reciprocating phenomenon of Raman intensity generated from polypeptide backbones
and side chains (Figure 8). In general, cross-β structures showed a characteristic Raman peak at
~1672 cm-1 originating from polypeptide backbones [28]. This peak was most defined and
intense with the TWG fiber axis perpendicular to the laser’s plane of polarization, which was
expected because in this arrangement polypeptide chains were parallel to the laser’s plane of
polarization. The same peak shifted and weakened when the fiber axis was parallel to the plane
of polarization, i.e., peptide chains were perpendicular to the plane of polarization. The opposite
Figure 8. Amide I and II regions of the Raman spectra of TWG fibers oriented perpendicular
(top), 45° (middle) and perpendicular (bottom) to the laser’s plane of polarization.
Page 95
81
was true for the side chain CH2 and CH3 peaks in the 1400-1500 cm-1 region. This observation
indicated that the side chains, which were arranged approximately perpendicular to main chains,
contained an abundance of CH2 and CH3 groups. Eighteen out of the 20 amino acids in Gd20 had
CH3 groups in the side chains so Gd20 peptides appeared to be an important part of the fiber
assembly.
Hydrophobic interactions promoted nano- and micro-scale self-assembly
The reaction was followed using FT-IR over 120 hours time period. Deconvolution and fitting of
the amide I band in the FT-IR spectra of TWG displayed a loss of α-helix with a concurrent gain
of cross-β structure as a function of time (Figure 9a). The change in secondary structure content
was concurrent with a change in the environment of aliphatic moieties on the peptide (Figure
9b,c). The aliphatic moieties of amino acid side chains gave rise to characteristic absorbance
bands around 2960 cm-1, 2872 cm-1, 2855 cm-1 and 1445 cm-1 representing asymmetrical CH3
stretching, νas(CH3), symmetrical CH3 stretching, νs(CH3), symmetrical CH2 stretching, νs(CH2),
and asymmetrical CH3 deformation, δas(CH3), respectively. Shifts of ν(CH2) and ν(CH3) to
lower wavenumber have been shown to correlate with an increasingly dehydrated environment
for these chemical groups [48]. The findings showed hydrophobic groups tightly packing
together and α-helices unraveling to form β-sheets.
Gd20 was 95% hydrophobic and adopted β-strand secondary structure (Figure 7). GtL75 adopted
an α-helix-rich secondary structure and contained ‘conformational switches’, which were
sequence regions that have both the propensity for α-helix and β-sheet [38]. FT-IR analysis of
the self-assembly kinetics showed how the two peptides cooperatively built large fibers. During
the first 12 hrs, α to β transition was paralleled by an increasingly hydrophobic environment as
Page 96
82
evidenced by shifting of νs(CH2). Eighteen out of the 20 amino acid side groups on Gd20
contained CH3 but 24/75 side groups on GtL75 had CH3. The absorbance ratios νs(CH3)/νas(CH3)
and νs(CH3)/δas(CH3) indicated CH3 packing by comparing symmetrical stretching (which was
enhanced by the constraint of packing) with asymmetrical and deformation vibration (Figure 9c).
The continuously increasing ratio after 30 hours showed tight packing of Gd20 amino acid side
groups and was a delayed process occurring after the α to β transition. Additionally, νs(CH3)
displayed substantial shifting to lower wavenumber after this time period (Figure 9b) suggesting
Gd20 side groups were not exposed to a hydrophobic environment until tightly packed.
Hydrophobic Gd20 peptides acted as nucleators for GtL75 α to β transitions, consistent with the
template-assisted model for cross-β fibril formation [46]. After about 30 hours, a sufficient
hydrophobic environment existed and all of the α to β transition had occurred. Further
Figure 9. a, Shift to lower wavenumber of νs(CH3) and νs(CH2) bands with time indicated an
increasing hydrophobic environment. b, Estimated secondary structure content as a function of
time, determined by fitting of the FT-IR amide I band. c, The intensity ratios νs(CH3) to νas(CH3)
and νs(CH3) to δas(CH3) indicated amount of hydrophobic packing.
Page 97
83
aggregation involved tight packing of CH3 on Gd20 amino acid side groups over many length
scales.
Figure 10. a, TWG Fibers embedded in epoxy resin matrix. b, Nanoindentation image of
longitudinal and transverse sections showing indentation imprints. c, Comparison of
nanoindentation results in longitudinal and transverse sections. Error bars represent 95%
confidence interval; n=9.
Nanometer- and micrometer-scale ordering imparted anisotropic mechanical properties to the
macroscopic fibers. Nanoindentation was used to measure mechanical properties in the directions
transverse and longitudinal to the fiber axis (Figure 10). Fibers were embedded in epoxy resin
and microtome sectioned to reveal either transverse or longitudinal sections (Figure 10a,b). The
test yielded reduced moduli of 10.6 ± 1.2 GPa in transverse sections and 6.8 ± 1.0 GPa in
longitudinal sections demonstrating mechanical anisotropy in the fiber. In general, cross-β
nanofibrils have Young’s moduli in the range 2-20 GPa [6, 7, 49]. The remarkable transfer of
nanoscale mechanical properties to the macroscopic scale indicated efficient contact between
adjacent structures over multiple length scales through hydrophobic packing.
Page 98
84
Figure 11. Schematic illustration of the templated self-assembly process leading to formation of
macroscopic cross-β fibers.
Is the templated assembly of α-helix rich peptides a natural motif?
Our results demonstrated that peptide self-assembly beyond the nanoscale was possible in vitro
under physiological conditions. Macroscopic protein fibers form when α-helix-rich peptides
undergo an α to β transition in the presence of a hydrophobic β-sheet template. A review of the
literature suggested that our identified self-assembly mechanism may be a route utilized by
several very different natural systems. Curli proteins in bacteria rely on templated self-assembly
Page 99
85
processes involving hydrophobic and long α-helix-rich peptides to assemble cross-β fibrils for
various biological functions[19]. Curli are cross-β fibers assembled by enteric bacteria such as
Escherichia coli and Salmonella spp. to be the major proteinaceous component of a complex
extracellular matrix and critical determinants of biofilm formation. Polymerization of the major
curli subunit protein, csgA is dependent on the csgB nucleator [50]. csgB has two chains with a
total of 151 amino acids. The 21 amino acid chain, csgB-21, has characteristics similar to Gd20
(Table 2).
The protein Mrcp-100k is believed to form the cross-β fibrils observed in barnacle cement [17,
18]. Mrcp-100k has two chains of 18 and 975 amino acids with [51] characteristics similar to
Gd20 and GtL75, respectively (Table 2,3). Interestingly, micrometer-scale cross-β ‘rod-like’
Table 2. Properties of templating peptides: Gd20 and the templating peptides in curli and
barnacle cement proteins.
Peptide Sequence length hydrophobicity1 (red indicates hydrophobic)
β-sheet propensity?2
β-aggregation propensity?3
Gd20 20
T F L I L A L L A I V A T T A T T A V R
3
-3
Y Y
CsgB(1-21) [50] 21
M K N K L L F M M L T I L G A P G I A A A
3
-3
Y Y
Mrcp-100k(1-18) [51]
18
M M R L S L V A V L L V T V S V T G
3
-3
Y Y
1 based on the work of Hopp and Woods [52]
2 predicted using the SSpro v 4.5 [53]and GORIV [47]secondary struture prediction methods.
3 evaluated using multiple prediction algorithms [37-41]
Page 100
86
features have been observed in barnacle cement8. The templating peptides in bacterial curli and
barnacle cement have characteristics similar to Gd20 in terms of hydrophobicity, secondary
structure and β-aggregation tendency (Table 2). The templated (polymerizable) peptides contain
conformational switches and have β-aggregation tendency similar to GtL75 (Table 3).
CONCLUSION
Our analysis suggested the following self-assembly mechanism, depicted schematically in Figure
11: In general, the self-assembly process was driven by hydrophobic interactions and the
structure stabilized through backbone hydrogen bonding and hydrophobic packing. Short
hydrophobic peptides self-assembled into (1) β-sheets. (2) An “elementary unit” of two β-sheets
stacked face-to-face was formed with inter-sheet distance mediated by the size of the of side
Table 3. Properties of templated peptides: GtL75 and templated peptides in curli and barnacle
cement proteins. Conformational switches in these peptides are triggered upon interaction with
hydrophobic templating peptides, thus promoting β-aggregation.
Peptide Sequence length
Secondary structure1
(blue: α-helix, purple: β-strand, orange: random coil )
α to β conformational switches?2
β-aggregation propensity? 3
GtL75
75 Y Y
CsgA [50] 151 Y Y
Mrcp-100k (19-993) [51]
975 Y Y
1 predicted using the SSpro v 4.5 [53]
2 based on the work by Hamodrakas et al [38]
3 evaluated using multiple prediction algorithms [37-41]
Page 101
87
chains, ~1.48 nm, and inter-peptide spacing of 0.47 nm. Our previous study with trypsin
hydrolyzed Gt and Gd [21] showed that control and TGt lacked the aggregating hydrophobic
sequence necessary to initiate the process, and TGd, while having the right sequence, did not go
beyond stage (2). The stacking distance in the elementary cross-β unit formed in TGd was 1.48
nm [21]. By contrast, in the presence of Gt peptides, the hydrophobic Gd20 β-sheets catalyzed
(3) an α to β transformation in long α-rich peptides, which then (4) joined the assembly through
backbone β-aggregation. The stacking distance changed with inclusion of the new peptides to ~
1.43 to 1.86 nm (Figure 5). (5) Aggregation proceeded in the vertical (stacking) and lateral (sheet
elongation) directions to eventually form nanoscale fibrils of diameter equal to multiples of β-
sheet stacking distance as seen in the EM micrograph. The fibrils then (6) interact and bundle
together to form large fiber with 10 - 15 µm in diameter and few hundred micrometers in length.
