University of Southern California Dissertations and Theses

Control the function, dynamics, aggregation of proteins with light illumination

(USC Thesis Other)

Control the function, dynamics, aggregation of proteins with light illumination

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
Control the Function, Dynamics, Aggregation of
Proteins with Light Illumination

by

Yimin Wang

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMICAL ENGINEERING)

May 2021

Copyright 2021 Yimin Wang

ii
Acknowledgements
Firstly, I would like to express my sincere gratitude to my PhD advisor, Dr. Lee. Without
his generous support and wise guidance for my PhD study and related research, I probably would
not be able to complete my work within 6 years. Besides immense professional knowledge in
chemical engineering field, I also learned a lot from Dr. Lee, including but not limited to rigorous
scholarship, critical thinking method, open-mind altitude, comprehensive personality. He can
always provide deep insights, helpful advice, creative views and continuous encouragement to me
when I encountered research difficulties not only in work but also in life. I’m really appreciated
for his generous help during my PhD pursuing time.
Secondly, I would like to thank Dr. Shing. Dr. Shing is a profound professor in our
department. She is always so rigorous and strict about research. She does not only provide me lots
of inspiring advice for my research, but also lots of consideration for my life. Her persistent and
integrity personality has always encouraged me to overcome difficulties during my PhD pursuing
time. I’m appreciated for her considerable help.
Thirdly, I would like to thank Dr. Nakano. Dr. Nakano is a knowledgeable professor in
many fields. He teaches me lots of knowledge about quantum chemistry and molecular modeling
in some other communication opportunities besides research. Dr. Nakano is always so warm and
generous when I’m seeking help from him. Dr. Nakano helps to broaden my eyes during my PhD
pursuing time. I’m appreciated for his enthusiastic help.
Besides, I would like to thank Zumra Seidel, who is my research partner. Zumra is always
so reachable to me when I’m seeking help. We comprehend, encourage and support each other
during our research work. We have spent a very nice and sweet time working together. I’m
appreciated for her accompany.
iii
Next, I would like to thank my families, especially my husband, Meng Wang, who have
always been so supportive to me during the six-year PhD pursuing time. And my parents and my
parents-in-law are always so comprehensive and loving me selflessly. They provide me such a
warm and love harbor so that I can always feel relax and happy when I return home. Without their
help, I may not be able to focus on my work and study. I’m appreciated for their love.
In the end, I want to thank my dear friends, Xian An, LinShan Huang, Ying Zhang, XiXi
Zhang, Wan Su and many others. They always encourage, comfort and support me whenever I
encounter obstacles. I also want to thank Mr. Li Jian, whose music and talent can always bring me
bright and hope, and Mr. Guo Degang, whose jokes can always provide comfort and joy to me
when I was upset.

iv
Table of Contents
Acknowledgements .................................................................................................. ii
List of Tables ........................................................................................................... vi
List of Figures ......................................................................................................... vii
Abstract .................................................................................................................... xi
1 Introduction .........................................................................................................1
1.1 Surfactant ..................................................................................................................................... 1
1.2 Proteins ........................................................................................................................................ 2
1.2.1 Structure and function and dynamic ....................................................................................................... 2
1.2.2 Catalase .................................................................................................................................................. 4
1.2.3 Ribonuclease A ...................................................................................................................................... 6
1.2.4 Insulin ..................................................................................................................................................... 7
1.3 Experiment methods and techniques ........................................................................................... 8
1.3.1 Dynamic light scattering ........................................................................................................................ 8
1.3.2 SANS ...................................................................................................................................................... 9
1.3.3 NSE ...................................................................................................................................................... 10
2 Enhanced Catalase Activity through Tetramer Dissociation in the Presence of a
Photoresponsive Surfactant .....................................................................................13
2.1 ABSTRACT ............................................................................................................................... 13
2.2 INTRODUCTION ..................................................................................................................... 13
2.3 EXPERIMENTAL METHODS ................................................................................................ 16
2.4 RESULTS AND DISCUSSION ................................................................................................ 19
3 Chapter 3: Ribonuclease-A and NSE ...............................................................32
3.1 Abstract ...................................................................................................................................... 32
3.2 Introduction ................................................................................................................................ 33
3.3 Experiments and methods .......................................................................................................... 34
3.3.1 Materials ............................................................................................................................................... 34
3.3.2 Neutron Spin Echo Spectroscopy ......................................................................................................... 35
3.3.3 Rigid-body model ................................................................................................................................. 36
3.3.4 Soft-linker model .................................................................................................................................. 37
3.3.5 Freely-jointed model ............................................................................................................................ 39
3.3.6 Kirkwood formula ................................................................................................................................ 41
v
3.4 Results and discussions .............................................................................................................. 42
3.4.1 DLS and SANS results ......................................................................................................................... 42
3.4.2 Kirkwood results .................................................................................................................................. 48
3.4.3 NSE results compared with results from rigid-body model and Kirkwood formula ........................... 49
3.4.4 Soft-linker model results ...................................................................................................................... 54
3.4.5 Freely-jointed model results ................................................................................................................. 56
3.5 Conclusions ................................................................................................................................ 58
4 Reveal Nanoscale Protein Motion of Bovine Serum Albumin by Neutron Spin
Echo Spectroscopy ..................................................................................................61
4.1 Introduction ................................................................................................................................ 61
4.2 Materials and methods ............................................................................................................... 63
4.2.1 Materials ............................................................................................................................................... 63
4.2.2 Light-Scattering Measurements ........................................................................................................... 64
4.2.3 Neutron spin echo spectroscopy ........................................................................................................... 64
4.3 Results and discussion ............................................................................................................... 66
4.4 Conclusion ................................................................................................................................. 70
5 Controlling Insulin Fibrillation with Light Using Photoresponsive Surfactant 71
5.1 Abstract ...................................................................................................................................... 71
5.2 Introduction ................................................................................................................................ 72
5.3 Materials and experiment methods ............................................................................................ 75
5.3.1 Materials ............................................................................................................................................... 75
5.3.2 Dynamic light scattering ...................................................................................................................... 76
5.4 Results and discussion ............................................................................................................... 77
5.5 Conclusion ................................................................................................................................. 84
6 Planned work ....................................................................................................85
6.1 Protein aggregation .................................................................................................................... 85
6.1.1 Insulin ................................................................................................................................................... 85
6.1.1.1 Introduction ................................................................................................................................. 85
6.1.1.2 Preliminary data .......................................................................................................................... 85
6.1.1.3 Planned work .............................................................................................................................. 87
References ...............................................................................................................89

vi
List of Tables
Table 2-1: Molecular weight and radius of gyration determination using Guinier and PDDF
analysis for catalase (2 mg/mL) as a function of azoTAB concentration under visible light. ...... 23
Table 3-1: Values of the radius of gyration (Rg) and I(0) determined from Guinier and PDDF
analyses of the SANS data, as well as hydrodynamic radii (RH) calculated from the shape
reconstructions using Kirkwood’s theory. .................................................................................... 43

vii
List of Figures
Figure 1-1:Structures of two isomers of azoTAB. .......................................................................... 1
Figure 1-2: UV-Vis absorption profile for the trans and cis azoTAB isomers. (1mM, 1mm
pathlength) ...................................................................................................................................... 2
Figure 1-3: Ribbon diagram structure of tetramer of bovine liver catalase. [PDB code: 8CAT] ... 5
Figure 1-4: Ribbon diagram structure of ribonuclease A. [PDB: 5rsa] .......................................... 7
Figure 1-5: Ribbon diagram structure of insulin. [PDB: 1ZNI] ..................................................... 8
Figure 2-1: Effect of visible- or UV-light adapted azoTAB on (a) the relative specific activity of
catalase at [H2O2] = 19.3 mM and (b) the initial velocity versus substrate concentration without
and with 1 mM azoTAB. Catalase activity in the presence of the conventional surfactants SDS
and DTAB is also shown in (a). The three data points at 19.3 mM in part (b) are repeated from (a)
for comparison. Error bars are typically within the symbols. ...................................................... 20
Figure 2-2: (a) SANS scattering data, (b) Guinier plots, and (c) calculated pair distance distribution
functions for solutions of catalase (2 mg/mL) mixed with various concentrations of visible- or UV-
light adapted azoTAB (closed circles and open squares, respectively). The same color scheme is
used throughout. Each successive spectra is offset by ´10 for clarity with the pure catalase
spectrum overlaid (small black circles). The experimental P(r) functions in (c) are compared to
those calculated from the crystallographic tetramer (PDB: 1TGU) and two potential dimers. .... 22
Figure 2-3: (a,b) X-ray crystallographic structures of the catalase tetramer and two possible dimers,
and (c) SANS-based solution structures of catalase as a function of azoTAB concentration under
visible light. Right-viewed images are reduced by half throughout for display purposes. .......... 27
Figure 2-4 Circular dichroism spectra of catalase as a function of azoTAB concentration and
illumination conditions. Inset: Difference spectra at each condition. .......................................... 29
viii
Figure 3-1: Schematic diagram of three models for protein internal motions: (a) a rigid-body model;
(b) a soft-linker model; and (c) a freely-jointed model. ................................................................ 36
Figure 3-2: The hydrodynamic radius of RNase A measured by DLS. [RNase A] = 0.93 mg/mL.
....................................................................................................................................................... 43
Figure 3-3. SANS data of RNase A-azoTAB solutions as a function of surfactant concentration
and isomeric state. (a) [RNase A] = 0.9 mg/mL (b) [RNase A] = 10 mg/mL. ............................. 45
Figure 3-4. Pair-distance distribution functions of RNase A-azoTAB solutions as a function of
surfactant concentration and light conditions. (a) [RNase A] = 0.9 mg/mL (b) [RNase A] = 10
mg/mL. .......................................................................................................................................... 46
Figure 3-5. In vitro conformations of RNase A in the presence of various azoTAB concentrations
in the trans (visible light) or cis (UV light) isomeric state, obtained from shape reconstruction
analysis of the SANS data (best fit shown in blue, consensus envelopes representing the average
of 10 runs shown in red), compared with the X-ray crystallographic structure of RNase A (PDB
1RBX).
112
[RNase A] = 10 mg/mL. .............................................................................................. 47
Figure 3-6. The hydrodynamic radius of ribonuclease A as a function of azoTAB concentrations
and light conditions. In the plot, the blue symbols (l) represents the hydrodynamic radius of pure
ribonuclease A, the black symbols (n) represents the hydrodynamic radius of ribonuclease A in
the presence of trans azoTAB, and the red symbols (u) represents the hydrodynamic radius of
ribonuclease A in the presence of cis azoTAB. [RNase A] = 10 mg/mL. .................................... 49
Figure 3-7. NSE data of RNase A effective diffusion coefficient with different azoTAB
concentrations and light conditions. In each plot, the black symbols (n) represent data obtained
from NSE spectroscopy, the blue symbols ( ▲) represent the diffusion coefficient calculated by
Kirkwood formula, and the solid red lines (—) represent rigid body calculations. ...................... 52
ix
Figure 3-8. comparison between the measured effective diffusion coefficients using NSE
spectroscopy and the estimated results based on soft linker model. In each plot, the black symbols
(n) represent data obtained from NSE spectroscopy, and the solid red lines (—) represent soft
linker calculations. ........................................................................................................................ 55
Figure 3-9: comparison between the measured effective diffusion coefficients using NSE
spectroscopy and the estimated results based on freely jointed model. In each plot, the black
symbols (n) represent data obtained from NSE spectroscopy, and the solid red lines (—) represent
freely jointed calculations. ............................................................................................................ 58
Figure 3-10: Ribbon diagram of the three-dimensional structure of ribonuclease A [PDB : 3rn3].
The inscriptions refer to the location of the eight cysteine residues that give rise to the four disulfide
bonds.
17
........................................................................................................................................ 60
Figure 4-1: (A)(C)(E) the shape reconstruction results from SANS of pure BSA, BSA in presence
of 2mM trans azoTAB and BSA in presence of 2mM cis azoTAB. (B)(D)(F) the effective diffusion
coefficients measured though NSE spectroscopy, which is represented by black square (n), is
compared with the effective diffusion coefficient measured by DLS, which is represented by a blue
diamond (¨), and the calculated results for rigid-body model, which is represented by a red solid
line (—). The estimated diffusion coefficients based on Kirkwood theory for different protein
configurations are also shown for illustration purpose. ................................................................ 68
Figure 5-1: structure and photoisomerization of azoTAB. ........................................................... 75
Figure 5-2: Time-dependent measurements of (a) the apparent hydrodynamic diameter determined
from DLS using the method of cumulants; (b) the scattering intensity counting rate (locally
averaged over a time span of ~1 min); and (c) the absorbance at 600 nm determined from UV-vis
spectroscopy measurements. Data were collected for pure insulin and insulin in the presence of
x
0.5 mM azoTAB and 1 mM azoTAB under visible light. [insulin] = 10 mg/mL, T = 60 °C, pH =
1.6.................................................................................................................................................. 78
Figure 5-3: df number of growth of fibrils of (a) pure insulin; (b) insulin with 0.5mM azoTAB in
the presence of visible light; (c) insulin with 1mM azoTAB in the presence of visible light. [insulin]
= 10mg/ml, T = 60 °C, pH = 1.6. .................................................................................................. 81
Figure 5-4: time-dependent measurements of (a) the apparent hydrodynamic diameter determined
from DLS using the method of cumulants, and (b) the average counting rate determined from DLS.
Data were collected for pure insulin and insulin in the presence of 0.5mM and 1mM azoTAB under
visible light. [insulin] = 10mg/ml, T = 50 °C, pH = 1.7. .............................................................. 82
Figure 5-5: df number of growth of fibrils of (a) pure insulin; (b) insulin with 0.5mM azoTAB in
the presence of visible light; (c) insulin with 1mM azoTAB in the presence of visible light. [insulin]
= 10mg/ml, T = 50 °C, pH = 1.6 ................................................................................................... 83

xi
Abstract
AzoTAB, which is a light sensitive surfactant, undergoes reversible photoisomerization
upon exposure to appropriate wavelength of light. When shining visible light onto it, azoTAB will
be in trans form, which is more hydrophobic, thus inducing a bigger degree of protein unfolding
than cis form, which is more hydrophilic under UV light. Because of such special and unique
character, azoTAB has been investigated as a great tool to photo control protein structure,
dynamics and function reversibly. For example, superactivity of catalase has been observed after
interacting with azoTAB under visible light, because catalase, which is a tetramer, undergoes a
slight structure change with trans isomer. What’s more, protein dynamic can also be photo
controlled after interacting with azoTAB. Ribnuclease A, which has expressed big difference in
translational and inner diffusion coefficient, has different dynamic performance after interacting
with azoTAB under different wavelength of light. Three different models, rigid body, soft-linker,
and freely jointed, have been employed to illustrate protein dynamic change. Besides, use of
azoTAB as a means to photo control protein aggregation has also been explored in inhibiting
amyloid fibrillation of insulin. The fibrillation of insulin is enhanced after interacting with azoTAB
in trans form, while aggregation of insulin is dramatically inhibited after interacting with azoTAB
in cis form. The early fibrillation stage of insulin has been studied in the presence of both trans
and cis azoTAB. A number of experimental techniques are used to determine the solution structure
of proteins, including dynamic light scattering, circular dichroism, Small Angle Neutron Scattering,
and Neutron Spin Echo.

1
1 Introduction
1.1 Surfactant
Surfactants are molecules that are consisted of a hydrophilic head group, which can be
ionic, zwitterionic or polar non-ionic, and a hydrophobic tail group, which can be hydrocarbon
chains. Surfactants are commonly used in biological system, including studying the protein
structures, dynamics, and functions, such as SDS and DTAB. AzoTAB (azobenzene-
trimethylammonium bromide) is a cationic photo responsive surfactant. The hydrophilic part of
azoTAB is a trimethylammonium head group, while the hydrophobic tail is composed of a photo
responsive azobenzene group and a spacer alkyl group. The light responsive surfactant undergoes
a photoisomerization from the relatively hydrophobic transisomer under visible light to the
relatively hydrophilic cis isomer upon UV illumination. AzoTAB can be synthetized according to
published procedures.
1,2

Figure 1-1:Structures of two isomers of azoTAB.

The dipole moment across the nitrogen double bond of the surfactant is significantly lower
in the planar transform, which is 0.5 Da, than the bent cis conformation, which is 3.1 Da.
2
Because
of higher dipole moment, the cis isomer is energetically less stable. azoTAB exhibits the maximum
absorption band of the trans form at 350 nm, while the maximum absorption band of the cis form
8
1.2 Photoresponsive Surfactants
Surfactants are molecules consisting of a hydrophilic head group and a
hydrophobic tail group. The head group may be ionic, zwitterionic or polar non-ionic
while the tail group generally consists of hydrocarbon chains. SDS and DTAB are
commonly used examples to study biological systems. AzoTAB (azobenzene-
trimethylammonium bromide), an analogous form of DTAB, is a cationic
photoresponsive surfactant that consists of a trimethylammonium head group providing
the hydrophilic part, a photoresponsive azobenzene group, a hydrophobic tail composed
of a spacer alkyl group between the head part and azobenzene group, and an alkyl tail
group providing the hydrophobic part of the molecule. The lightresponsive surfactant
undergoes a photoisomerization from the relatively hydrophobic trans isomer under
visible light to the relatively hydrophilic cis isomer upon UV illumination.

Figure 1.2: Chemical structure and photoisomerization of azoTAB surfactant.
Visible light
UV light
UV

Visible
2
at 415 nm. Trans isomers can be converted to cis isomers when illuminated by UV light, and cis
can be converted back to trans by illuminating visible light, high heat, or staying in black for 24
hours. Because of such special character, azoTAB can be utilized to study a variety system
including biological systems, surface tension, microemulsions, and aggregation properties.
The photoisomerization of the light-sensitive surfactant can be followed by spectral change
using UV-visible spectroscopy as shown in Figure 1-2. As it seen in Figure 1-2, the maximum
absorption band of the trans form occurs at 350 nm, while the cis isomer exhibits a characteristic
absorption band at 434 nm. Photoresponsive surfactants are utilized to study a variety system
including biological systems, surface tension, microemulsions, gels, aggregation properties, etc.
3–
5

Figure 1-2: UV-Vis absorption profile for the trans and cis azoTAB isomers. (1mM, 1mm
pathlength)
1.2 Proteins
1.2.1 Structure and function and dynamic
Proteins are organic macromolecules which exist everywhere in nature. They are
responsible for nearly all the functions in cells, including providing mechanical support, giving
9
The dipole moment across the nitrogen double bond of the surfactant is
significantly lower in the planar trans form (0.5 Da) relative to the bent cis conformation
(3.1 Da).(39) The cis isomer is energetically less stable. The photoisomerization of the
light-sensitive surfactant can be followed by spectral change using UV-visible
spectroscopy as shown in Figure 1.3. As it seen in Figure 1.3, the maximum absorption
band of the trans form occurs at 350 nm, while the cis isomer exhibits a characteristic
absorption band at 434 nm. Photoresponsive surfactants are utilized to study a variety
system including biological systems, surface tension, microemulsions, gels, aggregation
properties, etc.(35, 40-42)

Figure 1.3: UV-visible absorption spectra of azoTAB surfactant in 1 mm pathlength
cuvette cell, [azoTAB] = 1 mM.

3
immune protection, generating movement, transmitting nerve impulses, and controlling growth
and differentiation. Proteins formed by specific arrangements of amino acids are varied in size,
shape, charge, hydrogen bonding capacity, hydrophobic character and chemical reactivity,
therefore they are useful in many applications including therapeutic agents, catalysts, and materials.
Thus, understanding the protein structure-function relationship is very important in both academic
and industry areas.
There are four levels of protein structures, primary structure, secondary structure, tertiary
structure and quaternary structure. Through the translation process catalyzed by ribosome, a long
chain polymer of amino acid residues is produced by condensation reactions that repeatedly link
two amino acid residues by a peptide bond to form the main chain/backbone of a protein. The
primary structure of the protein is defined by the resulted specific sequential arrangement of the
amino acid residues.
Under suitable conditions, such as solvent, temperature, the one dimensional amino acid
sequence folds into a three dimensional structure, which is the native state, and the free energy is
lowest. Depending on the different degrees of polarity and water affinity to different amino acids,
hydrophilic amino acids, in most cases, are more likely to be in the core while hydrophobic amino
acids are on the surface of the protein. Through such kind of process, the strain on the relatively
rigid amide linkages is minimized, and the hydrogen-bonding potential of the main is satisfied.
This process forms the secondary structures, among which the earliest discovered ones are α-
helices and β-sheets.
6

α-helices have a spring-like structure formed by a set of continuous residues in a amino
acid sequence. β-sheets are composed of several independent sets of amino acid residue. Each
consecutive set (typically 5−10 residues) is a local structure, β-strands. Through connecting β-
4
strands with hydrogen bonds layer by layer, β-sheets are formed. Since then, other secondary
structures such as loops, which connecting helices and strands and other forms of helices were
discovered.
These secondary structures will arrange and form the final shape of the protein, or the
tertiary structure of the protein. There are various interaction forces that can facilitate this process,
for example, hydrophobic interactions inside the protein, disulfide bonds, hydrogen bonding, and
ionic interaction. Some proteins exist as oligomers which are composed of several identical
subunits or a set of different subunits. The three-dimensional organization of these subunits is the
quaternary structure of the protein.
These folding events leads to the formation of the native (or active) state of proteins, which
are essential for biological function. Take enzymes as an example. In the native state, an enzyme
folds to form an active site where substrates can be selectively recognized and efficiently bounded
and successfully reacted.
However, proteins are not static and regularly undergo different types of conformational
changes during reactions. Resulted from continuous conformational changes, protein motions can
lead to various kinds of protein dynamics, including subpicosecond atomic vibrations, pico- to
nanosecond backbone and side-chain fluctuations, and milli- to second conformational
rearrangement and breathing motions.
7
These protein motions are functionally important even
though mechanisms of protein function are not fully understood.