This latter process has origins in the packing of hydrophobic amino acid side chains.
The finding discussed in this chapter showed that multiscale hierarchical peptide self-assembly
was possible in vitro under physiological conditions. The self-assembled macroscopic fibers
from tryptic peptides were characterized and a mechanism by which they form was proposed.
We highlighted similarities between the TWG system and other naturally-occurring systems in
which templated self-assembly is used to create functional peptide materials. In the next chapter,
we show that the concept of multiscale self-assembly from a combination of β- and α- peptides
can be extended to other peptide systems. We also demonstrate the ability to control macroscopic
properties of the fiber by varying the ratio of constituent peptides. The ability to extend peptide
self-assembly beyond the nanoscale will have significant implications on design and fabrication
of new functional materials.
Page 102
88
REFERENCES
[1] Cherny, I., and Gazit, E., 2008, "Amyloids: Not Only Pathological Agents but Also
Ordered Nanomaterials13," Angew. Chem. Int. Ed., 47(22), pp. 4062-4069.
[2] Channon, K., and MacPhee, C. E., 2008, "Possibilities for 'smart' materials exploiting the
self-assembly of polypeptides into fibrils," Soft Matter, 4(4), pp. 647-652.
[3] Zhang, S., 2003, "Fabrication of novel biomaterials through molecular self-assembly,"
Nature Biotechnology, 21(10), pp. 1171-1178.
[4] Sunde, M., 1997, "Common core structure of amyloid fibrils by synchrotron X-ray
diffraction," J. Mol. Biol., 273, pp. 729-739.
[5] Nelson, R., Sawaya, M. R., Balbirnie, M., Madsen, A. Ø., Riekel, C., Grothe, R., and
Eisenberg, D., 2005, "Structure of the cross-β spine of amyloid-like fibrils," Nature, 435,
pp. 773-778.
[6] Smith, J. F., Knowles, T. P. J., Dobson, C. M., MacPhee, C. E., and Welland, M. E.,
2006, "Characterization of the nanoscale properties of individual amyloid fibrils," Proc.
Nat. Acad. Sci., 103(43), pp. 15806-15811.
[7] Knowles, T. P., Fitzpatrick, A. W., Meehan, S., Mott, H. R., Vendruscolo, M., Dobson,
C. M., and Welland, M. E., 2007, "Role of Intermolecular Forces in Defining Material
Properties of Protein Nanofibrils," Science, 318(5858), pp. 1900-1903.
[8] Gazit, E., 2007, "Self-assembled peptide nanostructures: the design of molecular building
blocks and their technological utilization," Chem. Soc. Rev., 36(8), pp. 1263-1269.
Page 103
89
[9] Scheibel, T., Parthasarathy, R., Sawicki, G., Lin, X.-M., Jaeger, H., and Lindquist, S. L.,
2003, "Conducting nanowires built by controlled self-assembly of amyloid fibers and
selective metal deposition," Proc. Nat. Acad. Sci., 100(8), pp. 4527-4532.
[10] Lashuel, H. A., LaBrenz, S. R., Woo, L., Serpell, L. C., and Kelly, J. W., 2000,
"Protofilaments, Filaments, Ribbons, and Fibrils from Peptidomimetic Self-Assembly:
Implications for Amyloid Fibril Formation and Materials Science," J. Am. Chem. Soc.,
122(22), pp. 5262-5277.
[11] Dobson, C. M., 2003, "Protein folding and misfolding," Nature, 426(6968), pp. 884-890.
[12] Fandrich, M., and Dobson, C. M., 2002, "The behaviour of polyamino acids reveals an
inverse side chain effect in amyloid structure formation," EMBO J., 21, pp. 5682-5690.
[13] Dobson, C. M., 1999, "Protein misfolding, evolution and disease," Trends Biochem. Sci.,
24(9), pp. 329-332.
[14] Uversky, V. N., and Fink, A. L., 2004, "Conformational constraints for amyloid
fibrillation: the importance of being unfolded," Biochim. Biophys. Acta, Proteins
Proteomics, 1698(2), pp. 131-153.
[15] Perutz, M. F., Johnson, T., Suzuki, M., and Finch, J. T., 1994, "Glutamine repeats as
polar zippers: their possible role in inherited neurodegenerative diseases," Proc. Nat.
Acad. Sci., 91, pp. 5355-5358.
[16] Chiti, F., and Dobson, C. M., 2006, "Protein Misfolding, Functional Amyloid, and
Human Disease," Annu. Rev. Biochem., 75(1), pp. 333-366.
Page 104
90
[17] Barlow, D. E., Dickinson, G. H., Orihuela, B., Kulp, J. L., Rittschof, D., and Wahl, K. J.,
2010, "Characterization of the Adhesive Plaque of the Barnacle Balanus amphitrite:
Amyloid-Like Nanofibrils Are a Major Component," Langmuir, 26(9), pp. 6549-6556.
[18] Sullan, R. M. A., Gunari, N., Tanur, A. E., Chan, Y., Dickinson, G. H., Orihuela, B.,
Rittschof, D., and Walker, G. C., 2009, "Nanoscale structures and mechanics of barnacle
cement," Biofouling: The Journal of Bioadhesion and Biofilm Research, 25(3), pp. 263 -
275.
[19] Wang, X., and Chapman, M. R., 2008, "Curli provide the template for understanding
controlled amyloid propagation," Prion, 2(2), pp. 57-60.
[20] Claessen, D., Rink, R., de Jong, W., Siebring, J., de Vreugd, P., Boersma, F. G. H.,
Dijkhuizen, L., and Wösten, H. A. B., 2003, "A novel class of secreted hydrophobic
proteins is involved in aerial hyphae formation in Streptomyces coelicolor by forming
amyloid-like fibrils," Gene Dev, 17(14), pp. 1714-1726.
[21] Athamneh, A. I., and Barone, J. R., 2009, "Enzyme-mediated self-assembly of highly
ordered structures from disordered proteins," Smart Mater Struct, 18(10), p. 104024.
[22] Oliver, W. C., and Pharr, G. M., 1992, "An improved technique for determining hardness
and elastic modulus using load and displacement sensing indentation experiments," J.
Mater. Res., 7(6), pp. 1564-1583.
[23] Babiker, E. F. E., Fujisawa, N., Matsudomi, N., and Kato, A., 1996, "Improvement in the
Functional Properties of Gluten by Protease Digestion or Acid Hydrolysis followed by
Microbial Transglutaminase Treatment," J. Agric. Food Chem., 44(12), pp. 3746-3750.
Page 105
91
[24] Kretschmer, C. B., 1957, "Infrared Spectroscopy and Optical Rotatory Dispersion of
Zein, Wheat Gluten and Gliadin," J. Phys. Chem., 61(12), pp. 1627-1631.
[25] Zandomeneghi, G., Krebs, M. R. H., McCammon, M. G., and Fandrich, M., 2004, "FTIR
reveals structural differences between native β-sheet proteins and amyloid fibrils,"
Protein Sci, 13(12), pp. 3314-3321.
[26] Tinti, A., Foggia, M. D., Taddei, P., Torreggiani, A., Dettin, M., and Fagnano, C., 2008,
"Vibrational study of auto-assembling oligopeptides for biomedical applications,"
Journal of Raman Spectroscopy, 39(2), pp. 250-259.
[27] Tuma, R., 2005, "Raman spectroscopy of proteins: from peptides to large assemblies,"
Journal of Raman Spectroscopy, 36(4), pp. 307-319.
[28] Xu, M., Shashilov, V., and Lednev, I. K., 2007, "Probing the Cross-β Core Structure of
Amyloid Fibrils by Hydrogen‚àíDeuterium Exchange Deep Ultraviolet Resonance
Raman Spectroscopy," J. Am. Chem. Soc., 129(36), pp. 11002-11003.
[29] Springall, H. D., 1954, The Structural Chemistry of Proteins, Academic Press, Inc., New
York.
[30] Fraser, R. D. B., and Mac Rae, T. P., 1973, Conformation in Fibrous Proteins and Related
Synthetic Polypeptides, Academic Press, New York.
[31] Kōmoto, T., Kim, K. Y., Ōya, M., and Kawai, T., 1974, "Crystallization of polypeptides
in the course of polymerization, 5. Effect of the steric hindrance on the crystal growth by
the side chains," Die Makromolekulare Chemie, 175(1), pp. 283-299.
Page 106
92
[32] Krebs, M. R. H., Bromley, E. H. C., and Donald, A. M., 2005, "The binding of thioflavin-
T to amyloid fibrils: localisation and implications," J. Struct. Biol., 149(1), pp. 30-37.
[33] Rauscher, S., Baud, S., Miao, M., Keeley, F. W., and PomËs, R., 2006, "Proline and
Glycine Control Protein Self-Organization into Elastomeric or Amyloid Fibrils,"
Structure, 14(11), pp. 1667-1676.
[34] DePace, A. H., Santoso, A., Hillner, P., and Weissman, J. S., 1998, "A critical role for
amino-terminal glutamine/asparagine repeats in the formation and propagation of a yeast
prion," Cell, 93, pp. 1241-1252.
[35] Fernandez-Escamilla, A.-M., Rousseau, F., Schymkowitz, J., and Serrano, L., 2004,
"Prediction of sequence-dependent and mutational effects on the aggregation of peptides
and proteins," Nature Biotechnology, 22(10), pp. 1302-1306.
[36] Caflisch, A., 2006, "Computational models for the prediction of polypeptide aggregation
propensity," Current Opinion in Chemical Biology, 10(5), pp. 437-444.
[37] Galzitskaya, O. V., Garbuzynskiy, S. O., and Lobanov, M. Y., 2006, "Prediction of
Amyloidogenic and Disordered Regions in Protein Chains," PLoS Comput Biol, 2(12), p.
e177.
[38] Hamodrakas, S. J., Liappa, C., and Iconomidou, V. A., 2007, "Consensus prediction of
amyloidogenic determinants in amyloid fibril-forming proteins," Int. J. Biol. Macromol.,
41(3), pp. 295-300.
[39] Zhang, Z., Chen, H., and Lai, L., 2007, "Identification of amyloid fibril-forming
Page 107
93
segments based on structure and residue-based statistical potential," Bioinformatics,
23(17), pp. 2218-2225.
[40] Lopez de la Paz, M., and Serrano, L., 2004, "Sequence determinants of amyloid fibril
formation," Proc. Natl Acad. Sci. USA, 101, pp. 87-92.
[41] Maurer-Stroh, S., Debulpaep, M., Kuemmerer, N., de la Paz, M. L., Martins, I. C.,
Reumers, J., Morris, K. L., Copland, A., Serpell, L., Serrano, L., Schymkowitz, J. W. H.,
and Rousseau, F., 2010, "Exploring the sequence determinants of amyloid structure using
position-specific scoring matrices," Nature Methods, 7(3), pp. 237-242.
[42] Rousseau, F., Schymkowitz, J., and Serrano, L., 2006, "Protein aggregation and
amyloidosis: confusion of the kinds?," Curr. Opin. Struct. Biol., 16(1), pp. 118-126.
[43] Okita, T. W., Cheesbrough, V., and Reeves, C. D., 1985, "Evolution and heterogeneity of
the alpha-/beta-type and gamma-type gliadin DNA sequences," J. Biol. Chem., 260(13),
pp. 8203-8213.
[44] Anderson, O. D., Greene, F. C., Yip, R. E., Halford, N. G., Shewry, P. R., and Malpica-
Romero, J. M., 1989, "Nucleotide sequences of the two high-molecular-weight glutenin
genes from the D-genome of a hexaploid bread wheat, Triticum aestivum L. cv
Cheyenne," Nucl. Acids Res., 17(1), pp. 461-462.
[45] Colot, V., Bartels, D., Thompson, R., and Flavell, R., 1989, "Molecular characterization
of an active wheat LMW glutenin gene and its relation to other wheat and barley
prolamin genes," Molecular and General Genetics MGG, 216(1), pp. 81-90.
Page 108
94
[46] Ding, F., LaRocque, J. J., and Dokholyan, N. V., 2005, "Direct Observation of Protein
Folding, Aggregation, and a Prion-like Conformational Conversion," J. Biol. Chem.,
280(48), pp. 40235-40240.
[47] Garnier, J., Gibrat, J.-F., and Robson, B., 1996, "[32] GOR method for predicting protein
secondary structure from amino acid sequence," Methods in Enzymology, F. D. Russell,
ed., Academic Press, pp. 540-553.
[48] Maeda, Y., 2001, "IR Spectroscopic Study on the Hydration and the Phase Transition of
Poly(vinyl methyl ether) in Water," Langmuir, 17(5), pp. 1737-1742.
[49] Kol, N., Adler-Abramovich, L., Barlam, D., Shneck, R. Z., Gazit, E., and Rousso, I.,
2005, "Self-Assembled Peptide Nanotubes Are Uniquely Rigid Bioinspired
Supramolecular Structures," Nano Letters, 5(7), pp. 1343-1346.
[50] Hammer, N. D., Schmidt, J. C., and Chapman, M. R., 2007, "The curli nucleator protein,
CsgB, contains an amyloidogenic domain that directs CsgA polymerization," Proc. Nat.
Acad. Sci., 104(30), pp. 12494-12499.
[51] Kamino, K., Inoue, K., Maruyama, T., Takamatsu, N., Harayama, S., and Shizuri, Y.,
2000, "Barnacle Cement Proteins," J. Biol. Chem., 275(35), pp. 27360-27365.
[52] Hopp, T. P., and Woods, K. R., 1981, "Prediction of protein antigenic determinants from
amino acid sequences," Proc. Nat. Acad. Sci., 78(6), pp. 3824-3828.
[53] Pollastri, G., Przybylski, D., Rost, B., and Baldi, P., 2002, "Improving the prediction of
protein secondary structure in three and eight classes using recurrent neural networks and
Page 109
95
profiles," Proteins, 47(2), pp. 228-235.
Page 110
CHAPTER FIVE
TEMPLATED SELF-
ASSEMBLY OF LARGE
FIBERS USING A MODEL
SYSTEM
ABSTRACT
In previous chapters we examined the self-assembly of macroscopic fibers from peptide mixtures
and hypothesized that the minimum requirements were a long α-helix-rich peptide and a template
formed by hydrophobic β-peptides rich with aliphatic amino acids. Here we present a model
system showing that this was indeed a system capable of spontaneously forming large peptide
fibers. Macroscopic fibers self-assemble when α-helix-rich peptides undergo α to β transition in
the presence of a hydrophobic β-sheet template. A kinetic model was developed to describe the
process of fiber formation and was fitted to experimental data. The model successfully predicted
the rate of formation of fibrillar species as a function of the ratio of constituent peptides.
Page 111
97
INTRODUCTION
Most studies of peptide aggregation are in the context of trying to understand amyloid formation
as it relates to neurological disorders such as Alzheimer’s disease or bovine spongiform
encephalopathy (BSE) [1]. Amyloid formation is generic to peptides as long as peptide chains
can be straightened to allow backbone hydrogen-bonding [2]. Amyloid typically appears in the
form of a nanofibril ~101-102 nm in diameter and >103 nm long [3-5]. The nanofibril has a
characteristic cross-β structure, where β-strands are arranged orthogonally to the fibril length [6,
7]. Traditional fibers made of β-sheets, such as silk or keratin, have the β-strands parallel to the
fiber axis [8]. X-ray diffraction shows that the cross-β structure has strands separated by 0.48
nm, compared to 0.46 nm in native or globular β-sheets [9]. Also characteristic of the cross-β
structure is the appearance of a FT-IR absorbance at about 1615-1621 cm-1, compared to 1630-
1643 cm–1 in traditional β-sheets [6]. The cross-β nanofibril is robust with excellent solvent
resistance and a modulus up to 10 GPa [10-12].
Cross-β structures are rare in nature. The most common occurrence is in the misfolded
pathogenic amyloid form. However, “functional” amyloids have been identified that are
beneficial cross-β structures. The egg stalk of the green lacewing fly suspends the eggs to protect
them [13, 14]. Curli proteins reside on the surface of bacteria to bind matrix and plasma proteins
[15, 16]. Chaplin proteins self-assemble into amyloid-like fibrils that enable microbial aerial
hyphae to grow into the air by lowering surface tension and covering the structure [17].
Barnacle adhesive is a composite of amyloid fibrils in a less ordered protein matrix [18, 19].
Amyloids as a motif in materials science overwhelmingly refer to the nanometer sized fibril form
Page 112
98
and sometimes flat tapes [20, 21]. However, we have shown that fibers ~10 µm in diameter and
many millimeters long can self-assemble from the right combination of peptides [22].
In a previous study, we reported self-assembly of macroscopic fibers from tryptic wheat gluten
peptides. We hypothesized that the minimum requirements for the formation of fibers were a
long α-helix-rich peptide and a template formed by β-peptides rich with aliphatic amino acids.
The first 20 N-terminal amino acid sequence of gliadin aggregated to form the template, while
the long α-helix-rich peptide came from glutenin. In this paper, we show that the concept of
multiscale self-assembly from a combination of β-sheet and α-helix peptides can be extended to
other peptide systems. The experiment was repeated using myoglobin, which had most of its
sequence arranged in well-defined α-helices [23], and gliadin. Macroscopic fibers formed in the
trypsin-myoglobin-gliadin (TMG) solution. To form large fibers, a solution of myoglobin and
gliadin was maintained at 37oC and pH 8. Trypsin was added to release the N-terminal 3-22
peptide of gliadin (Gd20), which formed the template needed to recruit α-helical myoglobin
peptides into the fiber assembly. Macroscopic fibers formed in the trypsin-myoglobin-gliadin
(TGM) solution and their morphology and number influenced by the gliadin to myoglobin ratio.