1.2.2 Catalase
Catalase is a heme containing enzyme found in the peroxisomes of eukaryotic cells that
protects the cell from the toxic effects of hydrogen peroxide by catalyzing H2O2 decomposition to
5
oxygen and water. It is a vital enzyme occurring in almost all aerobically respiring organisms.
8
In
its native state, the protein exists as a hydrophobically-bound tetramer of identical polypeptide
subunits, each over 500 residues long with a molecular weight of approximately 60 kDa.
9
The
molecule has an isoelectric point of 5.4, resulting in a negative charged at neutral pH. In spite of
the presence of 16 cysteine residues, tetrameric catalase does not form any disulfide bridges.
10
The
heme moiety and active site are buried in a hydrophobic pocket at the center of each monomer
approximately 20 Å below the molecular surface and the center of the tetramer.
11

Figure 1-3: Ribbon diagram structure of tetramer of bovine liver catalase. [PDB code: 8CAT]
Each catalase monomer is a four-domain protein. The first is a globular domain, containing
two α-helical arms involved in making quaternary contacts with other subunits. The second is a
β-barrel, consisting of two anti-parallel four-stranded β-sheets that interact with the heme group.
This corresponds to the most conformationally-rigid region of the polypeptide chain.
12
The third
is a largely helical domain that contributes to the formation of a hydrophobic channel that controls
the accessibility of the active site. The fourth domain is referred to as the wrapping domain, as it
extends around the outside of each subunit. Though the wrapping domain shows little discernible
secondary structure, one segment of this domain forms a short anti-parallel β-sheet with the
18

channel to this portion of the molecule. The third and fourth domains appear to be
involved in maintaining the enzyme’s quaternary structure. A globular domain with
two α-helical arms, and a ‘wrapping’ domain which extends around the outside of the
molecule and is capable of forming β-structured contacts with other monomer units,
stabilize the native tetrameric structure of the protein.[22, 23, 24] Changes in
structure resulting in dissociation of the native oligomeric protein have been shown to
yield dimers (both inactive and super-active forms have been observed) and inactive,
aggregation-prone monomers. [25, 26]

Figure 1.7 Ribbon diagram structure of the native catalase tetramer from
bovine liver. [PDB code 7CAT]

In the presence of azoTAB surfactants, the catalase tetramer dissociates into
smaller species. At low surfactant concentrations, a dimer with enhanced activity is
favored, and at higher surfactant concentrations the equilibrium shifts towards the
monomers, which go on to form aggregates. This effect is more pronounced in the
6
corresponding structural element on another monomer molecule, stabilizing the native quaternary
structure.
8,13

The instability of this oligomeric protein has been studied under a number of different
conditions, such as pH extremes and elevated concentrations of denaturants such as urea, GdmCl,
or SDS surfactant. Native catalase is known to dissociate into a number of observed protein
assemblies in solution, including the ellipical dimer and aggregation prone monomer.
9,14
While the
monomer is considered biochemically inactive, certain forms of the dimer have shown increased
activity relative to the tetramer.
9
We have performed a deep investigation on the effect of
increasing concentrations of azoTAB and changing light wavelength on the function and structure
of bovine liver catalase.

1.2.3 Ribonuclease A
Ribonuclease A (RNase A) is 124-residue, highly stable, monomeric enzyme that catalyzes
the cleavage of single-strand RNA. Ribonuclease A is a small protein, and contains 19 of the 20
natural amino acids, lacking only tryptophan. The molecular formula of the native, uncharged
enzyme is C575H907N171O192S12, corresponding to the molecular weight of ~ 13.7 kDa.
15

Ribonuclease A is an excellent model for protein studies since its amino acid sequence and three-
dimensional structure have been known.
16,17
There are two halves of the secondary structure of the
protein, three short α-helixes which is predominant, and a long anti-parallel β-sheet.
15
The active
site resides in a deep cleft. Overall structure of the molecule composed of 23% α-helix, 46% β-
sheet 21% β-turn and 10% random coil.
18,19
Ribonuclease A contains four disulfide bonds (cys26-
cys84, cys40-cys95, cys65-cys72, and cys58-cys110) which play a key roles in protein folding and
stability.
7

Figure 1-4: Ribbon diagram structure of ribonuclease A. [PDB: 5rsa]
Proteins undergo a variety of dynamical fluctuations that occur over a wide range of time
scales.
20
These time dependent fluctuations between different conformations of protein are
essential for protein folding and function. Their characterization and determination, therefore, is
of great importance. Ribonuclease A is a well-studied enzyme in protein three-dimensional
structures and thus a great model in studying protein dynamics.
1.2.4 Insulin
At neutral pH, the native state of insulin is primarily a toroid-shaped hexamer, but the
active form of the protein is monomer. Each insulin monomer has a molecular weight of ~5800
kDa, and is consisted of two peptide chains which are linked by disulfide bridges. Both peptide
chains contain large α-helixes, and the B chain, which is longer than the other, contains a short β-
sheet as well. Monomer is not the stable form of insulin. The destabilization of the monomer can
lead to self-association and initiation of the fibrillation process.
20,21
Insulin is an ideal model
protein for studying the fibrinogenesis mechanism among other proteins which may also
participate in the fibrillation process, since it is relatively cheap, accessible and well-studied.
8

Figure 1-5: Ribbon diagram structure of insulin. [PDB: 1ZNI]
The primarily α-helical secondary structure of insulin makes it a target for binding of the
azoTAB surfactants. In the presence of the trans isomer, the fibrilization rate is increased
compared with the native protein. In contrast, the cis isomer extends the lag phase and apparently
reduced the rate of fibrils formation. The effects of azoTAB surfactants on the fibrillation process
will be discussed in more details in a later chapter.

1.3 Experiment methods and techniques
1.3.1 Dynamic light scattering
Dynamic light scattering is a technique that can be used to determine the size distribution
profile of small particles in suspensions. When particles in the suspension are smaller compared
to the wavelength of light (usually below 250 nm), the light will scatter in all directions if light
hits on those small particles. Even if the light source in dynamic light scattering equipment is laser,
which is monochromatic and coherent, the intensity of scattering fluctuates over time. The
intensity fluctuation is due to the Brownian motion caused by small particles in the suspension, as
a result of which the distance between scatters in the solution is always changing over time. By
9
recording and autocorrelating the intensity fluctuation during the experiment, the dynamic
information of the particles is obtained. Once the autocorrelation data has been generated, there
are many different mathematical methods to determine the size of particles. The simplest approach
is to treat the autocorrelation function as a single exponential decay,
𝑔(𝑞;𝜏) = exp (−Γ𝜏) 1-1
where Γ is the decay rate, and 𝜏 is the decay time. The translational diffusion coefficient 𝐷
!
may
be derived at a single angle or at a range of angels depending on the wave vector 𝑞 ,
Γ = 𝑞
"
𝐷
!
1-2
with
𝑞 =
#$%
!
&
sin (
'
"
) 1-3
where 𝜆 is the laser wavelength, 𝑛
(
is the refractive index of the sample and 𝜃 is the angle at which
the detector is located with respect to the sample cell. Once the translational diffusion coefficient
is determined from particles dynamic information, the hydrodynamic diameter can be estimated
by using the Stokes-Einstein equation,
𝑅
)
=
*
"
+
, $-.
#
1-4
where 𝑘
/
is Boltzmann’s constant, 𝜂 is the solvent viscosity, and T is the temperature.
1.3.2 SANS
Small angel neutron scattering is an experimental technique that uses elastic neutron
scattering at small angles to investigate the structure of various substances. In a SANS experiment,
a beam of cold neutrons is directed at a sample. The neutrons are elastically scattered and have no
interaction with the electronic cloud of sample atoms as X-ray. Instead, neutrons interact with
nuclei and the interaction depends on the isotope, thus providing information about atomic
positions and sample structures. In light scattering, however, the length scale is determined by the
incident wavelength and scattering angel with the relationship 𝑑 =
𝜆
𝜃
9 , where 𝜆 is the
10
wavelength of light and 𝜃 is the scattering angle. In SANS experiment, a beam of cold and long
wavelength neutrons (~6 Å) and very small angels are employed, so that the structure of samples
with the size range of 60 Å to 1000 Å can be probed.
22

Guinier approximations provide roadmap for SANS data analysis, since it can provide
information about particle composition, shape and size. Generalization allows for analysis of
complex mixtures, allowing indentification of domains where each approximation applies. Guinier
approximation can be obtained from the following equation,
𝐼(𝑄) = 𝐼(0)𝑒
(%&
'
(
)
'
)
+
1-5
where Q is the measurement vector in the inverse space and 𝑄 =
4𝜋
𝜆
9 𝑠𝑖𝑛𝜃. In this expression,
𝜆 is the neutron wavelength and 𝜃 is the scattering angle. 𝐼(𝑄) is the extrapolated intensity at Q =
0, and Rg is the radius of gyration.
23
And the radius of gyration can be determined from the slope
of ln (𝐼(𝑄)) vs 𝑄
"
.
Another analysis technique applied to SANS data is the calculation of a pair distance
distribution function (PDDF). By calculating the probability (P(r)) of finding two scattering centers
separated by a distance r, the maximum dimension (Dmax) within the sample can be determined
with the following equation,
𝐼(𝑄) = 4𝜋∫ 𝑃(𝑟)
012 (56)
56
𝑑𝑟
.
,-.
(
1-6

1.3.3 NSE
Neutron spin echo (NSE), a quasielastic method, has been widely used in areas of studying
the dynamics of macromolecules. Compared with SANS, which investigates the static structures
of macromolecules, neutron spin echo spectroscopy can analyze the inelastic broadening of SANS
intensity and thereby analyzing the motion of macromolecules. Among all the neutron scattering
11
techniques, NSE has the advantage of providing a method to measure dynamics over a range of
both time and length scale. There are three sources of incoherent scattering: isotopic variation,
uncorrelated motions, and variation in the nuclear cross section due to the nuclear spin. In the
classical NSE spectrometer design, it is possible to measure the nuclear incoherent intermediate
structure factor Iincoh(Q,t). In practice, however, there are some difficulties need to be overcome.
One important reason is that the incoherent scattering intensity is low most of the times. Besides,
the nuclear spin-incoherent scattering reduces the polarization of the scattered beam, thus reducing
the echo amplitude. However, if the incoherent signal is strong enough, for example in the aqueous
systems, the incoherent scattering can be measured. With respect to a measurement performed
with a time-of-flight or a backscattering spectrometer, the main benefits are a better energy
resolution and the advantage of working in the time domain. The main drawback is that each Q
value as to be measured separately. In any case, long counting times are to be expected.
The main application of NSE spectroscopy is to measure the intermediate coherent
scattering function Icoh(Q,t), the coherent density fluctuations that correspond to some SANS
intensity pattern. This type of scattering may be orders of magnitude more intense than the
incoherent contributions. However, studies of the incoherent dynamics with NSE are possible and
in some cases have been successfully performed in the past.
In NSE experiment, after data reduction, the normalized intermediate scattering function
𝐼(𝑄,𝑡)
𝐼(𝑄,0)
I can be simplified into the following equation, when some higher-order cumulants
are neglected in the expansion of the normalized intermediate scattering function,

8(5,!)
8(5,()
= exp [−𝐷
:;;
(𝑄)𝑄
"
𝑡] 1-7
where Q is the momentum transfer which is determined by the incoming neutrons wavelength and
by the scattering angle experimentally.
12
If there is no internal dynamics and hydrodynamic interactions can be neglected, the
effective coefficient can be obtained from the following equation,
𝐷
:;;
(𝑄) = 𝐷
(
<
=(5)
1-8
where D0 is the diffusion coefficient when there are no interactions, and S(Q) is the static structure
factor. When Q values approaches to zero, the hydrodynamic limit of 𝐷
:;;
is the translational
diffusion coefficient, 𝐷
!6
, which can also be measured from the dynamic light scattering.
If the investigated sample has some kind of internal motions and those motions are
contributed to the observed dynamics, we can use the following equation to describe the
relationship between them since we can assume the two dynamics are not correlated with each
other,
𝐷
:;;
(𝑄) = 𝐷
!6
(𝑄)+ 𝐷
>%!:6%?@
(𝑄) 1-9
and the internal motions come from vibrations and rotations of the sample molecules.

13
2 Enhanced Catalase Activity through Tetramer Dissociation in the Presence of
a Photoresponsive Surfactant
2.1 ABSTRACT
Catalase is one of the fastest known enzymes, exhibiting an overall catalytic efficiency close to
the diffusion limit. In spite of this fact, herein we report increases in catalase activity in the
presence of the photoresponsive surfactant azobenzene trimethylammonium bromide (azoTAB).
The surfactant undergoes a reversible photoisomerization from the relatively hydrophobic trans
isomer under visible light to the relatively hydrophilic cis isomer upon UV illumination (350 nm).
This provides a means of controlling surfactant-protein interactions and, as a result, protein
structure and activity. In the presence of azoTAB, catalase is found to undergo a tetramer-to-dimer
dissociation, as revealed through small angle neutron scattering, dynamic light scattering, circular
dichroism, and fluorescence spectroscopy measurements. Unlike traditional “armed-wrapped”
dimers (i.e., dimers between chains A and C in the tetramer), which are typically inactive, the
solution structure of azoTAB-induced dimers appear to be of the A-B variety, which are shown to
have an activity of twice that of the native tetramer.
2.2 INTRODUCTION
Catalase is a ubiquitous enzyme found in the peroxisomes of almost all aerobically
respiring organisms
24
. The enzyme serves to protect cells from the toxic effects of hydrogen
peroxide by quickly catalyzing the decomposition of H2O2 into oxygen and water. In response to
this necessity, catalase is one of the fastest known enzymes, exhibiting an extremely high turnover
number (kcat = 4 ´ 10
7
s
-1
)
25
and a catalytic efficiency (kcat/KM = 4 ´ 10
7
s
-1
M
-1
)
26
approaching the
diffusion limit
27
. In the native state, catalase exists as a tetramer of identical 58 kDa polypeptide
subunits bound through hydrophobic and ionic interactions, and notably lacking disulfide bonds
10
.
14
In the crystallographic tetramer, a two-helix orthogonal bundle domain, acting as a bent “arm,”
hooks through a random/b-turn “wrapping” domain of the opposing Q-axis-related
24
molecule (i.e.,
the arm of chain A interlocks with the wrapping domain of chain C and vice versa, while chains B
and D undergo similar domain swapping).
13,24
This leads to a tetramer with three non-equivalent
protein-protein interfaces,
28
prompting some to refer to the enzyme as a “dimer of dimers.”
29

Interestingly, catalase-peroxidases typically exist as dimers in solution,
11
and dimers are more
stable than tetramers after dissociation.
30
The heme active site is located 20 Å below the molecular
surface in a beta-barrel domain, corresponding to the most conformationally-rigid region of the
polypeptide chain.
12
The active site is accessed through a long and narrow hydrophobic channel
perpendicular to the heme moiety,
31
which is formed by portions of the barrel domain combined
with an up-down bundle domain consisting of eight a-helices that reside on the enzyme surface.
As suggested above, the ability to dissociate catalase into dimers has long been known.
32

However, despite the fact that the heme groups in such dimers should be more readily accessible
to the substrate (as opposed to the restricted channel of the tetramer),
31
dimers are typically inactive,
such as when dissociated with acid,
33
base,
32
urea,
34
or SDS surfactant.
35
In a few rare cases,
however, a slightly “superactive” dimer (i.e., 120% of the activity of the native tetramer) has been
reported at low GdmCl concentrations prior to widespread denaturation.
9,36
Notably, the four
active sites within the oligomer are not identical,
37
unsurprising given the asymmetric subdomain
interfaces mentioned above, with so-called half-site reactivity with inhibitors frequently reported
(i.e., 50% maximum inhibition at high inhibitor concentrations).
38
Indeed, steric conflicts arising
from A-C contacts leads to so-called double conformations in X-ray crystallographic structures
near the active site channel (i.e., side chains of corresponding residues in the A and C molecules
alternately flipped 180° about the peptide bond), which may be the reason that only half of catalase
15
molecules are active at any one time.
39
These studies suggest that relieving the strain of quaternary
interactions in the tetramer (particularly A-C contacts) could make simultaneously available all
four heme groups and allow doubling of the enzyme activity.
The photoresponsive surfactant azobenzene trimethylammonium bromide (azoTAB) is
shown in Scheme 1. Under visible light, approximately 75% of azoTAB adopts the shown trans
conformation that, as a result of a lower dipole moment,
2
is more hydrophobic than the cis isomer
that is the predominant form (~90%) under UV light (350 nm). This provides a means to
photoreversibly control protein-surfactant interactions and the protein conformations that result.
3,40,41
Notable examples of controlling protein quaternary structure include photoreversible dimer-
to- monomer transitions leading to superactivity in a cellulase enzyme,
42
hexamer-to-dodecamer
transitions in a-chymotrypsin,
43
control of amyloid fibril formation with Ab40,
44
and dissociation
of the ribonuclease A-inhibitor complex.
45

Scheme 1. Chemical structure of azobenzene trimethylammonium bromide (azoTAB).

Azobenzene is a commonly-used moiety for the photo-control of biological
macromolecules.
46,47
Various azobenzene-modified ligands and polymers have been utilized to
initiate changes in protein unfolding,
48–50
often through conjugation to the protein backbone
51–53

or with replacement of an amino acid in the protein with phenylazophenylalanine.
54
Notable
fluctuations in enzyme activity with azobenzene photoisomerization include a nearly 3-fold change
observed with azobenzene-conjugated papain,
55
a 4-fold change in ribonuclease S activity using
CH
3
CH
2
N
N O(CH
2
)
4
N+(CH
3
)
3
Br-
16
phenylazophenylalanine,
56
an 7-fold change in lysozyme activity in the presence of azoTAB,
57
and
a 16-fold change in endonuclease activity using an azobenzene moiety to crosslink the enzyme.
53

In the present study, the effects of azoTAB on the quaternary structure and activity of
catalase are explored. Structural changes in the protein are observed through small angle neutron
scattering, providing shape-reconstructed images of the protein oligomers in solution that are
augmented with circular dichroism, dynamic light scattering, and fluorescence spectroscopy
measurements. Catalase superactivity in the presence of azoTAB is seen to result from
dissociation of the protein tetramers into a uniquely superactive dimeric structure.

2.3 EXPERIMENTAL METHODS
Materials. Catalase from bovine liver (catalog number C40), an 8.3 mM phosphate buffer
(P3288; pH 7.2), a 50 mM potassium phosphate dibasic trihydrate buffer (P5504), hydrogen
peroxide (30 wt %, H1009), 1 M hydrochloride acid (H1758), and all reagents for azoTAB
synthesis were purchased from Sigma-Aldrich at the highest purity and used as received. Both
buffers were prepared in ultrapure water (resistivity ³ 18 MW·cm) at 25 °C, with the latter (used
for the activity assay) adjusted to pH 7.0 using 1 M HCl. AzoTAB was synthesized according to
published methods via an azocoupling reaction of phenol with 4-ethylaniline, followed by
alkylation with 1,4-dibromobutane and quaternalization with trimethylamine.
1
The trans isomer
of azoTAB is more energetically favorable and is the predominant form under visible light (75/25
trans/cis) or in the dark (~100% trans after 24 h). When desired, a 2´15 W longwave (365 nm)
UV lamp (Spectroline, Model no. XX-15A) was used to pre-convert azoTAB to primarily the cis
state (10/90 trans/cis) prior to mixing with catalase. This pre-conversion of the surfactant was
17
necessary since even a small amount of UVB light can induce catalase aggregation and
deactivation,
58
and may also lead to the formation of reactive oxidant species.
59