Studying the kinetics of the reaction revealed a concentration-dependent process that involved a
competition between formation of non-fibrillar and fibrillar species.
MATERIALS AND METHODS
Growing the fibers. Myoglobin (Sigma Aldrich) and gliadin (TCI America) were incubated
together, at variable gliadin-to-myoglobin ratio w/w, in a 2.5% w/w aqueous suspension with
trypsin (Sigma Aldrich) at 1:100 enzyme-to-substrate ratio by weight. The solution was
maintained at 37 °C and pH 8 by manual addition of 1.0 M sodium hydroxide with continuous
Page 113
99
gentle stirring. Control reactions were carried out under identical conditions without adding
trypsin. Aliquots were frequently taken and dried under the fume hood at room temperature for
further analysis.
Size-exclusion chromatography (SEC). SEC was performed using the Superdex 200 HR
column on the ÄKTApurifier 10 FPLC system (GE Healthcare Biosciences, Pittsburgh, PA).
Myoglobin was dissolved in 50 mM potassium phosphate/ 0.15 NaCl, pH 7. Gliadin was
dissolved in in 0.1 M acetic acid, vortexed, centrifuged at 4000 rpm for 15 min, and 0.5 ml of the
supernatants injected onto the column. The column was conditions and eluted using the sample
buffer at 0.5 ml/min.
SDS-PAGE characterization of fibers. SDS-PAGE was performed using a Mini-PROTEAN
Tetra Electrophoresis System with a precast Tris-HCl gel (Bio-Rad Laboratories, Hercules, CA).
Reducing sample buffer was prepared by adding 2-mercaptoethanol (Sigma-Aldrich, Saint Louis,
MO) to Laemmli sample buffer (62.5 mM Tris-HCl, pH 6.8, 25% glycerol, 2% SDS, 0.01
Bromophenol Blue) for a final concentration of 5%. Electrophoresis was done at a constant
current of 20 mA for 100 min. The gel was stained with Biosfae Coomassie Stain (Bio-Rad
Laboratories, Hercules, CA).
Fourier Transform Infrared (FT-IR) Spectroscopy. Dried samples (5 mg) were mixed with
200 mg KBr and pressed into a disc using a die press. Transmission FT-IR spectra were recorded
on a Thermo Nicolet 6700 FT-IR Spectrometer (Thermo Fisher Scientific Inc., Waltham, MA)
using 256 scans at 4 cm-1 resolution from 525-4000 cm-1. Deconvolution of the amide I band into
individual components was performed using OMNIC v 7.3 software. The range 1725−1575 cm-1
of the original spectra was fitted with 13 Gaussian/Lorentzian peaks. The peaks were assigned to
Page 114
100
secondary structural components as described in the literature [24-26]. Initial peak positions
were determined by the automatic peak finding feature of the OMNIC program using low
sensitivity and full width at half-height of 3.857. All spectra were fitted using constant baseline
and a target noise of 10.0.
Raman Spectroscopy. Spectra were obtained using a Bruker Senterra dispersive Raman
spectrometer with confocal microscope (Bruker, Billerica, MA). Samples were excited with a
785 nm laser. Analyses were carried out with a spatial resolution of 9-15 cm-1, output laser
power of 100 mW, and 50 s acquisition time. Scanning sites were examined following each
measurement to make sure the laser beam did not cause any damage. Spectra were baseline
corrected in the OPUS software using the Concave Rubberband Method with 10 iterations and
64 baseline points.
Scanning Electron Microscopy (SEM). Twenty µl TGM solutions were deposited on a SEM
stub and allowed to dry under the fume hood. SEM micrographs were obtained using a LEO
Zeiss 1550 FESEM (Carl Zeiss SMT, Oberkochen, Germany) with a 7 mm working distance and
5 kV accelerating voltage.
Nanoindentation. Nanoindentation experiments were performed at room temperature using a
Hysitron Triboindenter (Minneapolis, MN) with a Berkovich diamond 142.3 degree 3-sided
pyramidal tip Indentations were carried out at ambient conditions under displacement rate
control of 100 nm s-1 and 1000 nm maximum displacement. The reduced modulus, Er, was
determined according to Oliver-Pharr [27], as
!
Er
="S
2 Ac
(1)
Page 115
101
where Ac was the contact area and S was the unloading stiffness, determined as the initial slope
of a polynomial function fit over 95-20% of the unloading curve. The fiber elastic modulus, Ef,
was related to the reduced modulus by
!
1Er
=1"# f
2
E f
+1"# t
2
Et (2)
where νf and νt were Poisson’s ratio of the fiber and indenter tip, respectively.
RESULTS AND DISCUSSION
Myoglobin α-helical peptides key to formation of macroscopic fibers
SEC and SDS-PAGE analyses showed that trypsin treatment had limited effect on myoglobin
(Figure 1). Some cleavage occurred but the bulk of the protein remained intact. Therefore, some
smaller peptides were released while the majority of myoglobin maintained its native α-helix
conformation. By contrast, gliadin was largely reduced to smaller fragments. Gliadin was not
soluble unless hydrolyzed to shorter peptides, and none of the soluble gliadin fragments formed
large fibers. Myoglobin was soluble but remained in the α-helix conformation and did not form
large fibers unless subjected to tryptic hydrolysis. However, adding unhydrolyzed myoglobin to
hydrolyzed gliadin, i.e. in the presence of the Gd20 template, did allow the formation of
macroscopic fibers (Figure 2).
Myoglobin had two possible tryptic fragments with properties similar to Gd20 (Table 1), which
explained why macroscopic fibers formed in tryptic myoglobin, 0% TMG. These fragments have
properties similar to Gd20 in terms of hydrophobicity profile, secondary structure and beta-
aggregation tendency as predicted multiple algorithms covering different aspects of the amino
Page 116
102
acid sequence (Table 1). Although limited, tryptic treatment was sufficient to release templating
peptides that enabled large fiber growth in hydrolyzed myoglobin when Gd20 peptides were not
available.
Figure 2. FE-SEM micrographs of macroscopic fibers formed after adding myoglobin to trypsin-
hydrolyzed gliadin.
17 kDa
> 100 kDa
14 kDa
10 kDa
37 kDa45 kDa
30 kDa
< 10 kDa
trypsin-hydrolyzed myoglobin
Myoglobin
trypsin-hydrolyzed gliadin
gliadin
Figure 1. Size-exclusion chromatograms of native and tryptic myoglobin and gliadin.
Page 117
103
Table 1. Properties of templating peptides from gliadin and myoglobin
Peptide Sequence length hydrophobicity1 (red indicates hydrophobic)
β-sheet propensity?2
β-aggregation propensity?3
Gd20 20
T F L I L A L L A I V A T T A T T A V R
3
-3
Y Y
Myoglobin (65-78) [28]
14 3
-3H G N T V L T A L G G I L K
Y Y
Myoglobin (104-119) [28]
16 3
-3Y L E F I S D A I I H V L H A K
Y Y
1 based on the work of Hopp and Woods [29]
2 predicted using the SSpro v 4.5 [30]and GORIV [31]secondary struture prediction methods.
3 evaluated using multiple prediction algorithms [32-36]
Kinetics of self-assembly was influenced by Gd20 peptide concentration
The spontaneous aggregation of Gd20 peptides into templating β-structures was a key driver of
the formation of the fibers. Accordingly, studying the kinetics of the reaction while varying the
concentration of Gd peptides provided important insight into the process. It was possible to vary
the concentration of Gd20 peptides by varying the ratio of gliadin-to-myoglobin. The experiment
was repeated with 0, 25, 50, 75 and 100% gliadin-to-myoglobin ratios and the reaction was
monitored over time using FT-IR spectroscopy. The number of fibers per 1 ml solution was
noted for each experiment and was found to be greatest in 50% TGM (Figure 3).
Page 118
104
0
5
10
15
0 20 40 60 80 100
num
ber o
f fib
ers
per m
l
gliadin to myglobin ratio, % w/w
more Gd20 template
Figure 3. Dependence of the number of self-assembled macroscopic fibers found in the solution
on the ratio of gliadin to myoglobin.
6
8
10
30
50
1 10 100
random coil!-helix"-sheet"-turn
y = 14 + 0.0051x R2= 0.002
y = 28 - 0.19x R2= 0.54
y = 10 + 0.13x R2= 0.58
y = 34 + 0.044x R2= 0.046
seco
ndar
y st
ruct
ure,
%
time, hr
Figure 4. Time-dependent change in secondary structure in 50% Trypsin-gliadin-myoglobin
solution calculated from FT-IR spectroscopy data.
Page 119
105
Figure 4 shows various secondary structure components as a function of time in 50% TGM
solution calculated by deconvolution and fitting of the FT-IR amide I band (see Materials and
Methods). There was no significant change (α=0.05) in the amount of random coil or β-turn
secondary structure in the solution over time. By contrast, there were significant changes
(α=0.05) in the amount of α and β secondary structure and the change was consistent with α to
β transformation. This suggested that, at 50% gliadin, any loss in α-helix was translated into gain
in β-sheets. The same was not true for other treatments with different gliadin levels. Although
significant (α=0.05) α to β transformation was observed at all gliadin levels, the amount of
random coil and/or β-turn also increased significantly (α=0.05) (Table 2). These results
suggested that when the optimum concentration of Gd20 peptide did not exist, the probability for
α-helical peptide to be recruited into the β-fiber assembly was reduced allowing peptides to fold
into random coils and β-turns instead. At 100% Gd, no fibers were found and the 5.1% net
increase in β-structure was due to non-fibrillar aggregates. At 100% Gd, β-content increased
Table 2. Net percent change in secondary structure content calculated from FT-IR spectroscopy
data for trypsin-gliadin-myoglobin solutions with variable gliadin content. Statistically
insignificant (α=0.05) values are not shown.