Small-angle neutron scattering. Small-angle neutron scattering experiments were
performed on the 30 m NG7 SANS instruments at NIST.
60
A neutron wavelength of l = 6 Å and
a detector offset of 25 cm with two sample-to-detector distances of 1.33 and 7.0 m were utilized
to achieve a Q range of Q = (4p/l)sin(q/2) = 0.00491–0.55 Å
-1
, where q is the scattering angle.
The net intensities were corrected for the background and empty cell (pure D2O), accounted for
detector efficiency using the scattering from an isotropic scatterer (Plexiglas), and converted to an
absolute differential cross section per unit sample volume (in units of cm
-1
) using an attenuated
empty beam. To achieve reasonable count rates, 2 mg/mL enzyme was employed.
The Guinier approximation 𝐼(𝑄) ≅ 𝐼(0)expN−𝑅
A
"
𝑄
"
3 ⁄ Q was employed to determine the
radius of gyration Rg and I(0), the scattering intensity extrapolated to Q = 0. A Q range of 0.023–
0.045 Å
-1
was employed to neglect scattering near the beam stop and to ensure Q×Rg < 1.8, since
the percent error of the Guinier approximation is 0.69(Q×Rg)
4
for spheres.
61
Pair distance
distribution functions, P(r), related to the probability that two scattering centers (nuclei for SANS)
within the protein macromolecule or oligomer are separated by a distance r, were calculated using
the program GNOM
62
through the equation 𝐼(𝑄) = 4𝜋∫ 𝑃(𝑟)
.
,-.
(
/01(&2)
&2
𝑑𝑟, where the maximum
intraparticle distance, Dmax, was selected to give a smooth return of P(r) to zero. Radii of gyration
were obtained from the P(r) curves using the equation 𝑅
B
"
= ∫ 𝑟
"
𝑃(𝑟)𝑑𝑟
.
,-.
(
2∫ 𝑃(𝑟)𝑑𝑟
.
,-.
(
9 .
Shape reconstructions were performed with the program GA_STRUCT.
63
Briefly, the program
utilizes a genetic algorithm to optimize the positions of the scattering centers to fit the experimental
data, with 10 independent runs for each sample compared for consistency. In the two latter analysis
18
methods, a Q range of 0.023–0.30 Å
-1
was used in order to exclude intermolecular interactions at
low Q and avoid length scales too small for protein continuity at high Q.
Circular dichroism spectroscopy. Circular dichroism (CD) measurements were
performed at the University of Southern California Bio-imaging Center on a Jasco J815 instrument.
Spectra were collected using a 1 mm path length cuvette at 25 °C. Due to the strong ultraviolet
absorption of azoTAB, a maximum surfactant concentration of 1 mM was used. Each spectrum
was averaged over four scans and was background-subtracted and normalized to molar ellipticity.
The spectra were then analyzed using BeStSel
64
to quantify the contributions of α-helical, β-sheet,
b-turn, and random coil secondary structural elements.
Dynamic light scattering. Dynamic light scattering (DLS) measurements were performed
on a Brookhaven BI-200SM instrument equipped with a BI-9000AT digital correlator
(Brookhaven Instrument Corporation) using a 35 mW helium-neon (HeNe) laser (Melles Griot,
model Number 05-LHP-928) with a wavelength of 632.8 nm (outside the absorbance spectrum of
azoTAB). The detector was at an angle of 90° to the incident beam and data were collected at a
temperature of 25 °C. Analysis was done using the non-negative least squares (NNLS) routine.
Fresh protein solutions were prepared and filtered through a 200 nm Anotop filter to remove large
structures immediately prior to data collection.
Fluorescence spectroscopy. Fluorescence spectroscopy was performed on a Photon
Technology International spectrofluorometer (model number QM-4). The fluorescent dye Nile
red (Sigma-Aldrich, catalog number N3013) was used as a micropolarity indicator. A small
aliquot of a concentrated solution of Nile red in ethanol (1 mM) was added to a 0.1 mg/mL catalase
solution giving a final Nile red concentration of 0.4 µM. Samples were stirred for 20 minutes prior
to data collection. For samples containing azoTAB, the corresponding azoTAB stock solution was
19
illumined with either visible or UV light prior to sample preparation to avoid photobleaching of
Nile red. Nile red was excited at 590 nm, with the emission monitored between 600 and 800 nm
(well beyond the absorption spectrum of azoTAB).
Activity measurements. A freshly prepared catalase solution (0.02 mg/mL) was incubated
with an equal amount of an azoTAB solution to achieve the reported surfactant concentrations.
For reference, catalase dissociation generally requires 2–3 min.
33,34
To initiate the reaction,
hydrogen peroxide (30 wt %) was diluted into 50 mM potassium phosphate buffer (pH 7.0) to a
concentration of 20 mM, followed by thermal equilibration for 5 min at 25 °C of 480 µL in a 2
mm path length quartz cuvette. Next, 20 µL of azoTAB-protein solution was added to the cuvette
followed by mixing through rapid pipetting. Catalase activity was determined from the initial
linear rate of decrease of the hydrogen peroxide concentration over the first 15s, determined from
the absorbance using 𝜖
"#,
C
'
D
'
= 0.031 mM
–1
cm
–1
, slightly offset from the more typical 240 nm to
coincide with an isosbestic point of azoTAB.
3,40
The specific activity of the pure enzyme was
determined to be 36,000 units/mg, in reasonable agreement with values previously reported for the
C40 enzyme preparation.
65
When determining the kinetic parameters kcat and KM, the above
procedure was employed with varying hydrogen peroxide concentrations (3.4–26.7 mM) and half
the final catalase concentration using a 1 cm quartz cuvette.
2.4 RESULTS AND DISCUSSION
The effect of azoTAB on the activity of catalase relative to the native enzyme is shown in
Figure 2-1a. The relative activity is seen to steadily rise up to a maximum of either 190% or 150%
at 1 mM visible- or UV-light adapted azoTAB, respectively. As shown below, this enhanced
activity results from azoTAB-induced dissociation of catalase into a specific dimer conformation
(A-B dimers), similar to, but more dramatic than, the 120% enhancements reported for catalase
20
dimers obtained at low (<0.3 M) GdmCl concentrations.
36,66
Further increases in azoTAB
concentration lead to a slow decline in activity, a result of progressive dissociation into
aggregation-prone and hence inactivated monomers (see DLS data in Figure S1).
67
Noteworthy is
the fact that the maxima in Figure 2-1a occur well below the critical micelle concentrations (CMCs)
of either visible- or UV-light adapted azoTAB (4 mM or >7 mM, respectively, measured in
identical buffered solutions; see Figure S2). Catalase activity in the presence of the conventional

Figure 2-1: Effect of visible- or UV-light adapted azoTAB on (a) the relative specific activity of
catalase at [H2O2] = 19.3 mM and (b) the initial velocity versus substrate concentration without
and with 1 mM azoTAB. Catalase activity in the presence of the conventional surfactants SDS
and DTAB is also shown in (a). The three data points at 19.3 mM in part (b) are repeated from
(a) for comparison. Error bars are typically within the symbols.

surfactants SDS and DTAB is also shown in Figure 2-1a, in both cases displaying a simple loss in
activity (~20%).
The enzymatic activity was also measured as a function of substrate concentration, as
shown in Figure 2-1b. Here it must be noted that standard analysis via Michaelis-Menten kinetics
in not possible
68
due to rapid inactivation of the enzyme when the H2O2 (substrate) concentration
75
100
125
150
175
200
0 1 2 3 4 5
relative specific activity (%)
surfactant concentration (mM)
azoTAB (visible)
azoTAB (UV)
(a)
SDS
DTAB
0
0.2
0.4
0.6
0.8
1
0 0.005 0.01 0.015 0.02 0.025 0.03
pure catalase
w/ azoTAB (vis)
w/ azoTAB (UV)
reaction velocity (mM/s)
[H O ] (M)
(b)
2

2
21
exceeds ~30 mM,
69
combined with complications arising from O2 (product) bubbles when using
an absorbance-based assay. These effects are particularly problematic given the large values of
KM » 1 M versus H2O2.
70
Thus, by necessity the substrate concentration [S] << KM in Figure 2-1b,
reducing the Michaelis-Menten equation to a linear equation (i.e., v = kcat[E]0[S]/(KM + [S]) »
(kcat/KM)[E]0[S], where [E]0 is the enzyme concentration). As a result, only the combined catalytic
efficiency kcat/KM can be determined as opposed to separately kcat and KM, a fact often overlooked.
Nevertheless, from the slopes of the plots in Figure 2-1b, visible-light adapted azoTAB (kcat/KM =
3.9 ´ 10
7
M
–1
s
–1
) is again seen to again result in a 2-fold increase in activity relative to the native
enzyme (kcat/KM = 1.8 ´ 10
7
M
–1
s
–1
), while UV-light adapted azoTAB (kcat/KM = 2.5 ´ 10
7
M
–1
s
–1
)
shows a 1.4-fold increase, both similar to Figure 2-1a.
It is interesting to note that the peak activities in Figure 2-1a occur at similar azoTAB
concentrations independent of light conditions. This implies that the surfactant-protein
interactions that trigger superactivity are not primarily hydrophobic in nature, and instead may be
driven by electrostatic interactions. Thus, the differences seen in Figure 2-1a between visible- and
UV-light adapted azoTAB may be a result of an increased steric hindrance for the interaction of
the bent cis azoTAB isomer with the protein relative to the linear trans form. The surface of bovine
liver catalase contains several locations of locally high concentration of negative charges.
24
Thus,
at neutral pH positively-charged azoTAB would be expected to bind to these surface regions,
thereby influencing the structure of the enzyme. Indeed, Gdm
+
cation binding to native catalase
leads to tetramer dissociation into active dimers.
9
Note, however, that the activity of catalase was
depressed in the presence of the traditional surfactant dodecyltrimethylammonium bromide
(DTAB) in Figure 2-1a, thus, the effect of azoTAB on catalase is not simply a result of electrostatic
interactions with the surfactant head group.
22
To gain insight into the effect of azoTAB on the quaternary structure of catalase, small
angle neutron scattering (SANS) data were collected as shown in Figure 2-2. Each spectrum
displays a shoulder (a common characteristics of oligomeric species
71
) at Q » 0.11 Å
–1
,
corresponding to

Figure 2-2: (a) SANS scattering data, (b) Guinier plots, and (c) calculated pair distance
distribution functions for solutions of catalase (2 mg/mL) mixed with various concentrations of
visible- or UV-light adapted azoTAB (closed circles and open squares, respectively). The same
color scheme is used throughout. Each successive spectra is offset by ´10 for clarity with the
pure catalase spectrum overlaid (small black circles). The experimental P(r) functions in (c) are
compared to those calculated from the crystallographic tetramer (PDB: 1TGU) and two potential
dimers.

an average separation distance between monomers of L = 2p/Q » 57 Å. For comparison, the heme
groups in chains A and C of the crystallographic tetramer are separated by ~45 Å. Guinier fits
(Figure 2-2b) over the range Q
2
= 0.0005–0.002 Å
–2
(i.e., ignoring the steep uptick in scattering
0.0001
0.01
1
100
10
4
10
6
0.01 0.1
I(Q) (cm
-1
)
Q (Å
-1
)
0.2
0.4
1.0
2.0
4.0 mM
pure
(a)
pure (÷5)
(10 mg/mL)
(2 mg/mL)
0 20 40 60 80 100
0
0.5
1
1.5
P(r)/I(0) x 10
3
r (Å)
(c)
A-B dimer
tetramer
(1TGU)
A-C dimer
0 0.001 0.002
-3
-2
-1
0
1
ln I
Q
2
(Å
-2
)
(b)
pure (÷1)
(10 mg/mL)
23
near the beam stop at Q < 0.01 Å
–1
, which may represent higher order aggregates, see DLS data in
Table S1) allowed determination of the radius of gyration and the scattering intensity I(Q)
extrapolated to Q = 0, as shown in Table 2-1. This latter parameter allows determination of the
weight-average molecular weight MW through the equation
72
𝐼(0) = 𝑐𝑀
E
𝜐̅ "
(Δ𝜌)
"
1000𝑁
F
⁄ ,
where c is the protein concentration (in units of mg/mL), 𝜐̅ is the protein specific volume (0.74
cm
3
/g),
73,74
Dr is the scattering length density difference or “contrast” between the protein and
D2O solvent (3.32 ´ 10
–6
Å
–2
, determined via a calibration curve),
42,75
and NA is Avogadro’s
number. As shown in Table 2-1, for pure catalase a MW of 190 kDa (at 2 mg/mL) or 180 kDa (at
10 mg/mL) was obtained, in approximate agreement with the value expected of the tetramer (233
kDa) when allowing for slight dissociation in vitro, as commonly observed under physiological
conditions.
76
With the addition of 2 mM azoTAB, the MW decreased to a minimum of 140 kDa
under visible light or 100 kDa under UV light, both consistent within error with tetramer
dissociation into dimers (expected MW = 117 kDa). Notably, with 4 mM azoTAB under visible
light (i.e., at the critical micelle concentration determined independently in Figure S1), the SANS
spectrum is markedly different than that of the pure enzyme especially at Q > 0.06 Å
–1
, confirming
the onset of micelles. The rebound of the MW back to a seemingly tetrameric value (210 kDa) in
this case may simply be a coincidence of micelle scattering. The spectrum obtained at 4 mM
azoTAB under UV light displays pronounced aggregation at low Q in Figure 2-1a (which appears
exaggerated due to the log scale of the abscissa), while Guinier analysis suggest a near monomeric
MW (80 kDa).
Table 2-1: Molecular weight and radius of gyration determination using Guinier and PDDF
analysis for catalase (2 mg/mL) as a function of azoTAB concentration under visible light.

Guinier PDDF
[azoTAB] I(0) MW Rg Rg
(mM) (cm
–1
) (kDa) (Å) (Å)
24
0 0.38 190 42 38
0
a
1.82 180 41 37
0.2 0.37 190 40 36
0.5 0.33 170 39 37
1.0 0.29 140 39 36
2.0 0.28 140 38 38
2.0
b
0.19 100 37 35
4.0 0.41 210 38 34
4.0
b
0.16 80 36 35
a
catalase concentration = 10 mg/mL,
b
under UV light

For quantitative analysis of the SANS data, pair distance distribution functions (PDDFs)
were calculated, as shown in Error! Reference source not found.. PDDFs are a model-
independent approach to determine the probability, P(r), that two scattering centers (i.e., atoms for
SANS) within the protein macromolecule or oligomer are separated by a distance r. The maximum
intraparticle dimensions obtained in Figure 2-2c (i.e., where each P(r) returns to zero) are in
agreement the edge-to-edge width of the crystallographic tetramer (i.e., 105 Å),
24
while the radius
of gyration obtained for the pure protein (38 Å) similarly agrees with published values (~39 Å).
75,77

With increasing azoTAB concentration, the radius of gyration decreases to a minimum of 34 Å,
again consistent with tetramer dissociation. Note that the Rg values obtained above from Guinier
analysis, which are z-averaged values (i.e., 𝑅
B,G
"
= ∑𝑦
>
𝑀
E,>
"
𝑅
B,>
"
∑𝑦
>
𝑀
E,>
"
9 ),
78
are consistently
larger than those obtained from the PDDFs, likely due to higher-order aggregates. Using the
correlation Rg µ MW
0.369
(i.e., nearly the MW
1/3
dependence expected for a perfectly compact
particle),
79
to a first approximation Rg = 34 Å would correspond to an ~40% drop in MW, again
consistent with dimer formation. Notably, these dimeric Rg values are substantially smaller than
those obtained when tetramer-to-dimer dissociation was initiated with conventional denaturants
(e.g., Rg = 42 Å for the low-pH dimer, a value higher than the native tetramer, implying significant
unfolding).
75
This could explain why catalase dimers obtained in the presence of azoTAB maintain
25
enzymatic activity (in fact, enhanced activity), as opposed to the typically inactive dimers observed
when denatured with acid,
33
base,
32
urea,
34
or SDS surfactant.
35

Returning to the P(r) curves in Figure 2-2c, the most-probable intraparticle distance for
catalase at low azoTAB concentrations is ~50 Å (recall the shoulder in Figure 2-2a), in near perfect
agreement with the peak of the P(r) function calculated from the crystallographic tetramer (PDB:
1TGU). At intermediate azoTAB concentrations (1–2 mM under visible light and 2–4 mM under
UV light), however, the most-probable distance decreases to ~40 Å, indicating a decrease in
oligomer size. Surprisingly, the shapes of these P(r) functions are more consistent with a dimer
formed from chains A and B of the crystallographic tetramer, as opposed to the “arm wrapped”
chains A and C, a fact that will be explored further below. Further increases in azoTAB
concentration (4 mM under visible light) eventually yield a most-probable distance of ~30 Å, with
a P(r) function now closely mimicking that calculated from the A-C crystallographic dimer.
However, as this latter case is now beyond the CMC, this agreement is likely a coincidence with
the new peak at 30 Å being simply a manifestation of azoTAB micelles. For comparison, pure
azoTAB beyond the CMC forms triaxial ellipsoid micelles (Ra = Rb = 15 Å and Rc = 30 Å).
80

To conclusively demonstrate the in vitro structure of the azoTAB-induced superactive
dimer, shape reconstruction analysis of the SANS data was performed. Since neutrons are
scattered by atomic nuclei, the scattering intensity from a given protein structure can be readily
calculated. In shape-reconstruction analysis (the so-called inverse scattering problem), the protein
structure is instead determined from the scattering data. The resolution of shape-reconstructed
images are limited by 2p/Qmax, namely the widest angle where reasonable S/N can be achieved
(e.g., Qmax » 0.3 Å
–1
in Figure 2-2, corresponding to a resolution of ~20 Å; but see Rambo and
Tainer
81
). At this resolution, individual protein domains
42
and enzyme active-site cavities
82
can
26
be visualized. The ill-posed nature of the inverse-scattering problem (i.e., fitting the positions of
thousands of scattering centers with only a few hundred data points) is overcome in the ab initio
methods of GASBOR
83
and GA_STRUCT
63
by comparing several independent runs for
consistency. Broadly, by noting that scattering is dominated by the protein-solvent interface, these
shape-reconstruction methods can be conceptualized as analogous to fitting the lengths of several
tens of 𝑟 ⃗ vectors extending to the irregular protein surface (as opposed to only three orthogonal Ra,
Rb, and Rc values when assuming a simple triaxial ellipsoid shape, for example). In the present
system, using SANS to determine protein structure has advantages over traditional techniques (e.g.,
X-ray crystallography and NMR) as crystal-packing constraints in the former can artificially
perturb protein conformations relative to that in solution,
84–88
while the latter is generally limited
to relatively small proteins.
Results of shape-reconstruction analysis of the SANS data are shown in Figure 2-3. As
seen in Figure 2-3a, each chain in the crystallographic tetramer consists of four domains. Two a-
helices in an orthogonal bundle domain (residues 3–67, shown in red) act as a bent “arm” that
hooks through an unordered/b-turn “wrapping” domain (residues 377–437, in yellow) of the
opposing molecule (i.e., the arm of chain A interlocks with the wrapping domain of chain C and
vice versa). The heme active site (shown space-filled in black) is located in a b-barrel domain
(residues 68–151 and 203–376, in blue), accessed through a hydrophobic channel formed by
portions of this domain combined with an eight-helix up-down bundle domain (residues 152–202
and 438–501, in green). The SANS-based solution structure of pure catalase exhibits an oblate
ellipsoid shape (i.e.,
27

Figure 2-3: (a,b) X-ray crystallographic structures of the catalase tetramer and two possible
dimers, and (c) SANS-based solution structures of catalase as a function of azoTAB
concentration under visible light. Right-viewed images are reduced by half throughout for
display purposes.

Rx = Ry > Rz), similar to the crystal structure. Furthermore, the symmetry of the crystallographic
tetramer is largely reproduced in the in vitro pure-catalase tetramer, which is symmetric in the xy
and yz planes and is only slightly asymmetric in the xz plane (i.e., front-back, left-right, and top-
bottom symmetry, respectively). The dimple seen in the center of the crystallographic tetramer is
mimicked in the solution structure by a region of low scattering-center density (i.e., the ability to
see through this portion of the shape-reconstructed image).
At intermediate azoTAB concentrations (0.2–0.5 mM), the lower-left portion of each
structure gets progressively smaller until completely disappearing at elevated azoTAB
concentrations (1–2 mM), resulting in a final “Y-shaped” dimer. These intermediate images are
likely not true oligomeric structures, but instead are weighted (z-averaged) superpositions
89
of the
simultaneously existing tetramer and Y-shaped dimer (the CD data below support such a two-state
28
transition). Interestingly, the final Y-shaped dimer is reminiscent of the “jellyfish-shaped” dimer
that would form from chains A and B of the crystallographic tetramer, rather than the “saddle-
shaped” A-C dimer, in agreement with the PDDFs above. In fact, both the 1 and 2 mM azoTAB
structures exhibit a twist similar to that seen in the “tentacles” of the A-B crystallographic dimer
(dashed blue box in Figure 2-3, discussed further below). Thus, the superactivity of catalase in
the presence of azoTAB results from formation of A-B dimers, rather than the typically inactivated
dimers seen with traditional denaturants. The formation of A-B dimers would eliminate the
aforementioned steric conflicts between side chains that arise from A-C contacts, which are
suspected to cause only half of catalase molecules to be active at any one time.
39

To augment these in vitro structures, circular dichroism was used to analyze changes in
catalase secondary structure during azoTAB-induced tetramer dissociation, as shown in Figure 2-4.
The pure catalase spectrum displays a negative peak near 222 nm and positive peak at ~195 nm,
consistent with a protein containing both substantial alpha and beta structures. Indeed, DSSP
calculations
90
from the crystallographic tetramer give 31% a-helical, 18% b-sheet, and 51% other
structures, while additionally 53 b-turns can be identified accounting for ~30% of the protein

-30
-20
-10
0
10
20
30
40
190 200 210 220 230 240 250
[q] (10
-3
deg cm
2
dmol
-1
)
Wavelength (nm)
pure catalase
0.2 mM
0.5 mM (UV)
0.5 mM
1.0 mM
-15
-10
-5
0
5
10
15
200 210 220 230 240 250
D[q]
Wavelength (nm)
29
Figure 2-4 Circular dichroism spectra of catalase as a function of azoTAB concentration and
illumination conditions. Inset: Difference spectra at each condition.

residues. Fitting the pure catalase spectrum using BeStSel, which was developed to lead to more
accurate estimates for beta-rich proteins,
64
gives 24% a-helical, 25% b-sheet, 9% b-turn, and 42%
other structures, consistent with the crystallographic calculations (although these fits suggests
about half of the b-sheets are of the parallel variety, while only ~2% are predicted
90
to be so from
the crystallographic structure). Notably, b turns are characterized by an intense positive band at
207.5 nm in CD spectra,
91
which in Figure 2-4 appears to shift the q208/q222 ratio of the pure enzyme
away from a more typical value near unity common for a helices in the absence of b structures.
With the addition of azoTAB, an isodichroic point develops near 214 nm, which is
suggestive of a two-state (i.e., tetramer-dimer) coil « beta transition
92
(as opposed to a two-state
coil « alpha transition, for example, which exhibits an isodichroic point near 204 nm).
93
Indeed,
BeStSel fits give a progressive loss in “other” (i.e., random coil) structures from 42% for the pure
enzyme to 25% (0.2 mM), 23% (0.5 mM), and ~0% (1 mM) with the addition of visible-light
adapted azoTAB, while the b-sheet amount increases to 53% over the same span (the b-turn
content remains constant at ~10%).
These secondary structure changes can be rationalized by considering the crystallographic
tetramer structure (Figure 2-3a). The two largest primarily-unordered regions in the tetramer are
residues 18–53 comprising the “elbow” of the two-helix arm domain, and residues 377–414 of the
wrapping domain. In other words, a certain degree of disorder appears necessary to facilitate arm
wrapping between opposing chains in the tetramer. These constraints would be liberated following
azoTAB-induced unraveling of the arm-wrapped A-C coupling to give the A-B dimers, freeing up
these unordered regions to now form local secondary structure (i.e., the unordered ® b-sheet
30
transition seen in Figure 2-4). Indeed, YASPIN analysis
94
predicts that when alone these two
unordered regions should instead exhibit several extended b-strands (colored in teal in the dashed
blue box of Figure 2-3b). Based on the high scattering-center density seen in this region in the in
vitro A-B dimers (cf. the dimple in the center of the in vitro tetramer), it would seem that these b-
strands may become “zipped up” into b-sheets, consistent with Figure 2-4. Similar increases in b-
sheet content have been observed as amyloid-beta (Ab) peptide fragments associate into the early
oligomeric precursors of amyloid fibrils,
95
while azoTAB has been shown to inhibit Ab fibril
formation and trap the peptides in this oligomeric state.
44
Intermolecular b-sheets of this type
could explain why azoTAB-promoted catalase dimers remain active, unlike the global unfolding
seen with traditional denaturants.