% change
gliadin content, % β-sheet α-helix random coil β-turn
0 47.6 -52.6 5.5 61.7
25 22.0 -24.5 55.5 -
50 95.2 -46.5 - -
75 109.3 -38.3 - 41.7
100 5.1 * -22.6 24.7 40.6
* No increase after 24 hours.
Page 120
106
33.4% during the first 8 hrs, but the right α-helical peptides were not available to collaborate and
form fibrillar aggregates. Instead, some non-fibrillar aggregation occurred and the rest of the
unstable β-peptides folded into random coils and β-turns.
The number of fibers in the solution was also found to be proportional to the rate of formation of
β-structures, dβ/dt (Figure 5d). The rates of change in α and β content, dα/dt and dβ/dt, were
determined by fitting the time-dependent secondary structure date for each gliadin level data to a
power function (Figure 5a,b). It was not clear whether a relationship between the rate of α-helix
unraveling and gliadin level existed or not (Figure 5c); as only weak association, R2 = 0.3, was
observed. The highest rate of decrease in α-helix content was observed in 100% Myoglobin and
probably had to do with the fact that myoglobin was more than 78% helix (starting from a high
baseline). By contrast, the rate of increase in β-content was influenced by the amount of gliadin
in the solution (Figure 5c). Interestingly, the maximum rate of increase in β-sheet content
coincided with the maximum number of fibers found per 1 ml of solution. Overall the number of
fibers in the solution increased linearly with the rate of increase in β-sheet content.
The fact that the number of self-assembled fibers was influenced by the ratio of gliadin to
myoglobin demonstrated the dependence of aggregation kinetics on the concentration of the
Gd20 template. There was an optimum gliadin level for the formation of fibers, which suggested
an inhibitory role played by Gd20 peptides in addition to their critical roles of recruiting α-
peptides into the fiber assembly. Initially, the number of fibers increased with more templates
becoming available. But at higher concentrations, the probability of Gd20 peptides interacting
with each other, and thus forming non-fibrillar β-aggregates, increased.
Page 121
107
10
100
1 10 100
0255075100
y = 39 * x^(-0.17) R= 0.77
y = 35 * x^(-0.1) R= 0.69
y = 33 * x^(-0.15) R= 0.81
y = 38 * x^(-0.14) R= 0.74
y = 26 * x^(-0.13) R= 0.69
!-h
elix
con
tent
, %
Time, hr
% Gliadin
a
1
10
100
1 10 100
0255075100
y = 7.1 * x^(0.092) R= 0.56
y = 6.9 * x^(0.096) R= 0.41
y = 7.7 * x^(0.2) R= 0.86
y = 5.9 * x^(0.22) R= 0.74
y = 20 * x^(0.019) R= 0.16
low
freq
uenc
y !-
shee
t con
tent
, %
time, hr
% Gliadin
b
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0
2
4
6
8
10
0 20 40 60 80 100
d!/dt d"/dt
rate
of i
ncre
ase
in !
-she
et c
onte
ct, d!/
dt
rate of decrease in "-sheet contect, d"
/dt
gliadin level, %
Y = 0.5 + .03*x - 0.0003*x2
R2 = 0.7
c
0
5
10
15
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
y = -2 + 9x R2= 0.7
num
ber o
f fib
ers
per m
l
rate of increase in !-sheet contect
d
Figure 5. a) Time dependent α-helix and (b) β-sheet content in trypsin-myoglobin-gliadin
solutions estimated using FT-IR spectroscopy. c) Dependence of the rate of α-helix unraveling
and β-sheet increase on gliadin content. d) Dependence of the number of fibers found in the
solution on the rate of β-sheet increase.
Page 122
108
Hydrophobic packing was a key driver of the multiscale self-assembly process
Hydrophobic packing seemed to play a critical role in the formation of macroscopic fibers in
TGM solutions. Following the reaction using FT-IR showed a change in the environment of
hydrophobic moieties on the peptide in addition to the change in secondary structure content.
The aliphatic moieties of amino acid side chains gave rise to characteristic FT-IR absorbance
bands around 2960 cm-1 and 1445 cm-1 representing asymmetrical CH3 stretching, νas(CH3) and
asymmetrical CH3 deformation, δas(CH3), respectively. Eighteen out of the 20 amino acid side
groups on Gd20 contained CH3. The absorbance ratios νs(CH3)/δas(CH3) indicated CH3 packing
by comparing symmetrical stretching (which was enhanced by the constraint of packing) with
deformation vibration (Figure 6). The ratio of the two intensities can thus indicate ‘hydrophobic
packing density’ which correlated with molecular organization and assembly. It was interesting
0
0.5
1
1.5
2
2.5
0 10 20 30 40 50 60 70
255075100
y = 1 - 0.0013x R2= 0.0025
y = 0.59 + 0.0084x R2= 0.71
y = 0.54 + 0.0069x R2= 0.59
y = 0.55 + 0.00045x R2= 0.0088
! s(CH
3)/"as
(CH
3)
time, hr
Figure 6. The intensity ratios (b) νs(CH3) to δas(CH3) indicated amount of hydrophobic packing.
Page 123
109
to note that sustained increase in the packing density in 50 and 75 % gliadin coincided with the
highest number of fibers in the solution. By contrast, at 100% gliadin, where no fibers formed,
there was essentially no change in hydrophobic packing density. At 25% gliadin, where some
fibers were noted, the initial increase in hydrophobic packing was not sustained after 24 hours,
indicating a critical template concentration needed to sustain the process of fiber self-assembly.
At 100% gliadin, although non-fibrillar aggregates occurred, there was no change in hydrophobic
packing density. These results suggested that the formation of fibrillar aggregates was associated
with a sustained change in hydrophobic packing, highlighting its role in micrometer scale self-
assembly.
Macroscopic properties can be controlled by varying the ratio of constituent peptides
SEM and nanoindentation analyses showed that the morphology and mechanical properties of
the fibers were influenced by the ratio of gliadin to myoglobin (Figure 7). It is known that,
despite a common core structure, cross-β nanofibrils often have heterogeneous morphology, size,
and mechanical strength depending on the constituent peptides [37]. With increased gliadin
content, fibers became shorter and more rigid. These observations were consistent with earlier
observations that shorter peptides assemble into nanofibrils with higher modulus [38]. It is also
interesting to note that the elastic modulus of micrometer scale “rod-like” objects observed in
barnacle cement ranged form 20 to 90 MPa similar to the values reported here for TGM fibers
[18]. Our observations demonstrated the utility of controlling the macroscopic properties by
varying the ratio of constituent peptides.
Page 124
110
Figure 7. a, Scanning electron micrographs of macroscopic fibers formed in trypsin-myoglobin-
gliadin solution (TMG) using different gliadin to myoglobin ratios. Scale bars: 100 µm. b,
Comparison of macroscopic properties of different TMG fibers. Aspect ratio values are averages
of up to 16 independent measurements. Reduced modulus values are averages of up to 24
independent measurements. Error bars represent standard error of the mean.
Kinetic model
Based on our understanding of the multiscale self-assembly of TGM fibers, a kinetic model
utilizing time-dependant secondary structure data was developed to describe the process. In this
model, a template (βt) was formed as a result of β-aggregation of short peptides rich with
aliphatic amino acids (Gd20) (Figure 8-1). The hydrophobicity of βt catalyzed an α to β
transformation in long α-rich peptides leading to formation of an intermediate structure,
αβ (Figure 8-3). From this point, two scenarios were possible. First, βt interacts with another βt
to form non-fibrillar aggregates, βnf, and together leave the solution. Second, βt interacts with αβ
Page 125
111
to form fibrillar aggregates, βf, which in turn form fibers. As such, βt performed an inhibitory
role with regard to the formation of fibers.
Figure 8. Schematic illustration of the templated self-assembly process leading to formation of
macroscopic cross-β fibers.
The proposed templated self-assembly process can be schematically represented as follows:
!
Gd20 +Gd20 k1" # " $t (3)
Page 126
112
!
"t +# k2 , k3$ % & & #" (4)
!
"t + "t kd# $ # "nf (5)
!
"t +#" k f$ % $ "f (6)
where βt represents the template resulting from aggregation of Gd20 peptides, αβ is an
intermediate complex comprising the partially unfolded α-helix peptide and the template, and βf
and βnf represent fibrillar and non-fibrillar aggregates, respectively. Assuming a first-order
reaction-rate expression for each step, the balances for each species lead to the following set of
ordinary differential equations (ODE)s:
!
d [Gd20]dt
= - k1 [Gd20][Gd20] (7)
!
d ["t ]dt
= k1 [Gd20][Gd20] # k2["t ][$] + k3[$"] # k f ["t ][$"] # kd["t ]["t ] (8)
!
d ["]dt
= - k2 [#t ]["] + k3["#] (9)
!
d ["#]dt
= k2 [#t ]["] $ k3["#] $ k f [#t ]["#] (10)
!
d ["nf ]dt
= kd["t ]["t ] (11)
!
d ["f ]dt
= k f ["t ][#"] (12)
Page 127
113
Equations 5–10 were solved numerically using the Runge-Kutta method. Kinetic constants were
estimated by performing a multidimensional parameter optimization using the steepest ascent
method [39]. The optimization procedure was automated using MATLAB (Appendix A.1). The
quality of fit of the model was assessed by calculating the mean square error, mse, between
experimental data points and simulation predictions,
!
mse =(xi " yi )
2
ni=1
n
# (12)
where xi was the simulation value at time = t, yi the raw data point value at time = t and n the
number of experimental data points.
The goal of the optimization procedure was to find the design vector x
!
x =
k1
k2
k3
kdk f
"
#
$ $ $
%
$ $ $
&
'
$ $ $
(
$ $ $
which minimizes the objective function, f (x) ) mse
To calculate mse, [αexp] and [βexp] measured using FT-IR spectroscopy for 50% TGM (Figure 9)
were used. Values of [αexp] were compared with those of [α] calculated from the model. Since βt
αβ, βf and βnf were all β structures, their sum, [βsim] was proportional to the amount of β-
structures measured with FT-IR, [βexp], i.e.,
!
["t ]+ [#"t ]+ ["nf ]+ ["f ] = ["sim ]$ ["ext ] (13)
Page 128
114
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1 10 100
y = 0.88 * x^(-0.15) R= 0.81
!-h
elix
con
tent
, [!
exp],
%
Time, hr
a
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1 10 100
y = 0.4 * x^(0.2) R= 0.86
!-sh
eet c
onte
nt, [! ex
p], %
time, hr
b
Figure 9. Time-dependent α-helix (a) and β-sheet (b) content in 50% TGM estimated from FT-
IR data.
To start the optimization procedure, initial estimates of the kinetic constants (k1, k2, k3, kd, kf)
were determined from the experimental data. The rate of decrease in α-helix content, calculated
by fitting the 50% TGM data to a power function was
!
d ["exp]dt
= - 0.132 t#1.15
(14)
where the quantity 0.132 represented the rate constant for the unraveling of α-helix, k2. Similarly
k1 was estimated using the normalized time-dependent β−content for 50% gliadin (Figure 1b).
The rate of increase in β−content was
!
d ["exp]dt
= 0.08 t0.8
(15)
Page 129
115
where the quantity 0.08 represented the rate constant for total β-aggregation. The initial design
vector was generated accordingly,
!
x =
0.080.1320.1320.1320.132
"
#
$ $ $
%
$ $ $
&
'
$ $ $
(
$ $ $
, and the initial objective function, f (x1) was calculated.
!
f (x1) = 0.1391
The following initial conditions were used:
[Gd200] =1, [βt0] = 0, [α0] = 1, [αβ0] = 0, [βnf0] = 0, [βf0] = 0
and after 8712 iterations, the following values were obtained:
!
x8712 =
0.26400.17100.13260.00630.0788
"
#
$ $ $
%
$ $ $
&
'
$ $ $
(
$ $ $
, f (x8712) = 0.04434
The model was then used to simulate the formation of fibril species as a function of gliadin level.
Results are summarized in Figure 9. The model was in good agreement with experimental data in
determining optimum gliadin level for fiber formation.
Page 130
116
0 20 40 60 80 100
simulated fibrillar aggregates, [!f]
experimental d!/dt
number of fibers found in solution
0
0.2
0.4
0.6
0.8
1
norm
aliz
ed u
nits
gliadin level, %
Figure 10. Comparison of experimental and simulated data.
CONCLUSION
The previous chapter of this dissertation discussed multiscale self-assembly of macroscopic
fibers from a combination of β- and α- tryptic wheat gluten peptides. Results discussed in this
chapter showed that the concept of multiscale self-assembly from a combination of β- and α-
peptides could be extended to other peptide systems. The TGM system presented in this chapter
provided the opportunity to further understand the multiscale self-assembly process by enabling
control over the concentration of templating peptide. Results suggested that nano- and micro-
scale self-assembly were both associated with hydrophobic interactions of the aliphatic amino
acids. The remarkable transfer of properties from the nanostructures to the macroscopic fibers
also indicated intimate molecular interactions such as hydrophobic packing. Using the TGM
Page 131
117
system, it was also possible to demonstrate the ability to control the macroscopic mechanical
properties by varying the ratio of constituent peptides.
Owing to the richness of the amino acid building blocks, peptides are highly versatile structural
and functional building blocks. The ability to extend, and control, peptide self-assembly over
multiple length scales is a significant leap toward incorporating peptide materials into dynamic
systems of higher complexity and functionality.
REFERENCES
[1] Prusiner, S. B., 1998, "Prions," Proc. Nat. Acad. Sci., 95(23), pp. 13363-13383.
[2] Dobson, C. M., 2003, "Protein folding and misfolding," Nature, 426(6968), pp. 884-890.
[3] Bouchard, M., Zurdo, J., Nettleton, E. J., Dobson, C. M., and Robinson, C. V., 2000,
"Formation of insulin amyloid fibrils followed by FTIR simultaneously with CD and
electron microscopy," Protein Science, 9(10), pp. 1960-1967.
[4] Sipe, J. D., and Cohen, A. S., 2000, "Review: history of the amyloid fibril," J. Struct.
Biol., 130, pp. 88-98.
[5] Jiménez, J. L., Nettleton, E. J., Bouchard, M., Robinson, C. V., Dobson, C. M., and
Saibil, H. R., 2002, "The protofilament structure of insulin amyloid fibrils," Proc. Nat.
Acad. Sci., 99(14), pp. 9196-9201.
[6] Kammerer, R. A., Kostrewa, D., Zurdo, J. s., Detken, A., García-Echeverría, C.,
Green, J. D., Müller, S. A., Meier, B. H., Winkler, F. K., Dobson, C. M., and Steinmetz,
M. O., 2004, "Exploring amyloid formation by a de novo design," Proc. Nat. Acad. Sci.,
Page 132
118
101(13), pp. 4435-4440.
[7] Sikorski, P., and Atkins, E., 2005, "New Model for Crystalline Polyglutamine
Assemblies and Their Connection with Amyloid Fibrils," Biomacromolecules, 6(1), pp.
425-432.
[8] Fraser, R. D. B., MacRae, T. P., Parry, D. A. D., and Suzuki, E., 1971, "The structure of
feather keratin," Polymer, 12(1), pp. 35-56.
[9] Perutz, M. F., Johnson, T., Suzuki, M., and Finch, J. T., 1994, "Glutamine repeats as
polar zippers: their possible role in inherited neurodegenerative diseases," Proc. Nat.
Acad. Sci., 91, pp. 5355-5358.
[10] Smith, J. F., Knowles, T. P. J., Dobson, C. M., MacPhee, C. E., and Welland, M. E.,
2006, "Characterization of the nanoscale properties of individual amyloid fibrils," Proc.
Nat. Acad. Sci., 103(43), pp. 15806-15811.
[11] Kol, N., Adler-Abramovich, L., Barlam, D., Shneck, R. Z., Gazit, E., and Rousso, I.,
2005, "Self-Assembled Peptide Nanotubes Are Uniquely Rigid Bioinspired
Supramolecular Structures," Nano Letters, 5(7), pp. 1343-1346.
[12] Athamneh, A. I., and Barone, J. R., 2010, "Multiscale peptide self-assembly."
[13] Geddes, A. J., Parker, K. D., Atkins, E. D., and Beighton, E., 1968, "“Cross-β”
conformation in proteins," J. Mol. Biol., 32, pp. 343-358.
[14] Parker, K. D., and Rudall, K. M., 1957, "The Silk of the Egg-Stalk of the Green Lace-
Wing Fly: Structure of the Silk of Chrysopa Egg-stalks," Nature, 179(4566), pp. 905-906.
Page 133
119
[15] Olsen, A., Wick, M. J., Morgelin, M., and Bjorck, L., 1998, "Curli, Fibrous Surface
Proteins of Escherichia coli, Interact with Major Histocompatibility Complex Class I
Molecules," Infect. Immun., 66(3), pp. 944-949.
[16] Barnhart, M. M., and Chapman, M. R., 2006, "Curli Biogenesis and Function," Annual
Review of Microbiology, 60(1), pp. 131-147.
[17] Elliot, M. A., Karoonuthaisiri, N., Huang, J., Bibb, M. J., Cohen, S. N., Kao, C. M., and
Buttner, M. J., 2003, "The chaplins: a family of hydrophobic cell-surface proteins
involved in aerial mycelium formation in Streptomyces coelicolor," Gene Dev, 17(14),
pp. 1727-1740.
[18] Sullan, R. M. A., Gunari, N., Tanur, A. E., Chan, Y., Dickinson, G. H., Orihuela, B.,
Rittschof, D., and Walker, G. C., 2009, "Nanoscale structures and mechanics of barnacle
cement," Biofouling: The Journal of Bioadhesion and Biofilm Research, 25(3), pp. 263 -
275.