CONCLUSIONS
Despite being one of the fastest known enzymes, the activity of catalase can be doubled
upon photosurfactant-induced dissociation of the enzyme tetramer into a specific dimer
conformation. While catalase dimers are commonly observed in the presence of conventional
denaturants, the dimers are nearly always inactive. In the presence of azobenzene
trimethylammonium bromide (azoTAB), however, in vitro solution structures obtained from
small-angle neutron scattering reveal that the enzyme dissociates into A-B type dimers. The
doubling of enzyme activity that occurs upon formation of these dimers is a result of the fact that
in the tetramer, only two of the four active sites are accessible at any given time (i.e., either chain
A or C). Secondary structure measurements indicate that the A-B dimers are likely stabilized by
the formation of intermolecular b-sheets from the descending regions of the Y-shaped dimer that
were previously disordered to facilitate arm-wrapping in the tetramer.
31

Supporting Information Description
hydrodynamic diameter of catalase in the presence of azoTAB; azoTAB critical micelle
concentration determined using Nile red fluorescence

ACKNOWLEDGEMENTS
We acknowledge the support of the National Institute of Standards and Technology, U. S.
Department of Commerce, in providing the neutron research facilities used in this work. This
material is based upon work supported by the National Science Foundation under Grant 1153699.
Any opinions, findings, and conclusions or recommendations expressed in this material are those
of the author(s) and do not necessarily reflect the views of the National Science Foundation.

32
3 Chapter 3: Ribonuclease-A and NSE
3.1 Abstract

A photoresponsive surfactant, azoTAB, has been used to control the structure and
dynamics of RNase A with light illumination. AzoTAB is a cationic azobenzene surfactant that
can undergo a reversible photoisomerization upon exposure to the appropriate wavelength of light.
Under visible light, azoTAB preferentially adopts a relative hydrophobic trans structure, while
under UV illumination the relative hydrophilic cis structure is preferred. According to previous
kinetic data, RNase A exhibits superactivity in the presence of trans azoTAB and reverses back to
native-like conditions after illumination with UV light. In order to explore the reasons behind these
differences, several techniques have been used to detect the structural changes occurring within
the protein. Dynamic light scattering (DLS) and small angle neutron scattering (SANS) results
show that there is a conformation change of Ribonuclease A after interacting with azoTAB under
both visible light and UV light. Both results prove that RNase A adopts a swollen form in presence
of trans azoTAB, which accompanies the enhancement of Ribonuclease A activity. At the same
time, the protein returns to a native like conformation in the presence of cis azoTAB. In order to
determine the internal dynamics of the protein when interacting with azoTAB, neutron spin echo
(NSE) spectroscopy was conducted. Three different models were employed to describe the internal
dynamics of the protein: a rigid-body model that assumes the protein moves as a whole and remains
rigid with no internal dynamics, a soft-linker model that treats the protein as being composed of
several subunits and considers the interaction between these subdomains as spring-like interaction,
and a freely jointed model that treats different subunits of the protein as “beads” in a necklace and
considers hydrodynamic interaction and restriction of the bonds connecting adjacent subdomains.
33
After comparing the NSE data with these model results, the most appropriate model and
parameters can be determined to describe the protein structure and dynamics in different cases.

3.2 Introduction
Proteins are large biomolecules that are widely distributed in living organism. Consisted
of one or more long chains of amino acid residues, proteins perform a vast range of functions,
including catalyzing metabolic reactions, DNA replication, cell production, and molecule
transportation. Basically, proteins play an important role in organism function. Proteins, with
structures that are determined by the amino acid sequences, are not static objects, however, and
can perform different dynamic events under various conditions, and notably undergo
conformational rearrangements during the course of activity.
96,97
Therefore, it is important to
investigate the dynamics of enzymes in coordination with structural determination.
There are various techniques to investigate the dynamic of proteins, such as static
crystallography,
98
neutron spectroscopy,
99,100
single-molecule measurements,
100
and molecular
dynamics simulation.
101–103
Among those, neutron spin echo (NSE) is an advanced technique
which can carry out measurements in both a wide range of length and time scale (from 10
-12
to 10
-
8
s). Therefore, NSE is a valuable technique to measure the internal motions of proteins.
AzoTAB, namely azobenzenetrimethylammonium bromide, is a light sensitive surfactant.
AzoTAB can undergo a conformation change when exposured to appropriate wavelengths of light.
Under visible light, azoTAB exists primarily in trans form (75/25 trans/cis), which is more
hydrophobic and has a stronger binding affinity to proteins than the cis isomer, which is the
predominant form of azoTAB under UV light (10/90 trans/cis). Therefore, azoTAB has been
utilized as a tool to photocontrol the structure, function, and dynamics of proteins.
3,43,82,104–106

34
Ribonuclease A (RNase A) is an excellent model protein since its amino acid sequence and
three-dimensional structure are well known.
16,17
RNase A is a monomeric enzyme that catalyzes
the cleavage of single-stranded RNA. RNase A consists of 124 amino acid residues giving the
molecule a molecular weight of ~13.7 kDa. The enzyme folds into two halves: a predominantly α-
helical N-terminal half (residues 1-60) and a primarily β-sheet C-terminal half (residues 65-124).
The active site is located in a deep cleft. The secondary structure of the molecule is composed of
23% α-helix, 46% β-sheet, 21% β-turn, and 10% random coil structures.
18,19
RNase A is a basic
protein (pI =8.63) containing four disulfide bonds (cys26-cys84, cys40-cys95, cys65-cys72, and
cys58-cys110) that are essential for protein folding and stability.
In this study, neutron spin echo is used to examine the internal motions of RNase A in the
presence of both trans and cis azoTAB isomers. Three models have been employed to approximate
the dynamics of RNase A, a rigid-body model, a soft-linker model, and a freely jointed model. The
Kirkwood formula has also been used to calculate the translational diffusion coefficient of the
protein from the SANS-based solution structures. Results from these three models are compared
with neutron spin echo experimental data to better understand the dynamics of RNase A after
interacting with different concentrations of azoTAB and under exposure of different wavelengths
of light.

3.3 Experiments and methods
3.3.1 Materials
Ribonuclease A from bovine pancreas (catalog number R5500) was purchased from
Sigma-Aldrich. Deuterium oxide (item number DLM-4-100) used in neutron spin echo
35
measurements was purchased from Cambridge Isotope Laboratories, Inc. The photosensitive
surfactant, 4-ethyl-4’(trimethylamino-butoxy) azobenzene bromide (azoTAB, Mw = 420 g/mol),
was synthesized according to published procedures.&&When desired, an 84 W long wave (365 nm)
UV lamp (Spectroline, Model no. XX-15A) was used to convert azoTAB to the cis state prior to
mixing with the protein. During neuron spin echo experiments, a 200 W Mercury arc lamp (Oriel,
6283) was used to maintain azoTAB in cis state by continuously illuminating UV light onto the
samples during the experiment (~24 h). UV light was obtained by filtering the light from a mercury
arc lamp using both an IR light and heat absorbing filter (Newport Corporation, FSQ-KG3) and a
320 nm bandpass filter (Newport Corporation, FSQ-UG5). In order to convert azoTAB back to
trans state, a 400 nm longpass filter (Newport Corporation, FSQ-GG400) was used in place of the
320 nm bandpass filter. The filtered light was focused by a fiber-bundle focusing assembly (Oriel,
77800) and transported to the sample through a light guide (Oriel, 77557).
3.3.2 Neutron Spin Echo Spectroscopy
The neutron spin echo experiments were conducted on the NG-5 neutron spin echo
spectrometer at NIST.
60
The neutron wavelength was 𝜆 = 8 Å. The data were collected over the
range of 0.046 Å
H<
≤ 𝑄 ≤ 0.246 Å
H<
. The path length of the sample cell was 4 mm. The NSE
experiment was conducted at 25 °C. The ribonuclease A-azoTAB mixture was made in buffered
D2O solution (8.3 mM sodium phosphate, pH 7.2) at 25 °C. The final concentration of protein was
10 mg/mL and concentrations of azoTAB from 4 mM to 10 mM were employed. To obtain
normalized intermediate scattering functions, 𝐼(𝑄,𝑡) 𝐼(𝑄,0) ⁄ , the data of empty cell, the solvent
(D2O buffer) only, and the instrumental resolution were collected at different Q. The DAVE
36
software was the used to reduce and analyze the data. &&The effective diffusion coefficient of the
protein was obtained by fitting the data to a single-exponential decay as shown in the equation

8(5,!)
8(5,()
= 𝑒𝑥𝑝e−𝐷
:;;
(𝑄)𝑄
"
𝑡f. 3-1
In order to investigate the dynamics of the protein, three models were employed to examine
the Q dependence of Deff, including a rigid-body model,
102
a soft-linker model,
102
and a freely-
jointed model.
107
The schematic diagram of the three models are shown below in Figure 3-1. By
comparing these model results with the effective diffusion coefficient measurement obtained by
NSE spectroscopy, the best model fit can be used to gain insight in the enzyme internal motions
that accompany enhancements in activity.

Figure 3-1: Schematic diagram of three models for protein internal motions: (a) a rigid-body
model; (b) a soft-linker model; and (c) a freely-jointed model.

3.3.3 Rigid-body model
The rigid-body model treats the protein as a rigid entity and does not consider any interactions
between subunits, as shown in Figure 3-1a. Thus, this model only examines the translational and
rotational diffusion of the molecule as a whole. These rigid-body calculations were applied to both
37
the protein crystal structure (PDB ID 3rn3) and the solution structure of the protein with PDB files
generated from shape reconstruction of SANS data. In the rigid-body model, the Q-dependent
diffusion coefficient 𝐷
:;;
(𝑄) is calculated by the following equation
102

𝐷
:;;
(𝑄) =
*
"
+
5
'
∑ 〈K
3
K
4
L5)
5
5MN(O))
(
N(@)P:
6&72
3
% 2
4
9
〉
34
∑ 〈K
3
K
4
:
6&72
3
% 2
4
9
〉
34
, 3-2

where bj and bl are the neutron scattering lengths of effective scattering centers j and l, respectively.
The sum was taken over effective residues j and l, with the center of each effective residue taken
as the average coordinate of the atoms in the effective residue and with the neutron scattering
length b of the effective residual being the sum of neutron scattering lengths of all atoms in a
residue. For the solution structure of Ribonuclease A obtained by GA_STRUCT analysis of the
SANS data, b values were assumed to be identical for all scattering centers. 𝐿(𝑗) = 𝑄 × 𝑟
O
is the
angular momentum vector, and 𝐻
+
and 𝐻
R
are the translational and rotational mobility tensor,
respectively. The three principal-axis translational diffusion coefficients 𝐷
S
+
, 𝐷
T
+
and 𝐷
U
+
in 𝐻
+
,
and the three principal-axis rotational diffusion coefficients 𝐷
S
R
, 𝐷
T
R
, and 𝐷
G
R
in 𝐻
R
were obtained
from the software HYDROPRO.
108,109

3.3.4 Soft-linker model
The soft-linker model, in contrast to the rigid-body model, systematically includes internal
normal modes.
102
In the soft-linker model, the protein is treated as being composed of N rigid
domains, whose coordinates vary little from the crystal structures. These domains are connected
by “soft-linkers” which are much like springs, as shown in Figure 3-1b. The Q-dependent diffusion
coefficient 𝐷
:;;
(𝑄) was calculated from the following equation
102

38
𝐷
:;;
(𝑄) =
∑ .
6
=
6
:
6;<
(5)
=
(:-=> ?
(5)
, 3-3
where 𝐷
>
=
𝑘
/
𝑇
𝜉
>
9 with 𝜉
>
being the friction constant of the ith domain, and i = 1,2, …, N. 𝑆
>
(𝑄)
and 𝑆
RV?W:X
(𝑄) are the rotationally-averaged static form factors, which can be estimated from the
coordinate information obtained from SANS and calculated by
𝑆
>
(𝑄) = ∑ 𝑏
Y
𝑏
%
012[5|6
,
H 6
@
|]
5|6
,
H 6
@
|
Y,% ∈ >
3-4
and

𝑆
RV?W:X
(𝑄) = ∑ 𝑏
Y
𝑏
%
012[5|6
,
H 6
@
|]
5|6
,
H 6
@
|
V
<
MV
'
M⋯MV
:
Y,%_<
, 3-5

where bm and bn are neutron scattering length of residual m and n in the ith domain, rm and rn are
the coordinates of residual m and n, and |𝑟
Y
− 𝑟
%
| is the distance between residual m and n. In
equation 1-5, the sum is taken over the Ni atoms in domain i, and Ni is the number of atoms in
domain i. Therefore, 𝑁
<
+ 𝑁
"
+⋯+ 𝑁
V
is the total number of scattering centers in ribonuclease
A.
Compared with the friction constant 𝜉
>
of each individual domain calculated by the
Kirkwood-Riseman formula,
110
it is necessary to increase this value by a factor of ~1.25 because
of the fluid displaced between each domain by their relative motions. Similarly, Bu et at
102
found
it necessary to increase the friction constants by a factor of ~2. Besides, the number of domains,
N, is the key parameter that will affect the values of Q-dependent effective diffusion coefficients.
Since domain motions have been considered in this model, the smallest value for N is 2. From the
shape reconstruction results obtained from the SANS experiments, it is clear that there are 6
domains at most when Ribonuclease A was partially unfolded in the presence of azoTAB in the
39
solution. Therefore, the valid value of N could be any integer from 2 to 6. In the sections below,
the influence of number of domains on the effective diffusion coefficients will be discussed.

3.3.5 Freely-jointed model
In the freely-jointed model, the protein is treated as a “necklace” with the subunits acting
as “beads” connected by freely-jointed bonds of constant length,
107
as shown in Figure 3-1c.
Therefore, both hydrodynamic interactions between two different subunits and the bond length
constraints are considered in this model. All the subunits are treated as the same, so the detailed
conformational information of the protein is not necessary here unlike the previous two models.
Ribonuclease A was considered as N + 1 subunits connected by N bonds and b was the average
fixed bond length between two subunits. The Deff(Q) values were calculated from the equation
107

𝐷
:;;
(𝑄) = 𝑘
/
𝑇
`(5)
=(5)
, 3-6
where 𝑘
/
𝑇 is the thermal energy, and 𝑆(𝑄) is the static structure factor calculated by
𝑆(𝑄) = (𝑁+1){1+2
V
VM<
O
!
(5K)
!
(5K)
r1−
O
!
(5K)
!
(5K)
!
(5K)
:
V
s}, 3-7
where 𝑗
(
is the zeroth order spherical Bessel function. Also, 𝜇(𝑄) is the mobility tensor given by
𝜇(𝑄) = 𝜇
<
(𝑄)− 𝜇
"
(𝑄), 3-8

𝜇
<
(𝑄) = 𝑗
(
(𝑄𝑏)
|`H a|
𝑯
`a
, 3-9
and
𝜇
"
(𝑄) =
b
"K
'
([𝑯𝑪]
`O
×x𝑒𝑥𝑝N𝑖𝑸∙𝑹
`a
QN𝑸
|
∙𝒃
O
Q
"
~[𝑪
𝑻
𝑯]
Oa
, 3-10
where summation over 𝜇,𝜈 = 0,1,…,𝑁 and 𝑗 = 1,…,𝑁 is implied. Furthermore,
〈𝑒𝑥𝑝N𝑖𝑸∙𝑹
`a
QN𝑸
|
∙𝒃
O
Q
"
〉 = (𝑏
"
/3)𝑗
(
(𝑄𝑏)
|`H a|
3-11
when 𝑹
`a
does not contain 𝒃
O
or −𝒃
O
, or conversely [make this a proper equation by adding the
angular bracket term to the left-hand side]
40
= 𝑏
"
𝑗
(
(𝑄𝑏)
(|`H a|H<)
[𝑗
(
(𝑄𝑏)−2𝑗
<
(𝑄𝑏)/𝑄𝑏] 3-12
otherwise. Finally, 𝑪 is an (𝑁+1) × 𝑁 matrix defined by
𝐶
`O
= 𝛿
`O
− 𝛿
`,OH<
. 3-13
In these expressions, the tensor 𝑯
𝝁𝝂
represents the hydrodynamic interactions among subunits,
which in the freely-jointed case is given by
𝐻
`a
= 𝜉
H<
e𝛿
`a
+ N1− 𝛿
`a
Q𝜏(2/𝜋)
∫ 𝑑𝑥𝑗
(
(𝑥)
|`H a|
g
(
f, 3-14
where 𝑗
(
(𝑥) =
sin𝑥
𝑥
9
and 𝜏 =
𝜉
6𝜋𝜂𝑏
9 , 𝜉 is the friction coefficient of subunit, and 𝜂 is the
viscosity of solvent. In order to calculate the friction coefficient of a subunit, 𝜉 is given by
𝜉 = 6𝜋𝜂𝑅 3-15
where R is the radius of one subunit. Therefore, 𝜏 is the ratio of subunit radius to the fixed bond
length, namely
𝜏 =
𝑅
𝑏
9 . 3-16
The freely-jointed model by Akcasu et al.
107
has taken both bond length constraints and
hydrodynamic interaction into account, and the model is suitable for an arbitrary chain length. The
dynamics of a freely-jointed chain, including both internal and external motions, is completely
described by these equations above.
107
Since the relatively large values of Q employed here
correspond to relatively small length scales (i.e., Q = 2p/L), neutron spin echo is able to observe
the separate motions of the individual protein subdomains (unlike DLS, where the length scale is
so large so as to render the protein as a single dot).
107

In order to better understand the freely-jointed model and improve the agreement between
theory and neutron spin echo experimental results, a systematic analysis was done by changing the
values of the bond length b, the ratio of subunit radius over bond length 𝜏, and the number of bonds
N, which also represents the number of subdomains (N + 1). The value of b, which is the average
bond length between all domains, can be estimated from Kirkwood theory
19,111
using small angle
neutron scattering results, and compared to the dynamic light scattering results. From both
41
experiments and theory, a reasonable range of b values from 30 Å to 50 Å can be assumed. For the
value of 𝜏, the upper limit is 0.5, corresponding to a bond length of twice the radius of the
subdomains (i.e., the subdomains are directly adjacent to each other, corresponding to a strong
hydrodynamic interaction between each subdomain).
107
The lower limit for 𝜏, in theory, can be 0,
meaning that the distance between two subdomains approaches infinity (i.e., no hydrodynamic
interaction between the domains).
107
And as for the value of N, it can be any integer between 1 and
5, which means that there are at least two, and at most six, subdomains. When N equals to 1, for
example, the protein appears as a dumbbell. For the values of Q, 18 data points were selected
uniformly from 0.05 Å
H<
to 0.25 Å
H<
, which is consistent with the NSE data. Both 2D and 3D
figures were generated to illustrate the relationship between these parameters and the effective
diffusion coefficients of the protein as a function of Q using Matlab by employing the equations
listed above. Data and figures are shown in the Results and Discussion section below.
When comparing with experiment data measured by neutron spin echo, results calculated
from the freely-jointed model were multiplied by a factor of 1.25, accounting for an approximately
5 Å thick hydration layer around each domain.
105

3.3.6 Kirkwood formula
Kirkwood theory
19,110
allows calculation of the hydrodynamic radius of a protein by
treating it as a chain of n units, information of which are obtained through SANS. The diffusion
coefficient obtained by Kirkwood theory is the mean translational diffusion constant and is given
by
𝐷

= 𝑘
/
𝑇r
<
%b
+
<
, $%
'
-
∑
<
R
4=

%
@,W_<
s , 3-17
42
where n is the number of structural units. 𝑹
@
and 𝑹
W
are vectors which specify the positions of two
structure units l and s in 3-dimensional space and 𝑅
@W
is the distance between them, therefore the
summations go over from 𝑙,𝑠 = 1,2,…,2000. The information of 𝑹
@
can be obtained from the
PDB files generated by shape-reconstruction of the SANS data. 𝜉 is the friction coefficient of the
structural unit and is given by
𝜉 = 6𝜋𝜂𝑅
Y>%
, 3-18
where 𝑅
Y>%
is the minimum radius of the structure unit of the protein and is estimated by the
following equation

#
h
𝜋𝑅
Y>%
h
×𝑛 =
i
j
∙ l
V
?
3-19
where 𝑉

is the specific volume of the protein, M is the molecular weight of the protein, a NA is
Avogadro’s number.