[19] Barlow, D. E., Dickinson, G. H., Orihuela, B., Kulp, J. L., Rittschof, D., and Wahl, K. J.,
2010, "Characterization of the Adhesive Plaque of the Barnacle Balanus amphitrite:
Amyloid-Like Nanofibrils Are a Major Component," Langmuir, 26(9), pp. 6549-6556.
[20] Lamm, M. S., Rajagopal, K., Schneider, J. P., and Pochan, D. J., 2005, "Laminated
Morphology of Nontwisting β-Sheet Fibrils Constructed via Peptide Self-Assembly," J.
Am. Chem. Soc., 127(47), pp. 16692-16700.
[21] Davies, R. P. W., Aggeli, A., Beevers, A. J., Boden, N., Carrick, L. M., Fishwick, C. W.
G., McLeish, T. C. B., Nyrkova, I., and Semenov, A. N., 2006, "Self-assembling β-Sheet
Page 134
120
Tape Forming Peptides," Supramol. Chem., 18(5), pp. 435-443.
[22] Athamneh, A. I., and Barone, J. R., 2009, "Enzyme-mediated self-assembly of highly
ordered structures from disordered proteins," Smart Mater Struct, 18(10), p. 104024.
[23] Kendrew, J. C., Dickerson, R. E., Strandberg, B. E., Hart, R. G., Davies, D. R., Phillips,
D. C., and Shore, V. C., 1960, "Structure of Myoglobin: A Three-Dimensional Fourier
Synthesis at 2 Å Resolution," Nature 185, pp. 422 - 427.
[24] Tinti, A., Foggia, M. D., Taddei, P., Torreggiani, A., Dettin, M., and Fagnano, C., 2008,
"Vibrational study of auto-assembling oligopeptides for biomedical applications,"
Journal of Raman Spectroscopy, 39(2), pp. 250-259.
[25] Zandomeneghi, G., Krebs, M. R. H., McCammon, M. G., and Fandrich, M., 2004, "FTIR
reveals structural differences between native β-sheet proteins and amyloid fibrils,"
Protein Sci, 13(12), pp. 3314-3321.
[26] Kretschmer, C. B., 1957, "Infrared Spectroscopy and Optical Rotatory Dispersion of
Zein, Wheat Gluten and Gliadin," J. Phys. Chem., 61(12), pp. 1627-1631.
[27] Oliver, W. C., and Pharr, G. M., 1992, "An improved technique for determining hardness
and elastic modulus using load and displacement sensing indentation experiments," J.
Mater. Res., 7(6), pp. 1564-1583.
[28] Evans, S. V., and Brayer, G. D., 1988, "Horse heart metmyoglobin. A 2.8-A resolution
three-dimensional structure determination," J. Biol. Chem., 263(9), pp. 4263-4268.
[29] Hopp, T. P., and Woods, K. R., 1981, "Prediction of protein antigenic determinants from
Page 135
121
amino acid sequences," Proc. Nat. Acad. Sci., 78(6), pp. 3824-3828.
[30] Pollastri, G., Przybylski, D., Rost, B., and Baldi, P., 2002, "Improving the prediction of
protein secondary structure in three and eight classes using recurrent neural networks and
profiles," Proteins, 47(2), pp. 228-235.
[31] Garnier, J., Gibrat, J.-F., and Robson, B., 1996, "[32] GOR method for predicting protein
secondary structure from amino acid sequence," Methods in Enzymology, F. D. Russell,
ed., Academic Press, pp. 540-553.
[32] Galzitskaya, O. V., Garbuzynskiy, S. O., and Lobanov, M. Y., 2006, "Prediction of
Amyloidogenic and Disordered Regions in Protein Chains," PLoS Comput Biol, 2(12), p.
e177.
[33] Hamodrakas, S. J., Liappa, C., and Iconomidou, V. A., 2007, "Consensus prediction of
amyloidogenic determinants in amyloid fibril-forming proteins," Int. J. Biol. Macromol.,
41(3), pp. 295-300.
[34] Zhang, Z., Chen, H., and Lai, L., 2007, "Identification of amyloid fibril-forming
segments based on structure and residue-based statistical potential," Bioinformatics,
23(17), pp. 2218-2225.
[35] Lopez de la Paz, M., and Serrano, L., 2004, "Sequence determinants of amyloid fibril
formation," Proc. Natl Acad. Sci. USA, 101, pp. 87-92.
[36] Maurer-Stroh, S., Debulpaep, M., Kuemmerer, N., de la Paz, M. L., Martins, I. C.,
Reumers, J., Morris, K. L., Copland, A., Serpell, L., Serrano, L., Schymkowitz, J. W. H.,
Page 136
122
and Rousseau, F., 2010, "Exploring the sequence determinants of amyloid structure using
position-specific scoring matrices," Nature Methods, 7(3), pp. 237-242.
[37] Marshall, K. E., and Serpell, L. C., 2009, "Structural integrity of β-sheet assembly,"
Biochem Soc T, 037(4), pp. 671-676.
[38] Knowles, T. P., Fitzpatrick, A. W., Meehan, S., Mott, H. R., Vendruscolo, M., Dobson,
C. M., and Welland, M. E., 2007, "Role of Intermolecular Forces in Defining Material
Properties of Protein Nanofibrils," Science, 318(5858), pp. 1900-1903.
[39] Antoniou, A., and Lu, W.-S., 2007, Practical Optimization: Algorithms and Engineering
Applications, Springer, New york.
Page 137
CHAPTER SIX
CONCLUSIONS
The overall goal of this work was to enhance our understanding of protein-based structural
materials and the way they assemble in nature. The work was divided into two parts focusing on
protein hydration and self-assembly. Part one was devoted to discussing the use of agricultural
proteins as a feedstock for bio-based polymers. Particularly, it focused on the effect of hydrogen
bonding, or lack thereof, between proteins as mediated by hydration or plasticization. The effect
of varying plasticizer (glycerol) levels on mechanical properties of a series of proteins was
discussed in the context of primary and secondary structure of proteins. We found that the extent
to which a protein can be plasticized was dependent on its primary structure and not simply
molecular weight, as it was often assumed in previous studies. In fact, the effect of plasticizer
was complex. It increased order (relative β-sheet content) at low concentration by allowing
protein chains to align, but the addition of low molecular weight plasticizer counteracted the
ordering to reduce elastic modulus. At higher plasticizer concentrations, an overall drop in elastic
modulus was observed because protein-protein hydrogen bonding was progressively taken away
in favor of protein-solvent interactions. The sensitivity to plasticization was higher for proteins
with more polar amino acids, especially those with carbonyl groups, i.e, glutamine, asparagine,
glutamic acid, and aspartic acid. Conversely, the presence of cysteine made the protein less
sensitive to plasticization due to formation of a robust cystine cross-linked network.
Page 138
124
The second part of the dissertation focused on the study of protein self-assembly as a way to
make useful macroscopic materials. There are major efforts to study protein self-assembly for
various medical and industrial reasons. Despite huge progress, most studies have focused on
nanoscale self-assembly but the crossover to the macroscopic scale remains a challenge. In this
work, we have shown that multiscale peptide self-assembly was possible in vitro under
physiological conditions. We were able to assemble macroscopic fibers from a combination of α-
and β- peptides. Much of the work was focused on characterizing the fibers and trying to
understand the mechanism by which they form. The fibers had cross-β structural arrangement
and their formation was a cooperative process involving α-helix-rich peptide and a template
formed by hydrophobic β-peptides. A review of the literature suggested that our identified self-
assembly mechanism might be a route utilized by several very different natural systems. Finally,
We demonstrated the ability to control macroscopic properties by varying the ratio of constituent
peptides and to conserve properties from the molecular to the macroscopic scale.