3.4 Results and discussions
3.4.1 DLS and SANS results
Besides neutron spin echo, several other techniques were previously used to examine the
hydrodynamic radius and radius of gyration of Ribonuclease A by a previous student from our
group, including dynamic light scattering, SANS and Guinier. His data is listed below in order to
better compare with data from NSE and the three models. Combined together, they can provide
insight into the dynamics of Ribonuclease A in the presence of azoTAB in both trans and cis form.
43

Figure 3-2: The hydrodynamic radius of RNase A measured by DLS. [RNase A] = 0.93 mg/mL.

Table 3-1: Values of the radius of gyration (Rg) and I(0) determined from Guinier and PDDF
analyses of the SANS data, as well as hydrodynamic radii (RH) calculated from the shape
reconstructions using Kirkwood’s theory.

As seen in Table 3-1, for the case of 0.9 mg/mL RNase A under visible light, both the radius of
gyration (Rg) and I(0) values increase with increasing azoTAB concentration from 13.28 Å to
20.23 Å and from 0.0115 cm
-1
to 0.4367 cm
-1
, respectively. Under visible light, progressive protein
swelling occurs with increasing surfactant concentration. However, upon UV light illumination
1.9
2
2.1
2.2
2.3
2.4
2.5
0 2 4 6 8 10
visible
UV
44
using up to 6.12 mM azoTAB, the values obtained for Rg and I(0) are consistent with those of the
pure protein (13.46 ± 0.26 Å and 0.0138 ± 0.0027 cm
-1
, respectively), indicating that RNase A
exhibits a native structure under UV light illumination. Even at the highest surfactant concentration
(8.3 mM) under UV light, the protein is significantly refolded (i.e., 15.04 Å versus 20.23 Å under
visible light), although not quite back to the native state.
Similar results are observed for the case of 10 mg/mL RNase A. Again, under visible light, Rg and
I(0) increase with increasing azoTAB concentration from 14.82 Å to 18.33 Å and from 0.106 cm
-
1
to 0.404 cm
-1
, respectively. At 10 mg/mL protein, the value of I(0) is about an order of magnitude
higher than that of the pure protein at 0.9 mg/mL, which is as expected. However, in the presence
of 8.33 mM trans azoTAB, the I(0) values are approximately the same for both protein
concentrations, implying that conformational changes only depend on the surfactant concentration
and not the protein concentration. These results indicate that some sort of surfactant saturation
level is reached. Similar effects were observed for lysozyme; namely, once a concentration of 12
mM is reached no further conformational changes were observed.
45

Figure 3-3. SANS data of RNase A-azoTAB solutions as a function of surfactant concentration
and isomeric state. (a) [RNase A] = 0.9 mg/mL (b) [RNase A] = 10 mg/mL.
(a)
0.01 0.1
pure RNase-A
2.17 mM visible
2.17 mM UV
4.56 mM visible
4.56 mM UV
6.12 mM visible
6.12 mM UV
8.33 mM visible
8.33 mM UV
10
-3
10
-2
10
-1
Q / Å
-1
0.01 0.1
pure RNaseA 10mg/mL
2.01mM visible
4.16 mM visible
6.21 mM visible
8.13 mM visible
2.01 UV
4.16 mM UV
8.13 mM UV
0.001
0.01
0.1
Q / Å
-1
(b)
46

Figure 3-4. Pair-distance distribution functions of RNase A-azoTAB solutions as a function of
surfactant concentration and light conditions. (a) [RNase A] = 0.9 mg/mL (b) [RNase A] = 10
mg/mL.
0 10
0
5 10
-4
1 10
-3
1.5 10
-3
2 10
-3
2.5 10
-3
3 10
-3
3.5 10
-3
4 10
-3
0 8 16 24 32 40 48 56
pure RNase A 10mg/mL
2.01 mM visible
2.01 mM UV
4.16 mM visible
4.16 mM UV
6.21 mM visible
8.33 mM visible
8.33 mM UV
r / Å
0 10
0
5 10
-4
1 10
-3
1.5 10
-3
2 10
-3
2.5 10
-3
3 10
-3
3.5 10
-3
0 8 16 24 32 40 48 56
pure RNase A 0.9mg/mL
2.17 mM visible
2.17 mM UV
4.56 mM visible
4.56 mM UV
6.12 mM visible
6.12 mM UV
8.33 mM visible
8.33 mM UV
r / Å
(b)
(a)
47

Figure 3-5. In vitro conformations of RNase A in the presence of various azoTAB concentrations
in the trans (visible light) or cis (UV light) isomeric state, obtained from shape reconstruction
analysis of the SANS data (best fit shown in blue, consensus envelopes representing the average
of 10 runs shown in red), compared with the X-ray crystallographic structure of RNase A (PDB
1RBX).
112
[RNase A] = 10 mg/mL.

The SANS data were also analyzed by calculation of pair-distance distribution functions
(PDDFs) that represent the probability, P(r), of finding two scattering centers within the protein
(i.e., atomic nuclei) a distance r apart, providing detailed information about protein conformation.
The PDDF results are displayed in Figure 3-4. With this technique the radius of gyration as well
as the maximum dimension within the protein Dmax can be obtained and, thus, it can be determined
whether azoTAB-RNase A complexes adopt a globular structure under a given experimental
conditions. As seen in Figure 3-4, the most common dimension within the protein (i.e., the r value
where P(r) is a maximum) slightly increases with addition of azoTAB, indicating protein swelling.
However, the probability, P(r), of this dimension (the height of the maximum peak), and also the
maximum dimension within the protein, Dmax (i.e., the r value where P(r) returns to zero), both
remain essentially constant. This indicates that the protein is indeed swelling (i.e., remaining
primarily globular) rather than unfolding and becoming elongated. The PDDF results for the two
48
different protein concentrations are quite comparable with only a slight difference in the rate of
increase in r. At 0.9 mg/mL protein under visible light, there is a marked transition between 2.17
mM to 4.56 mM azoTAB, following by steady increases with further increases in surfactant
concentration, which is probably due to the surfactant saturation limit. However, in both protein
concentrations under UV illumination, the obtained PDDFs largely agree with that of the pure
protein (with the exception of 8.3 mM), with Dmax remainsing unchanged. This result supports the
idea that trans azoTAB induces progressive protein swelling, while under UV light protein
swelling is reversed to a compact native structure.

3.4.2 Kirkwood results

The hydrodynamic radii of the protein using Kirkwood formula were calculated and the
result are shown in Figure 3-6. The PDB data files used here are from shape reconstruction of the
SANS data. The hydrodynamic radii calculated using Kirkwood formula agrees with the DLS and
SANS results shown above in Table 3-1. The hydrodynamic radius of ribonuclease A is shown to
expand progressively with increasing concentration of trans azoTAB. The hydrodynamic radius
of the native protein is 14.47 Å, while after interacting with trans azoTAB, the hydrodynamic
radius increases to 15.74 Å, 16.48 Å, 18.06 Å, and 19.05 Å when the azoTAB concentration is 2
mM, 4 mM, 6 mM and 8 mM respectively. These results indicate that ribonuclease A has
undergone a swelling progress in the presence of trans azoTAB, which matches with the results
from DLS and SANA as shown above. Upon illumination of UV light, however, the hydrodynamic
radius of protein stays very close to that of native protein, giving 14.39 Å and 14.15 Å (comparing
to 14.47 Å for the native protein) when the azoTAB concentration is 2 mM and 4 mM. respectively.
49
When the concentration of azoTAB is increased to 8 mM under UV light, the hydrodynamic radius
does increase slightly to 15.71 Å.

Figure 3-6. The hydrodynamic radius of ribonuclease A as a function of azoTAB concentrations
and light conditions. In the plot, the blue symbols (l) represents the hydrodynamic radius of
pure ribonuclease A, the black symbols (n) represents the hydrodynamic radius of ribonuclease
A in the presence of trans azoTAB, and the red symbols (u) represents the hydrodynamic radius
of ribonuclease A in the presence of cis azoTAB. [RNase A] = 10 mg/mL.

3.4.3 NSE results compared with results from rigid-body model and Kirkwood
formula

Neutron spin echo spectroscopy was performed to examine the effective diffusion
coefficients of ribonuclease A in presence of azoTAB under both visible and UV light illumination,
50
as shown in Figure 3-7. The concentration of ribonuclease A used in here is 10 mg/mL. For
comparison, the effective diffusion coefficients measured at Q = 0.05 Å
-1
can be considered as
essentially the translational diffusion coefficients of the protein. Together with these NSE result,
effective diffusion coefficients calculated using Kirkwood formula (i.e., the blue triangle data
points at Q = 0), and diffusion coefficient values calculated from rigid-body model in the whole Q
range (red curves), are compared.
51

52
Figure 3-7. NSE data of RNase A effective diffusion coefficient with different azoTAB
concentrations and light conditions. In each plot, the black symbols (n) represent data obtained
from NSE spectroscopy, the blue symbols (▲) represent the diffusion coefficient calculated by
Kirkwood formula, and the solid red lines (—) represent rigid body calculations.
The intramolecular interference of a rigid body is known to cause oscillations of the
diffusion coefficient as a function of Q, therefore, the contributions of translational and rotational
diffusion to Deff(Q) was calculated.
102,113
These rigid-body calculations for ribonuclease A
determined by shape reconstruction of the SANS data are plotted as solid red lines at each
concentration of azoTAB. The fact that rigid-body calculations for pure ribonuclease A from both
crystal and the solution structures are similar and agree with the experiment data confirms that the
native protein behaves as a rigid body, exhibiting little measurable nanosecond internal motions.
In the presence of azoTAB, however, the NSE data shown in Figure 3-7 exhibit different
degrees of variation from the rigid-body calculations at higher Q values, implying that the protein
undergoes dynamic motions in addition to simple intramolecular translational and rotational
motions. Furthermore, the pure Ribonuclease A data starts to show deviation from the rigid-body
calculations when Q values is about 0.2 Å
-1
, while in the presence of trans azoTAB deviations
from rigid-body calculations occur at lower values of Q of about 0.15 Å
-1
. Furthermore, the degree
of deviation from the rigid-body calculations progressively increases with increasing azoTAB
concentration, indicating that the protein is undergoing increasing internal motions. In the presence
of cis azoTAB, however, the degree of deviation from the rigid-body calculation is only slightly
greater than that of the pure protein. This indicates that ribonuclease A is more like a native protein
in the presence of small concentration of cis azoTAB, and does not start to swell until at a higher
concentration of cis azoTAB (e.g., 8 mM).
The shape reconstruction of SANS has shown that in the presence of trans azoTAB,
ribonuclease A is swollen compared to the pure protein to a progressively greater degree with
53
increasing concentrations of azoTAB, while upon exposure to UV light the protein shape looks
very similar to that of the pure protein. This result matches with NSE data very well.
Since Ribonuclease A exhibits internal motions of different degrees upon interacting with
azoTAB, both a soft-linker and free-jointed models were employed to account for these internal
motions. The influence of the different parameters of each of these two models on the effective
diffusion coefficient will first be discussed, followed by then attempting to fit the experimental
NSE data.
54
3.4.4 Soft-linker model results

55
Figure 3-8. comparison between the measured effective diffusion coefficients using NSE
spectroscopy and the estimated results based on soft linker model. In each plot, the black symbols
(n) represent data obtained from NSE spectroscopy, and the solid red lines (—) represent soft
linker calculations.

Figure 3-8 shows the comparison between the calculated effective diffusion coefficients
based on soft linker model and the measured diffusion coefficients using the NSE spectroscopy.
To obtain the estimated results, the PDB file of the solution structure of pure ribonuclease A and
ribonuclease A-surfactant complex obtained from SANS was used as input in the code. As can be
seen in Figure 3-8, even though the calculated effective diffusion coefficients match with NSE
results when Q is very small (Q = 0.05 Å
-1
), which represents the translational and rotational
diffusion coefficients, the increasing slope of the model results do not match with that of NSE data
points when value of Q rises (0.05 Å
-1
< Q < 0.15 Å
-1
). This may be because in this model, the
“spring-like” interaction between domains is too soft to describe the internal motions of the protein.
Within this range of Q values, the protein is still smaller than the length scale and only translational
and rotational diffusion can be measured, so it is not a surprise if the values are smaller than the
model results which include the “spring-like” internal motions. Furthermore, there is no peak
existed in NSE data points within the overall range of Q, but the calculated effective diffusion
coefficients start to decrease when Q > 0.15 Å
-1
. Besides, in the conditions where azoTAB
concentration is high (6mM and 8 mM), 4 domains and 5 domains assumption were applied in soft
linker model, which has no physical meaning to the protein, since there are 2 domains in
ribonuclease A. Therefore, the soft-linker model is not suitable to describe the internal motion of
Ribonuclease A in the presence of trans and cis azoTAB.

56
3.4.5 Freely-jointed model results
The freely-jointed model was been employed to investigate the internal dynamics of the
protein, where as described above bond length constraints and hydrodynamic interaction between
each two subunits are considered. In the freely-jointed model, detailed structure information about
the protein is not required. There are three parameters considered in this model, which are the bond
length b, the number of bond N (giving N + 1 subdomains), and the ratio of subunit radius over
bond length τ. The primary radius R is determined by the number of subdomains, according to the
equation 𝑀
!
𝑉

= (𝑁+1)
#
h
𝜋𝑅
h
, where 𝑉

is the specific volume of protein(𝑉

= 0.708 𝑚𝑙/𝐿).
After the primary R is obtained, the value of b and the relevant R can be determined by adjusting
the value of τ. Compared with the effective diffusion coefficients measured by NSE spectroscopy,
the estimated effective diffusion coefficients calculated from freely jointed model are shown in
Figure 3-9 below.
57

58
Figure 3-9: comparison between the measured effective diffusion coefficients using NSE
spectroscopy and the estimated results based on freely jointed model. In each plot, the black
symbols (n) represent data obtained from NSE spectroscopy, and the solid red lines (—)
represent freely jointed calculations.
Depending on the values of these three parameters, the effective diffusion coefficients
calculated from the freely-jointed model can vary substantially. As can be seen from the figure
above, when ribonuclease A is in presence of trans azoTAB, the diffusion coefficients calculated
from freely-jointed model do not match with NSE data points very well, which is similar to the
condition where ribonuclease A is modeled using the soft-linker model, and when ribonuclease A
is in presence of cis azoTAB, the matching degree is better than that of trans isomer case. But in
overall, the data points from NSE do not match with freely jointed model results. When in small
Q values range (Q < 0.15 Å
-1
), the estimated diffusion coefficients calculated from freely-jointed
model are in good match with NSE data points, which is not a surprise because when Q is small,
that is when the length scale is big, it makes sense if the protein behaves like “jointed beads”. In
higher Q values range (Q > 0.15 Å
-1
), however, the trend lines from freely jointed model start to
decrease, while the NSE data points rise to even greater values, indicating that when the length
scale becomes smaller, the amplitude of the protein internal motion is more greater than what is
expected in freely-jointed model. And for some cases where azoTAB concentrations are high
(6mM and 8mM), multiple domains are assumed in this model, which, again, does not have any
physical meaning to the protein. Therefore, the dynamics of the ribonuclease A cannot be described
using freely-jointed model no matter the protein is in the presence of trans or cis azoTAB.

3.5 Conclusions
According to NSE spectroscopy, ribonuclease A exhibits different degrees of internal
dynamics in the presence of different concentrations of azoTAB under either visible or UV light.
59
The effective diffusion coefficient measured at low Q matches well with that obtained from the
hydrodynamic radius measured by DLS or calculated using Kirkwood formula. Shape
reconstruction of the SANS data also demonstrates that there is significant swelling to the protein
in the presence of trans azoTAB, while conversely the protein stays more in a nativelike state in
the presence of cis azoTAB. In order to investigate the inner motions of the protein, three models
have been employed and implemented using Matlab. In rigid-body model, the modeling results fit
the NSE data points the best when Q < 0.2 Å
-1
, which is very reasonable. From SANS data, the
diameter of the ribonuclease A is 38 Å (native state) or 50 Å (azoTAB swollen state). When Q <
0.2 Å
-1
, the minimum length scale of valid NSE measurement is
2𝜋
𝑄
9 = 31 Å, which is close to
the size of pure protein, indicating that the protein behaves like a rigid globe in overall. When Q
becomes larger, which means when the length scale is smaller, the NSE data points deviate from
the rigid body model results, indicating that there is internal motion undergoing inside the protein.
Therefore, when Q is greater than 0.2 Å
-1
, the internal motion of the protein is observed besides
translational and rotational diffusion. The soft-linker and freely-jointed models, however, do no
match well with NSE data points in higher Q range (Q > 0.2 Å
-1
) either, therefore the intenal
motions of ribonuclease A cannot be described by these two models.
Additional point to mention is that ribonuclease A is cross-linked by four disulfide bonds,
which involve all eight of its cysteine residues, as shown in Figure 3-10 below
17
. These four
disulfide bonds in ribonuclease A are critical to the stability of the native protein. Therefore, the
protein does not dissociate into subdomains even when it is in presence of high concentration of
trans azoTAB, let alone cis isomers. Instead, the protein is just swelled in presence of azoTAB,
beside of which the internal motion of the protein is enhanced and observed through the NSE
spectroscopy.
60

Figure 3-10: Ribbon diagram of the three-dimensional structure of ribonuclease A [PDB : 3rn3].
The inscriptions refer to the location of the eight cysteine residues that give rise to the four
disulfide bonds.
17

61
4 Reveal Nanoscale Protein Motion of Bovine Serum Albumin by Neutron Spin
Echo Spectroscopy
4.1 Introduction
Proteins are undergoing different kinds of motions all the time. Through protein dynamics
proteins are able to adapt different conformations in order to bind and release various of ligands
or to form signaling complexes
114,115
. Therefore, protein dynamics is essential and important for
protein functions
116
. Tons of studies have shown that protein dynamics are of various levels,
occurring on timescales ranging form femtoseconds to longer than seconds, and on length scales
ranging from angstroms to micrometers
99,116
. There are various of techniques to characterize
protein dynamics in different time and length scales. Neutron spin echo spectroscopy (NSE) is a
unique technique among them because it can determine nanoscale dynamic motions.
NSE, one of the quasi-elastic neutron scattering techniques, is substantially different from the
conventional inelastic neutron scattering techniques. In fact, it allows for the detection of small
velocity changes of the neutrons during the scattering process. By measuring the incident and the
scattering neutrons velocities, NSE determines the energy change between neutrons and molecules
with motions, thus obtaining the information of molecule dynamics
117
. The trick is to employ the
Larmor precession of neutron spins in a magnetic guide field as an “internal” clock, so that
extremely small changes in velocities of scattering neutrons are able to be detected
118
. NSE
spectroscopy is the best technology to investigate the thermal fluctuations of molecules and in fact,
it allows the investigation of length scale up to the tenths of nanometer and relaxation times from
the fraction of nanosecond to one hundred nanoseconds.
However, not all proteins are qualified to measure the internal motions by NSE. In order to
observe the internal motions successfully, the proteins have to meet a few criteria according to
62
Biehl, R et al. One is that the protein has well separated domains because greater scale thermal
fluctuations are easier to be detected by the NSE spectroscopy
119
. For smaller proteins, the
amplitude of domain motions are on a smaller scale and move out of accessible length scale range.
Another key point is that a high purity protein solution must be guaranteed and contaminations
from other proteins or fragments must be excluded during the measurement. In another word, the
proteins must have a defined state during the measurement. The aggregation process during the
beam time is not accessible.
Bovine serum albumin (BSA) can be a good candidate protein to study the protein internal
motion using the NSE spectroscopy. The BSA is a relatively big protein with a molecular weight
of 66.5 kDa consisting of three domains (I, II and III) in the primary structure, and each domain
contains two subdomains. Held together by 17 disulfide bonds BSA consists of nine loops, so that
alternatively each domain can also be seen as containing one small and two large loops
120–122
.
According to Lee, T. et al, the BSA exhibits a “heart-shaped” structure in solution, similar to that
obtained from X-ray crystallography
3,123
. Two additional conformations BSA in solution are also
observed with decreasing pH. Beginning with the “heart shape” normal “N” form (triangular shape
with 80 Å edges and 30 Å thick) at neutral pH, BSA changes into the fast “F” form (40 × 129 Å)
below a pH of about 4, and concludes with the expanded “E” form (21×250Å) for pH less than
about 3
120,122
. This provides a wide possible range of length scale for NSE spectroscopy.
According to Lee, T. et al, the conformation change of BSA can also be achieved through the
use of a light-responsive azobenzene surfactant, which undergoes a reversible trans « cis
photoisomerization upon exposure to visible or UV light
1,2
. The BSA exhibits different
conformations in the presence of different surfactant concentrations and light illumination
3
. Since
the dipole moment is larger across the azo group in the nonplanar form, the trans (planar) form of
63
the surfactant is more hydrophobic compared to the cis (bent) form
2
, resulting in stronger
hydrophobic interactions between the protein and surfactant. As a consequence, the hydrophobic
interactions between the protein and surfactant can be controlled by light illumination and the
conformation change of the protein can be photo-reversible
3
. The structure change of BSA in the
presence of various azoTAB concentrations and light illumination has been explored using
dynamic light scattering and small angel neutron scattering technology
3
, but the protein dynamics
in the presence of azoTAB remains unknown and curious. In the present work, the internal domain
motions are explored by the NSE spectroscopy.
4.2 Materials and methods
4.2.1 Materials
An azobenzene trimethylammonium bromide surfactant (azoTAB) was synthesized
according to published procedures
1,2
, with a purity of 99% as determined by gas chromatography
and NMR. In order to convert the trans form to the cis form of azoTAB, the 365 nm UV line from
a 200 W Mercury arc lamp (Oriel, model no. 66942) isolated using a 320 nm band-pass filter (Oriel,
model no. 59800) was used to illuminate the solution. In order to convert from the cis form back
to trans form, a 400 nm long-pass filter (Oriel, model no. 59472) was used instead. In both cases,
a heat-absorbing filter (Oriel, model no. 59042) was placed in the beam path to absorb the IR light
produced by the lamp. It should be noted that absorption measurements indicate that under visible-
light conditions the surfactant exhibits an approximately 75:25 trans/cis equilibrium, while under
UV light, the surfactant is primarily of the cis. Throughout this paper, the terms “trans form” and
“cis form” indicate that the surfactant is primarily trans or cis, respectively.
Highest quality lyophilizate BSA, guaranteed monomeric for 1 year as a solid and for 6
months in solution, was purchased from Roche and used as received. The low ionic strength
64
phosphate buffer (pH 7.2, 8.3 mM) was obtained from Sigma. All other chemicals were obtained
from Aldrich. A protein concentration of 10 mg/mL was employed in the neutron scattering
experiments, while a concentration of 0.66 mg/mL was used for the dynamic light scattering.