Owing to the richness of the amino acid building blocks, peptides are highly versatile structural
and functional building blocks. The ability to extend and control peptide self-assembly over
multiple length scales is a significant leap toward incorporating peptide materials into dynamic
systems of higher complexity and functionality. Future work may address the problem of
function
Page 140
126
A.1 MODEL FITTING AND OPTIMIZATION
New MATLAB program: Optimization_withODE.m
% Optimization.m % Adopted and modified by Ahmad Athamneh, 7/5/2010, form Bioprocess % Engineering course by Dr. Ryan S. Senger % This program performs a steepest ascent algorithm on % Reaction_ODE_solver.m % This program runs the optimization algorithm for a specified % number of iterations % The design vector is [k1; k2; ; kd ; kf ; kd] % The objective function is the mse between simulations and raw data % points clear; %Initiating data matrix used to calculate MSE % Experimental data. % Time column data(:,1)=[1;3;4;8;21;25;26;50;70]; % Normalized alpha helix content data estimated from FTIR data(:,2)=[1;0.851023657;0.966342115;0.777099518;0.637667709;0.660813006;0.712423896;0.673429712;0.496456602]; % Normalized beta sheet content data estimated from FTIR data(:,3)=[0.560689588;0.443489868;0.541068341;0.635500829;0.946547648;0.990779616;0.878435593;1.041674017;1]; %LFB 50% % initiate the design vector and kinetic constants % the design vector is [k1; k2; kd; kf; k3] design_vector=[.08 ; 0.132 ; 0.132 ; 0.132 ; 0.132 ]; delta=0.1; %constant used to approximate partial derivatives omega=0.0001; %step-size in the optimization program number_of_iterations=10000; %number of times to recalculate design vector %Initial conditions Gd0=1; %initial Gd20 concentration Tm0=0; %initial template concentration a0=1; %initial alpha species concentration aTm0=0; %initial alpha-beta complex (intermediate) concentration bd0=0; %initial nonfibrillar aggregates concentration f0=0; %initial fibrillar aggregates concentration
Page 141
127
%initiate the gradient vector delf=zeros(size(design_vector,1),1); for z=1:number_of_iterations k10=design_vector(1,z); k20=design_vector(2,z); kd0=design_vector(3,z); kf0=design_vector(4,z); k30=design_vector(5,z); %running simulations to calculate finite differences k1=k10; k2=k20; kd=kd0; kf=kf0; k3=k30; Reaction_ODE_solver % this program solves ODEs for each set of
% kinetic constants and returns MSE mse1=mean_square_error; k1=k10+delta*k10; k2=k20; kd=kd0; kf=kf0; k3=k30; Reaction_ODE_solver mse2=mean_square_error; k1=k10-delta*k10; k2=k20; kd=kd0; kf=kf0; k3=k30; Reaction_ODE_solver mse3=mean_square_error; k1=k10; k2=k20+delta*k20; kd=kd0; kf=kf0; k3=k30; Reaction_ODE_solver mse4=mean_square_error; k1=k10; k2=k20-delta*k20; kd=kd0; kf=kf0; k3=k30;
Page 142
128
Reaction_ODE_solver mse5=mean_square_error; k1=k10; k2=k20; kd=kd0+delta*kd0; kf=kf0; k3=k30; Reaction_ODE_solver mse6=mean_square_error; k1=k10; k2=k20; kd=kd0-delta*kd0; kf=kf0; k3=k30; Reaction_ODE_solver mse7=mean_square_error; k1=k10; k2=k20; kd=kd0; kf=kf0+delta*kf0; k3=k30; Reaction_ODE_solver mse8=mean_square_error; k1=k10; k2=k20; kd=kd0; kf=kf0-delta*kf0; k3=k30; Reaction_ODE_solver mse9=mean_square_error; k1=k10; k2=k20; kd=kd0; kf=kf0; k3=k30+delta*k30; Reaction_ODE_solver mse10=mean_square_error; k1=k10; k2=k20; kd=kd0; kf=kf0; k3=k30-delta*k30; Reaction_ODE_solver mse11=mean_square_error; %calculating del f delf(1,z)=(mse2-mse3)./(2.*delta.*k10); delf(2,z)=(mse4-mse5)./(2.*delta.*k20); delf(3,z)=(mse6-mse7)./(2.*delta.*kd0);
Page 143
129
delf(4,z)=(mse8-mse9)./(2.*delta.*kf0); delf(5,z)=(mse10-mse11)./(2.*delta.*k30); error_total(1,z)=mse1; %printing results to the command window fprintf(['iteration ' num2str(z) ' of ' num2str(number_of_iterations) '.The mse value is: ' num2str(error_total(1,z)) '.' '\n']); %determining if the mse value is increasing (program is stopped if it %increases) if z>1 if error_total(1,z)>=error_total(1,z-1) fprintf(['The mse is increasing or identical to the last iteration. Optimization is finished.' '\n']); break end end %calculating the new design vector design_vector(:,z+1)=design_vector(:,z)-omega*delf(:,z); end %print the final results fprintf(['The final value of k1 is: ' num2str(design_vector(1,size(design_vector,2))) '\n']); fprintf(['The final value of K2 is: ' num2str(design_vector(2,size(design_vector,2))) '\n']); fprintf(['The final value of Kd is: ' num2str(design_vector(3,size(design_vector,2))) '\n']); fprintf(['The final value of Kf is: ' num2str(design_vector(4,size(design_vector,2))) '\n']); fprintf(['The final value of K3 is: ' num2str(design_vector(5,size(design_vector,2))) '\n']); %plotting the raw data points as circles subplot(2,1,1) plot(data(:,1),data(:,2),'o'); xlabel(['time']); ylabel(['[alpha-helix]']); hold on subplot(2,1,2) plot(data(:,1),data(:,3),'o'); xlabel(['time']); ylabel(['Beta Indicator']); hold on %plotting final simulation results (blue line) subplot(2,1,1) plot(t,a); subplot(2,1,2) plot(t,betaindicator) %re-calculate and plot the simulation results of the initial guess k1=design_vector(1,1); k2=design_vector(2,1); kd=design_vector(3,1); kf=design_vector(4,1);
Page 144
130
k3=design_vector(5,1); Reaction_ODE_solver subplot(2,1,1) plot(t,a,'r'); legend('experimental','Optimized simulation','initial simulation'); subplot(2,1,2) plot(t,betaindicator, 'r'); legend('experimental','Optimized simulation','initial simulation'); fprintf(['done!' '\n']); end;
New MATLAB program: Reaction_ODE_solver.m
% Reaction_ODE_solver.m % The built-in MATLAB 'ODE45' function was used to solve the % reaction ODEs. % The M file equations.m evaluates the right-hand-side expresion of & the ODE. [t,y]=ode45(@(t,y) equations(t,y,k1,k2,kd,kf,k3),[0 75],[Gd0 Tm0 a0 aTm0 bd0 f0]); Gd=y(:,1); Tm=y(:,2); a=y(:,3); aTm=y(:,4); bd=y(:,5); f=y(:,6); beta_indicator=Tm+f+bd+aTm+0.560689588; %calculating the mse data(:,4)=0; %this column will be the simulation value of a data(:,5)=0; %this column will be the simulation value of b data(:,6)=0; %this column will be the delta a data(:,7)=0; %this column will be the delta b data(:,8)=0; %this column will be delta a + delta b for i=1:size(data,1) target=data(i,1); id=0; for j=1:size(t,1) if target>=t(j,1) id=j; end end if id>0
Page 145
131
data(i,4)=a(id,1); data(i,5)=beta_indicator(id,1); else fprintf(['simulation data points do not exist for all raw data points' '\n']); Return end end data(:,6)=(data(:,2)-data(:,4)).^2; data(:,7)=(data(:,3)-data(:,5)).^2; data(:,8)=data(:,6)+data(:,7); mean_square_error=sum(data(:,8)./size(data,1)); end;
New MATLAB program: Equations.m
% equations.m % This function evaluates the right-hand-side expression of ODEs 7- % 12 on page 112 on dissertation. function dydt=equations(t,y,k1,k2,kd,kf,k3) dydt=zeros(size(y)); Gd20=y(1); Tm=y(2); a=y(3); aTm=y(4); bd=y(5); f=y(6); dydt(1)=-k1*Gd20*Gd20; dydt(2)=k1*Gd20*Gd20-k2*Tm*a+k3*aTm-kf*aTm*Tm-kd*Tm*Tm; dydt(3)=-k2*a*Tm+k3*aTm; dydt(4)=k2*Tm*a-k3*aTm-kf*aTm*Tm; dydt(5)=kd*Tm*Tm; dydt(6)=kf*Tm*aTm; end;
Page 146
132
A.2 SIMULATION: FORMATION OF FIBRILLAR AGGREGATES AS A FUNCTION OF
GLIADIN LEVEL.
New MATLAB program: simulation.m
% simulation.m % written by Ahmad Athamneh, 7/28/2010 % This program simulates the formation of fibrillar aggregates as a % function of Gliadin level clear; %Initial conditions Tm0=0; %initial template concentration aTm0=0; %initial alpha-beta complex (intermediate) concentration bd0=0; %initial nonfibrillar aggregates concentration f0=0; %initial fibrillar aggregates concentration %Rate constant from Model fitting k1=0.2665; k2=0.17219; k3=.074138; kd=0.13044; kf=0.0010165; %time span, hrs time=75; for z=0:100; Gd0=z*2/100; % gliadin level varies from 0 to 2 (0 to 100%) a0=2-Gd0; % alpha species varies from 2 to 0 (100 to 0%) simulation_ODE_solver % This program runs the simulation for each % gliadin level %calculating the new design vector and store results. result(z+1,1)=Gd0; %Bf after 75 hours result(z+1,2)=final_f; %bd after 75 hours result(z+1,3)=final_bd; %bd after 75 hours result(z+1,4)=final_Tm; end
Page 147
133
subplot(3,1,1); plot (result(:,1),result(:,2)); legend(' fibrillar aggregates '); subplot(3,1,2); plot (result(:,1),result(:,3)); legend(' nonfibrillar aggregates '); subplot(3,1,3); plot (result(:,1),result(:,4)); legend(' template '); end;
New MATLAB program: simulation_ODE_solver.m
% simulation_ODE_solver.m % The built-in MATLAB 'ODE45' function was used to solve the % reaction ODEs. % The M file equations.m evaluates the right-hand-side expresion of & the ODE. [t,y]=ode45(@(t,y) equations(t,y,k1,k2,kd,kf,k3),[0 time],[Gd0 Tm0 a0 aTm0 bd0 f0]); Gd=y(:,1); Tm=y(:,2); a=y(:,3); aTm=y(:,4); bd=y(:,5); f=y(:,6); final_f=f(size(f,1),1); final_Tm=Tm(size(Tm,1),1); final_bd=bd(size(bd,1),1);