4.2.2 Light-Scattering Measurements
DLS measurements at 25 °C were performed at an angle of 90° on a Brookhaven model
BI-200SM instrument (Brookhaven Instrument Co.) with an argon ion laser operating at 514 nm.
A relatively low laser power (<100 mW) was used to avoid conversion of the azoTAB cis isomer
into the trans form during the course of the experiments. The data were analyzed with both the
NNLS and CONTIN routines (difference < 2 Å) using a BI-9000AT digital correlator (Brookhaven
Instrument Corp.). The protein-buffer solutions were passed through a 0.2 μm filter prior to the
measurements, while the surfactant was added after filtration to avoid loss of the surfactant through
adsorption on the membrane. A lower protein concentration (10-5 M or 0.66 mg/mL) was used in
the light- scattering experiments to measure the infinite dilution diffusion coefficient of the protein
more accurately.
4.2.3 Neutron spin echo spectroscopy
The neutron spin echo experiments were conducted on the NG-5 neutron spin echo
spectrometer at NIST
60
. The neutron wavelength was 𝜆 = 8 Å. The data were collected over the
range of 0.046 Å
H<
≤ 𝑄 ≤ 0.246 Å
H<
. The path length of the sample cell was 4 mm. The NSE
experiment was conducted at 25 °C. The BSA-azoTAB mixture was made in buffered D2O
solution (8.3 mM sodium phosphate, pH 7.2) at 25 °C. The final concentration of protein was 10
mg/mL and concentrations of azoTAB is 2mM. To obtain normalized intermediate scattering
functions, 𝐼(𝑄,𝑡) 𝐼(𝑄,0) ⁄ , the data of empty cell, the solvent (D2O buffer) only, and the
65
instrumental resolution were collected at different Q. The DAVE software was the used to reduce
and analyze the data. The effective diffusion coefficient of the protein was obtained by fitting the
data to a single-exponential decay as shown in the equation
𝐼(𝑄,𝑡)
𝐼(𝑄,0)
= 𝑒𝑥𝑝e−𝐷
:;;
(𝑄)𝑄
"
𝑡f

Rigid-body model
The rigid-body model is employed is investigate the internal motion of BSA in the presence of
azoTAB under visible or UV light. The rigid-body model is originally developed to describe the
dynamics of random coil polymers
107
, but the one used here is generalized to include both
translational and rotational motions. These rigid-body calculations were applied to the solution
structure of the protein with PDB files generated from shape reconstruction of SANS data. In the
rigid-body model, the Q-dependent diffusion coefficient 𝐷
:;;
(𝑄) is calculated by the following
equation
𝐷
:;;
(𝑄) =
*
"
+
5
'
∑ 〈K
3
K
4
L5)
5
5MN(O))
(
N(@)P:
6&72
3
% 2
4
9
〉
34
∑ 〈K
3
K
4
:
6&72
3
% 2
4
9
〉
34
,
where bj and bl are the neutron scattering lengths of effective scattering centers j and l, respectively.
The sum was taken over effective residues j and l, with the center of each effective residue taken
as the average coordinate of the atoms in the effective residue and with the neutron scattering
length b of the effective residual being the sum of neutron scattering lengths of all atoms in a
residue. For the solution structure of BSA obtained by GA_STRUCT analysis of the SANS data,
b values were assumed to be identical for all scattering centers. 𝐿(𝑗) = 𝑄 × 𝑟
O
is the angular
momentum vector, and 𝐻
+
and 𝐻
R
are the translational and rotational mobility tensor, respectively.
The three principal-axis translational diffusion coefficients 𝐷
S
+
, 𝐷
T
+
and 𝐷
U
+
in 𝐻
+
, and the three
66
principal-axis rotational diffusion coefficients 𝐷
S
R
, 𝐷
T
R
, and 𝐷
G
R
in 𝐻
R
were obtained from the
software HYDROPRO
108,109
.
4.3 Results and discussion
Before NSE experiment, DLS, SANS and shape reconstruction have been done to
investigate the conformation change of BSA in the presence of various concentrations of azoTAB
and under visible or UV light. The normal “N” form, or the “heart shape” of the protein was
observed in the pure BSA solution. With increased azoTAB concentrations, a decrease in diffusion
coefficient was observed from dynamic light scattering measurement, which indicates an increase
of the overall size of the complex. Especially, the trans form surfactant results in a more rapid
increase in size than does the cis form
111
, that is, the trans form surfactant can cause a larger degree
of protein unfolding.
For illustrative purpose, estimated diffusion coefficients for different potential
conformations of BSA are also shown besides the effective diffusion coefficients measured by
dynamic light scattering. These calculations were based on the theory of Kirkwood
19,111
, assuming
that the protein has a chain of six identical spherical subunits, which is reasonable since BSA is
known to have three domains and each domain contains two subdomains. Each subunit was
assumed to be a compact sphere and the minimum radius of it was calculated based on the subunit
molecular weight (𝑅
Y>%
= 0.66 𝑀𝑊
= 15 Å) with the addition of a 3 Å thick water hydration
shell. When there was pure BSA in solution, the initial protein structure was assumed to be a “heart
shape” triangle. With the addition of low concentration trans azoTAB, the protein structure was
assumed to be a special configuration, which corresponds to the unfolding of the protein. These
estimations and assumptions were verified from shape reconstruction results obtained through
small angel neutron scattering (SANS). GA_STRUCT algorithm was applied to SANS data at
67
corresponded azoTAB concentration and light illumination conditions in order to obtain the
structure information of the protein. The results from shape reconstruction were shown in the left
side of each figure.
Neutron spin echo spectroscopy was performed to examine the effective diffusion
coefficients of pure BSA and BSA-azoTAB complex under visible or UV light. The concentration
of BSA was 10mg/ml. The effective diffusion coefficient measured at 𝑄 = 0.05 Å
H<
can be
considered as the translational diffusion coefficient of the protein, to be compared with the
effective diffusion coefficient measured by DLS. Together with NSE results, rigid-body model
based on the structure information from SANS was also shown and compared.
68

Figure 4-1: (A)(C)(E) the shape reconstruction results from SANS of pure BSA, BSA in
presence of 2mM trans azoTAB and BSA in presence of 2mM cis azoTAB. (B)(D)(F) the
effective diffusion coefficients measured though NSE spectroscopy, which is represented by
black square (n), is compared with the effective diffusion coefficient measured by DLS, which
is represented by a blue diamond (¨), and the calculated results for rigid-body model, which is
represented by a red solid line (—). The estimated diffusion coefficients based on Kirkwood
theory for different protein configurations are also shown for illustration purpose.

Overall, the effective diffusion coefficients calculated from rigid-body model agree with the
effective diffusion coefficients measured from Neutron spin echo spectroscopy for all the three
69
cases, which indicates that the protein behaves like a rigid body in these three conditions. For BSA
alone in solution, the effective diffusion coefficient measured at low Q (Q = 0.05 Å
H<
) using NSE
technology matches with the one measure by DLS, and with increasing Q values, the effective
diffusion coefficients measured by NSE agree with rigid-body model results, which means that
translational and rotational diffusion dominate the protein dynamics and thermal fluctuation is not
obvious. In the presence of trans azoTAB, BSA is unfolding because of the hydrodynamic
interaction between the protein and surfactant. From the shape reconstruction result on the left,
BSA has been elongated and transformed from “N” form to “F” form, and both the effective
diffusion coefficients measured from DLS and the one measured by NSE at Q = 0.05 Å
H<
are
smaller than those of pure protein, verifying that the protein size was enlarged. What’s more, the
effective diffusion coefficient measured by DLS is even smaller than the estimated diffusion
coefficient based on Kirkwood theory, which assumes that the protein is like a chain connecting
six identical subunits in a straight line. This may be because there is small distance between two
adjacent subdomains and the subdomains are connecting loosely to each other rather than tightly
in the estimated Kirkwood model. That the results from rigid-body model agree well with the
effective diffusion coefficients measured by NSE indicates the protein still behaves like a rigid
body in the presence of trans azoTAB, instead of gaining more internal motions. Since cis form
azoTAB is less hydrophobic than trans form azoTAB, the hydrophobic interaction between the
protein and cis form surfactant is weaker than trans form surfactant, thus resulting in a smaller
degree of protein unfolding, which can be seen from the shape reconstruction result. Both the
effective diffusion coefficients measured from DLS and from NSE are greater than those of the
protein in presence of trans azoTAB, which indicates that the protein has a smaller size, that is a
smaller degree of protein unfolding. And the effective diffusion coefficient measured by DLS is
70
close to the estimated value of “straight line” shape model, indicating that the protein unfolds into
the F form but in a smaller degree than the case of protein in presence of trans azoTAB. With
increasing Q values, the effective diffusion coefficients measured by NSE match with the results
from rigid-body model even better, indicating that the protein still behaves like a rigid-body in this
case. Overall, BSA behaves like a rigid body no matter it is alone or in presence of trans or cis
form azoTAB, and translational and rotational diffusion dominate the dynamics of the protein but
in degrees of small difference.

4.4 Conclusion
Even though BSA has met all the criteria to be a good candidate protein for Neutron spin
echo spectroscopy, obvious domain motion was not observed no matter for the pure protein or the
protein surfactant complex. And results from rigid-body model have also verified that translational
and rotational diffusion dominate the dynamics of the protein, since the rigid-body model results
match well with the effective diffusion coefficients measured by NSE. This conclusion is not a
surprise since the protein is unfolded in the presence of both trans form and cis form surfactant
which can be told from SANS results. Rather than being a good candidate protein for Neutron spin
echo spectroscopy to investigate protein dynamics, but a great model protein to examine the
hydrodynamic interaction between the protein and surfactants.

71
5 Controlling Insulin Fibrillation with Light Using Photoresponsive Surfactant
1

5.1 Abstract
Amyloid fibrinogenesis, a process by which proteins self-aggregate into insoluble, fibril-like
structures, has been implicated in the progression of a number of neurodegenerative diseases.
Morphological similarities in the fibrils formed from various non-homologous proteins suggest
that fibrillation occurs via a shared molecular mechanism. Insulin is one such protein, which
shows a tendency to form fibrils under certain conditions. Existing as a hexamer in the native state
at pH 7, the dimeric form of insulin and a small fraction of aggregation-prone monomers, however,
becomes the predominant species at low pH (~2), leading to significant fibrillation. In this study,
the various intermediates that are formed during insulin fibrillization will be identified and
characterized with different techniques, including dynamic light scattering and small angel neutron
scattering. The photosensitive surfactant azoTAB (azobenzene trimethylammonium bromide)
undergoes a reversible trans « cis photoisomerization upon exposure to visible or UV light. Under
visible light, the relatively hydrophobic trans isomer shows a greater tendency to bind with
proteins compared to the relatively hydrophilic cis isomer predominant under UV light. Thus, it is
possible to photo-reversibly control protein-surfactant interactions, thereby controlling protein
conformation and aggregation with light exposure. Herein we apply this property to control the
amyloid fibrillation pathway of insulin. Early pre-fibrillar oligomeric species formed in the
presence of trans azoTAB are shown to have a greater tendency towards lateral aggregation than
those formed with pure protein, leading to enhanced fibrillation rates in the presence of trans
azoTAB. In the presence of cis azoTAB, however, fibrillization is largely inhibited. This provides

1
Portions of this chapter previously appeared in the dissertation of Khiza Mazwi.

72
for a novel method to halt, and potentially reverse, aggregation, with light illumination.
Furthermore, the “point of no return” (e.g., dodecamer formation), beyond which aggregation can
no longer be halted, will be identified.
5.2 Introduction
Insulin is a peptide hormone that controls the uptake of glucose in the bloodstream by target
cells and plays a vital role in the regulation of the metabolism of fats and carbohydrates.
124
The
insulin monomer is a 51 residue protein with a molecular weight of approximately 5.8 kDa, and is
comprised of two peptide chains (chain A with 21 amino acids and chain B with 30 amino acids
for most species) connected by two disulfide linkages, while a third disulfide bond links two
cysteine residues within chain A.
125
Chain A contains two antiparallel helices (i.e., an α-helix and
a 310-helix) connected by a loop, while chain B contains an α-helix and an extended section
analogous to a β-strand near the C-terminus. Hydrogen bonds between this later C-terminal region
lead to head-to-tail dimer formation through intermolecular β-sheet formation in solutions between
pH = 2 – 8.
126
The dimer interface consists of six aromatic side chains (two tyrosine and one
phenylalanine residue from each monomer) that have been shown to pack in a quasi-crystalline
arrangement.
127

In the presence of zinc or other divalent metallic ions between pH 5 and pH 8, these dimers
further associate into toroid-shaped hexamers containing two metal ions each coordinated with
three B10 histidine residues and three water molecules.
128
When produced in pancreatic islets in
vivo, hexamer formation allows for a more closed-packed arrangement of insulin and lowers the
osmotic impact. In vitro, hexamer formation is necessary to increase the shelf life of the drug since
insulin monomers, which are in rapid equilibrium with dimers,3 readily form insoluble fibers via
amyloid formation.
129,130
However, when insulin is injected subcutaneously, dissociation of the
73
hexamer into dimers/monomers must first occur to facilitate insulin transport from the
subcutaneous tissue to the bloodstream. Specifically, steric hindrance to transport across capillary
membranes leads to an inverse relationship between oligomer size and the subcutaneous
absorption rate, with the hexamer generally consider not to be able to traverse at appreciable
rates.
131
Thus, a delay of up to two hours between injection and peak plasma concentration can
occur,
132
meaning insulin would have to be administered well before eating. In contrast, various
so-called rapid-acting insulin analogues, which are designed to exist as stable monomers/dimers
primarily through modification of the C-terminus of the B chains, have been developed to bypass
this dissociation step and lead to faster transport of insulin into the bloodstream. Despite extensive
studies of both the dissociation of insulin hexamers and the aggregation of monomeric insulin into
fibrils, the instability of the drug during processing, storage, and delivery continues to be a major
challenge,
129
with various additives including traditional surfactants utilized in an attempt to
stabilize insulin solutions from aggregation
125,133
The effects of insulin aggregation can sometimes
be catastrophic, such as in the rare but serious medical condition known as injection amyloidosis,
where insulin fibrils accumulate at the site of frequent injections.
134

Currently there are approximately 30 human proteins that have been shown to deposit as
amyloid fibrils in various diseases,4 including islet amyloid polypeptide (IAPP) that forms
amyloid deposits in the pancreatic islets of patients with diabetes mellitus type 2 and is co-
expressed and co-secreted with insulin. 5,6 Perhaps the most notorious of these diseases are late
onset neuro-degenerative disorders such as Alzheimer’s disease, which is projected to affect as
many as 1 in 85 people globally by 2050.
127
However, there are currently no effective treatment
strategies for these diseases, due at least in part to the fact that the structures of the pre-fibrillar
oligomeric species (and thus the pathogenesis of these disorders) remain largely unknown, a
74
consequence of these early aggregates being ill-suited to structural determination with NMR
(generally too large) or crystallography (prone to uncontrolled aggregation). This is particularly
unfortunate given that these early intermediates have become increasingly viewed as the primary
pathogenic species,
125,126
with synaptic activity shown to be affected by soluble intermediates well
before amyloid aggregation into fibril plaques.
129
Thus, inhibition or interruption of protein
aggregation, particularly early in the process, is considered a possible therapeutic route in treating
these diseases.
In the present study, a photoresponsive surfactant is used to control both the insulin hexamer
® dimer/monomer transition and the monomer ® amyloid aggregation processes. The
photoresponsive surfactant offers the ability to tune with light illumination the interaction of the
surfactant azobenzene group (which adopts a planar trans conformation under visible light and
bent cis structure under UV light) with the various protein moieties responsible for association into
both oligomers and fibrils. Characterization of the prefibrillar intermediates is achieved through
a combination of techniques, including dynamic light scattering (DLS), small angle neutron
scattering (SANS), atomic force microscopy (AFM), and circular dichroism (CD). Notably, SANS
allows the solution structure of early aggregates to be determined. The formation of insulin fibrils
in the presence of the trans surfactant occurs by a nucleation and growth process, similar to that
seen in the pure protein, while aggregation in the presence of the cis isomer occurs much slower
and leads to non-fibrillar, amorphous aggregates. As a result, in situ UV illumination can be used
to halt insulin fibrillization.
75
5.3 Materials and experiment methods
5.3.1 Materials
The surfactant 4-ethyl-4′(trimethylaminobutoxy) azobenzene bromide (azoTAB) shown as
in Figure 5-1 was synthesized according to previously published literatures.
1,5
All chemicals used
in synthesizing azoTAB were purchased from Sigma-Aldrich in the highest purity and used as
received.

Figure 5-1: structure and photoisomerization of azoTAB.

AzoTAB undergoes a reversible trans « cis photoisomerization upon exposure to visible (434
nm) or UV (350 nm) light, respectively. Under visible light, azoTAB primarily adopts a planar
trans form (75/25 trans/cis), which is more hydrophilic compared to the bent cis isomer under UV
light (10/90 trans/cis). The trans isomer can be recovered upon subsequent re-exposure to visible
light, or completely recovered in the dark (100% trans isomers after ~24 h at 25 °C) or at high
temperatures (e.g., ~100% trans isomers after ~5 min at 60 °C). To prepare cis samples, azoTAB
solutions were pre-incubated under an 84 W long wave (365 nm) UV lamp (Spectroline, Model
no. XX-15A) for more than half an hour before mixing with insulin. Furthermore, during the
experiments at high temperature (e.g., 60 °C), it was necessary to continuously illuminate the
samples with a high-intensity UV lamp to avoid thermal conversion back to the trans state. Here
–
CH
3
CH
2
O(CH
2
)
4
N(CH
3
)
3
Br
350 nm
434 nm
–
CH
2
CH
3
O(CH
2
)
4
N(CH
3
)
3
Br
trans isomer
cis isomer
76
a 200 W mercury arc lamp (Oriel, 6283) was used in conjunction with an IR light, heat absorbing
filter (Newport Corporation, FSQ-KG3) and a 320 nm bandpass filter (Newport Corporation, FSQ-
UG5). In order to convert azoTAB back to cis state, a 400 nm longpass filter (Newport Corporation,
FSQ-GG400) was instead used. The filtered light was focused by a fiber-bundle focusing assembly
(Oriel, 77800) and transported to the sample through a light guide (Oriel, 77557).
Insulin from bovine was purchased from Sigma-Aldrich (catalog number I5500). Protein
solutions at 10 mg/mL were freshly prepared for each experiment in aqueous solutions containing
0.1 M NaCl and adjusted to pH 1.6 by addition of HCl.

5.3.2 Dynamic light scattering
DLS measurements were performed on a Brookhaven BI-200SM instrument equipped with
a BI-9000AT digital correlator (Brookhaven Instrument Corporation). The incident beam was a
632.8 nm, 35 mW helium neon laser (Melles Griot, model Number 05-LHP-928). This wavelength
does not affect the surfactant photostationary state. Measurements were performed at a scattering
angle of 90°. Data were analyzed using both the z-averaged effective diameter (i.e., the method of
cumulants) and a non-negative least squares (NNLS) routine. Prior to data collection, the protein
solutions were filtered through a 450 nm Anotop filter to remove large structures that could act as
nucleation centers. As aggregation is a dynamic process, time-dependent DLS analysis was
performed by collecting separate autocorrelation functions approximately every minute during
fibrillation at a relatively high protein concentration (10 mg/mL) to achieve reliable fits of the
autocorrelation function. The temperature was maintained at 60 °C, and the illumination state was
maintained and changed in situ through the use of the light guide. The hydrodynamic diameter,
dH, was calculated assuming a spherical shape, according to the Stokes-Einstein equation, namely
77
𝑑
)
= 𝑘
/
𝑇/3𝜋𝜂𝐷, where 𝑘
/
is Boltzmann’s constant, T, is the temperature, 𝜂 is the solvent
viscosity, and D is the experimentally-determined diffusion coefficient.

5.4 Results and discussion
The fibrillation of pure insulin and insulin in the presence of azoTAB under visible light was
monitored using both dynamic light scattering (DLS) and UV-vis spectroscopy measurements, as
shown in Figure 5-2. The time-dependent hydrodynamic diameters determined by DLS (Figure
5-2a) monitor the formation of relatively-small aggregates that develop early in the fibrillation
process, while the UV-vis measurements (Figure 5-2c) correlate with the formation of relatively-
large species since light scattering becomes appreciable when the particle size approaches about
10% of the wavelength of light (notice the lower time scales in Figure 5-2a relative to Figure 5-2c).
In each case in Figure 5-2a, aggregation appears to start instantly and the apparent sizes begin to
quickly increase and reach a plateau at specific times depending on the system (~12 mins for pure
insulin versus 8 mins and 6 mins when in the presence of 0.5 mM and 1 mM azoTAB under visible
light respectively. The hydrodynamic diameter started to increase from ~5 nm or even ~10 nm,
while the insulin hexamer hydrodynamic diameter is ~5.6 nm.
135
This means insulin begins to
aggregate instantly from the hexamer state and there is no lag phase for this case.
78

Figure 5-2: Time-dependent measurements of (a) the apparent hydrodynamic diameter
determined from DLS using the method of cumulants; (b) the scattering intensity counting rate
(locally averaged over a time span of ~1 min); and (c) the absorbance at 600 nm determined from
UV-vis spectroscopy measurements. Data were collected for pure insulin and insulin in the
presence of 0.5 mM azoTAB and 1 mM azoTAB under visible light. [insulin] = 10 mg/mL, T =
60 °C, pH = 1.6.

The average counting rate (ACR) measured from light scattering (Figure 5-2b) also provide
similar information, since light scattering intensity is proportional to the square of the particle
79
volume (i.e., radius to the 6
th
power for spherical aggregates). The average count rates are shown
to increase and then reach plateau at different times depending on the system as well (~15 min for
pure insulin versus 13 minutes and 9 minutes when in the presence of 0.5 mM and 1 mM cis
azoTAB, respectively), which agrees with data obtained from hydrodynamic diameters. For
comparison, a lag time of approximately three hours was reported at an insulin concentration of 5
mg/mL in 50 mM KCl/HCl pH 1.6 solutions at 60 °C.
136
But note that to a first approximation the
lag time should be proportional to the concentration raised to the power of n, where n is the number
of monomers in the nucleus.
137,138
Thus, at the present protein concentration of 10 mg/mL and
assuming n = 6 for hexamers, the lag time for pure insulin obtained here would be predicted to be
a factor of 2
6
shorter than that reported above (i.e., on the order of a few minutes).
As the aggregate sizes in Figure 5-2a continue to grow, the samples become measurably
turbid, as seen in the time-dependent effective absorbance at 600 nm shown in Figure 5-2c. This
wavelength was selected so as to not be influenced by the absorbance of azoTAB (i.e., 𝜆
Y?S
!6?%W
=
350 nm and 𝜆
Y?S
n>W
= 434 nm).
5
Note that while light attenuation in Figure 5-2c is a combination of
scattering (off the growing aggregates) and absorption (from the tails of the azobenzene peaks
above), as is convention we simply report the effective absorbance calculated from the negative of
the common logarithm of the measured transmittance. Notably, when the aggregates sizes
approach about one-tenth of the wavelength of light in Figure 5-2a (i.e., at approximately 46 min
for pure insulin, 49 min and 45 min with 0.5 mM and 1 mM azoTAB under visible light), the
effective absorbance values in Figure 5-2c are first observed to increase, in accordance with the
well-known rule of thumb.
Collectively, the data in Figure 5-2 demonstrate that the rate of insulin aggregation is slightly
increased in the presence azoTAB under visible light compared to the pure protein, while the effect
80
of azoTAB concentrations is limited. This is perhaps not surprising given that the relatively-
hydrophobic cis form of azoTAB has been previously shown to unfold or dissociate a variety of
proteins or protein oligomers,
5,40,41,44,57,139
while insulin fibrillization is generally considered to
originate from a slightly-unfolded monomeric protein conformation and be driven by hydrophobic
interactions (i.e., mutations to polar residues have been shown to increase the lag time).
125

Furthermore, while the conditions in Figure 5-2 highly favor insulin aggregation, at lower
temperatures (i.e., 25 °C or 15 °C) no measurable insulin aggregation could be found even after
several weeks.
Figure 5-3 illustrates how the fractal dimension df can obtained from the insulin
fibrillization data in Figure 5-2. The intensity of scattered light is given by
140

, 5-1
where ni is the number of aggregates of type i and IS,i is the scattering from particles of type i.
Noting that IS,i is proportional to the mass of the particle mi, and further that mi is proportional to
the particle diameter D raised to the power of df, then IS,i is proportional to 𝐷
"o
A
. Finally, noting
that the number of particles (i.e., aggregates) would decrease to 𝐷
Ho
A
, we arrive at
. 5-2
Thus, a log-log plot of the average count rate versus the hydrodynamic diameter would have a
slope of df. As seen in Figure 5-3, during the early stages of insulin fibrillization a fractal
dimension of ~3 is obtained, suggesting these early aggregates grow equally in all three
dimensions. Once the aggregates emerge from the lag phase at the onset of rapid fibrillization,
however, the fractal dimension decreases to values slightly less than 1, suggesting that now particle
growth occurs only in one dimension, likely in an end-to-end fashion. This result is not surprising
since both the experiment temperature (T = 60 °C) and protein concentration are very high.
𝐼
W
= 𝑛
>
(𝑡)𝐼
W,>
(𝑡)
p
>_<

𝐼
W
∝ 𝐷
o
A

81

Figure 5-3: df number of growth of fibrils of (a) pure insulin; (b) insulin with 0.5mM azoTAB in
the presence of visible light; (c) insulin with 1mM azoTAB in the presence of visible light. [insulin]
= 10mg/ml, T = 60 °C, pH = 1.6.

The Figure 5-4 shows the insulin fibrillation process under a lower temperature (T = 50
°C). As shown in Figure 5-4 (a), in the case of pure insulin and insulin in the presence of 0.5mM
azoTAB under visible light, the hydrodynamic diameters (dH) exhibited at initial time were ~3 nm,
consistent with the insulin dimer (dH = 3.5 nm) and or monomer (dH = 2.6 nm)
135
. Following this
lag phase, insulin hydrodynamic diameter begins to increase quickly at specific times depending
on the system (~22 min for pure insulin and 15 min when in the presence of 0.5 mM azoTAB
under visible light). For the case of insulin in the presence of 1 mM azoTAB under visible light,
the apparent size begins to increase instantly, same as those cases under higher temperature.
82

Figure 5-4: time-dependent measurements of (a) the apparent hydrodynamic diameter determined
from DLS using the method of cumulants, and (b) the average counting rate determined from DLS.
Data were collected for pure insulin and insulin in the presence of 0.5mM and 1mM azoTAB under
visible light. [insulin] = 10mg/ml, T = 50 °C, pH = 1.7.

The average counting rate determined from DLS, as shown in Figure 5-4 (b), provides
similar information about the lag time. The average counting rate for pure insulin starts to increase
rapidly from about 22 minutes, while the average counting rate for insulin in the presence of 0.5
mM azoTAB under visible light starts to increase from about 15 minutes. And for the case of
insulin in the presence of 1mM azoTAB under visible light, the average counting rate starts to
increase instantly when the measurement begins. Since the average counting rate represents the
light scattering intensity, which is proportional to particle’s size, the time when it starts to raise
can also provide information when insulin start to aggregate. And the lag time agrees with results
from hydrodynamic diameter determined from DLS.
Figure 5-5 shows the fractal number, df, during different stages of insulin fibrillation. As
shown in Figure 5-5 (a), during the early stage of pure insulin fibrillization, the fractal dimension
is less than 1, suggesting that particle growth happens only in one dimension. And later on a fractal
83
number of ~2 and ~3 are obtained, suggesting that insulin fibrils are growing in two dimensions,
perhaps planarly, first and three dimensions later during different stages of fibrillization process.
Similarly, for the case of insulin in the presence of 0.5 mM azoTAB under visible light, different
fractal numbers are obtained during different stages, and they are ~1, ~3, ~1 and less than 1
respectively. This suggests that particles are growing only in one dimension at early stage, and all
the three dimensions equally in the second place, and then fibrils are elongated in one dimension
only again. In the end, df value is much less than 1, suggesting that fibrils are growing in one
dimension only. For the third case, insulin in the presence of 1 mM azoTAB under visible light,
as shown in Figure 5-5 (c), fractal dimension values are similar to those under a higher temperature
(T = 60 °C). The initial df is of ~3 suggests that insulin aggregates in all the three dimensions
equally during the early fibrillization and once aggregates emerge from the early rapid
fibrillization, the df becomes less than 1, which means that the fibrils are growing in one dimension.

Figure 5-5: df number of growth of fibrils of (a) pure insulin; (b) insulin with 0.5mM azoTAB in
the presence of visible light; (c) insulin with 1mM azoTAB in the presence of visible light. [insulin]
= 10mg/ml, T = 50 °C, pH = 1.6

84
5.5 Conclusion
Through dynamic light scattering and UV-vis spectroscopy, it has been demonstrated that
insulin fibrillation is accelerated when in presence of azoTAB under visible light. Previous student
work has shown that insulin fibrillation process can be delayed or controlled when in presence of
azoTAB under UV light, which needs to be confirmed in future work. It is believed that cis isomers,
which are more hydrophobic, have stronger interaction with protein thus inducing a bigger degree
of protein unfolding and aggregating. This leads to the formation of unstable nuclei in the early
stage of fibrillization. The cis isomers, however, are more hydrophilic and supposed to interrupt
the interaction between surfactant and protein, as a result which the fibrillation of protein is delayed.
If this assumption can be confirmed in future, we can realize controlling insulin fibrillization by
using photosensitive surfactant.

85
6 Planned work
6.1 Protein aggregation
6.1.1 Insulin
6.1.1.1 Introduction
As discussed above, insulin aggregation process is a function of temperature, azoTAB
concentration, insulin concentration and so on. Since trans isomers are more hydrophobic and
more likely to interact with proteins, it is not surprising to see insulin aggregation rate increased
in the presence of trans azoTAB. The bent cis isomers, however, are more hydrophilic and the
interaction with proteins is non-specific, it is expected to see insulin aggregation rate is slowed in
the presence of cis isomers.
6.1.1.2 Preliminary data
The process of insulin aggregation in the presence of zaoTAB has been measured using
dynamic light scattering was also measured by a previous student in our group. His experiment
temperature is 50 °C and azoTAB concentration is 0.5mM. From his data, as shown in Figure 6-,
it looks like that under UV light cis isomers are able to inhibit insulin hexamers from aggregating
into large fibrils. Both the hydrodynamic diameters of pre-fibrils in the early stage measured by
DLS and the large fibrils measured at 600nm by UV-Vis are pointing to the same conclusion.
86

Figure 6-1: Time-dependent measurements of (a) the hydrodynamic diameter determined from
DLS using the method of cumulants, and (b) the absorbance at 600 nm determined from UV-vis
spectroscopy absorbance measurements. Data were collected for pure insulin and insulin in the
presence of 0.5 mM azoTAB under either visible or UV light. [insulin] = 10 mg/mL, T = 60 °C,
pH = 1.6.

Besides of these, it was also observed that when transfer visible light to UV light, the rapid
aggregation could be inhibited. As mentioned before, cis isomers are able to slow down the insulin
fibrillation process, and our group has observed that in situ trans ® cis conversion at 10 min could
stop the rapid aggregation but in situ trans ® cis conversion at 20 minutes could not. Thus, a
“point of no return” exists beyond which the aggregation cascade cannot be stopped as shown in
Figure 6-.
87

Figure 6-2: Growth of amyloid fibrils for insulin with trans or cis azoTAB, where trans-to-cis
conversion halts further aggregation. Inset: Time-resolved DLS measurements.

6.1.1.3 Planned work
In future, we will continue the systematic research on different effects such as temperature,
azoTAB concentration and light conditions on insulin fibrillation process. Under each condition,
the time-dependent hydrodynamic diameters of the overall particles during the process will be
measured by dynamic light scattering. Besides of this, the size distribution of intermediates as a
function of time will also be analyzed. Based on the particle diameters and scattering intensity, the
fractal dimension growth of particles will be analyzed and discussed. In the end, the systematic
research about the relationship between fractal dimension and experiment condition will be
obtained. In addition to pro-fibrils in the early aggregation stage, large fibrils growth will be
measured by UV-Vis at 600nm. Furthermore, we will continue to identify the “point of no return”
under different conditions and the necessary requirement that insulin fibrillation can be stopped.
When we have grasped a big picture of how to control insulin aggregation by photoresponsive
surfactant, we will use the small angle neutron scattering techniques to characterize both
88
oligomeric intermediates and the resulting fibrils that develop during the course of aggregation,
especially during the early stage of fibrillation.

89
References
1. Hayashita, T., Kurosawa, T., Miyata, T., Tanaka, K. & Igawa, M. Effect of structural
variation within cationic azo-surfactant upon photoresponsive function in aqueous solution.
Colloid Polym. Sci. 272, 1611–1619 (1994).
2. Shang, T., Smith, K. A. & Hatton, T. A. Photoresponsive surfactants exhibiting unusually
large, reversible surface tension changes under varying illumination conditions. Langmuir
19, 10764–10773 (2003).
3. Lee, C. T., Smith, K. A. & Hatton, T. A. Photocontrol of protein folding: The interaction of
photosensitive surfactants with bovine serum albumin. Biochemistry 44, 524–536 (2005).
4. Eastoe, J. & Vesperinas, A. Self-assembly of light-sensitive surfactants. Soft Matter 1, 338–
347 (2005).
5. Lee, C. T., Smith, K. A. & Hatton, T. A. Photoreversible viscosity changes and gelation in
mixtures of hydrophobically modified polyelectrolytes and photosensitive surfactants.
Macromolecules 37, 5397–5405 (2004).
6. Pauling, L. & Corey, R. B. Two hydrogen-bonded spiral configurations of the polypeptide
chain. J. Am. Chem. Soc. 72, 5349 (1950).
7. Boehr, D. D., Dyson, H. J. & Wright, P. E. An NMR perspective on enzyme dynamics.
Chem. Rev. 106, 3055–3079 (2006).
8. Murthy, M., Reid, T., Sicignano, A. & Tanaka, N. Strucuture of Beef Liver Catalase. J. Mol.
Biol. 152, 465–499 (1981).
9. Prakash, K. Unique oligomeric intermediates of bovine liver catalase. Protein Sci. 11, 46–
57 (2002).
10. Kirkman, H. N. & Gaetani, G. F. Catalase: A tetrameric enzyme with four tightly bound
90
molecules of NADPH. Proc. Natl. Acad. Sci. U. S. A. 81, 4343–4347 (1984).
11. Diaz, A., Loewen, P., Fita, I. & Carpena, X. Thirty years of heme catalases structural
biology. Arch. Biochem. Biophys. 102–110 (2012).
12. Fita, I. & Rossmann, M. G. The active center of catalase. J. Mol. Biol. 185, 21–37 (1985).
13. Hillar, A. & Nicholls, P. A mechanism for NADPH inhibition of catalase compound II
formation. FEBS Lett. 314, 179–182 (1992).
14. Huang, M. Q. & Zhou, H. M. Alkaline unfolding and salt-induced folding of aminoacylase
at high pH. Enzym. Protein 48, 229–237 (1994).
15. Raines, R. T. Ribonuclease A. Chem. Rev. 98, 1045–1065 (1998).
16. Oleinikova, A., Sasisanker, P. & Weingärtner, H. What can really be learned from dielectric
spectroscopy of protein solutions? A case study of ribonuclease A. J. Phys. Chem. B 108,
8467–8474 (2004).
17. Merlino, A., Vitagliano, L., Ceruso, M. A., Di Nola, A. & Mazzarella, L. Global and local
motions in ribonuclease A: A molecular dynamics study. Biopolymers 65, 274–283 (2002).
18. Cole, R. & Loria, J. P. Evidence for flexibility in the function of ribonuclease A.
Biochemistry 41, 6072–6081 (2002).
19. Kirkwood, J. G. The general theory of irreversible processes in solutions of macromolecules.
J. Polym. Sci. Part B Polym. Phys. 34, 597–610 (1996).
20. Mccammon, J. A., Harvey, S. C. & Wolynes, P. G. Dynamics of Proteins
and Nucleic Acids . Phys. Today 41, 105–106 (1988).
21. Nayak, A., Sorci, M., Krueger, S. & Belfort, G. A universal pathway for amyloid nucleus
and precursor formation for insulin. Proteins Struct. Funct. Bioinforma. 74, 556–565 (2009).
22. Hammouda, B. Probing Nanoscle Structure - SANS Toolbox: Small-Angle Neutron
91
Scattering Polymers Complex Fluids Biology Materials Science. (2016).
23. Stenstam, A., Montalvo, G., Grillo, I. & Gradzielski, M. Small angle neutron scattering
study of lysozyme-sodium dodecyl sulfate aggregates. J. Phys. Chem. B 107, 12331–12338
(2003).
24. Murthy, M. R. N., Reid III, T. J., Sicignano, A., Tanaka, N. & Rossmann, M. G. Structure
of beef liver catalase. J. Mol. Biol. 152, 465–499 (1981).
25. Bonnichsen, R. K., Chance, B. & Theorell, H. Catalase activity. Acta chem. scand 1, 685–
709 (1947).
26. Ogura, Y. Catalase Activity at High Concentration of Hydrogen Peroxide. (1955).
27. Fersht, A. Structure and mechanism in protein science: a guide to enzyme catalysis and
protein folding. (Macmillan, 1999).
28. FINCH, J. T. Electron Microscopy of Proteins. in (eds. NEURATH, H. & HILL, R. L. B.
T.-T. P. (Third E.) 413–497 (Academic Press, 1975). doi:https://doi.org/10.1016/B978-0-
12-516301-9.50010-3.
29. Sugadev, R., Ponnuswamy, M. N. & Sekar, K. Structural analysis of NADPH depleted
bovine liver catalase and its inhibitor complexes. Int. J. Biochem. Mol. Biol. 2, 67 (2011).
30. Zamocky, M. & Koller, F. Understanding the structure and function of catalases: clues from
molecular evolution and in vitro mutagenesis. Biophys. Mol. Biol. 72, 19–66 (1999).
31. Putnam, C. D., Arvai, A. S., Bourne, Y. & Tainer, J. A. Active and inhibited human catalase
structures: ligand and NADPH binding and catalytic mechanism. J. Mol. Biol. 296, 295–
309 (2000).
32. Tanford, C. & Lovrien, R. Dissociation of Catalase into Subunits. J. Am. Chem. Soc. 84,
1892–1896 (1962).
92
33. Samejima, T. & Yang, J. T. Reconstitution of Acid-denatured catalase. J. Biol. Chem. 730,
(1962).
34. SCHERZ, B., KUCHINSKAS, E. J., WYSS, S. R. & AEBI, H. Heterogeneity of erythrocyte
catalase: dissociation, recombination and hybridization of human erythrocyte catalases. Eur.
J. Biochem. 69, 603–613 (1976).
35. Jones, M. N., Manley, P., Midgley, P. J. W. & Wilkinson, A. E. Dissociation of Bovine and
Bacterial Catalases by Sodium n-dodecyl Sulfate. Biopolymers 21, 1435–1450 (1982).
36. Takeda, A., Hirano, K., Shiroya, Y. & Samejima, T. On the denaturation of porcine
erythrocyte catalase with alkali, urea, and guarndine hydrochloride in relation to its subunit
structure. J. Biochem. 93, 967–975 (1983).
37. Nicholls, P. Classical catalase: ancient and modern. Arch. Biochem. Biophys. 525, 95–101
(2012).
38. Padovani, D., Hessani, A., Castillo, F. & Liot, G. sulfheme formation during homocysteine
S-oxygenation by catalase in cancers and neurodegenerative diseases. Nature
communications (2016).
39. Murshudov, G. N. et al. The structures of Micrococcus lysodeikticus catalase, its ferryl
intermediate (compound II) and NADPH complex. Acta Crystallogr. Sect. D Biol.
Crystallogr. 58, 1972–1982 (2002).
40. Mirarefi, P. & Lee, C. T. Photo-induced unfolding and inactivation of bovine carbonic
anhydrase in the presence of a photoresponsive surfactant. Biochim. Biophys. Acta -
Proteins Proteomics 1804, 106–114 (2010).
41. Wang, S. C., Mirarefi, P., Faraone, A. & Lee, C. T. Light-controlled protein dynamics
observed with neutron spin echo measurements. Biochemistry 50, 8150–8162 (2011).
93
42. Seidel, Z. P. & Lee Jr, C. T. Enhanced Activity of the Cellulase Enzyme β-Glucosidase
upon Addition of an Azobenzene-Based Surfactant. ACS Sustain. Chem. Eng. 8, 1751–1761
(2020).
43. Hamill, A. C., Wang, S. C. & Lee, C. T. Solution structure of an amyloid-forming protein
during photoinitiated hexamer-dodecamer transitions revealed through small-angle neutron
scattering. Biochemistry 46, 7694–7705 (2007).
44. Hamill, A. C. & Lee, C. T. Photocontrol of β-amyloid peptide (1-40) fibril growth in the
presence of a photosurfactant. J. Phys. Chem. B 113, 6164–6172 (2009).
45. Mirarefi, P. & Ted Lee Jr, C. Reversible control of enzyme‐inhibitor interactions with light
illumination using a photoresponsive surfactant. Proteins Struct. Funct. Bioinforma. 87,
715–722 (2019).
46. Beharry, A. A. & Woolley, G. A. Azobenzene photoswitches for biomolecules. Chem. Soc.
Rev. 40, 4422–4437 (2011).
47. Serrano, A. L., Waegele, M. M. & Gai, F. Spectroscopic studies of protein folding: linear
and nonlinear methods. Protein Sci. 21, 157–170 (2012).
48. Kameta, N., Akiyama, H., Masuda, M. & Shimizu, T. Effect of Photoinduced size changes
on protein refolding and transport abilities of soft nanotubes. Chem. Eur. J. 22, 7198–7205
(2016).
49. Martin, N., Ruchmann, J. & Tribet, C. Prevention of aggregation and renaturation of
carbonic anhydrase via weak association with octadecyl-or azobenzene-modified poly
(acrylate) derivatives. Langmuir 31, 338–349 (2015).
50. Sun, J. et al. Unfolding of cytochrome c upon interaction with azobenzene-modified
copolymers. Biomacromolecules 13, 3736–3746 (2012).
94
51. Hoppmann, C., Maslennikov, I., Choe, S. & Wang, L. In situ formation of an azo bridge on
proteins controllable by visible light. J. Am. Chem. Soc. 137, 11218–11221 (2015).
52. Ihalainen, J. A. et al. Folding and unfolding of a photoswitchable peptide from picoseconds
to microseconds. Proc. Natl. Acad. Sci. 104, 5383–5388 (2007).
53. Schierling, B. et al. Controlling the enzymatic activity of a restriction enzyme by light. Proc.
Natl. Acad. Sci. 107, 1361–1366 (2010).
54. Liu, D., Karanicolas, J., Yu, C., Zhang, Z. & Woolley, G. A. Site-specific incorporation of
photoisomerizable azobenzene groups into ribonuclease S. Bioorg. Med. Chem. Lett. 7,
2677–2680 (1997).
55. Willner, I., Rubin, S. & Riklin, A. Photoregulation of papain activity through anchoring
photochromic azo groups to the enzyme backbone. J. Am. Chem. Soc. 113, 3321–3325
(1991).
56. James, D. A., Burns, D. C. & Woolley, G. A. Kinetic characterization of ribonuclease S
mutants containing photoisomerizable phenylazophenylalanine residues. Protein Eng. 14,
983–991 (2001).
57. Wang, S. C. & Lee, C. T. Enhanced enzymatic activity through photoreversible
conformational changes. Biochemistry 46, 14557–14566 (2007).
58. Zigman, S., Reddan, J., Schultz, J. B. & McDaniel, T. Structural and functional changes in
catalase induced by near-UV radiation. Photochem. Photobiol. 63, 818–824 (1996).
59. Heck, D. E., Shakarjian, M., Kim, H. D., Laskin, J. D. & Vetrano, A. M. Mechanisms of
oxidant generation by catalase. Ann. N. Y. Acad. Sci. 1203, 120 (2010).
60. Glinka, C. J. et al. The 30 m Small-Angle Neutron Scattering Instruments at the National
Institute of Standards and Technology. J. Appl. Crystallogr. 31, 430–445 (1998).
95
61. Feigin, L. A. & Svergun, D. I. Structure analysis by small-angle X-ray and neutron
scattering. vol. 1 (Springer, 1987).
62. Svergun, D. I. Determination of the regularization parameter in indirect-transform methods
using perceptual criteria. J. Appl. Crystallogr. 25, 495–503 (1992).
63. Heller, W. T., Krueger, J. K. & Trewhella, J. Further insights into calmodulin - Myosin light
chain kinase interaction from solution scattering and shape restoration. Biochemistry 42,
10579–10588 (2003).
64. Micsonai, A. et al. Accurate secondary structure prediction and fold recognition for circular
dichroism spectroscopy. Proc. Natl. Acad. Sci. 112, E3095–E3103 (2015).
65. Gauthier, K. M. et al. Soluble epoxide hydrolase contamination of specific catalase
preparations inhibits epoxyeicosatrienoic acid vasodilation of rat renal arterioles. Am. J.
Physiol. Physiol. 301, F765–F772 (2011).
66. Deisseroth, A. & Dounce, A. L. Nature of the change produced in catalase by lyophilization.
Arch. Biochem. Biophys. 120, 671–692 (1967).
67. Prajapati, S., Bhakuni, V., Babu, K. R. & Jain, S. K. Alkaline unfolding and salt‐induced
folding of bovine liver catalase at high pH. Eur. J. Biochem. 255, 178–184 (1998).
68. Weydert, C. J. & Cullen, J. J. Measurement of superoxide dismutase, catalase and
glutathione peroxidase in cultured cells and tissue. Nat. Protoc. 5, 51–66 (2010).
69. Aebi, H. Catalase in vitro. Methods Enzymol. 105, 121–126 (1984).
70. Ogura, Y. & Yamazaki, I. Steady-state kinetics of the catalase reaction in the presence of
cyanide. J. Biochem. 94, 403–408 (1983).
71. Glatter, O. The interpretation of real-space information from small-angle scattering
experiments. J. Appl. Crystallogr. 12, 166–175 (1979).
96
72. Guinier, A., Fournet, G. & Yudowitch, K. L. Small-angle scattering of X-rays. (1955).
73. Deutsch, H. F. The properties of various crystalline horse erythrocyte catalase preparations.
Acta. Chem. Scand 6, 1516–1521 (1952).
74. Price, V. E. & Greenfield, R. E. Liver catalase II. Catalase fractions from normal and tumor-
bearing rats. J. Biol. Chem. 209, 363–376 (1954).
75. joHN RANDALL, S. I. R., STARLING, D. & KoNRAD, I. The Scattering of Neutrons by
Catalase: A Study of Molecules, Subunits, and Tubules. NEUTRON Scatt. Anal. Biol. Struct.
78.
76. Wiemer, E. A. C. et al. Production and characterisation of monoclonal antibodies against
native and disassembled human catalase. J. Immunol. Methods 151, 165–175 (1992).
77. KUNTZ, G., STÖCKEL, P. & HEIDRICH, H.-G. The conformer nature of the multiple
forms of beef liver catalase as obtained by biochemical and small-angle X-ray scattering
experiments. A model for the quaternary structure of the beef liver catalase molecule.
Hoppe-Seyler´ s Zeitschrift für Physiol. Chemie 359, 959–974 (1978).
78. Achter, E. K. & Swan, I. D. A. Conformation of lysozyme and α-lactalbumin in solution.
Biochemistry 10, 2976–2978 (1971).
79. Smilgies, D. M. & Folta-Stogniew, E. Molecular weight-gyration radius relation of globular
proteins: A comparison of light scattering, small-angle X-ray scattering and structure-based
data. J. Appl. Crystallogr. 48, 1604–1606 (2015).
80. Lee Jr, C. T., Smith, K. A. & Hatton, T. A. Small-angle neutron scattering study of the
micellization of photosensitive surfactants in solution and in the presence of a
hydrophobically modified polyelectrolyte. Langmuir 25, 13784–13794 (2009).
81. Rambo, R. P. & Tainer, J. A. Accurate assessment of mass, models and resolution by small-
97
angle scattering. Nature 496, 477–481 (2013).
82. Hamill, A. C., Wang, S. C. & Lee, C. T. Probing lysozyme conformation with light reveals
a new folding intermediate. Biochemistry 44, 15139–15149 (2005).
83. Svergun, D. I., Petoukhov, M. V. & Koch, M. H. J. Determination of domain structure of
proteins from x-ray solution scattering. Biophys. J. 80, 2946–2953 (2001).
84. Rosenbusch, J. P., Lustig, A., Grabo, M., Zulauf, M. & Regenass, M. Approaches to
determining membrane protein structures to high resolution: do selections of subpopulations
occur? Micron 32, 75–90 (2001).
85. Bowie, J. U. Stabilizing membrane proteins. Curr. Opin. Struct. Biol. 11, 397–402 (2001).
86. Chin, C.-N., von Heijne, G. & de Gier, J.-W. L. Membrane proteins: shaping up. Trends
Biochem. Sci. 27, 231–234 (2002).
87. Tatulian, S. A. Attenuated total reflection Fourier transform infrared spectroscopy: a
method of choice for studying membrane proteins and lipids. Biochemistry 42, 11898–
11907 (2003).
88. White, S. H. Translocons, thermodynamics, and the folding of membrane proteins. FEBS
Lett. 555, 116–121 (2003).
89. Vanhoudt, J., Abgar, S. d, Aerts, T. & Clauwaert, J. A small‐angle X‐ray solution scattering
study of bovine α‐crystallin. Eur. J. Biochem. 267, 3848–3858 (2000).
90. Kabsch, W. & Sander, C. Dictionary of protein secondary structure: pattern recognition of
hydrogen‐bonded and geometrical features. Biopolym. Orig. Res. Biomol. 22, 2577–2637
(1983).
91. Brahms, S., Brahms, J., Spach, G. & Brack, A. Identification of β, β-turns and unordered
conformations in polypeptide chains by vacuum ultraviolet circular dichroism. Proc. Natl.
98
Acad. Sci. 74, 3208–3212 (1977).
92. Maeda, H. & Ooi, K. Isodichroic point and the β–random coil transition of poly (S‐
carboxymethyl‐l‐cysteine) and poly (S‐carboxyethyl‐l‐cysteine) in the absence of added salt.
Biopolym. Orig. Res. Biomol. 20, 1549–1563 (1981).
93. Wallimann, P., Kennedy, R. J., Miller, J. S., Shalongo, W. & Kemp, D. S. Dual wavelength
parametric test of two-state models for circular dichroism spectra of helical polypeptides:
anomalous dichroic properties of alanine-rich peptides. J. Am. Chem. Soc. 125, 1203–1220
(2003).
94. Lin, K., Simossis, V. A., Taylor, W. R. & Heringa, J. A simple and fast secondary structure
prediction method using hidden neural networks. Bioinformatics 21, 152–159 (2005).
95. Barrow, C. J., Yasuda, A., Kenny, P. T. M. & Zagorski, M. G. Solution conformations and
aggregational properties of synthetic amyloid β-peptides of Alzheimer’s disease: analysis
of circular dichroism spectra. J. Mol. Biol. 225, 1075–1093 (1992).
96. Kelemen, B. R. et al. Hypersensitive substrate for ribonucleases. Nucleic Acids Res. 27,
3696–3701 (1999).
97. Park, C. et al. Fast, facile, hypersensitive assays for ribonucleolytic activity. Methods in
Enzymology vol. 341 (Elsevier Masson SAS, 2001).
98. Smith, J. C. Protein dynamics: Comparison of simulations with inelastic neutron scattering
experiments. Q. Rev. Biophys. 24, 227–291 (1991).
99. Daniel, R. M., Dunn, R. V., Finney, J. L. & Smith, J. C. The role of dynamics in enzyme
activity. Annu. Rev. Biophys. Biomol. Struct. 32, 69–92 (2003).
100. Gabel, F. et al. Protein dynamics studied by neutron scattering. Q. Rev. Biophys. 35, 327–
367 (2002).
99
101. Hayward, S., Kitao, A. & Berendsen, H. J. C. Model-free methods of analyzing domain
motions in proteins from simulation: A comparison of normal mode analysis and molecular
dynamics simulation of lysozyme. Proteins Struct. Funct. Genet. 27, 425–437 (1997).
102. Bu, Z., Biehl, R., Monkenbusch, M., Richter, D. & Callaway, D. J. E. Coupled protein
domain motion in Taq polymerase revealed by neutron spin-echo spectroscopy. Proc. Natl.
Acad. Sci. U. S. A. 102, 17646–17651 (2005).
103. Arnold, G. E. & Ornstein, R. L. Protein hinge bending as seen in molecular dynamics
simulations of native and M6I mutant T4 lysozymes. Biopolymers 41, 533–544 (1997).
104. Eisenberg, H. Forward scattering of light, X-rays and neutrons. Q. Rev. Biophys. 14, 141–
172 (1981).
105. Lehmann, M. S. & Zaccai, G. Neutron Small-Angle Scattering Studies of Ribonuclease in
Mixed Aqueous Solutions and Determination of the Preferentially Bound Water.
Biochemistry 23, 1939–1942 (1984).
106. Nsppert, A., Gast, K., Mfiller-frohne, M., Zirwer, D. & Damaschun, G. Reduced-denatured
ribonuclease A is not in a c o m p a c t state. 380, 179–182 (1996).
107. Akcasu, A. Z., Hammouda, B., Stockmayer, W. H. & Tanaka, G. First cumulant for chains
with constraints. J. Chem. Phys. 85, 4734–4743 (1986).
108. García De La Torre, J., Huertas, M. L. & Carrasco, B. Calculation of hydrodynamic
properties of globular proteins from their atomic-level structure. Biophys. J. 78, 719–730
(2000).
109. Carrasco, B. & De La Torre, J. G. Hydrodynamic properties of rigid particles: Comparison
of different modeling and computational procedures. Biophys. J. 76, 3044–3057 (1999).
110. Kirkwood, J. G. & Riseman, J. The intrinsic viscosities and diffusion constants of flexible
100
macromolecules in solution. J. Chem. Phys. 16, 565–573 (1948).
111. Bloomfield, V., Dalton, W. O. & Van Holde, K. E. Frictional coefficients of multisubunit
structures. I. Theory. Biopolymers 5, 135–148 (1967).
112. Dunbar, J., Yennawar, H. P., Banerjee, S., Luo, J. & Farber, G. K. The effect of denaturants
on protein structure. Protein Sci. 6, 1727–1733 (1997).
113. Gebe, J. A. & Schurr, J. M. Intramolecular interference effects in dynamic light scattering
from rigid rings. Biopolymers 33, 1757–1764 (1993).
114. Boehr, D. D., Nussinov, R. & Wright, P. E. The role of dynamic conformational ensembles
in biomolecular recognition. Nat. Chem. Biol. 5, 789–796 (2009).
115. Otzen, D. Protein-surfactant interactions: A tale of many states. Biochim. Biophys. Acta -
Proteins Proteomics 1814, 562–591 (2011).
116. Frauenfelder, H., Sligar, S. G. & Wolynes, P. G. The energy landscapes and motions of
proteins. Science (80-. ). 254, 1598 LP – 1603 (1991).
117. Bee, M. Quasielastic neutron scattering : principles and applications in solid state
chemistry, biology, and materials science. (Adam Hilger, 1988).
118. Montes, H. et al. Neutron spin echo investigation of the concentration fluctuation dynamics
in melts of diblock copolymers. J. Chem. Phys. 110, 10188–10202 (1999).
119. Biehl, R., Monkenbusch, M. & Richter, D. Exploring internal protein dynamics by neutron
spin echo spectroscopy. Soft Matter 7, 1299–1307 (2011).
120. Carter, D. C. & Ho, J. X. Structure of serum albumin. Adv. Protein Chem. 45, 153–176
(1994).
121. Rosenoer, V. M., Oratz, M. & Rothschild, M. A. Albumin Structure, Function and Uses.
(1977). doi:10.1136/gut.19.2.159-b.
101
122. Peters, T. J. All about Albumin. (Academic Press, 1996).
123. He, X. M. & Carter, D. C. Atomic structure and chemistry of human serum albumin. Nature
358, 209–215 (1992).
124. Jang, W. G. et al. Glucocorticoid receptor mediated repression of human insulin gene
expression is regulated by PGC-1α. Biochem. Biophys. Res. Commun. 352, 716–721 (2007).
125. Garriques, L. N., Frokjaer, S., Carpenter, J. F. & Brange, J. The effect of mutations on the
structure of insulin fibrils studied by Fourier transform infrared (FTIR) spectroscopy and
electron microscopy. J. Pharm. Sci. 91, 2473–2480 (2002).
126. Chang, X., Jørgensen, A. M. M., Bardrum, P. & Led, J. J. Solution structures of the R6
human insulin hexamer. Biochemistry 36, 9409–9422 (1997).
127. Brookmeyer, R., Johnson, E., Ziegler-Graham, K. & Arrighi, H. M. Forecasting the global
burden of Alzheimer’s disease. Alzheimer’s Dement. 3, 186–191 (2007).
128. Ross, C. A. & Poirier, M. A. Protein aggregation and neurodegenerative disease. Nat. Med.
10, S10 (2004).
129. Sluzky, V., Klibanov, A. M. & Langer, R. Mechanism of insulin aggregation and
stabilization in agitated aqueous solutions. Biotechnol. Bioeng. 40, 895–903 (1992).
130. Waugh, D. F. A Fibrous Modification of Insulin. I. The Heat Precipitate of Insulin. J. Am.
Chem. Soc. 68, 247–250 (1946).
131. Brange, J., Owens, D. R., Kang, S. & Vølund, A. Monomeric insulins and their experimental
and clinical implications. Diabetes Care 13, 923–954 (1990).
132. Barnett, A. H. & Owens, D. R. Drug profile: Insulin analogues. Lancet 349, 47–51 (1997).
133. Lee, J. & Pilch, P. F. the insulin receptor: strucuture, function, and signaling. Am. J. Phys.
Cell Physiol. 266, 319–334 (1994).
102
134. Ivanova, M. I., Thompson, M. J. & Eisenberg, D. A systematic screen of β2-microglobulin
and insulin for amyloid-like segments. Proc. Natl. Acad. Sci. U. S. A. 103, 4079–4082
(2006).
135. Hvidt, S. Insulin association in neutral solutions studied by light scattering. Biophys. Chem.
39, 205–213 (1991).
136. Mauro, M. et al. Kinetics of Different Processes in Human Insulin Amyloid Formation. J.
Mol. Biol. 366, 258–274 (2007).
137. Hofrichter, J. Kinetics of sickle hemoglobin polymerization. III. Nucleation rates
determined from stochastic fluctuations in polymerization progress curves. J. Mol. Biol. 189,
553–571 (1986).
138. Hofrichter, J., Ross, P. D. & Eaton, W. A. Supersaturation in sickle cell hemoglobin
solutions. Proc. Natl. Acad. Sci. U. S. A. 73, 3035–3039 (1976).
139. Wang, S. C. & Ted Lee, C. Protein secondary structure controlled with light and
photoresponsive surfactants. J. Phys. Chem. B 110, 16117–16123 (2006).
140. Connor, A. J. O. & Hatton, T. A. Dynamics of Micelle - Vesicle Transitions in Aqueous
Anionic / Cationic Surfactant Mixtures. langmuir 13, 6931–6940 (1997).
141. Creighton, T. E. & Freeman, W. H. Proteins: structures and molecular properties (2nd
edition). (1993).
142. IBEL, K. et al. Protein‐decorated micelle structure of sodium‐dodecyl‐sulfate–protein
complexes as determined by neutron scattering. Eur. J. Biochem. 190, 311–318 (1990).
143. Turro, N. J., Lei, X. G., Ananthapadmanabhan, K. P. & Aronson, M. Spectroscopic Probe
Analysis of Protein-Surfactant Interaciion: The BSA/SDS System. Langmuir 11, 2525–
2533 (1995).
103
144. Peters, T. Serum Albumin. Adv. Protein Chem. 37, 161–245 (1985).

Abstract (if available)

Abstract AzoTAB, which is a light sensitive surfactant, undergoes reversible photoisomerization upon exposure to appropriate wavelength of light. When shining visible light onto it, azoTAB will be in trans form, which is more hydrophobic, thus inducing a bigger degree of protein unfolding than cis form, which is more hydrophilic under UV light. Because of such special and unique character, azoTAB has been investigated as a great tool to photo control protein structure, dynamics and function reversibly. For example, superactivity of catalase has been observed after interacting with azoTAB under visible light, because catalase, which is a tetramer, undergoes a slight structure change with trans isomer. What’s more, protein dynamic can also be photo controlled after interacting with azoTAB. Ribnuclease A, which has expressed big difference in translational and inner diffusion coefficient, has different dynamic performance after interacting with azoTAB under different wavelength of light. Three different models, rigid body, soft-linker, and freely jointed, have been employed to illustrate protein dynamic change. Besides, use of azoTAB as a means to photo control protein aggregation has also been explored in inhibiting amyloid fibrillation of insulin. The fibrillation of insulin is enhanced after interacting with azoTAB in trans form, while aggregation of insulin is dramatically inhibited after interacting with azoTAB in cis form. The early fibrillation stage of insulin has been studied in the presence of both trans and cis azoTAB. A number of experimental techniques are used to determine the solution structure of proteins, including dynamic light scattering, circular dichroism, Small Angle Neutron Scattering, and Neutron Spin Echo.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Controlling the form-dynamics-function relationship of proteins with light illumination

PDF

A molecular dynamics study of interactions between the enzyme lysozyme and the photo-responsive surfactant azobenzene trimethylammonium bromide (azoTAB)

PDF

Molecular dynamics studies of protein aggregation in unbounded and confined media

PDF

The interaction of photo-responsive surfactants with biological macromolecules

PDF

Molecular dynamics simulation study of initial protein unfolding induced by the photo-responsive surfactants, azoTAB

PDF

Photocontrol of protein conformation through the use of photoresponsive surfactants, investigated by small angle neutron scattering

PDF

Photocontrol of protein conformation and enzyme activity in the presence of light-responsive surfactants

PDF

Controlling membrane protein folding with light illumination and catanionic surfactant systems

PDF

Enhancement of biofuel enzyme activity and kinetics with azoTAB surfactants

PDF

Stability and folding rate of proteins and identification of their inhibitors

PDF

Controlling membrane protein folding using photoresponsive surfactant

PDF

Photocontrol of DNA compaction using light-responsive surfactants

PDF

Protein-surfactant adsorption on solid surfaces

PDF

High-throughput nanoparticle fabrication and nano-biomembrane interactions

PDF

Dynamic modeling and simulation of flapping-wing micro air vehicles

PDF

Structure – dynamics – function analysis of class A GPCRs and exploration of chemical space using integrative computational approaches

PDF

Structural studies of the IAPP membrane-mediated aggregation pathway

PDF

Understanding protein–DNA recognition in the context of DNA methylation

PDF

Quantum molecular dynamics and machine learning for non-equilibrium processes in quantum materials

PDF

Kinetic studies of collisionless mesothermal plasma flow dynamics

Asset Metadata

Creator Wang, Yimin (author)

Core Title Control the function, dynamics, aggregation of proteins with light illumination

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Chemical Engineering

Publication Date 03/16/2021

Defense Date 11/19/2020

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag light-sensitive,OAI-PMH Harvest,protein aggregation,protein dynamics,protein structure,surfactant

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Lee, Ted (committee chair), Nakano, Aiichiro (committee member), Shing, Katherine (committee member)

Creator Email amywang432@gmail.com,yiminw@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-427516

Unique identifier UC11668244

Identifier etd-WangYimin-9323.pdf (filename),usctheses-c89-427516 (legacy record id)

Legacy Identifier etd-WangYimin-9323.pdf

Dmrecord 427516

Document Type Dissertation

Rights Wang, Yimin

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA