University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Molecular Dynamics Simulation Study of Initial Protein Unfolding Induced by the Photo-
Responsive Surfactants, AzoTAB

by
Chih-Ying Lin

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 2014

Copyright 2014 Chih-Ying Lin

ii

Dedication

This dissertation is dedicated to my parents and my sister, brother
for their constant support and love.

iii

Acknowledgements

Wow! I am standing on the shoulders of the two Giants, Prof. Katherine Shing and Prof.
Ted Lee! With their strong academic leadings, any dude-dwarf like me can not only see much
farther and also see much more clearly than other students.
Thank you very much to my Chair Advisor, Prof. Katherine Shing! Without her, I could
not go through the tough research work. She is the one understanding my weakness and giving
me her consistent support. I learn the critical thinking from her, not letting the any of
unreasonable issues go and so I can reach the research goals step by step.
I also thank to my co-advisor, Prof. Ted Lee. Well, I have to admit that his project is
really challenging but interesting. In the beginning of research work, I was slow and he
encouraged me to read as many papers as possible. He showed me three big boxes of papers he
read when he was a Ph.D. student. Moreover, he sincerely reminded me that research is an
independent work and now I am fully convinced that research is a truly independent work. In
fact, life is a lonely journal most of time even though you have family or not and you have to
work and think and judge and make a decision independently at least 50% time of your life. No
one can guarantee you a lucky and happy life in the future from the financial point of view, as
Prof. Shing said to Class CHE 405 in the Fall 2012; therefore, thinking independently and
making good decisions are to guarantee good lives for yourselves.
Many thanks to Prof. Aiichiro Nakano as he is my committee member and he instructed
me on both molecular dynamics simulation and parallel computing in class. I also thanks for
Prof. Priya Vashishta’s instruction of fundamental theories of MD techniques as well as hands-
iv

on training of running simulations in class. And, thank you for Tao Wei since he initiated me
and helped me to start up the simulation work.
Otherwise, thank you for all of my friends around to keep me warm and to encourage
me to overcome the difficulties. Especially, I want to say you to the Pastor and his wife from
the Power of Praise Church with my full heart as they concern about me with their endless
love.
Many thanks to the technical support from USC High-Performance computing center
and comfortable learning environment at USC.
In summary, I learned to think critically and think independently. Work hard and work
smart!

v

Table of Contents

Dedication ........................................................................................................................... ii
Acknowledgements ............................................................................................................ iii
List of Tables .................................................................................................................... vii
List of Figures .................................................................................................................... ix
Abstract ............................................................................................................................. xvi
Chapter 1: Introduction ....................................................................................................... 1
1.1 Protein Structures ..................................................................................................... 1
1.2 Molecular Dynamics ................................................................................................ 4
1.3 Photoresponsive Surfactant (azoTAB)..................................................................... 6
1.4 Protein Folding / Unfolding ..................................................................................... 9
1.4.1 Levinthal's Paradox ......................................................................................... 9
1.4.2 Detection of Protein Intermediate ..................................................................10
1.5 Lysozyme ............................................................................................................... 18
1.6 Ribonuclease A (RNase A) .................................................................................... 23
1.7 α-Lactalbumin ........................................................................................................ 25
1.8 References .............................................................................................................. 28

Chapter 2: Microsecond Simulations of Lysozyme Unfolding Induced by
Photoresponsive Surfactant................................................................................... 34
2.1 Introduction ............................................................................................................ 35
2.2 Simulation Details .................................................................................................. 42
2.3 Results and Discussion........................................................................................... 45
2.3.1 Dynamic Affinity and Binding Sites ............................................................. 45
2.3.2 Analysis of Secondary Structures ................................................................. 51
2.3.3 Comparison with Experimental Results........................................................ 56
2.3.4 Internal Dynamics of Lysozyme ................................................................... 58
2.3.5 Binding affinity of different parts of azo-TAB toward lysozyme................. 74
2.4 Conclusion ............................................................................................................. 75
2.5 References .............................................................................................................. 77

Chapter 3: Massively Parallel Simulations of Early Unfolding α Domain of
Lysozyme Induced by Surfactants, azoTABs ....................................................... 84
3.1 Introduction ............................................................................................................ 85
3.2 Simulation Details .................................................................................................. 90
3.3 Results and Discussion........................................................................................... 93
3.3.1 Relative Unfolding Effects on the α / β Domains ......................................... 93
vi

3.3.2 Loose Helix Packing ....................................................................................100
3.3.3 Analysis of Secondary Structures ............................................................... 102
3.3.4 Residence Time ........................................................................................... 107
3.4 Conclusions .......................................................................................................... 111
3.5 References ............................................................................................................ 113

Chapter 4: Ribonuclease A – Structure and Dynamics................................................... 117
4.1 Introduction .......................................................................................................... 117
4.2 Simulation details................................................................................................. 121
4.3 Results and Discussion......................................................................................... 124
4.3.1 Dynamic Affinity and Binding Sites ........................................................... 124
4.3.2 Analysis of Secondary Structures ............................................................... 130
4.3.3 Comparison with Experimental Results...................................................... 135
4.4 References ............................................................................................................ 145

Chapter 5: Future Work .................................................................................................. 152
5.1 Enhanced enzymatic activity and selectivity of lysozyme through
surfactants-induced conformational changes ................................................. 152
5.2 Microseconds simulation of unfolding α-lactalbumin Induced by azoTABs ...... 153
5.3 References ............................................................................................................ 158

Bibliography ................................................................................................................... 159

Appendices...................................................................................................................... 174
vii

List of Tables

Table 1.1. The 20 standard amino acids. Some side chains carry charges; others
do not. [1]. .............................................................................................................. 3

Table 1.2. Force field parameters of azo-functional units [12]........................................... 8

Table 1.3.Current experimental techniques used to characterize the protein
intermediates [28]. ................................................................................................ 12

Table 2.1. Summary of simulation systems. Dimension of box is average after
equilibrium.. .......................................................................................................... 43

Table 2.2. Calculated binding affinity (Pi values) of azo-TAB towards each
residue of lysozyme. Average is taken overall trans and cis systems
through recorded trajectories from 900ns-1000ns.. .............................................. 49

Table 2.3. Average percentages of secondary structures contents of lysozyme
intermediates during simulation course, 900 ns – 1000 ns.. ................................. 51

Table 2.4. Average structural properties of the intermediates of HEW lysozyme.
In the system T1~T10 and C1 ~ C10, the time range for averaging is over
protein trajectories 900ns - 1000ns and we adopt one protein structure per
0.5 ns. The system, P, is simulating pure protein solvated in water. The
average protein structures are over 100ns-200ns. Domain distance is
defined as the distance between two centers of mass from α and β
domains. All solvent access surface area is calculated using Gromacs
program, which employs the Lee and Richards algorithm with a probe
radius of 0.14m. Transition point is read from all plots. The reference
structure of RMSD calculation is crystal protein.................................................. 72

Table 2.5. Number of residue intra-molecular contacts and number of hydrogen
bonds in the intermediates of HEW lysozyme. Average is over 900ns-
1000ns for system T1~T10 and C1~C10 while 100ns-200ns for system P.
Two residues are considered to have contacts with each other when their
Cα atoms are within 0.6 nm. Hydrogen bonds are determined based on
cutoff radius, 0.35 nm (acceptor – donor) and cutoff angle, 30
o
(acceptor -
donor - hydrogen).. ............................................................................................... 73

Table 3.1. Summary of simulation systems. Dimension of box is average after
equilibrium.. .......................................................................................................... 91
viii

Table 3.2. Average percentages of secondary structures contents of lysozyme
intermediates overall simulation course, ~27 ns.. ............................................... 104

Table 3.3. (A), (C), (a), (c) are the Residence (%) of which the distance between
any atom on mainchain (sidechain) of the α domain/the β domain/the
protein and any atom of surfactants is within 0.4 nm. (B), (D), (b), (d) are
average residence time (ns) while the distance between any atom on
mainchain (sidechain) of the α domain/the β domain/the protein and any
atom of surfactants is within 0.4 nm.. ................................................................. 109

Table 4.1. Summary of 21 simulation system for RNase; ten trans-azoTABs
system (T1~T10); ten cis-azoTABs (C1~C10) system; one pure protein in
water (P). Dimension of box is average after equilibrium.................................. 128

Table 4.2. List of the average dynamic affinity, in which trans or cis azoTABs
interact with the protein molecules, RNase, at different extent.. ........................ 129

Table 4.3. Average percentages of secondary structures contents of Ribonuclease A
intermediates induced by the surfactants during all of simulation courses
collecting from every simulation system............................................................. 130

Table 4.4. Average structural properties of the intermediates of Ribonuclease A.
In the system T1~T10 and C1 ~ C10, the time range for averaging is over
protein trajectories 1,200ns – 1,300ns and we adopt one protein structure
per 0.5 ns. The system, P, is simulating pure protein solvated in water and
its average protein structures are over all simulation time. α-β layer
distance is defined as the distance between two centers of mass from α and
β layers. All solvent access surface area is calculated using Gromacs
program, which employs the Lee and Richards algorithm with a probe
radius of 0.14m. Transition point is read from all plots. The reference
structure of RMSD calculation is crystal protein................................................ 143

Table 4.5. Number of residue intra-molecular contacts and number of hydrogen
bonds in the intermediates of HEW lysozyme. Average is over 1,200ns-
1,300ns for system T1~T10 and C1~C10 while total simulation course for
system P. Two residues are considered to have contacts with each other
when their Cα atoms are within 0.6 nm. Hydrogen bonds are determined
based on cutoff radius, 0.35 nm (acceptor – donor) and cutoff angle, 30
o

(acceptor - donor - hydrogen).. ........................................................................... 144

Table 5.1. Summary of simulation systems (Alpha-Lactalbumin). Dimension of
box is average after equilibrium... ...................................................................... 155
ix

List of Figures

Figure 1.1. The hierarchy of protein structure. Primary (amino acid), secondary
(α-helix or β-sheet), tertiary (three-dimensional structure of a folded
protein), and quarternary(mixture of protein molecules). [1-2]........................……4

Figure 1.2. Structure of azobenzene trimethylammonium bromide surfactant
(azoTAB) upon the UV light or visible light exposure.[5] .................................... 6

Figure 1.3. UV-vis absorption spectrum of azoTAB.[11]................................................... 7

Figure 1.4. Structures of trans (left) and cis (right) azobenzene H5C6N=NC6H5 [12]........ 8

Figure 1.5. Schematic illustration of experiment measurement on detecting the
characteristics of protein intermediates [27].......................................................... 11

Figure 1.6. Illustration of the solvent accessible surface area (SASA). The probe
is given in blue and the targeted molecule is colored in red shown as the
van der Waal surface. As the probe rolls over the van der Waal surface, its
center has the trace of the dashed lines forming the accessible surface.[31]........ 16

Figure 1.7. Structure of native lysozyme (6LYZ.pdb). The alpha and beta domains
are shaded red and green, respectively. The active site is located at the
interface of these domains. The locations of the six tryptophan (28, 62, 63,
108, 111, 123) side chains and four disulfide bonds (6-127, 30-115, 64-80,
76-94) are indicated.[5].......................................................................................... 19

Figure 1.8. Schematic free-energy surface representing features of the folding of
hen lysozyme as present by Dinner et al. [37] The yellow trajectory
represents a “fast track” where both domains of the protein form
concurrently (forming I α β intermediates). The red trajectory represents a
“slow track” where the protein becomes trapped in a long-lived
intermediate (I α) with persistent structure in only the  domain. ......................... 21

Figure 1.9. Crystal structure of Ribonuclease A molecule. The 3D structure of the
protein is drawn as New Ribbons with the VMD program [5]. ............................ 23

Figure 1.10. Crystal structure of α-Lactalbumin molecule. The 3D structure of the
protein is drawn as New Ribbons with the VMD program [5]. ............................ 25

Figure 2.1. Crystal structure of hen egg white (HEW) lysozyme adopted from
x

6LYZ.pdb. The α domain is colored in red and it has the α-helices A, B, C,
D plus the C-terminal 310 helix. The β domain is colored in green with a
three-stranded β-sheet and a 310 helix. Each disulfide bonds are shown as a
pair of two disulfide atoms (yellow spheres). There are two disulfide
bonds in the α domain, one in the β domain, the other connecting α and β
domain. And, six Trp residues (28, 62, 63, 108, 111, 123) are indicated.
VMD program [30] is adopted.............................................................................. 37

Figure 2.2. Three-dimensional structures of trans-azoTaB (left) and cis-azoTAB
(right). Alkyl group CH3 and CH2 are treated as typical atoms in Gromos
force field. ............................................................................................................. 45

Figure 2.3. Average binding affinity (Pi value) of azo-TAB toward each residue of
lysozyme. Data are listed in Table 2.2. Trans-azoTAB binding affinity is
shown as the black line and cis-azoTAB binding affinity is the red line.
Secondary structure of crystal lysozyme (native state) is indicated along
the axis of residue number.. .................................................................................. 48

Figure 2.4 (a) Binding Affinity of trans-azoTAB toward lysozyme. (b) Binding
Affinity of cis-azoTAB toward lysozyme. Residues with high binding
affinity, Pi > 1.0, are colored in red; low binding affinity, Pi <0.7 in blue;
and moderate binding affinity, 0.7< Pi <1.0, in yellow.(c) Hinge is in
violet color. (d) C-subsite is in purple color. All plots are shown as New
Ribbons with VMD program [30]......................................................................... 50

Figure 2.5. Time evolution of the protein secondary structure in T5 system with
DSSP program. On the right hand side, we indicates location of helix A,
helix B, helix C, helix D, the two 310 helices and the β sheet............................... 53

Figure 2.6. SANS images compared with P and T5 simulation results. (a) A
simulation snapshot of protein in P system at 200ns. (b) SANS image of
pure protein in water. (c) Simulation snapshot of protein in T5 system at
1300ns. (d) SANS image of protein in 12.2 mM azo-TAB solution under
visible light. Plots of simulation snapshot are obtained by VMD program
[30].. ...................................................................................................................... 56

Figure 2.7. Time evolution of the solvent access surface area (SASA) calculation
in T5 system. Blue line represents the protein solvent access surface area
and the green line stands for the interface between protein and ten trans-
azoTABs. The transition point happens around 450ns, and higher
amplitude (fluctuation) of SASA (protein) but much less amplitude
(fluctuation) of protein-azoTAB interface after 450 ns. ....................................... 60

Figure 2.8. Root mean-square deviation (in nm) of all atoms of protein as a
function of time (ns) with respect to the crystal structure for T5 and P
system ................................................................................................................... 62
xi

Figure 2.9. Root mean-square deviation (in nm) of all atoms of protein as a
function of time (ns) with respect to the crystal structure for P system.
This plot is zoomed in from lower left side of Figure 2.8 .................................... 63

Figure 2.10. Radius of gyration (nm) of the whole protein, the α domain, and the β
domain for the protein in T5 sytem as a function of time (ns). The
transition point happens around 450ns, and higher amplitude (fluctuation)
after 450 ns............................................................................................................ 64

Figure 2.11. Domain Distance (protein in T5 system) – distance (nm) between the
centers of mass of the α-domain and the β-domain as a function of time
(ns). The transition point happens around 450ns, and higher amplitude
(fluctuation) after 450 ns....................................................................................... 65

Figure 2.12. Percentage of contacts (%) of the whole protein in T5 system. The
percentage is based on the total native contacts of the crystal lysozyme.
Total contacts are summed up of native contacts and non-native contacts.
The transition point happens around 450ns, and higher amplitude
(fluctuation) after 450 ns....................................................................................... 67

Figure 2.13. Number of total contacts of the whole protein, the α domain, the β
domain, and the interface between α-β domain for T5 system. The
transition point happens around 450ns, and higher amplitude (fluctuation)
after 450 ns............................................................................................................ 67

Figure 2.14. Number of native contacts of the whole protein, the α domain, the β
domain, and the interface between α-β domain for T5 system. The
transition point happens around 450ns, and higher amplitude (fluctuation)
after 450 ns............................................................................................................ 68

Figure 2.15. S-S bond length (CYS6 and CYS 127, both are in the α domain) is
calculated for protein in T5 system. The transition point happens around
450ns, and higher amplitude (fluctuation) after 450 ns. ....................................... 68

Figure 2.16. S-S bond length (CYS30 and CYS 115, both are in the α domain) is
calculated for protein in T5 system. The transition point happens around
450ns, and higher amplitude (fluctuation) after 450 ns. ....................................... 69

Figure 2.17. S-S bond length (CYS64 and CYS 80, both are in the β domain) is
calculated for protein in T5 system. The transition point happens around
450ns, and higher amplitude (fluctuation) after 450 ns. ....................................... 69

Figure 2.18. S-S bond length (CYS76 in the α domain and CYS 94 in the β
domain) is calculated for protein in T5 system. The transition point
happens around 450ns, and higher amplitude (fluctuation) after 450 ns. ............. 70
xii

Figure 2.19. Number of hydrogen bonds of the whole protein, the α domain, the β
domain, and the interface between α-β domains in T5 system. The
transition point happens around 450ns, and higher amplitude (fluctuation)
after 450 ns............................................................................................................ 71

Figure 2.20. The relative effective collisions in percentage of azotabs to
lysozyme. There are two sets of percentages; the percentage shown inside
the brackets is based on the effective collision distance of 0.3 nm while the
other is based on 0.4 nm. The phenyl rings have higher relative collision
percentage for both trans and cis structures.. ........................................................ 75

Figure 3.1. Crystal structure of hen egg white (HEW) lysozyme adopted from
6LYZ.pdb. The α domain is colored in red and it has the α-helices A, B, C,
D plus the C-terminal 310 helix. The β domain is colored in green with a
three-stranded β-sheet and a 310 helix. Each disulfide bonds are shown as a
pair of two disulfide atoms (yellow spheres). There are two disulfide
bonds in the α domain, one in the β domain, the other connecting α and β
domain. And, six Trp residues (28, 62, 63, 108, 111, 123) are indicated.
VMD program [2] is adopted................................................................................ 86

Figure 3.2. Time evolution of the relative unfolding effects of α / β domains (ΔRg
(%) and ΔRMSD(nm) ). Time evolution of the number of hydrogen bonds
are displayed in (c) and (h) and that of the total contacts are displayed (d)
and (i) between α – β domain interface. The α – β domain distance with
time is shown on (e) and (j). Blue cross in (a)~(e) stands for the 25994
intermediates induced by 10 trans-azoTABs. Blue cross in (f)~(j) stands
for the 25746 intermediates induced by 10 cis-azoTABs. The red cross
represents the 524 lysozyme intermediates in water without azoTABs ............... 94

Figure 3.3. The α and β domains of lysozyme intermediates. Increase difference
of protein intermediates ΔRg (%) - (a),(b) in the α and β domain.
RMSD_α (nm) and RMSD_β (nm) are shown in (c) and (d). Blue cross in
(a) and (b) stands for the 25994 intermediates induced by 10 trans-
azoTABs. Blue cross in (c) and (d) stands for the 25746 intermediates
induced by 10 cis-azoTABs. The green cross represents the 524 lysozyme
intermediates in water without azoTABs. Red lines are simply the
diagonals where ΔRg_ α (%) = ΔRg_ β (%) and RMSD_ α (nm) =
RMSD_ β (nm). .................................................................................................... 97

Figure 3.4. The normalized distributions are based on the 25994 intermediates
induced by 10 trans-azoTABs, the 25746 intermediates induced by 10 cis-
azoTABs and the 524 lysozyme intermediates in water without azoTABs.
The black squares represent the crystal structure. The curves in the right
are showing the α domains and the curves in the left are showing the β
domains. ................................................................................................................ 98
xiii

Figure 3.5. The normalized distributions are based on the 25994 intermediates
induced by 10 trans-azoTABs, the 25746 intermediates induced by 10 cis-
azoTABs and the 524 lysozyme intermediates in water without azoTABs.
The black squares represent the crystal structure. The curves in the right
are showing the α domains and the curves in the left are showing the β
domains. The widely spread of the helix angle distribution implies the
large degrees of motions of the helices, making the loose packing of
helices in the α domain. ...................................................................................... 102

Figure 3.6. Time evolution of the protein secondary structure in T5 system with
DSSP program. On the right hand side, we indicates location of helix A,
helix B, helix C, helix D, the two 310 helices and the β sheet............................. 106

Figure 4.1. The picture of Ribonuclease A molecule. Three helices are shown in
purple and from left to right, they are helix A, helix B, and helix C. The β
sheets are printed in yellow. The α-layer is in front but the β-layer is in the
back. We keep the same orientation of the protein in this chapter,
including the protein shape reconstruction from SANS data.............................. 119

Figure 4.2. Molecular structures of the surfactants, azoTAB. Trans structure is on
the left and the right is the cis structure. As Gromos force field is applied,
alkyl group CH3 and CH2 are treated as typical atoms. ...................................... 124

Figure 4.3. Average Dynamic Affinity of the surfactants, azoTABs, on the protein
Ribonuclease A all over ten simulation system across whole simulation
duration. The black line stands for the the dynamic affinity of trans-
azoTABs toward RNase while the red line is the dynamic affinity for cis-
azoTABs on RNase. Distribution of the secondary structures of crystal
RNase is along the axis of the residue number. The moderate dynamic
affinity is defined between 0.7~1.0, in where two horizontal dashed lines
are shown in light blue. ....................................................................................... 128

Figure 4.4. (a) Binding Affinity of trans-azoTAB toward Ribonuclease A. (b)
Binding Affinity of cis-azoTAB toward Ribonuclease A. Residues with
high binding affinity, Pi > 1.0, are colored in red; low binding affinity, Pi
<0.7 in blue; and moderate binding affinity, 0.7< Pi <1.0, in yellow. .............. 130

Figure 4.5. Time evolution of the protein secondary structure in C6 system with
DSSP program. On the right hand side, we indicates location of helix A,
helix B, helix C, and the βI, βII, βIII sheets........................................................ 133

Figure 4.6. RNase: SANS images compared with P and C2 simulation results. (a)
Crystal Structure (1RBX) (b) SANS image of pure protein (RNase) in
water. (c) Simulation snapshot of protein in C2 system at 1300ns. (d)
SANS image of protein (RNase) in 8.3 mM azo-TAB solution under
xiv

visible light. Plots of simulation snapshot are obtained by VMD program
(64). ..................................................................................................................... 135

Figure 4.7. Time evolution of RMSD in T8 system. The three curves specifically
represent RMSD of whole protein (blue line), RMSD of α-layer (red line)
and RMSD of β-layer (green line). The RMSD of α-layer (red line) is
leading among the three and the RMSD of β-layer is under the other two
curves. ................................................................................................................. 137

Figure 4.8. Time evolution of radius of gyration in T8 system. We have calculated
the gyration radius for the α-layer (red line) and β-layer (green line) and
the whole protein (blue line). Due to the small portion of the α-layer
within the protein, the gyration radius of α-layer (red line) is lower than
the other two curves but has the most increase than the other two ..................... 137

Figure 4.9. Time evolution of number of contacts within the protein in T8 system.
The plot is showing total contacts (blue curve), native contacts (red curve)
and non-native contacts (green curve). Total contacts remain almost ~300
because the decrease of the native contacts (~200) is compensated by the
increase of non-native contacts (~100). After 800ns, the number of native
contacts or non-native contacts persists in a range of constants. ........................ 138

Figure 4.10. Time evolution of number of native contacts within the protein in T8
system. The number of the total native contacts within the crystal structure
of RNase molecule are ~300; however, it decreases to ~200 after 800ns
(blue curve), in which the native contacts of the α-layer drops down more
than that of the β-layer.. ...................................................................................... 139

Figure 4.11. Time evolution of distance between the two layers, α layer and β
layer in T8 system. The distance decreases over most of time although it
increase during 200~300 ns.. .............................................................................. 139

Figure 4.12. Time evolution of the number of the hydrogen bonds within in the
protein (blue line), the α layer (red line), the β layer (green line) and the
interface between the two layer (purple line) in T8 system. Obviously, the
number of the hydrogen bonds in the β layer increases but that in the α-β
layer interface decreases. .................................................................................... 140

Figure 4.13. The time evolution of solvent accessible surface area (SASA) in T8
system. The blue line represents the SASA calculation of the protein and
the purple line represents the interface between the protein and 10
surfactants. .......................................................................................................... 140
xv

Figure 5.1. (a) Crystal structure of α-lactalbumin (b) Crystal structure of hen egg
white lysozyme. The VMD program is used to show the protein secondary
structures in New Ribbons style. [2]................................................................... 155

Figure 5.2. Binding Affinity of trans-azoTAB toward alpha-lactalbumin (black
line). Binding Affinity of cis-azoTAB toward alpha-lactalbumin (red line). ..... 156

Figure 5.3. (a) Binding Affinity of trans-azoTAB toward α-lactalbumin. (b)
Binding Affinity of cis-azoTAB toward α-lactalbumin. Residues with high
binding affinity, Pi > 1.0, are colored in red; low binding affinity, Pi <0.7
in blue; and moderate binding affinity, 0.7< Pi <1.0, in yellow. The two
New Ribbons plots are constructed by VMD program [2]. ................................ 156

Figure 5.4. Time evolution of the protein secondary structure in C5 system with
DSSP program [5-6]. .......................................................................................... 157

xvi

Abstract

How proteins fold and unfold has been a great focus for decades. Techniques of
molecular dynamics simulations provide the atomic insight of protein folding/unfolding.
Proteins solvated in water remain well at the native structures under room temperature.
Being perturbed by a small amount of photoresponsive surfactants, azoTAB, at room
temperature, protein molecules, such as lysozyme, ribonuclease A, and α-lactalbumin,
encounter the conformational changes and partially unfold, especially in the α domain.
We conduct molecular dynamics simulation in microseconds and through analysis of the
structural properties of protein intermediates as functions of time, we demonstrate that
the surfactant-unfolded intermediates of protein molecules, owning the unfolded α-
domain but the relatively intact β-domain, although the hydrophobic interaction is higher
in the α domain than the β domain. The increased internal dynamics of partially-unfolded
protein molecules induced by azoTABs is potentially contributed to the increase
enzymatic activity of protein. Molecular dynamics simulation has offered supporting
evidence to better understand experimental phenomena.

1

Chapter 1: Introduction

1.1 Protein Structures
Proteins are fundamental building blocks of all living cells such as enzymes,
hormones, and antibodies. They are also essential diet of animals for the growth and
repair of tissue. A protein is made of amino acids arranged in a chain as the primary
structure and then folded into a globular form, the tertiary structure. Each type of protein
has its own unique sequence of amino acids and any two amino acids are connected by
the peptide bond. There are 20 standard amino acids. Some of their side chains are
nonpolar and hydrophobic (water-fearing); others are polar and hydrophilic (water-
favoring) (Table 1.1). The distribution of the polar and nonpolar amino acids governs the
folding of any protein. To avoid contact with the water that surrounds them, the nonpolar
(hydrophobic) side chains in a protein tend to cluster in the interior of the molecule. In
contrast, polar side chains tend to arrange themselves near the outside of the molecule,
where they can form hydrogen bonds with water and with other polar molecules. The
folding of a protein chain is further constrained by the different sets of weak noncovalent
bonds, such as hydrogen bonds, ionic bonds, and van der Waals attractions. Although the
individual noncovalent bonds are 30–300 times weaker than the typical covalent bonds,
they can act in parallel to hold two regions of a polypeptide chain tightly together and
stabilize the folded shaped of the protein.
2

An illustration of hierarchy (four levels) of protein structure is shown below
(Figure 1.1). The primary structure of the protein is defined by the specific sequential
arrangement of the amino acid residues. The repeated link of two amino acid residues by
a peptide bond forms the main chain/backbone of a protein. The hydrogen-bonding
between the N–H and C=O groups in the polypeptide backbone results in two regular
folding patterns, α helix and β sheet. The α helix is characterized by hydrogen bonds
along a chain while the β sheet is characterized by hydrogen bonds crossing between
chains. A right handed helical structure has average torsion angles Φ = -57
◦
and Ψ = -47
◦

while a parallel pleated sheet structure has Φ=-119
◦
and Ψ=113
◦
. The α helix and β sheet
are so called the secondary structure of the protein and they are further folded into a
compact globule, the tertiary structures, which refer to three-dimensional structure of a
single protein molecule, a native state. Interactions stabilizing the tertiary structure are
disulfide bonds, hydrophobic interactions, hydrogen bonds and ionic interactions. Single
polypeptides can associate with each other to form larger protein complexes, called
quaternary structures. Individual polypeptides in protein complexes are referred to as
subunits. Most enzymes are complexes of proteins and the symmetry and stoichiometry
of the complexes is crucial for their activity.

3

Table 1.1: The 20 standard amino acids. Some side chains carry charges; others do not.
[1]

Amino Acid

Primary Structure – “Amino Acid Sequence”

Secondary Strucure – “Regular Sub-Structure”

α-helix β-sheet
4

Tertiary Struture – “three-dimensional structure”

Quarternary – “Complex of protein molecules”

Figure 1.1 : The hierarchy of protein structure. Primary (amino acid), secondary (α-helix
or β-sheet), tertiary (three-dimensional structure of a folded protein), and
quarternary(mixture of protein molecules). [1-2]

The structure of a protein largely determines its biological function. A great deal
of work has been investigating the form-function relationship.

1.2 Molecular Dynamics
Molecular dynamics (MD) is a computer simulation technique. The time
evolution of a set of interacting atoms is followed by integrating their equations of
motion. It is based on the laws of classical mechanics, and notably Newton's law.
Fi = mi ai
5

Where Fi is the force acting on atom i with mass, mi, and ai is its acceleration. The
starting point is at t=t0, and the positions of the atoms are xi(t0) and the velocities of them
are vi(t0). Within a very small time change, δt, one can calculate xi(t0+ δt) through .
Knowing the new positions and velocities, the procedures can be repeated again and
again. The force Fi at time t is determined by the potential energy function V :
Fi = V
i
 
Where V
i
 is the gradient of the potential energy V with respect to the position
coordinates of atom i. In the Gromos force field, the potential energy can be calculated as
the sum of six terms:
) ( ) ( ) ( ) ( ) ( ) ( ) ( x V x V x V x V x V x V x V
dih imp angle bond lj ex
     
) (x V
ex
describes the Coulombic interaction between electrostatic charges. ) (x V
lj
is
the Lennard Jones interaction energy, which reflects the van der Waals interaction
between atoms. ) (x V
bond
and ) (x V
angle
represent the bond stretching and bond angle
potential energy. ) (x V
imp
keeps groups of atoms into a spatial configuration, such as
planar or tetrahedral. Last, ) (x V
dih
is a dihedral torsional angle potential. Gromacs manual
[3] has more details of each term. The time step δt is limited by the highest bond-
stretching motions; hence, the bond length constraint algorithm (SHAKE [3-4]) is applied
to keep bond length rigid some or less.

6

1.3 Photoresponsive Surfactant (azoTAB)
Azobenzene trimethylammonium bromide surfactant (azoTAB) (Figure 1.2) is a
type of cationic surfactant. It is composed of a trimethylammonium head group owning
the hydrophilic capacity and a hydrophobic tail, which includes the photo-switchable azo-
benzene groups, trans and cis-azobenzene. Upon illumination with visible light (434 nm)
or UV light (350 nm), azoTAB undergoes a photoisomerization. The AzoTAB solution
exist most trans isomers (75:25, trans:cis) under visible light (75:25, trans:cis) but the
dominant cis isomers (>90% cis) in the UV light.

Figure 1.2 : Structure of azobenzene trimethylammonium bromide surfactant (azoTAB)
upon the UV light or visible light exposure. The 3D two structural types (trans and cis) of
the surfactants are plotted in VDW drawing style in the VMD program. [5].

UV-Vis spectroscopy is as shown in Figure 1.3, with the trans isomer exhibiting a
maximum absorbance at 350 nm and the cis isomer at 434 nm. The dipole moment across
the –N=N– is ~0.5D for the trans isomer (planar) compared to ~3.1D for the cis form
(bent). As a result, the trans isomer is significantly more hydrophobic than the cis form.
Therefore, trans-azoTAB has a higher affinity for protein binding while cis-azoTAB has
7

a higher affinity for water. These photoresponsive surfactants have shown the ability to
control the protein structure (secondary [6], tertiary [7] and quaternary structure [8]) and
further affect its functions [6-10]. Other tunable properties of the photoresponsive
surfactants like surface tension and electric tension are applied on the oil-water interface
system and microemulsions.

Figure 1.3 : UV-vis absorption spectrum of azoTAB. [11]
In our simulation, the GROMOS 45a3 force field was applied on both lysozyme
and the surfactant, azoTAB. The tail of the surfactant, azoTAB, carries the photo-
switchable azo functional unit (H5C6N=NC6H5), see Figure 1.4. The whole set of
reparametrization to this azo unit is from Böckmann’s previous work [12]. They tried to
derive the MM force field of trans-AB (trans-azobenzene, H5C6N=NC6H5) and cis-AB
(cis-azobenzene, H5C6N=NC6H5) starting from the well-tested GROMOS 45a3 force
field. First, they carry out all-QM of ab initio molecular dynamics simulations of trans-
and cis-AB in the gas phase. Secondly, all-MM of classical force field simulations of
8

trans- and cis-AB in the gas phase are performed. Thirdly, Böckmann et al. treated the
photoswitch –N=N– fully quantum mechanically in a QM/MM framework. This allows
for maximum compatibility between a force field (MM) description and an electronic
structure based (QM) description of the photoswitch. Fourthly, the MM force field
parameters are adapted for bonded interactions and partial charges so as to yield the best
agreement between the simulation results of all-QM and all-MM. Fifthly, the whole set of
this derived force field is employed on the 4,4′-dioctyloxyazobenzene liquid crystal to
perform all-MM simulations of its phase transition. The results are directly compared
with experimental observations to test the validation of the force field parameters of the
azo functional unit (H5C6N=NC6H5), which are summarized on Table 1.2.

Figure 1.4 : Structures of trans (left) and cis (right) azobenzene H5C6N=NC6H5 [12].

Table 1.2: Force field parameters of azo-functional units [12].
9

1.4 Protein Folding / Unfolding
1.4.1 Levinthal's Paradox
The Levinthal paradox [13-15] illustrates that a protein with 101 amino acids could
exist in 3
100
= 5 x 10
47
configurations considering each bond connects amino acids can
have three possible states. If the protein is able to sample new configurations at the rate
of 10
13
per second, or 3 x 10
20
per year, it will take 10
27
years to try them all. In fact,
proteins do fold in a time scale of seconds or less. Levinthal concluded that random
searches are not an effective way of finding the correct state of a folded protein. They do
not explore exhaustively their conformational space on the way to their native structure.
Protein folding is speeded and guided by the rapid formation of local interactions which
then determine the further folding of the peptide; this suggests local amino acid
sequences which form stable interactions serve as nucleation points in the folding
process.
A simple model [15] considers the connecting bond between two neighboring
amino acids can be characterized as "correct" or "incorrect." By imposing an energy cost
of a few kT for locally incorrect bond configurations, the first-passage time to the fully
correct state become very much shorter. Mathematical analysis shows a few kT can
reduce Levinthal's time to a biologically significant size. Monte Carlo folding simulations
[16] start with 200 randomly-interacted sequences. The global minimum (native state) is
known and the chain does not get trapped in local minima. Folding starts by a rapid
collapse from a random-coil state to a random semi-compact globule. It then proceeds by
10

a slow, rate-determining search through the semi-compact states. The folding mechanism
led to the resolution of the Levinthal paradox: the conformations that need to be searched
in the semi-compact globule are reduced.

1.4.2 Detection of Protein Intermediate
According to ‘Levinthal paradox’, it is impossible for proteins to search all over
structural conformations to find their final native states and there must exist ‘folding
pathways’. Intermediates, which formed between native state and totally denatured state,
play a particular role to help us understand how protein folds. Since the early 1970s, the
detection and studies of intermediates has been an important aspect to investigate the
mechanism of protein folding [17-23]. Later, it is found that not all the intermediates are
on folding pathway, some others are in fact trapped in the local minimum states, which is
on off-pathway of folding [16, 23-26]. A detected intermediate of ribonuclease A is
proved on its late folding pathway. Also, the intermediates of lysozyme (I α) were
demonstrated on the slow folding pathway while the intermediates (I αβ) were found in the
fast folding pathway. The intermediates of lysozyme (I α) is characterized as α domain
folded, β domain unfolded and the intermediates (I αβ) is with both α and β domain
partially folded. Combination of both experimental work and simulations gives us highly
comprehensive information about protein folding because the simulations contribute the
details which could not be reached by current experimental techniques.

11

(1) Experimental Measurement of Protein Intermediate Structure
As shown in Figure 1.5 and Table 1.3 , the use of a set of probes starts to provide a
not-blurred picture of protein intermediate structures. A single probe characterizes a
specific feature of unfolding protein and therefore, combination of traits delineated by
different probes give the complementary properties of proteins. With the time evolution
of intermediate structures during the unfolding process, the mechanism of protein
unfolding will be revealed.

Figure 1.5 : Schematic illustration of experiment measurement on detecting the
characteristics of protein intermediates [27]

12

Method Information
Hydrogen exchange labeling and
NMR
Formation of persistent hydrogen bonds, burial
from solvent at individual sites
Hydrogen exchange labeling and
ESMS
Detection of transient intermediates and folding
populations
Far UV CD Formation of secondary structure
Near UV CD Immobilization of aromatic residues in tertiary
structure
Intrinsic fluorescence Environment of Trp and Tyr residues
Fluorescence quenching Solvent accessibility of fluorophores
ANS binding Exposure of hydrophobic surfaces
Inhibitor binding Formation of native state
Absorbance Environment of aromatic residues
Table 1.3: Current experimental techniques used to characterize the protein intermediates
[28]

(2) Structural properties of intermediate derived from the MD simulation
In molecular dynamics simulation, all atom positions of protein are recorded for
each time step. With the positions of every protein atom, we could derive the time
evolutionary of structural properties of protein. Further, we study the unfolding / folding
event from the changing structures of the protein intermediates under specific
environment. The time scale of protein folding is normally within millisecond but some
very fast protein folding is complete in just a few microseconds. However, it is still a
challenge for the current computer technology to run the MD simulations over
13

microseconds. Some controlling factors, such as high temperature, are imposed on the
MD systems to much increase the unfolding rate. The following math definitions of the
structural properties are shown in Gromacs Manual [29].
Root Mean Square Deviations in Structure
This is a deviation measure of the unfolding protein with its crystal structure. The
root mean square deviation (RMSD) of certain atoms in a molecule with respect to a
reference structure can be calculated by least-square fitting the structure to the reference
structure (for t2 = 0, crystal structure) and subsequently calculating the RMSD.[29]

2
1
2
2 1
1
2 1
) ( ) (
1
) ; (






 


t r t r m
M
t t RMSD
i i
N
i
i

Where



N
i
i
m M
1
and ) (t r
i
is the position of atom i at time t.

Radius of Gyration
This measure gives an indication of the shape of the molecule at each time. To
have a rough measure for the compactness of a structure, we can calculate the radius of
gyration, [29]
2
1
2











i
i
i
i i
g
m
m r
R
where mi is the mass of atom i and ri the position of atom i with respect to the center of
mass of the molecule. It is especially useful to characterize polymer solutions and
proteins.
14

Domain Distance
We define domain distance as the distance between the center of mass of each
domain. For example, there are two domains of lysozyme, α domain and β domain. And,
the domain distance between α and β domains is defined as the distance along the center
of mass of α and β domain.

Mean Square Displacement
This measure gives an diffusion coefficient /diffusion rate of a molecule.
To determine the self-diffusion coefficient DA of particles A one can use the Einstein
relation, [29]
  t D r t r
A
A i
i i
t
6 0 ) ( lim
2
 

 

For molecules consisting of more than one atom, ri can be taken as the center of
mass positions of the molecules.

Root Mean Square Fluctuation
This measure gives an indication of the fluctuation of each atom with respect to its
reference position within certain time.
15

The mean square fluctuation (MSF) is a measure of the deviation between the position
of particle i and some reference position. [29-30]


 
T
t
i j i
j
x t x
T
MSF
1
2
)
~
) ( (
1
,
where T is the time over which one wants to average, and
i
x
~
is the reference
position of particle i. Typically this reference position will be the time-averaged position
of the same particle i, ie.
i
x .
Note that where the difference between RMSD and RMSF is that with the latter
the average is taken over time, giving a value for each particle i. With RMSD the average
is taken over the particles, giving time specific values.

Number of Native Contacts
This measure gives an indication of how much the native state tertiary structure of
an unfolding protein is reserved. Tertiary structure of a protein molecule refers to its three
dimensional arrangement for each secondary structure and the relative positions of the
protein atoms. The precise three-dimensional tertiary structure is required for a protein to
perform its function correctly. We consider that there exists native contacts between two
different residues i, j if the distance between the Cα atoms of the two residues, i, j, is
16

within 0.6nm. The total native contacts of the crystal structure of lysozyme from
6LYZ.pdb are 365.

SASA (Solvent Accessible Surface Area)

Figure 1.6 : 2-D plot of demonstrating the solvent accessible surface area (SASA). The
probe is given in blue and the targeted molecule is colored in red shown as the van der
Waal surface. As the probe rolls over the van der Waal surface, its center has the trace of
the dashed lines forming the accessible surface. [31]

The hydrophobic interactions make an important role to bundle the nonpolar
amino acids inside the folded structure of the protein while immersed in water to avoid
the unfavorable polar-nonpolar interaction. The polar amino acids with charged side
chains typically reside on the surface of the protein and have quick access to water. Once
the protein is denatured, the bundled interior of the protein could be exposed to water.
17

The SASA (solvent accessible surface area) is the surface area of a protein
molecule which is possibly and easily accessible to a solvent. Lee and Richards first
described SASA in 1971 [32]. The algorithm of SASA is to use a 'rolling ball' and probe
over the surface of the protein molecule, as shown in Figure 1.6. Normally, the 'rolling
ball' is set to the similar radius as a water molecule, say 0.14 nm. The calculation of
SASA reveals the unfolding extent of a protein where there are more nonpolar residues
are exposed in water. The maximum charge of a hydrophobic atom is set to 0.2 e (|e|
≤0.2).
Secondary Structure (DSSP Program)
Generally, an unfolding protein loses tertiary structure first and followed by the
damage of secondary structure. Once the secondary structure is broken, the protein is
going to be totally unfolded. The secondary structure is formed by hydrogen bonds.
Different hydrogen bond patterns form different secondary structures. DSSP program
[33] is often used on determining the protein secondary structure from the MD simulation
trajectories. Although the exact definition of a hydrogen bond should be critical, DSSP
provides an electrostatic model to identify formation of hydrogen bond. The electrostatic
energy is given by
. / 332 *
1 1 1 1
2 1
mol kcal
r r r r
q q E
CN OH CH ON






   
18

If E is less than -0.5 kcal/mol, a hydrogen bond exists. The partial charges of the
carbonyl carbon and oxygen is assigned as e q 42 . 0
1
  , respectively, and charges of the
amide nitrogen and hydrogen is . 20 . 0
2
e q  
DSSP characterizes eight types of secondary structure based on different
hydrogen bond patterns. G = 3-turn helix (310 helix), H = 4-turn helix (α helix), I = 5-
turn helix (π helix), T = turn, E = extended strand in β-sheet conformation, B = β-bridge,
S = bend, C = coil.

1.5 Lysozyme
Hen Egg White Lysozyme has 129 amino acid residues with a molecular weight of
~14 kDa. Lysozme is positively charged (+8 charges) at neutral pH (pI = 11.0). It has a
wide distribution, including mammals, birds, reptiles, fishes, and insects. Lysozyme is an
enzyme which functions to hydrolyze the β(1-4) glycosidic bond between residues of N-
acetylmuramic acid (NAM) and N-acetylglucosamine (NAG) in certain polysaccharides
of bacterial cell walls. In this way, lysozymes can protect the host from bacterial
infections.
19

Figure 1.7 : Structure of native lysozyme (6LYZ.pdb). The alpha and beta domains are
shaded red and green, respectively. The active site is located at the interface of these
domains. The locations of the six tryptophan (28, 62, 63, 108, 111, 123) side chains and
four disulfide bonds (6-127, 30-115, 64-80, 76-94) are indicated. The 3D structure of the
protein is drawn as New Ribbons with the VMD program [5].
The mechanism of lysozyme folding is well studied through the characterization
of intermediates either detected in experimental work or the computer simulation. In the
lab, lysozyme molecules are first denatured and refolded again upon removing the
denaturing condition, for example, diluting the protein solution. During the refolding
process, lysozyme molecules rapidly turn to hydrophobic collapsed states [34]; then, the
native-like secondary structure are formed and the secondary elements further assemble
into two separate domains, ultimately docking two domains together and leading to the
fully native state.
Proton exchangeability revealed that different parts of lysozyme has different
folding kinetics [35]. Even in the same solution, different lysozyme molecules fold along
kinetically distinct pathways. It has been concluded that folding of lysozyme involves
6-127
30-115
64-80
76-94
28
62
108
63
111
123
20

multiple parallel pathways. As shown in Figure 1.8, two folding pathways are marked in
red (slow-track) and in yellow (fast-track) on the free-energy surface. Along slow track,
it encounters a minimum while the free energy of fast track decreases monotonically.
Among the folding molecules, 70% folds on the slow pathway where the -helical
domain folds faster than the -sheet domain [36]. The intermediate, I α, accumulates on
the free-energy minimum of the slow-track with the structured α-domain but an unstable
β-domain. 20% of folding intermediates, I αβ, are found on the fast pathway where a
native-like level of protection against exchange in both domains is found [37]. About
10% of molecules refold in a very slow reaction because of cis-trans isomerisation of
proline residues [38]. The final step, also the rate-limiting step, in folding process is to
correctly dock the residues from one domain into the other and form the α-β interdomain
interface.

21

Figure 1.8 : Schematic free-energy surface representing features of the folding of hen
lysozyme as present by Dinner et al. [37] The yellow trajectory represents a “fast track”
where both domains of the protein form concurrently (forming I αβ intermediates). The red
trajectory represents a “slow track” where the protein becomes trapped in a long-lived
intermediate (I α) with persistent structure in only the  domain.

As major amount of intermediates are I α, most of previous experimental work
detected intermediates with an unfolded β-domain but a structured α-domain [39-41]. It
has been considered that the β-domain is less stable than the α-domain in lysozyme [42]
and the α-domain is more tightly bounded than the β-domain. On the other hand, the
unfolding simulations of hen egg white lysozyme at 500 K [43] had the β-domain
destabilized first and then the α-domain follows. Under urea solution, lysozyme was
denatured [44] while the greater increase of mean radius of gyration of the unfolding β-
22

domain (~0.5 nm) than the α-domain (~0.2 nm). Such an interpretation is supported by
the results of the MD simulations [45-46]. Surfactants usually act as denaturants which
can unfold proteins totally; nevertheless, at low concentrations of surfactants, azoTAB,
stabilize the intermediately-folded states of lysozyme in solution[7]. The SANS data
combined with FT-IR spectroscopic analysis of the protein secondary structure
characterize the azotab-induced intermediates as Iαβ owing the unfolding α-domain but
the relatively intact β-domain. Upon visible light, azoTAB enhance the activity of
lysozyme up to 8 fold [10] by increasing the flexibility of lysozyme and hinge-bending
motion.
In the present study, we conducted a series of the molecular dynamic simulations
of lysozyme-surfactants solution (pH=7.0) at 300K and then the one microsecond-
trajectory data are analyzed. The simulation results agree well with experimental work as
the binding surfactants, azoTAB, increase the internal dynamics of lysozyme and also the
surfactants unfold the α domain of lysozyme while keeping its β domain relatively intact.
It is considered that the α-domain of the native state of lysozyme is much more highly
bounded than its β domain but the azoTAB unfolds the α domain of lysozyme and can
stabilize the intermediately folded states.

23

1.6 Ribonuclease A (RNase A)

Figure 1.9 Crystal structure of Ribonuclease A molecule. The 3D structure of the protein
is drawn as New Ribbons with the VMD program [5].
Ribonuclease A (RNase A) is one of the most studied proteins in the 20
th
century,
not merely because of its small size (124 residues, ~13.7 kDa) but also its important
function of cleaving single-stranded RNA. The “A” is meant to be the predominant form
of enzyme from the pancreas of Bos Taurus. [47] RNase A has an kidney-like shape with
the active site in the cleft, as shown in Figure 1.9. It is characterized as a two-layer α+β
protein, the α layer having three α helices (residues 3-13, 24-34, and 50-60) from the N-
terminal half and the β layer consisting of three β-hairpins (residues 61-74, 79-104 and
105-124) from C-terminal half. Four disulfide bonds (Cys26-Cys84, Cys58-Cys110,
Cys40-Cys95, and Cys65-Cys72) are formed in the native state of RNase molecule. The
first two disulfide bonds are connecting between the α layer and the β layer, rendering a
small hydrophobic pocket in its vicinity as well as the essential conformational folding.
24

The latter two are less essential because the native structure of RNase A will not be
affected even if one of the two disulfide bonds is reduced.
RNase A is positive charge protein under neutral pH solution with pI=9.63. The
three essential residues for catalysis are His12, Lys 41, and His 119. These histidine (His
12 , His 119) can both accept electrons (acid) and donate electrons (base) and render the
catalyst reaction dependent on pH values in the solution. As RNase A and lysozyme are
very similar proteins, it has fast and slow track under certain unfolding conditions. At
65 °C and low pH 1.5~3.8, RNase A molecules encounter thermal denaturalization and
appear no stable secondary structures since there is no significant hydrogen-exchange
protection based on NMR study [48]. This thermal denaturation also has been studied and
proven as a reversible, two-state transition [49-51]. On the other hand, the chemical
denaturants can induce more unfolding of RNase molecules than the thermal denaturants
as the increase of the gyration radius of the RNase molecules in the chemical denaturants
is larger than that of RNase under thermal denaturants [52].
In this study, we will perform molecular dynamics simulations in microseconds to
investigate the unfolding event of RNase A by the effect of the surfactants, azoTABs,
either trans or cis. RNase molecules carry positive charges and have a small size of
hydrophobic groups. The surfactants, azoTABs, in trans form have higher hydrophobicity
than in the cis form. Our simulations are tracing the trajectories of each atom of the
RNase molecules and reveal the structures of the protein intermediates which are
partially-unfolded and the α-layers are unfolding quicker than the β domain under the
25

very low concentration of surfactants in cis form (<~10mM), and even lower
concentration of surfactants in the trans form.

1.7 α-Lactalbumin

Figure 1.10 Crystal structure of α-Lactalbumin molecule. The 3D structure of the protein
is drawn as New Ribbons with the VMD program [5].
The protein molecules, α-lactablumin (see Figure 1.10), share a common three
dimensional tertiary structures with c-type lysozyme molecules mainly because of
35%~40% sequence homology. However, these two proteins perform different functions.
Lysozyme bind and cleave the glycosidic bond in sugars; on the other hand, α-
lactalbumin does not bind sugar but involved in the synthesis of lactose. The main
structural characterizations of these two proteins are their two domains, the α domain and
the β domain which are separated by a cleft. As human α-lactablumin an example, it has
26

123 residues, four α-helices (A, B, C and D) from the residues 1~38 and 83~123 form the
α domain while three β-strands (S1, S2, S3) and the loop from the residues 39~82
constitute the β domain. Calcium binding is important for the human α-lactalbumin
molecule to maintain its structural integrity [53-56] and the calcium binding site is
situated at loop on the bottom of the cleft. There are two hydrophobic cores within α-
lactablumin molecules, one is in the α domain including helix A, B, 310 ; the other is
made up of the residues from not only the α domain ( helix C and D ) but also the β
domains.
Not every protein folds through molten globule intermediates; however, under
various conditions, the α-lactalbumin molecules can fall into the molten globule states
[57-59]. In its molten globule state, the α domain is ordered but the β domain is lack of
order [60-62]. Protein dissection technique is to characterize the protein intermediate
states using peptide models and the investigation of α-lactalbumin molecules
demonstrates that a minimum core structure of 58 residues can sufficiently form a molten
globule state [63]. However, deletion of the A, D, or C-terminal 310 helical regions
destroys the stability of the hydrophobic core, much increasing the difficulty of the
formation of the molten globule state. In addition, helix D plays an important role of
stabilizing the molten globule intermediate. Otherwise, the distribution of the
hydrophobic and hydrophilic residues may determine if a protein can form into a molten
globule intermediate [64-65].
27

The α-lactalbumin molecules have strong binding with ions such as Ca2+, Mn2+,
Mg+2, Zn2+, K+ and Na+, resulting in increases of protein stability. These ions bindings
modulate the interactions of the α-lactalbumin molecules with peptides, membranes,
proteins, and substrates.
It had been shown that the reversible control of lysozyme and RNase A’s
structures with the photo-responsive surfactants, azoTABs, allows for manipulating
control of their functions reversibly [11]. Due to 35%~40% sequence homology with
lysozyme molecules, α-lactalbumin molecules have been interested and been studied to
understand the issue of how proteins fold decoding from the residue sequence while
comparing with the lysozyme molecules. We adopted molecular dynamics simulation to
investigate the unfolding event of the α-lactalbumin molecules triggered by the
surfactants, azoTABs. Although the α-lactalbumin molecules share common 3D
structures with lysozymes molecules, our simulations discover the different binding sites
of the surfactants toward them and therefore different unfolding pathways are revealed.
The time evolution of secondary structure analysis has shown that helix C of the α-
lactalbumin molecules encounters dramatic early unfolding by perturbation of azoTABs;
on the other hand, helix C of lysozyme molecules remain almost intact. As a result,
different residue sequences can determine the unfolding pathway of proteins even though
they have similar 3D tertiary structures.

28

1.8 References
1. Alberts, B., Johnson, A., Lewis., J., et al. , Molecular Biology of the The Cell.
http://www.ncbi.nlm.nih.gov/books/NBK26830/, 2002.

2. LadyofHats, Main protein structures levels.
http://en.wikipedia.org/wiki/File:Main_protein_structure_levels_en.svg, 2008.

3. van Gunsteren, W.F.a.B.H.J.C., Gromos manual. BIOMOS, Biomolecular
Software, Laboratory of Physical Chemistry, University of Groningen, The
Netherlands, 1987.

4. Ryckaert, J.P., G. Ciccotti, and H.J.C. Berendsen, Numerical integration of
cartesian equations of motion of a system with constraints - Molecular dynamics
of n-alkanes. Journal of Computational Physics, 1977. 23(3): p. 327-341.

5. Humphrey, W., A. Dalke, and K. Schulten, VMD: Visual molecular dynamics.
Journal of Molecular Graphics, 1996. 14(1): p. 33-&.

6. Wang, S.C. and C.T. Lee, Protein secondary structure controlled with light and
photoresponsive surfactants. Journal of Physical Chemistry B, 2006. 110(32): p.
16117-16123.

7. Hamill, A.C., S.C. Wang, and C.T. Lee, Probing lysozyme conformation with
light reveals a new folding intermediate. Biochemistry, 2005. 44(46): p. 15139-
15149.

8. Hamill, A.C., S.C. Wang, and C.T. Lee, Solution structure of an amyloid-forming
protein during photoinitiated hexamer-dodecamer transitions revealed through
small-angle neutron scattering. Biochemistry, 2007. 46(26): p. 7694-7705.

9. Lee, C.T., K.A. Smith, and T.A. Hatton, Photocontrol of protein folding: The
interaction of photosensitive surfactants with bovine serum albumin.
Biochemistry, 2005. 44(2): p. 524-536.

10. Wang, S.C. and C.T. Lee, Enhanced enzymatic activity through photoreversible
conformational changes. Biochemistry, 2007. 46(50): p. 14557-14566.

11. Wang, S.-C., Controlling the form-dynamics-function relationship of proteins
with light illumination. DIssertation_University of Southern California, 2008.

12. Bockmann, M., et al., Atomistic force field for azobenzene compounds adapted for
QM/MM simulations with applications to liquids and liquid crystals. Journal of
Chemical Theory and Computation, 2007. 3(5): p. 1789-1802.
29

13. Rooman, M., et al., What is paradoxical about Levinthal paradox? Journal of
Biomolecular Structure & Dynamics, 2002. 20(3): p. 327-329.

14. Levintha.C, Are there pathways for protein folding. Journal De Chimie Physique
Et De Physico-Chimie Biologique, 1968. 65(1): p. 44-&.

15. Zwanzig, R., A. Szabo, and B. Bagchi, Levinthals paradox. Proceedings of the
National Academy of Sciences of the United States of America, 1992. 89(1): p.
20-22.

16. Sali, A., E. Shakhnovich, and M. Karplus, How does a protein fold. Nature, 1994.
369(6477): p. 248-251.

17. Creighton, T.E., Protein folding. Biochemical Journal, 1990. 270(1): p. 1-16.

18. Kim, P.S. and R.L. Baldwin, Intermediates in the folding reactions of small
proteins. Annual Review of Biochemistry, 1990. 59: p. 631-660.

19. Dobson, C.M., Characterization of protein folding intermediates : Current
Opinion in Structural Biology 1991, 1:22-27. Current Opinion in Structural
Biology, 1991. 1(1): p. 22-27.

20. Fersht, A.R., Protein folding and stability - the pathway of folding of barnase.
Febs Letters, 1993. 325(1-2): p. 5-16.

21. Matthews, C.R., Pathways of protein folding. Annual Review of Biochemistry,
1993. 62: p. 653-683.

22. Karplus, M. and A. Sali, Theoretical studies of protein folding and unfolding.
Current Opinion in Structural Biology, 1995. 5(1): p. 58-73.

23. Miranker, A.D. and C.M. Dobson, Collapse and cooperativity in protein folding.
Current Opinion in Structural Biology, 1996. 6(1): p. 31-42.

24. Chan, H.S. and K.A. Dill, Transition states and folding dynamics of proteins and
heteropolymers. Journal of Chemical Physics, 1994. 100(12): p. 9238-9257.

25. Sosnick, T.R., et al., The barriers in protein folding. Nature Structural Biology,
1994. 1(3): p. 149-156.

26. Baldwin, R.L., The nature of protein-folding pathways - the classical versus the
new view. Journal of Biomolecular Nmr, 1995. 5(2): p. 103-109.

30

27. Dobson, C.M., A. Sali, and M. Karplus, Protein folding: A perspective from
theory and experiment. Angewandte Chemie-International Edition, 1998. 37(7): p.
868-893.

28. Dobson, C.M., P.A. Evans, and S.E. Radford, Understanding how proteins fold -
the lysozyme story so far. Trends in Biochemical Sciences, 1994. 19(1): p. 31-37.

29. D. van der Spoel, E.L., B. Hess, A. R. van Buuren, E. Apol, P. J. Meulenhoff,,
A.L.T.M.S. D. P. Tieleman, K. A. Feenstra, R. van Drunen and H. J. C., and
Berendsen, Gromacs User Manual version 3.3, www.gromacs.org. 2005.

30. Wikipedia, http://en.wikipedia.org/wiki/Root_mean_square_fluctuation.

31. Callenberg, K., Illustration of solvent accessible surface in relation to the van der
Waals surface of an atom
http://en.wikipedia.org/wiki/File:Accessible_surface.svg, 2010.

32. Lee, B. and F.M. Richards, Interpretation of protein structures - estimation of
static accessibility. Journal of Molecular Biology, 1971. 55(3): p. 379-&.

33. Kabsch, W. and C. Sander, Dictionary of protein secondary structure - pattern -
recognition of hydrogen-bonded and geometrical features. Biopolymers, 1983.
22(12): p. 2577-2637.

34. Bachmann, A., D. Segel, and T. Kiefhaber, Test for cooperativity in the early
kinetic intermediate in lysozyme folding. Biophysical Chemistry, 2002. 96(2-3): p.
141-151.

35. Radford, S.E., C.M. Dobson, and P.A. Evans, The folding of hen lysozyme
involves partially structured intermediates and multiple pathways. Nature, 1992.
358(6384): p. 302-307.

36. Kiefhaber, T., Kinetic traps in lysozyme folding. Proceedings of the National
Academy of Sciences of the United States of America, 1995. 92(20): p. 9029-
9033.

37. Dinner, A.R., et al., Understanding protein folding via free-energy surfaces from
theory and experiment. Trends in Biochemical Sciences, 2000. 25(7): p. 331-339.

38. Matagne, A., S.E. Radford, and C.M. Dobson, Fast and slow tracks in lysozyme
folding: Insight into the role of domains in the folding process. Journal of
Molecular Biology, 1997. 267(5): p. 1068-1074.

31

39. Mark, A.E. and W.F. Vangunsteren, Simulation of the thermal denaturation of
hen egg white lysozyme - Trapping the molten globule state. Biochemistry, 1992.
31(34): p. 7745-7748.

40. Hunenberger, P.H., A.E. Mark, and W.F. Vangunsteren, Computational
approaches to study protein unfolding - hen egg white lysozyme as a case study.
Proteins-Structure Function and Genetics, 1995. 21(3): p. 196-213.

41. Williams, M.A., J.M. Thornton, and J.M. Goodfellow, Modelling protein
unfolding: hen egg-white lysozyme. Protein Engineering, 1997. 10(8): p. 895-903.

42. Pepys, M.B., et al., Human lysozyme gene mutations cause hereditary systemic
amyloidosis. Nature, 1993. 362(6420): p. 553-557.

43. Kazmirski, S.L. and V. Daggett, Non-native interactions in protein folding
intermediates: Molecular dynamics simulations of hen lysozyme. Journal of
Molecular Biology, 1998. 284(3): p. 793-806.

44. Chen, L.L., K.O. Hodgson, and S. Doniach, A lysozyme folding intermediate
revealed by solution X-ray scattering. Journal of Molecular Biology, 1996. 261(5):
p. 658-671.

45. Smith, L.J., et al., Molecular dynamics simulations of peptide fragments from hen
lysozyme: Insight into non-native protein conformations. Journal of Molecular
Biology, 1998. 280(4): p. 703-719.

46. Zhou, R.H., et al., Destruction of long-range interactions by a single mutation in
lysozyme. Proceedings of the National Academy of Sciences of the United States
of America, 2007. 104(14): p. 5824-5829.

47. Raines, R.T., Ribonuclease A. Chemical Reviews, 1998. 98(3): p. 1045-1065.

48. Robertson, A.D. and R.L. Baldwin, Hydrogen-exchange in thermally denatured
Ribonuclease A. Biochemistry, 1991. 30(41): p. 9907-9914.

49. Hermans, J. and H.A. Scheraga, Structural studies of Ribonuclease V. Reversible
change of configuration. Journal of the American Chemical Society, 1961. 83(15):
p. 3283-&.

50. Ginsburg, A. and W.R. Carroll, Some specific ion effects on the conformation and
thermal stability of Ribonuclease. Federation Proceedings, 1965. 24(2P1): p. 412-
&.

32

51. Tsong, T.Y., et al., A calorimetric study of thermally induced conformational
transitions of Ribonuclease A and certain of its derivates. Biochemistry, 1970.
9(13): p. 2662-&.

52. Tanford, C., Chemical basis for antibody diversity and specificity. Accounts of
Chemical Research, 1968. 1(6): p. 161-&.

53. Haezebrouck, P., et al., Hydrophobic interaction of lysozyme and alpha-
lactalbumin from equine milk whey. Biochimica Et Biophysica Acta, 1992.
1122(3): p. 305-310.

54. Vanderheeren, G., et al., Thermodynamic characterization of the partially
unfolded state of Ca2+-loaded bovine alpha-lactalbumin: Evidence that partial
unfolding can precede Ca2+ release. Biochemistry, 1996. 35(51): p. 16753-
16759.

55. Wu, L.C., et al., Disulfide determinants of calcium-induced packing in alpha-
Lactalbumin. Biochemistry, 1996. 35(3): p. 859-863.

56. Kuhlman, B., et al., Calcium binding peptides from alpha-lactalbumin:
Implications for protein folding and stability. Biochemistry, 1997. 36(15): p.
4607-4615.

57. Creighton, T.E. and J.J. Ewbank, Disulfide-rearranged molten globlue state of
alpha-lactalbumin. Biochemistry, 1994. 33(6): p. 1534-1538.

58. Dolgikh, D.A., et al., Alpha-lactalbumin - compact state with fluctuating tertiary
structure. Febs Letters, 1981. 136(2): p. 311-315.

59. Kronman, M.J., Cerankow.L, and L.G. Holmes, Inter and intramolecular
interactions of alpha-lactalbumin 3 special changes at acid PH. Biochemistry,
1965. 4(3): p. 518-&.

60. Baum, J., et al., Characterization of a partly folded potein by NMR methods -
studies on the molten globule state of guinea-pig alpha-lactalbumin. Biochemistry,
1989. 28(1): p. 7-13.

61. Peng, Z.Y., L.C. Wu, and P.S. Kim, Local structural preferences in the alpha-
lactalbumin molten globule. Biochemistry, 1995. 34(10): p. 3248-3252.

62. Wu, L.C., Z.Y. Peng, and P.S. Kim, Bipartite structure of the alpha-lactalbumin
molten globule. Nature Structural Biology, 1995. 2(4): p. 281-286.

33

63. Demarest, S.J., et al., Defining the core structure of the alpha-lactalbumin molten
globule state. Journal of Molecular Biology, 1999. 294(1): p. 213-221.

64. Davidson, A.R. and R.T. Sauer, Folded proteins occur frequently in libraries of
random amino-acid-sequences. Proceedings of the National Academy of Sciences
of the United States of America, 1994. 91(6): p. 2146-2150.

65. Raleigh, D.P., S.F. Betz, and W.F. Degrado, A de-novo designed protein mimics
the native-state of natural proteins. Journal of the American Chemical Society,
1995. 117(28): p. 7558-7559.

34

Chapter 2: Microsecond Simulations of Lysozyme Unfolding
Induced by Photoresponsive Surfactant

Previous experimental work has shown that lysozyme unfolding can be
photoreversibly controlled using low concentration (<12mM) of photoresponsive
azobenzene surfactants (azoTAB), which adopts a relatively-hydrophobic trans structure
under visible-light and a relatively-hydrophilic cis structure under UV-light. A series of 1
µs molecular dynamic simulations of the explicit solution of pH = 7.0 at 300 K reveal
that at low (1:10) lysozyme : surfactant ratios (either all trans or all cis), lysozyme is
swollen primarily in α-helical regions, while the β-sheets remain relatively intact, in
agreement with previous experimental observations. The high binding affinity toward
both trans and cis azoTAB are Trp residues, helix D, C-subsite of the active cleft, and
lower part of hinge, which connects the α and β domains. Therefore, azoTABs assist
protein unfolding primarily in helix D and loose lower part of hinge. The enhaced
internal dyanamics is reached right after the transition point where the more close
contacts between lysozyme and azoTAB but less native contacts within lysozyme are
observed. It is followed by the fast increase of the RMSD value of the α domain along
with the greater increase of the α-domain gyration radius than the β domain indicating
that the α-domain is more puffy than the β-domain, as seen in the previous SANS results.
The α-domain is then showing more dynamic than the β-domain with the higher
frequency of hydrogen bonds breaking and reforming and larger fluctuation of several
35

calculation of structural properties. Around 1 µs, the number of native contacts reduced
to half number of its native state (crystal form) in the interface between the α-β domain;
however, the total number of contacts and the total number of hydrogen bonds are more
than those of the native state in the domain interface, implying that the α domain is not
leaving the β domain apart and therefore the β domain can still maintain most of its
secondary structure by the end of 1µs simulation. The phenol rings of the azo-TAB are
found to have higher effective collision with lysozyme within either 0.4 nm or 0.3 nm of
collision radius than the charged head group of azo-TAB. The potential dominant
interaction between lysozyme and azo-TAB can be non-bonding / hydrophobic
interactions instead of electrostatic interactions. The results have clearly demonstrated
that the surfactants, azo-TAB (either trans or cis), induce the unfolding of lysozyme
mainly in the tight bounded α domain while leaving the relatively intact of β sheet
through the hydrophobic interaction in the room temperature.

2.1 Introduction
Proteins are the most abundant molecules in living systems and how proteins fold
correctly has been thought to be closely related to how proteins function properly. Such a
problem is existing but fundamental for decades. Proteins are peptide chains of the amino
acid sequences and it is believed that to decode the secret for protein folding is to
investigate these essential sequences [1]; moreover, the correct folding of protein within
the cell makes those assembly of highly ordered folding structures. Routinely, the protein
36

folding is not processed through random search of all possible conformations [2]; instead,
it must be through a specific pathway or multiple pathways to fold correctly within
seconds [3]. In the refolding experiment, the denatured protein forms back to its native
states and the intermediates, or partially folded structures, are found on the folding
pathways. These intermediate conformations contain some or less native-like secondary
structures and, therefore, can elucidate folding process of what parts of protein form
earlier [4]. The attempt to understand folding pathways is approached by the extended
effort on characterizing transient folding intermediates not merely by the experimental
work [5-13] but also by the folding simulations [14-15].
Hen egg-white (HEW) lysozyme (Figure 2.1) is a 129-residue protein with two
domains (α and β). The α domain includes residues 1 to 39 and residues 88 to 129 while
the β domain has residues 40 to 87. The active site cleft is between the two domains and
there are a total of four disulfide bonds within the lysozyme molecule, where two of them
are situated in the intra-α-domain (C6-C127 and C30-C115), one is in the intra-β-domain
(C64-C80) and the other (C76-C94) is bonding α and β domains together. The α-domain
is basically composed of helices, four α-helices (helices A-D) and one C-terminal 310
helix. The β-domain is mainly a three-stranded anti-parallel β sheet, followed by a 310
helix, and an irregular loop. The HEW lysozyme is the first enzyme whose crystal
structure is resolved under X-ray facility [16] and it is also the most well-studied protein
whose intermediates are delicately characterized so that a good understanding about its
hierarchical folding pathway has been achieved [17]. It is discovered that the secondary
structure constructed before the formation of tertiary structure in the refolding
37

experiments of hen lysozyme [18-20]. A wide variety of complementary techniques have
combined to provide detailed structural pictures of the intermediates on the kinetic
folding pathway of HEW lysozyme to interpret its folding procedures [17]. For instances,
far-UV CD can monitor the time-evolutionary development of protein secondary
structure [18, 20-22]; near-UV CD probes the packing of aromatic residues in native
structures [21, 23-24]; inhibitor binding detects the formation of native state [19, 24-25];
ANS binding tests the exposure extent of the hydrophobic surfaces [24]; H/D exchange
along with NMR measures how many hydrogen bonds form [21, 26-29]; and intrinsic
fluorescence could be contributed from Trp, Tyr, or Phe residues [24].

Figure 2.1 : Crystal structure of hen egg white (HEW) lysozyme adopted from 6LYZ.pdb.
The α domain is colored in red and it has the α-helices A, B, C, D plus the C-terminal 310
helix. The β domain is colored in green with a three-stranded β-sheet and a 310 helix.
Each disulfide bonds are shown as a pair of two disulfide atoms (yellow spheres). There
38

are two disulfide bonds in the α domain, one in the β domain, the other connecting α and
β domain. And, six Trp residues (28, 62, 63, 108, 111, 123) are indicated. VMD program
[30] is adopted.

Hen lysozyme folds in multiple pathways [21]. Structural studies of folding
intermediates from both theory and experiment reveal that there are approximately 20%
of the lysozyme molecules, as the I αβ intermediates, forming the α and β domains at the
same time through a ‘fast track’ [31], then quickly folding into its native state. On the
free energy surfaces [32], those 20% molecules follow along the monotonically-
decreasing curve toward the native states with the hydrogen-bonding network fast built
inside the entire lysozyme molecules. On the other hand, there are near 10% molecules
fold very slowly due to their likely involvement of the proline isomerism. The remaining
molecules (70%), the majority ones, are folding back to the native structures through a
‘slow track’ [31], where there is a minimum on the free-energy landscape [32] and those
trapped intermediates, I α, have only the persistent helix structure in the α domain
resulting from the early formation of the hydrogen bonds in the α domain. Along with the
‘slow track’, the helix formation of the α domain precedes the β-sheet formation of the β
domain. The final step, also the rate-limiting step, in folding process is to correctly dock
the residues from one domain into the other and form the α-β interdomain interface. It is
particularly important to insert residues Leu 55 and Ile 56 of the β-domain into a
hydrophobic slot in the α domain [33]. All-atom molecular dynamics simulations have
been developed to study the folding pathways of proteins by offering the intermediate
structures at atomic resolution. The size of a lysozyme molecule is small with 129
39

residues so the computational cost is affordable. Some simulation work about unfolding
of lysozyme has indicated that the observed intermediates are with an unfolded β-domain
but a structured α-domain [34-36] in agreement with those intermediates found on the
dominant slow-track pathway. The β-domain has shown less stable than the α-domain in
lysozyme [37]. Under 500 K of unfolding simulations, hen egg white lysozyme is thermal
denatured rendering loss of the β-domain before the α-domain unfolds [38]. Moreover,
the simulations of the separate α-domain peptide and β-domain peptide; the β-sheet
structure disappeared earlier than the α-helix at high temperature [39]. It was investigated
that the lysozyme molecules were unfolded in the urea solution. Through X-ray
scattering, the overall increase of the radius gyration of the denatured lysozyme induced
by urea was monitored. The mean increase of the β-domain gyration radius is about 0.5
nm while there is around 0.2 nm increase in α-domain. This suggested that the β-domain
unfolds much faster than the α-domain [40]. The MD simulations have the same results
and support the same conclusion for the lysozyme-urea solution [39, 41]. In contrast to
the large population of the intermediates (70%) trapped in a free-energy minimum on the
slow track, another 20% of the intermediates are expected to be accumulated along the
fast track of folding [31]. The key to stabilize the α-domain is to maintain its hydrophobic
contacts between residues; nevertheless, side chain-side chain hydrogen bonding and salt
bridges together with both hydrophobic and polar contacts all play significant roles to
resist the possibly dynamic fluctuations leading the β-domain to unfold. Some other
unfolding simulations are showing relatively persistent β-sheets. A modeling procedure,
which introduces more water molecules penetrating into the core of lysozyme, led the
40

protein into the unfolding pathway similar to its fast track on free-energy diagram [32].
Subsequently, the helices fluctuate rapidly and further the compactness of the α domain is
lost; nevertheless, the native β-sheet is essentially intact [36]. With force field
GROMOS87, two simulation works performed by Mark and van Gunsteren et al. [34-35]
are under high temperature, high pressure, force controlled, or kinetic energy controlled.
Both of the simulations have lysozyme with persistent β-sheets but unstructured α-
helices.
As chemical denaturants, like urea and GdmCl [5, 42], unfold proteins,
surfactants are also used to control the unfolding extent of proteins. For each surfactant,
its hydrophilic head attracts water but its hydrophobic tail repels water simultaneously. In
surfactant-protein solution, hydrophobic tails of the surfactants interact with uncharged or
nonpolar amino acids and reduce the unfavorable exposure of the protein in water.
Therefore, a surfactant with a great degree of hydrophobicity can have high chance to
induce large extent of unfolding in proteins [43]. The surfactants, azoTAB, have been
proven to stabilize the partially-folded states, the intermediates, of lysozyme in solution
[12]. It carries a head, the trimethyl ammonium, and a big photoresponsive tail, two
phenol rings. With different light illumination, the relative position of the two phenol
rings can be adjusted, which results in two different structures/forms of the azoTAB,
trans (visible light) and cis (UV-light). Because the two forms of azoTAB, trans and cis,
have different extent of interaction with lysozyme under low concentration of azoTAB
(<12mM), it was demonstrated that the unfolding event of lysozyme can be controlled by
light illumination. Basically, trans-azoTAB induces much more unfolding conformation
41

changes of lysozyme than cis-azoTAB at such low surfactant concentration. As a result,
the unfolding mostly happens in the α-domain of the lysozyme, which makes the
apparent swelling of its α-domain leaving a relatively intact structure of its β-domain
through the observation via SANS combined with the analysis of the protein secondary
structure via FT-IR spectroscopy. These induced intermediates by azo-TAB show
separate characteristics than those well-studied lysozyme intermediates that have an
intact folded α domain but unfolded β domain.
In the present study, we conducted a series of the molecular dynamic simulations
at 300K and found that a little amount of either trans or cis azo-TAB can unfold the
lysozyme and further stabilize the intermediately folded states of lysozyme in solution.
At the end of the simulation time (longer than 1µs), the β-sheet are relatively reserved
while part of the α-helix unfolded. This corresponds to the experimental observation,
where lysozyme unfolds even under low concentration (<12mM) of azo-TAB and more
significant unfolding happens in the α domain. It is found that the nonbonding /
hydrophobic interactions could potentially dominate in the lysozyme-azoTAB solutions.
There is a transition point happening in each single simulation system. Before the
transition point, the lysozyme behaves like its native state. After the close contact
between lysozyme and azo-TAB but less native contacts within the protein, the transition
point is achieved resulting in the increase of the internal dynamics; especially, the α-
domain is showing more dynamic than the β-domain. Finally, we characterize the
intermediate of lysozyme during 900ns to 1000ns.
42

2.2 Simulation Details
The Gromacs 3.3.3 package [44-48] is used to perform our simulation. The
GROMOS 45a3 force field [49] is chosen for protein lysozyme and azotab (both trans
and cis structure, shown in Figure 2.2) along with the TIP3P [50-51] (TIP3P stands for
transferable intermolecular potential, three position model.) water model. The whole set
of reparametrization to the photoswitchable azo functional unit is from Böckmann’s
previous work [52] and CH3 and CH2 are typical ‘atom type’ in Gromos force field.
The starting lysozyme structure is the x-ray crystal structure (6LYZ.pdb [53]).
One lysozyme is first centered in the box of 6nm * 6nm * 6nm and 10 azotabs (either all
trans or all cis structures, see Figure 2.2) are randomly put around the lysozyme. If any
azotab molecule overlaps with other azotab molecules or lysozyme, it will be replaced by
another randomly distributed azotab. The experimental system [12, 54] which we are
trying to understand with MD simulation is conducted as very low concentration (<12
mM). From the previous SANS data, the effective molecular weight determines that the
surfactant-protein complex has the approximate 12 – 25 bounded azoTAB to one
lysozyme [12]. Also, the optimal molar ratio of CTAB to protein was 10 to obtain good
refolding yield and avoid the lysozyme aggregates [55]. Hence, it is reasonable to enclose
one lysozyme and 10 azo-TAB in each of our simulation box. Next, add solvent
molecules as the distance between any atom of the solvent molecule and any atom of the
solute molecule(s) is greater than the sum of the Van der Waals radii of both atoms or
solvent molecules are removed from this box [44]. Eighteen CL- counter ions are then
added to neutralize the pH=7 solution. In total, twenty simulation boxes are created and
43

each is with 6nm*6nm* 6nm dimension. Within the simulation box, there are one
lysozyme centered and 10 different initial-randomly distributed azotabs (either all trans
or all cis). Ten boxes / simulation runs are for all trans azotabs immersed (system:
T1~T10) and another ten boxes / simulation runs are for all cis azotabs immersed
(system: C1~C10). One additional simulation run, for comparison, is made to mimic pure
lysozyme solution (system: P) where one lysozyme in the box with dimension of 6nm *
6nm * 6nm and TIP3P water model is also applied.
System Dimension
of box
No. of
water molecules
Azotab
Structure
No. of
azoTAB
Simulation
time (ns)
P 6.16 6396 none 0 211
T1 6.15 6318 all trans 10 1382
T2 6.15 6316 all trans 10 1231
T3 6.15 6303 all trans 10 1301
T4 6.15 6298 all trans 10 1284
T5 6.15 6325 all trans 10 1364
T6 6.15 6305 all trans 10 1247
T7 6.15 6299 all trans 10 1260
T8 6.14 6296 all trans 10 1298
T9 6.15 6300 all trans 10 1285
T10 6.15 6302 all trans 10 1334
C1 6.14 6296 all cis 10 1307
C2 6.14 6307 all cis 10 1316
C3 6.15 6313 all cis 10 1250
C4 6.15 6314 all cis 10 1334
C5 6.16 6306 all cis 10 1325
C6 6.14 6287 all cis 10 1315
C7 6.14 6308 all cis 10 1327
C8 6.15 6305 all cis 10 1233
C9 6.14 6278 all cis 10 1248
C10 6.14 6297 all cis 10 1208

Table 2.1: Summary of simulation systems. Dimension of box is average after
equilibrium.

The energy minimization procedures are performed in the beginning and the
steepest descent algorithm is adopted to minimize the energy of each system. Next, the
44

two-stage procedure is carried out to equilibrate each system. In the first stage, the
solvents are allowed to move freely while keeping the atoms of the proteins fixed to its
original positions; therefore, only the solvent molecules are equilibrated. In the second
stage, all atoms are free moving and the whole system is equilibrated. Following is the
NPT (i.e. isothermal and isobaric) simulation at 300 K and 1 atm with time step 2 fs.
LINCS algorithm [56] was employed to fix the all of chemical bonds. For the long-range
electrostatic interactions, we use Reaction-Field (RF) method [57], the cutoff is 14
angstroms. It would be better to use Particle-Mesh Ewald (PME), which calculates all of
the electrostatic interactions in the infinite periodic system leading more accurate results
than RF. But, PME is much more expensive than RF. We setup 20 simulation systems,
each of which represents separate lysozyme–azoTAB solution, and we expect that no less
than 1µs to partially unfold the lysozyme by azo-TAB under room temperature;
moreover, the chosen GROMOS force fields are developed using RF. Therefore, the RF-
method is chosen for our simulation runs. The multimillion-atom simulations on the
petascale supercomputers exists good agreement for calculating the long range
electrostatics in water with method of RF and PME [58]. For the van der Waals
interaction, 14 angstroms cutoff is applied. The Brendsen coupling algorithm is used [59]
to maintain a constant temperature and pressure during simulations. Periodic boundary
conditions are employed throughout our simulations. We calculate the root-mean-square
deviation (RMSD), the radius of gyration (Rg), solvent-accessible surface area (SASA),
and counts of hydrogen bonds with Gromacs analysis programs.

45

Figure 2.2 : Three-dimensional structures of trans-azoTaB (left) and cis-azoTAB (right).
Alkyl group CH3 and CH2 are treated as typical atoms in Gromos force field. All plots are
shown as VDW drawing style with VMD program [30].

2.3 Results and Discussion
2.3.1 Dynamic Affinity and Binding Sites
The current techniques to predict the protein-ligand binding sites are, for
examples, docking( such as AutoDock software), organic solvent mapping [60],
calculation of free energy of binding [61]. Docking considers proteins as rigid bodies or
permits few conformation changes of proteins and also low degrees of rotational freedom
of ligands are allowed. Organic solvent mapping uses small organic molecules as probes
on the protein surface but it could result in false prediction of binding sites.
Thermodynamics integration (TI) and free-energy perturbation (FEP) techniques have
been applied on this calculation of the binding free energy; however, computational cost
is high [46]. New technique uses molecular dynamics simulations to determine the ligand
binding sites [62]. One protein and one ligand are put in the simulation box at one time
46

and performed 2-5 ns MD simulation to finally locate the binding sites. In this work, we
conducted the MD simulations longer than 1 µs and further recognized the binding sites
of azotabs around lysozyme with the dynamic affinity [62], which is proportional to the
number of the collisions between them. Azotabs are expected to collide more frequently
with binding sites of lysozyme. The simulations started as one lysozyme in the center of
box (6nm * 6nm * 6nm) and ten azotabs (either all trans or all cis) distributed randomly.
Ten simulations were setup for all trans azotab systems (T1~T10); another ten simulation
were setup for all cis azotab systems (C1~C10). The effective collision happened
between the lysozyme and azotab when any atom of azotab is within a distance of 0.4 nm
or less from any atom of the protein residue through the simulation trajectories. The
dynamic affinity is defined as n n P
i i
/  [62], where
i
n indicates the number of
collisions with the ith residue. N M n /  is the average number of collisions for a
single residue,



N
i
i
n M
1
is the total number of collisions, N is the number of residues of
lysozyme. There are 129 residues for hen egg white lysozyme ( 129  N ). We adopted the
900
th
ns to the 1000
th
ns simulation paths averaged over the ten simulation runs for either
all trans or all cis system (average over either 100 trans-azotabs or 100 cis-azotabs) on
the defined dynamic affinity,
i
P . The binding affinity is shown in (Figure 2.3, 0 (a),(b) );
red color represents Pi > 1.0(high binding affinity); blue represents Pi <0.7 (low affinity);
and yellow is in between 0.7< Pi <1.0 (moderate affinity).
47

In Figure 2.3, it shows that both trans and cis azotab have high binding affinity on
beta sheet I, helix D, and 310 helix but have low binding affinity on loop and helix C.
Otherwise, helix A and helix B have moderate binding affinity toward both trans and cis
azotabs. Residue 64 is located inside the lysozyme; therefore, either trans or cis azotabs
can hardly reach it and results in the lowest binding affinity, 0.008 (trans) and 0.027 (cis).
For residues 76, 80, and 94, the trans and cis azotabs have the very low binding affinity
of less than 0.3. Residue 64 and 76 are situated on the loop and residue 94 is on helix C.
Generally azotabs prefer to bind to hydrophobic site. For instance, the hydrophobic
residues Asn106, Ala107, Trp108, Val 109 are part of helix D and are involved more
interaction with both trans and cis azotabs. There are two catalytically active residues,
Asp52 and Glu35, in the hen-egg lysozyme and both of them have strong binding on
azotabs for trans and cis. Moreover, either trans or cis azotabs have frequent interaction
with the residues, Gln57, Ile58, Asn59, Trp62, Trp63, Ala107 and Trp108, which form
the C-subsite in the active cleft of hen-egg lysozyme. The large C-subsite (shown in 0(d))
allows more translational/rotational degrees of freedom of ligands and has several
hydrogen bond donor or acceptor groups. It is also a hydrophobic pocket created by
hydrophobic side chains. Due to its low binding free energy, C-subsite attracts a large
variety of organic solvents, such as ethanol, propanol, butanol and pentanol [63-64] and
some denaturants, like urea [65-66] and DMSO [67], and guanidinium hydrochloride
[67]. In Figure 2.3 , dynamic affinity >0 for all residues, that implies that azotabs can
penetrate into the lysozyme and resides in grooves and cavities of lysozyme. The phenol
rings of azotabs interact with Trp groups. Cis-azotabs have the strong binding affinity
48

with all of Trp groups (Trp28, Trp62, Trp63, Trp108, Trp111, Trp123) while trans-
azotabs have the strong binding affinity only with Trp62, Trp63, Trp108 and Trp123. As
shown in Figure 2.1, α domain is colored in red and β domain is colored in green; hinge
is in between the two domains. The residues (87-101, 35, 36, 39-42, 55, 56) are defined
as hinge [68], see 0(c). In 0 (a) and (b), back hinge area is colored in red (high binding
affinity) for both trans and cis azotabs but the front hinge area (helix C) is colored in blue
(low binding affinity). The high binding affinity in the back hinge area might explain that
azotabs (both trans and cis) increase the activity of lysozyme [54]. The structural
difference between trans and cis affects the significantly different binding affinity toward
some residues (23, 28, 46, 47, 99, 105, 111, 116, 117, 118). Front half of the β domain is
the cold binding sites.

Figure 2.3 : Average binding affinity (Pi value) of azo-TAB toward each residue of
lysozyme. Data are listed in Table 2.2. Trans-azoTAB binding affinity is shown as the
49

black line and cis-azoTAB binding affinity is the red line. Secondary structure of crystal
lysozyme (native state) is indicated along the axis of residue number.
Res. No. Trans Cis Res. No. Trans Cis
1 0.353 0.450 66 0.501 0.717
2 1.253 0.889 67 0.309 0.606
3 1.486 1.395 68 0.584 0.748
4 0.994 0.550 69 0.444 0.459
5 0.949 0.602 70 0.399 0.415
6 0.740 0.822 71 0.424 0.321
7 1.451 1.033 72 0.487 0.497
8 1.203 1.327 73 0.826 0.726
9 0.548 0.700 74 0.684 0.621
10 0.983 0.893 75 1.378 0.991
11 0.985 0.952 76 0.176 0.271
12 0.592 0.917 77 0.570 0.850
13 0.694 0.849 78 0.212 0.738
14 0.783 0.758 79 0.321 0.802
15 0.634 0.863 80 0.082 0.282
16 0.514 0.634 81 0.596 0.840
17 0.657 0.817 82 0.404 0.627
18 1.062 0.912 83 0.582 0.462
19 1.337 1.006 84 1.212 1.011
20 1.128 1.401 85 1.096 1.025
21 1.023 1.222 86 1.501 1.483
22 1.267 1.407 87 0.966 1.462
23 1.165 2.152 88 1.045 1.383
24 1.164 1.001 89 0.307 0.673
25 0.949 0.841 90 0.151 0.695
26 0.641 0.504 91 0.330 0.644
27 0.897 1.055 92 0.264 0.688
28 0.572 1.488 93 0.186 0.523
29 0.776 0.818 94 0.093 0.203
30 0.819 0.784 95 0.417 0.618
31 0.192 0.631 96 0.540 0.884
32 0.497 0.945 97 0.735 0.681
33 1.421 1.262 98 0.961 1.152
34 1.928 1.515 99 0.890 1.666
35 2.691 1.869 100 1.092 1.128
36 2.013 1.381 101 1.409 1.469
37 2.433 1.643 102 1.007 1.311
38 2.463 2.165 103 2.054 2.313
39 1.761 1.522 104 1.472 1.586
40 1.395 1.078 105 0.997 1.931
41 1.557 1.229 106 1.846 2.399
42 1.138 0.996 107 2.429 2.152
43 1.626 1.245 108 1.872 2.259
44 1.952 1.821 109 1.252 1.045
45 1.429 0.900 110 0.646 0.537
46 2.323 1.680 111 0.701 1.623
47 1.260 0.689 112 0.623 0.797
48 1.665 1.251 113 0.323 0.433
49 0.849 0.611 114 0.966 0.646
50 1.407 1.253 115 1.117 0.815
51 0.671 0.561 116 1.323 0.629
52 1.640 1.738 117 1.206 0.550
53 0.853 0.805 118 1.316 0.768
54 0.550 0.552 119 1.128 0.945
55 1.479 1.660 120 1.022 0.931
56 1.202 0.977 121 0.810 0.793
57 1.889 1.945 122 0.646 0.612
58 0.516 0.426 123 1.370 1.063
59 1.309 1.369 124 0.896 0.765
60 0.160 0.169 125 0.491 0.560
61 0.837 0.722 126 0.514 0.541
62 2.474 2.122 127 1.054 1.024
63 1.793 1.257 128 0.827 0.624
64 0.008 0.027 129 1.536 1.230
65 0.322 0.640

Table 2.2 : Calculated binding affinity (Pi values) of azo-TAB towards each residue of
lysozyme. Average is taken overall trans and cis systems through recorded trajectories
from 900ns-1000ns.
50

(a)

(b)

(c)

(d)

Figure 2.4 : (a) Binding Affinity of trans-azoTAB toward lysozyme. (b) Binding Affinity
of cis-azoTAB toward lysozyme. Residues with high binding affinity, Pi > 1.0, are
colored in red; low binding affinity, Pi <0.7 in blue; and moderate binding affinity, 0.7<
Pi <1.0, in yellow.(c) Hinge is in violet color. (d) C-subsite is in purple color. All plots
are shown as New Ribbons with VMD program [30].

51

2.3.2 Analysis of Secondary Structures
Helix (%) Strand /β-Sheet (%) β-Turn (%) Coil (%)
Regions residue Pure Trans Cis Pure Trans Cis Pure Trans Cis Pure Trans Cis
HelixA (4-15) 87 75 78 1 8 3 5 6 6 7 11 10
HelixB (24-36 ） 63 67 61 19 13 16 11 10 7 7 10 17
HelixC (89-99 ） 93 85 84 3 1 3 1 8 6 2 6 6
HelixD (108-115) 61 33 27 5 19 14 14 23 15 20 25 44
Beta I (41-45) 0 0 0 64 55 59 9 5 3 26 40 38
Beta II (50-53) 0 0 0 77 76 75 0 3 0 23 22 25
Beta III (58-60) 0 1 6 64 64 55 33 27 26 2 8 13
Loop (61-78) 0 2 4 51 42 44 23 26 20 25 31 33
3-10 (80-84) 83 65 54 3 7 13 14 16 13 1 11 20
3-10 (120-124) 72 42 45 6 14 25 17 20 10 5 24 21

Table 2.3: Average percentages of secondary structures contents of lysozyme
intermediates during simulation course, 900 ns – 1000 ns.
The secondary structures are determined with the DSSP program [69]. The
average percentage of secondary structure calculation over the 900
th
-1000
th
ns of 10trans
(system:T1~T10) /cis systems (system: C1~C10) is shown in Table 2.3. One snapshot per
0.5 ns was taken over the 100 ns – trajectories. The secondary structure conformations
are considered as helix, strand, β-turn as well as coil and the percentage average of each
structure was calculated over each specific regions. For comparison, the average
percentage of secondary structure calculation was taken over the 100
th
-200
th
ns of the
pure lysozyme in water (system: P). Most helix structures are lost in helix D for both
trans and cis system, which may result from the high binding affinity of both trans and cis
azotab toward helix D. But, helices A, C, and B have less loss of helical structure. Helix
B is embedded inside the α-domain core and therefore there is less chance for azo-TAB
to reach it. Figure 2.3 is showing low binding affinity of both trans and cis azo-TAB on
helix B and the calculation of the average secondary structure reveals that helix B keeps
its high percentage helix structure. Although helix C is exposed on the surface of the
52

lysozyme, both trans and cis azo-TAB have little interaction with it and so it is less
perturbed and maintains its helix structure well. Helix A is also on the surface of the
lysozyme but showing a bit higher binding affinity toward azo-TAB than helix C, which
explains more helix loss of helix A than helix C. The β domain is composed of three β-
strands (β-sheet). The β sheet I located on the front position of the active cleft has
experienced high attachment with both trans and cis azo-TAB and thus a little few
percents of strand structure are lost; nevertheless, β sheets II and III appeared more
stable.
Trans azotab maintains more helical structure in the alpha domain and more
strand structures in the β domain than cis azotab. On the other hand, in both α and β
domain, trans azotab induces more β-turn structures than cis while cis azotab induces
more coil structure than trans. During folding course, β-turns occur earlier than the
formation of helices and sheets [70], and thus the overall higher β-turns induced by trans-
azotab indicates the relative instability before trans azotabs unfolds hen lysozyme.
Although cis unfold lysozyme more than trans based on our simulation results (more coil
found in the cis systems), the fact that more β-turns induced by trans could foresee more
extent of unfolding induced by trans with extension of simulation time. Loop structure
moves from strand to coil. 310 helix lost more helix structure and have more coil structure
afterall for both trans and cis. Overall, as a result of interaction with azotab, lysozyme is
swollen primarily in the alpha domain, while the β-sheets remain relatively intact. Hamill
et al. [12] conducted both SANS and FT-IR experiment of lysozyme in the presence of
<12 mM azoTAB upon exposure of either visible (trans : cis = 75 : 25) or UV (trans : cis
53

= 10 : 90) revealed the loss of the α-helix and increase of the unordered structure, which
could correspond to the coil structures as the results from the simulation. However, the β
structures maintain relatively stable and unchanged. The helix-to-unordered transition
indicates that under the very low concentration of azoTAB, the very little amount of these
photo-controlled surfactants can swell the α domain of lysozyme while keeping β domain
intact. Our present simulation work supports these phenomena observed from the
experimental work.

Figure 2.5 : Time evolution of the protein secondary structure in T5 system with DSSP
program [69]. On the right hand side, we indicates location of helix A, helix B, helix C,
helix D, the two 310 helices and the β sheet.
Figure 2.5 displays the time evolution of the protein secondary structure
calculated from the trajectories of the T5 system, in which one lysozyme and ten trans-
54

azo-TAB are immersed in water over 1300 ns. Overall view, there are more disruptions
on the α domain (residues 1 to 35; 85 to 129) than on the β domain (residues 36 to 84) by
the end of the simulation time, 1300 ns. In the beginning, 310 helix (120-124)
disappeared in 200 ns and followed by helix D, which unfolds and reforms within 1000
ns but totally unfolds after that. Helix A is perturbed earlier and more than helix C due to
higher binding affinity of azo-TAB for either trans or cis than helix C. The helical
structure of helix A is totally lost in 1100 ns but 50% of helical structure of helix C
remains at the 1300
th
ns. Helix B is quite intact during the simulation course because it is
situated in the core of the α domain and hence has low binding affinity toward either
trans or cis azo-TAB. It is found that β-sheet I (41-45) was gone around the 400
th
ns in T5
system but still exists through the entire simulation course under other simulation
systems. β-sheet II (50-53) and β-sheet III (58-60) are relatively stable. All other
trajectories derived from either trans or cis systems share the similar results of the
secondary structure deformation as discussed above although the lyozyme in each system
behaves differently and unfolds at different timing.
Under high temperature (500K or 700K) simulation, the unfolding behavior of the
lysozyme has shown different events [38-39, 71]. The β-sheet starts to disappear earlier
than the α-helix. Helix C is lost rapid at 700K within 500 ps [39] while the significant
residual helicity in helix D and 310 helix (120-124) exist after that. Found in the
conformation of ~80% of the refolding molecules [27-28], the amide hydrogen atoms of
α domain are protected before those of β domain. Other 20%, the amide hydrogen atoms
in both domains were at the equally protected rate. The further experimental studies [72-
55

75] suggest that the significant hydrogen exchange protection for amides within helix A,
B and D in the absence of helix C. The hen lysozyme denatured in 8M urea has the
earlier unfolding of β domain than α domain, which is observed from both the
experiments [40, 76] and the computer simulations [41]. The helix C is first denatured by
urea along other helices and hence the ensemble of the denatured lysozyme molecules
only contains the native helical structures of helix A, B, D and 310; however, without
helix C [76]. Through the detailed non-native contact analysis [39], it is considered that
the contacts between α domain and β domain can stabilize the helix C; particularly, the
burial of hydrophobic groups (either main chain or side chain) are importantly involved
in this process. As shown in Table 2.5, Figure 2.13, and Figure 2.14, total contact counts
and hydrogen bond counts of the interface between α-β domains increase over 900
th
ns to
1000
th
ns for the average intermediates in the systems T1 ~ T10 and C1 ~ C10 compared
with the lysozyme crystal structure, which explains that the helix C maintain most of its
helix structure. On the other hand, the binding affinity of the helix C toward either trans
or cis azo-TAB is low; therefore, the helix C is little disturbed by azo-TAB and its helical
structure is stable.

56

2.3.3 Comparison with Experimental Results
(a) Lysozyme in water after 200 ns
(system P)

(b) pure lysozyme in water

(c) T5 after 1300ns

(d)12.2 mM visible light (75% trans-
azoTAB)

Figure 2.6 : SANS images compared with P and T5 simulation results. (a) A simulation
snapshot of protein in P system at 200ns. (b) SANS image of pure protein in water. (c)
Simulation snapshot of protein in T5 system at 1300ns. (d) SANS image of protein in
12.2 mM azo-TAB solution under visible light. Plots of simulation snapshot are obtained
by VMD program [30].

SANS technique is successfully applied on study of protein unfolding [11-13, 77].
Combined with shape-reconstruction methods, the unfolding intermediate of protein
induced by surfactants can be visualized. Figure 2.6(b) is showing the native state of
lysozyme in solution, generally its shape and size agree with the X-ray crystal structures
Figure 2.1 (PDB code 6LYZ) [53]. Although the resolution of SANS is not as high as X-
ray crystallography, the active-site cleft, which is between the α and β domains, was still
clearly observed in the upper right of the SANS molecule image. Under very low
concentration of surfactants (<5mM), lysozyme does not show the unfolding event no
57

matter under either visible or UV light exposure. Increasing the surfactant concentration,
lysozyme started to unfold with visible-light illumination (trans:cis = 75:25). As seen in
Figure 2.6 (d), the conformations of lysozyme intermediates in surfactant solution (12.2
mM) under visible light were determined from shape reconstruction of the SANS data
and the lower-left side of the protein was notably observed - progressive swelling [12].
SANS is an ensemble technique, which average overall the fluctuated conformations of
protein intermediates being in solution, and so taking rotational averaging makes it
impossible to determine which domain (α or β) is unfolding using the SANS data alone.
The analysis on the secondary structure loss of α helix from FT-IR elucidated that the
surfactants denature the α domain of lysozyme and generally unfold the α helices. With
surfactant addition under visible light, the lower-left side of the protein in Figure 2.6 (d)
appears swelling, it seems that unfolding of either helix A or helix C in the α domain may
be primarily responsible for the conformation of the swollen lysoyzme intermediates.
Figure 2.6(c) was taken from the snapshot of lysozyme from the T5 system at the 1300
th

ns. With secondary structure analysis (Figure 2.5), it is clear that both helix A and helix C
are lost in result of the much less compacted lysozyme corresponding to the puffy lower-
left side of the protein molecule seen in SANS (Figure 2.6 (d)). One simulation was
conducted on the pure lysozyme in water over 200 ns and for comparison, its structure is
shown Figure 2.6 (a), which is very close to the X-ray crystal structures, Figure 2.1 and
so the force field we use is consistent/stable.

58

2.3.4 Internal Dynamics of Lysozyme
The internal dynamics of protein in the T5 simulation is seen in Figure 2.8~
Figure 2.19 as one microsecond trajectory of each atom of the protein collected per 0.5 ns
reveals the time evolution of protein structure as well as protein motion. A structural
transition is apparently observed occurring around 450 ns in T5 run and other simulations
also appear the structural transitions. For systems, T1~T5, the structural transition
happens around 410ns~500ns while for C1~C10, it is around 340ns~450ns, listed in
Table 2.4. Once the transition point is reached, azoTABs interact more closely with the
protein, shown in Figure 2.7 and, therefore, the atoms within the protein behave more
dynamic, resulting in a decrease in the overall native contacts within the protein. At this
time, the protein loses 40% of its tertiary contacts and moves into a swollen state / molten
globule state. Details are described below and we take T5 as an example here. Other
systems have shown similar phenomena.
The interface between lysozyme and azo-TAB in T5 trajectory is shown in Figure
2.7. We first calculate solvent accessible surface area (SASA) of lysozyme (S L),
surfactants (SS), and lysozyme-surfactants complex (SC). The interface is defined as
½*(SL+SS-SC) and the radius of solvent probe is set to 0.14 nm. The solvent accessible
surface area of lysozyme in crystal structure (6LYZ) is 90.20 nm
2
and SASA of pure
lysozyme in water (SL) is around 90~95 nm
2
. SASA of the unfolding lysozyme molecules
induced by azoTAB, SL, is around 120~140 nm
2
for all simulation cases (see Table 2.4,
we have SASA-hydrophobic and SASA-hydrophilic separately). The increase of SASA
of the unfolding lysozyme corresponds to the increase of its gyration radius, where the
59

lysozyme molecule becomes swollen. As seen in Figure 2.7, from the beginning to 450
th

ns, the value of the interface have dramatic fluctuations, in which it is jumping up and
down between 0 to 20 nm
2
. At 450
th
ns where the transition state took place, the protein
structure encountered a conformation change as both its native contacts (Figure 2.12) and
number of hydrogen bonds (Figure 2.19) had a quick decrease since 400
th
ns
accompanied with an increase of its RMSD (Figure 2.8), gyration radius (Figure 2.10)
and domain distance (Figure 2.11). Also, loss of the secondary structure is already seen at
the transition state (Figure 2.5) and all of the four disulfide bonds became unstable right
after transition state, including two disulfide bonds in the α domain (Figure 2.15; Figure
2.16) and one in the β domain (Figure 2.17) and another connecting α and β domains
(Figure 2.18), implying the loose tertiary structures. This conformational transition brings
the proteins from the native state to a state – a molten globule alike since similar
structural properties are found compared with a molten globule. As described above, the
surfactants-induced structures of lysozyme remain a significant amount of secondary
structure but without the specific tertiary structure (loss of native contacts and four
unstable disulfide bonds). There is 10% - 30% increase in the radius of gyration of the
swollen lysozyme (Table 2.4) corresponding to the dramatic increase of SASA (SL)
(Figure 2.7 and Table 2.4). Also the packed hydrophobic core of the α domain of
lysozyme was loose and unfolded by the surfactants, azo-TAB. All these are the common
characteristics of the molten globule state as indicated by Arai and Kuwajima [78] and so
we could understand that the surfactants were making the lysozyme molecules move to
the molten globule states after the conformation transition point. It is known that the
60

molten globule has liquid-like interior which might result in the close attachment of azo-
TAB on lysozyme molecule observed after the transition point (shown as Figure 2.7).
Before the transition point, the azoTABs have not found the most appropriate binding
sites and have not bound on the protein tightly, resulting in the values of the interface
between lysozyme and surfactants have the range 0~20 nm
2
. After transition point,
azoTABs bind on the protein tightly and so the small range of the interface value 13~17
nm
2
is obtained. Because the sizes and the structures of the surfactants should not be
negligible, it would be easier for surfactants to attach on a liquid-like interior of lysozyme
molecule than a rigid-native state of lysozyme. Salt ions could also attach closely on a
protein at a molten globule state [79].

Figure 2.7 : Time evolution of the solvent access surface area (SASA) calculation in T5
system. Blue line represents the protein solvent access surface area and the green line
stands for the interface between protein and ten trans-azoTABs. The transition point
happens around 450ns, and higher amplitude (fluctuation) of SASA (protein) but much
less amplitude (fluctuation) of protein-azoTAB interface after 450 ns.
61

We calculated the root mean square deviations (RMSDs) of the unfolding protein
molecule in the azo-TAB solution to measure the structural differences compared with its
initial crystalline form. Figure 2.8 displays the RMSDs values of the all atoms in the
whole lysozyme molecule, the all atoms of the α domain, and the all atoms of the β
domain in the T5 system over the entire simulation trajectory. Also, the RMSD-time
evolution of the backbone atoms (C, N and Cα) of the proteins in T5 system is shown in
the same plot. It is clear that the RMSDs curve of the α domain is on the top while the
RMSDs curve of the β domain lies under the other three curves. In the crystal structure,
the α domain has been considered as the more rigid region than the β domain inside the
molecule. Moreover, the lysozyme molecule in aqueous solution at room temperature
remains significant structural features from the crystal molecules [80] although the
flexibility of the small segments of lysozyme is affected by the surrounded water
molecules. As shown in Figure 2.9, the pure lysozyme in water (system P) has the little
higher RMSD value of the β domain than that of the α domain. The same results was
obtained by Sinha, S.K. and S. Bandyopadhyay [81], displaying the less structured β
domain than the α domain. Our simulations of lysozyme in azoTAB solution have been
also performed in room temperature. In the beginning, the surfactants, azoTABs, are
randomly distributed around lysozyme. Within 50ns, they moves closely to the lysozyme
and replaces the positions of the water molecules surrounding the protein, then further
searches and bounds on the binding pockets of lysozyme by interacting with the protein.
At the transition point, the 450
th
ns, a sharp increase of RMSD is observed for the whole
protein and the α domain but a slow increase of RMSD for the β domain. This implies
62

azo-TAB induce more internal motion as well as internal flexibility in the α domain than
in the β domain. The two RMSD curves of the whole protein and the α domain are almost
overlapped and it explains that the RMSDs of the whole protein are mainly contributed
from the disordered helical structures in the α domain. It is clear that the RMSD curves
for the backbone atoms(C, N and Cα) agrees well with the RMSD curve of the all atoms
of the protein and therefore, the significant change of RMSDs did not arise from the
active motions of the residue side chains. As we found in the time evolution of the
secondary structure seen in Figure 2.5, the Helix A lost 40% of its helical conformation
around the 450
th
ns. This could explain the jump in RMSD curves for the α domain and
the whole protein at the 450
th
ns; however, RMSDs values stop increasing and reach
around 1.0 at the 550
th
ns. On the other hand, there is a slow increase of RMSD values of
the β domain since 500
th
ns and the lysozyme did not keep its intact structure for β sheet I
at the same time. The RMSD values of the β domain levels off at 0.58 after 800
th
ns. The
fluctuation amplitudes of the four RMSD curves did not increase after the 450
th
ns.

Figure 2.8 : Root mean-square deviation (in nm) of all atoms of protein as a function of
time (ns) with respect to the crystal structure for T5 and P system.
63

Figure 2.9 Root mean-square deviation (in nm) of all atoms of protein as a function of
time (ns) with respect to the crystal structure for P system. This plot is zoomed in from
lower left side of Figure 2.8.

As shown in Figure 2.10, the gyration radius of the whole protein structure
decreases while the gyration radius of each α and β domain increases during the early
450ns. The protein, lysozyme, is composed of the two domains. Since the gyration radius
is measuring the size of the molecule, the larger gyration radius of the two domains, the
bigger sizes of each domains. However, the whole gyration radius of the overall protein
become smaller, which predicts the closer distance between the α and β domain during
this early simulation time period. The domain distance is defined as the distance between
the two centers of mass of the two domains, α and β and obviously, we can see the
decrease of the domain distance in Figure 2.11. Also, the observing movement of the
Helix B and the Helix D toward β sheet results in the closer domain distance as revealed
64

by the trajectory movie. The three gyration curves exhibit two subtle jumps, one at the
450
th
ns and the other at the 750
th
ns approximately. There is also a jump for the domain
distance at the 450
th
ns while a decrease round the 750
th
ns. The increase gyration radius
of both α and β domain as well as the increase distance between the two domains result in
the overall increase of the gyration radius of the whole protein at the 450
th
ns. The
fluctuation amplitudes of the gyration radius and the domain distance increase after the
450
th
ns. The enhanced amplitudes of fluctuation showing on the calculation of three
gyration radiuses and the domain distance may imply the improved internal dynamics and
the more flexible conformation of the protein after the 450
th
ns.

Figure 2.10 : Radius of gyration (nm) of the whole protein, the α domain, and the β
domain for the protein in T5 sytem as a function of time (ns). The transition point
happens around 450ns, and higher amplitude (fluctuation) after 450 ns.
65

Figure 2.11 : Domain Distance (protein in T5 system) – distance (nm) between the
centers of mass of the α-domain and the β-domain as a function of time (ns). The
transition point happens around 450ns, and higher amplitude (fluctuation) after 450 ns.

Tertiary structure of a protein molecule refers to its three dimensional
arrangement for each secondary structure. Calculation of the preserved native contacts
versus time of the unfolding proteins provides information about how much the atoms of
the protein rearrange their relative positions or how its secondary structures are
dislocated. We consider that there exists native contacts between two different residues i,
j if the distance between the Cα atoms of the two residues, i, j, is within 0.6nm. The total
native contacts of the crystal structure of lysozyme from 6LYZ.pdb are 365. There are
224 native contacts in the α domain, 125 native contacts in the β domain and 16 native
contacts in the α-β domain interface. Figure 2.12 shows the percentage of contacts of the
unfolding lysozyme versus time of the T5 system based on the overall native contacts
(365) of the crystal structure. Obviously, the total contacts and native contacts in the
66

lysozyme molecule decreases; especially around the transition point- 450
th
ns, there is a
drop where the close contacts between lysozyme and azo-TABs are formed, as seen from
Figure 2.7. Non-native contacts steadily increase through the simulation course. After the
transition point, the much larger fluctuation of calculation of total contacts and native
contacts are found within the protein molecule but there is slightly more fluctuation of
counting non-native contacts. Therefore, it has shown that the protein molecule has
increased internal dynamics while closely interacting with azo-TAB, either trans or cis.
Figure 2.14 displays that near the 1000
th
ns, 63%-67% native contacts left within the
whole protein, 60%-65% native contacts left in the α domain and 65-70 % native contacts
left in the β domain. The dislocation of α helix is found so the α domain has less
percentage of native contacts left than the β domain and this occurrence is noticed in all
other simulation runs (Table 2.5). Again, it is an indication that azo-TAB (either trans or
cis) induce more unfolding in the α domain than in the β domain. Also, the fluctuations of
percentage of native contacts both in the α domain and in the β domain become larger
after the transition point. However, there are higher fluctuations in the α domain than in
the β domain after the transition point. It is predictable that there are higher internal
dynamics in the α domain than in the β domain while lysozyme molecules are unfolding
due to biding and interacting with azo-TAB (either trans or cis).
67

Figure 2.12 : Percentage of contacts (%) of the whole protein in T5 system. The
percentage is based on the total native contacts of the crystal lysozyme. Total contacts are
summed up of native contacts and non-native contacts. The transition point happens
around 450ns, and higher amplitude (fluctuation) after 450 ns.

Figure 2.13 : Number of total contacts of the whole protein, the α domain, the β domain,
and the interface between α-β domain for T5 system. The transition point happens around
450ns, and higher amplitude (fluctuation) after 450 ns.
68

Figure 2.14 : Number of native contacts of the whole protein, the α domain, the β
domain, and the interface between α-β domain for T5 system. The transition point
happens around 450ns, and higher amplitude (fluctuation) after 450 ns.

Figure 2.15 : S-S bond length (CYS6 and CYS 127, both are in the α domain) is
calculated for protein in T5 system. The transition point happens around 450ns, and
higher amplitude (fluctuation) after 450 ns.
69

Figure 2.16 : S-S bond length (CYS30 and CYS 115, both are in the α domain) is
calculated for protein in T5 system. The transition point happens around 450ns, and
higher amplitude (fluctuation) after 450 ns.

Figure 2.17 : S-S bond length (CYS64 and CYS 80, both are in the β domain) is
calculated for protein in T5 system. The transition point happens around 450ns, and
higher amplitude (fluctuation) after 450 ns.
70

Figure 2.18 : S-S bond length (CYS76 in the α domain and CYS 94 in the β domain) is
calculated for protein in T5 system. The transition point happens around 450ns, and
higher amplitude (fluctuation) after 450 ns.

The total number of hydrogen bonds of the entire protein and its α domain start
decreasing around the 300
th
ns (Figure 2.19). The number of hydrogen bonds in the β
domain begins decreasing at the 400
th
ns accompanied with slow loss of native contacts
in the β domain in the early simulation (0
th
to 400
th
ns) (Figure 2.14). The transition point
is at 450
th
ns after which relatively large fluctuations are found. The total number of
hydrogen bonds of the crystal lysozyme (6LYZ.pdb) is 89, including 49 hydrogen bonds
in the α domain and 35 in the β domain, 5 at the interface of α-β domains. The
approximate number of hydrogen bonds of pure lysozyme in water is about 85~105. The
number of the hydrogen bonds in the interface of α and β domain increases a few and
71

then decrease; basically, the number is around 10-15, higher than the value of the crystal
form, as shown in Table 2.5.

Figure 2.19 : Number of hydrogen bonds of the whole protein, the α domain, the β
domain, and the interface between α-β domains in T5 system. The transition point
happens around 450ns, and higher amplitude (fluctuation) after 450 ns.

72

Transiti
on point
(~X
th

ns)

RMSD-
protein
(nm)

RMSD-
α
domain
(nm)
RMSD-
β
domain
(nm)
Radius of
gyration -
protein
(nm)
Radius of
gyration -
α domain
(nm)
Radius of
gyration -
β domain
(nm)
domain
distance
(nm)

Protein –
azotabs-
interface
(nm
2
)
SASA-
hydrophobic
(nm
2
)

SASA-
hydrophilic
(nm
2
)

Crystal X 0.0 0.0 0.0 1.40 1.18 0.99 1.76 X 50.96 39.24
P X 0.48 0.45 0.46 1.37±0.01 1.22±0.01 1.02±0.02 1.55 X 53.44±1.72 39.12±1.48
T1 500 0.90 0.97 0.55 1.42±0.02 1.26±0.02 1.00±0.02 1.67 16.63±1.24 69.92±9.37 50.01±7.27
T2 470 1.09 1.17 0.60 1.46±0.02 1.23±0.02 1.02±0.02 1.83 13.45±1.30 72.57±8.55 54.44±6.83
T3 450 0.89 0.95 0.57 1.43±0.02 1.27±0.02 1.02±0.02 1.66 12.95±1.18 69.03±8.18 50.56±6.12
T4 480 0.93 0.99 0.58 1.47±0.02 1.25±0.03 1.06±0.02 1.80 15.73±1.80 69.53±8.19 52.57±6.35
T5 450 0.92 0.99 0.57 1.45±0.02 1.36±0.02 1.09±0.02 1.50 14.78±1.21 72.22±9.05 50.19±7.54
T6 460 0.99 1.04 0.56 1.44±0.02 1.26±0.02 1.01±0.02 1.73 15.52±1.20 72.34±9.37 54.20±7.39
T7 470 0.99 1.02 0.52 1.47±0.02 1.24±0.02 1.05±0.02 1.82 16.08±1.08 69.28±8.21 52.98±6.46
T8 460 0.95 0.93 0.56 1.45±0.02 1.32±0.02 1.05±0.02 1.60 18.52±1.15 71.41±7.92 52.29±6.63
T9 450 1.01 1.04 0.63 1.52±0.02 1.33±0.02 1.00±0.02 1.90 17.39±1.34 72.14±8.31 52.11±6.38
T10 410 0.98 1.03 0.64 1.47±0.01 1.34±0.02 0.99±0.02 1.71 19.14±1.21 76.91±6.87 57.01±5.56
C1 420 0.96 1.05 0.62 1.46±0.02 1.25±0.02 1.00±0.02 1.83 13.65±1.49 66.62±8.74 51.07±6.77
C2 400 1.06 1.15 0.66 1.46±0.02 1.31±0.03 1.01±0.02 1.69 18.59±1.27 70.89±7.56 50.94±5.78
C3 400 1.11 1.15 0.67 1.53±0.03 1.33±0.03 1.10±0.01 1.83 19.42±1.53 74.10±6.72 51.28±5.31
C4 450 0.87 0.95 0.58 1.46±0.02 1.31±0.02 1.02±0.02 1.70 17.71±1.15 72.05±9.64 52.97±7.14
C5 340 1.00 1.08 0.66 1.52±0.02 1.29±0.02 1.02±0.02 1.92 19.69±1.39 76.61±9.06 56.48±7.13
C6 400 1.01 1.06 0.65 1.47±0.02 1.34±0.02 1.04±0.02 1.65 15.43±1.35 79.07±9.03 56.95±6.93
C7 400 0.97 1.01 0.62 1.50±0.02 1.29±0.02 1.04±0.02 1.86 17.55±1.24 77.35±7.80 54.47±6.65
C8 350 0.95 1.05 0.59 1.54±0.02 1.41±0.02 1.05±0.02 1.74 19.18±1.17 80.29±8.25 58.36±6.32
C9 370 0.80 0.85 0.52 1.42±0.02 1.34±0.02 1.05±0.02 1.42 15.72±1.40 69.24±6.97 50.92±5.33
C10 340 1.08 1.14 0.67 1.56±0.02 1.37±0.02 1.04±0.02 1.90 18.74±1.30 77.00±7.52 54.49±6.14

Table 2.4: Average structural properties of the intermediates of HEW lysozyme. In the
system T1~T10 and C1 ~ C10, the time range for averaging is over protein trajectories
900ns - 1000ns and we adopt one protein structure per 0.5 ns. The system, P, is
simulating pure protein solvated in water. The average protein structures are over 100ns-
200ns. Domain distance is defined as the distance between two centers of mass from α
and β domains. All solvent access surface area is calculated using Gromacs program,
which employs the Lee and Richards algorithm with a probe radius of 0.14m. Transition
point is read from all plots. The reference structure of RMSD calculation is crystal
protein.

73

#Native
Contacts -
protein
(ratio%)
#Native
Contacts-
α-domain
(ratio%)
#Native
Contacts-
β-domain
(ratio%)
#Native
Contacts-
α-β
interface
#Total
Contacts-
α-β
interface
#Hbond-protein
(ratio%)

#Hbond-
α-domain
(ratio%)

#Hbond-
β-domain
(ratio%)

#Hbond-
α-β
interface
(ratio%)
Crystal 365 224 125 16 16 89 49 35 5
P 283±6(78±2) 174±4(78±2) 101±3(81±3) 9±1 9±2 89±6(100±6) 50±4(101±8) 31±3(87±10) 9±2
T1 268±12(74±3) 160±8(72±4) 100±6(80±5) 9±1 14±3 85±7(96±8) 47±5(97±10) 30±4(86±13) 7±2
T2 241±10(66±3) 145±8(65±4) 89±5(71±4) 7±1 8±2 74±7(83±8) 39±5(79±10) 29±4(82±12) 6±2
T3 244±10(67±3) 152±7(68±3) 84±4(67±3) 8±1 17±2 83±7(93±8) 46±5(94±10) 25±4(73±10) 11±2
T4 251±11(69±3) 159±9(71±4) 86±4(69±3) 6±1 8±1 81±7(91±8) 48±5(98±10) 29±3(83±10) 5±2
T5 236±10(65±3) 145±7(65±3) 84±4(67±3) 7±1 29±4 73±6(82±6) 38±4(78±7) 24±3(68±10) 11±3
T6 256±12(70±3) 153±8(68±4) 95±6(76±4) 8±1 17±3 81±7(91±8) 44±5(89±11) 29±4(82±12) 8±2
T7 239±9(66±2) 148±6(66±3) 83±4(66±3) 8±1 12±2 79±7(89±8) 48±5(97±10) 25±4(71±11) 7±2
T8 251±10(69±3) 152±7(68±3) 91±5(73±4) 8±1 13±2 76±6(85±7) 38±4(78±8) 26±4(74±11) 12±2
T9 240±10(66±3) 154±8(69±3) 81±4(64±3) 5±1 14±2 81±6(91±7) 44±5(90±10) 28±4(79±10) 9±2
T10 270±8(74±2) 161±6(72±3) 102±5(82±4) 7±1 9±2 78±6(87±7) 41±4(83±8) 30±4(86±11) 7±2
C1 273±12(75±3) 159±8(71±4) 104±6(83±5) 10±2 11±2 83±6(93±7) 44±4(91±9) 32±4(91±12) 7±2
C2 241±9(66±3) 138±7(61±3) 97±5(78±4) 6±1 11±2 79±7(89±8) 42±5(87±10) 31±4(89±11) 5±2
C3 248±10(68±3) 158±7(71±3) 83±5(66±4) 6±1 28±3 79±6(89±7) 44±5(90±9) 21±4(60±11) 14±3
C4 262±12(72±3) 164±8(73±4) 91±4(72±3) 8±1 18±3 82±7(92±8) 44±5(89±10) 27±3(78±10) 11±2
C5 231±9(63±3) 145±8(65±4) 80±4(64±3) 6±1 12±2 72±7(81±8) 43±5(88±10) 23±4(66±11) 6±2
C6 259±12(71±3) 158±9(71±4) 92±5(74±4) 8±1 12±2 77±7(86±8) 45±5(91±11) 25±4(71±10) 7±2
C7 242±9(66±2) 153±7(68±3) 82±5(66±4) 7±1 14±2 77±7(87±8) 39±5(80±9) 30±4(86±12) 8±2
C8 217±8(60±2) 130±6(58±3) 81±4(65±4) 6±1 14±3 69±7(78±8) 38±4(78±9) 24±4(69±11) 7±2
C9 247±9(68±2) 153±7(68±3) 88±5(70±4) 6±1 27±3 78±7(87±8) 39±4(80±8) 23±4(66±11) 15±3
C10 227±8(62±2) 144±7(64±3) 79±4(63±3) 4±1 14±2 75±6(85±7) 38±5(78±10) 29±3(82±10) 8±2

Table 2.5: Number of residue intra-molecular contacts and number of hydrogen bonds in
the intermediates of HEW lysozyme. Average is over 900ns-1000ns for system T1~T10
and C1~C10 while 100ns-200ns for system P. Two residues are considered to have
contacts with each other when their Cα atoms are within 0.6 nm. Hydrogen bonds are
determined based on cutoff radius, 0.35 nm (acceptor – donor) and cutoff angle, 30
o

(acceptor - donor - hydrogen).

74

2.3.5 Binding affinity of different parts of azo-TAB toward lysozyme
We have been examining what part(s) of the surfactant bind to the protein using
the relatively effective collision times with protein. We test the effective collision of the
specific atom of azotabs to the protein within a distance of 0.4 nm from any atom of the
protein residue through the overall simulation trajectories. For better resolution, we
conduct another collision test; however, the shorter distance (0.3 nm) of the effective
collision is applied. The relative collision times in percentage based on either 0.4 nm or
0.3 nm (shown inside the brackets) are indicated in Figure 2.20. The most probable
interaction site of azoTAB with lysozyme is the tail group, the phenyl rings, opposite to
the head group (-N(CH3)3). The phenyl rings are considered having hydrophobic
interactions with lysozyme while the (-N(CH3)3 with positive charges is conducting
electrostatic interactions with lysozyme. As a result, the increase of the hydrophobic
SASA is larger than the increase of the hydrophilic SASA, see Table 2.4. There exists
similar action of some penicillins with HSA. Thus, while alkyl surfactants are generally
considered to interact non-specifically with proteins, azoTAB binding appears to exhibit
quasi-specific, potential -  stacking interactions.

75

Figure 2.20 : The relative effective collisions in percentage of azotabs to
lysozyme. There are two sets of percentages; the percentage shown inside the brackets is
based on the effective collision distance of 0.3 nm while the other is based on 0.4 nm.
The phenyl rings have higher relative collision percentage for both trans and cis
structures. The plots of the two structural types of the surfactants are shown as VDW
drawing style with VMD program [30].

2.4 Conclusion
Our MD simulations illustrate that a little amount of surfactants, azo-TAB (either
trans or cis), can unfold the lysozyme at room temperature, where lysozyme is swollen
and stabilized as α-helices are partly unfolding, while β-sheets is relatively intact. This
corresponds to the results of the previous unfolding experiment of lysozyme under low
concentration of azo-TAB (<12 mM) upon either visible light or UV light exposure and
the induced-intermediates represent the conformations with the swollen, unfolding α-
domain but more intact β domain. The primary denaturing begins from helix D (residues
108-115), followed by helix A (residues 4-15) as suggested by the SANS techniques
combined FI-IR measurements. Helix C results in few unfolding due to the low binding
affinity toward both trans and cis azo-TAB and the high total contacts in the interface of
the α and β domains. Helix B has little changes since it is the most imbedded helix or the
76

core of the hydrophobic region, where the surfactants hardly reach it. The bottom part of
the hinge linking the two domains, α and β, attracts much of azo-TAB binding and
becomes loose during the simulation course. Both of trans-azotab and cis-azotab have
high binding affinity on the hydrophobic residues of lysozyme, such as Trp. The number
of total contacts and hydrogen bonds between α-β domain interface increase during the
simulation course; therefore, it could explain the relatively intact structure of β domain.
The lysozyme molecule exhibits increased internal motion once the transition point is
reached. Azo-TABs interact more closely with the protein first and then the atoms within
the protein behave more dynamic, resulting in a decrease in the overall native contacts
within the protein. At this time, the protein loses 30%~40% of its tertiary contacts and
moves into a swollen state/ molten globule state. Particularly, the more improved
dynamics is observed in the α domain than the β domain. This simulation work has
revealed that the ability of azo-TAB to unfold the α domain via the potential nonbonding
- hydrophobic interaction. Thus, the unfolding pathway of lysozyme in azo-TAB solution
can be further tracked through characterizing these intermediates owning more structured
β domain than α domain.

77

2.5 References
1. Anfinsen, C.B., Principles that govern folding of protein chains. Science, 1973.
181(4096): p. 223-230.

2. Levintha.C, How to fold graciously. In Proceedings of a Meeting Held at Allerton
House, Monticello, IL (Debrunner, P., Tsibris, J. C. M. & MuÈ nck, E., eds), pp.
22-24, University of Illinois Press, Urbana., 1969.

3. Karplus, M. and A. Sali, Theoretical studies of protein folding and unfolding.
Current Opinion in Structural Biology, 1995. 5(1): p. 58-73.

4. Roder, H. and W. Colon, Kinetic role of early intermediates in protein folding.
Current Opinion in Structural Biology, 1997. 7(1): p. 15-28.

5. Matthews, C.R., Pathways of protein folding. Annual Review of Biochemistry,
1993. 62: p. 653-683.

6. Rothwarf, D.M. and H.A. Scheraga, Regeneration of bovine pancreatic
ribonuclease - A steady state distribution. Biochemistry, 1993. 32(10): p. 2671-
2679.

7. Evans, P.A. and S.E. Radford, Probing the structure of folding intermediates.
Current Opinion in Structural Biology, 1994. 4(1): p. 100-106.

8. Roder, H.a.G.A.E., in Mechanisms of Protein Folding (Pain, R. H., Ed.) pp 26-54,
Oxford University Press, New York. 1994.

9. Baldwin, R.L., The nature of protein-folding pathways - the classical versus the
new view. Journal of Biomolecular Nmr, 1995. 5(2): p. 103-109.

10. Fersht, A.R., Mapping the structures of transition states and intermediates in
folding - delineation of pathways at high resolution. Philosophical Transactions of
the Royal Society of London Series B-Biological Sciences, 1995. 348(1323): p.
11-15.

11. Lee, C.T., K.A. Smith, and T.A. Hatton, Photocontrol of protein folding: The
interaction of photosensitive surfactants with bovine serum albumin.
Biochemistry, 2005. 44(2): p. 524-536.

12. Hamill, A.C., S.C. Wang, and C.T. Lee, Probing lysozyme conformation with
light reveals a new folding intermediate. Biochemistry, 2005. 44(46): p. 15139-
15149.

78

13. Hamill, A.C., S.C. Wang, and C.T. Lee, Solution structure of an amyloid-forming
protein during photoinitiated hexamer-dodecamer transitions revealed through
small-angle neutron scattering. Biochemistry, 2007. 46(26): p. 7694-7705.

14. Brooks, C.L., Simulations of protein folding and unfolding. Current Opinion in
Structural Biology, 1998. 8(2): p. 222-226.

15. Duan, Y. and P.A. Kollman, Pathways to a protein folding intermediate observed
in a 1-microsecond simulation in aqueous solution. Science, 1998. 282(5389): p.
740-744.

16. Blake, C.C.F., et al., Structure of hen egg-white lysozyme - A 3-dimensional
fourier synthesis at 2A resolution. Nature, 1965. 206(4986): p. 757-&.

17. Dobson, C.M., P.A. Evans, and S.E. Radford, Understanding how proteins fold -
the lysozyme story so far. Trends in Biochemical Sciences, 1994. 19(1): p. 31-37.

18. Ikeguchi, M., et al., Evidence for identity between the equilibrium unfolding
intermediate and a transient folding intermediate - a comparative study of the
folding reactions of alpha-lactalbumin and lysozyme. Biochemistry, 1986. 25(22):
p. 6965-6972.

19. Kato, S., et al., Spectral evidence for a rapidly formed structural intermediate in
the refolding kinetics of hen egg white lysozyme. Biochemistry, 1981. 20(5): p.
1080-1085.

20. Kuwajima, K., et al., Comparison of the transient folding intermediates in
lysozyme and alpha-lactalbumin. Biochemistry, 1985. 24(4): p. 874-881.

21. Radford, S.E., C.M. Dobson, and P.A. Evans, The folding of hen lysozyme
involves partially structured intermediates and multiple pathways. Nature, 1992.
358(6384): p. 302-307.

22. Chaffotte, A.F., Y. Guillou, and M.E. Goldberg, Kinetic resolution of peptide
bond and side chain far-UV circular dichroism during the folding of hen egg
white lysozyme. Biochemistry, 1992. 31(40): p. 9694-9702.

23. Acharya, K.R., et al., Crystal structure of human alpha-lactalbumin at 1.7 A
resolution. Journal of Molecular Biology, 1991. 221(2): p. 571-581.

24. Itzhaki, L.S., et al., Tertiary interactions in the folding pathway of hen lysozyme -
kinetic studies using fluorescent probes. Biochemistry, 1994. 33(17): p. 5212-
5220.

79

25. Kato, S., N. Shimamoto, and H. Utiyama, Identification and characterization of
the direct folding process of hen egg white lysozyme. Biochemistry, 1982. 21(1):
p. 38-43.

26. Buck, M., S.E. Radford, and C.M. Dobson, A partially folded state of hen egg
white lysozyme in trifluoroethanol - structural characterization and implications
for protein folding. Biochemistry, 1993. 32(2): p. 669-678.

27. Miranker, A., et al., Demonstration by NMR of folding domains in lysozyme.
Nature, 1991. 349(6310): p. 633-636.

28. Miranker, A., et al., Detection of transient protein folding populations by mass
spectrometry. Science, 1993. 262(5135): p. 896-900.

29. Hooke, S.D., et al., Cooperative elements in protein folding monitored by
electrospray ionization mass spectrometry. Journal of the American Chemical
Society, 1995. 117(28): p. 7548-7549.

30. Humphrey, W., A. Dalke, and K. Schulten, VMD: Visual molecular dynamics.
Journal of Molecular Graphics, 1996. 14(1): p. 33-&.

31. Matagne, A., S.E. Radford, and C.M. Dobson, Fast and slow tracks in lysozyme
folding: Insight into the role of domains in the folding process. Journal of
Molecular Biology, 1997. 267(5): p. 1068-1074.

32. Dinner, A.R., et al., Understanding protein folding via free-energy surfaces from
theory and experiment. Trends in Biochemical Sciences, 2000. 25(7): p. 331-339.

33. Gilquin, B., C. Guilbert, and D. Perahia, Unfolding of hen egg lysozyme by
molecular dynamics simulations at 300K: Insight into the role of the interdomain
interface. Proteins-Structure Function and Genetics, 2000. 41(1): p. 58-74.

34. Mark, A.E. and W.F. Vangunsteren, Simulation of the thermal denaturation of
hen egg white lysozyme - Trapping the molten globule state. Biochemistry, 1992.
31(34): p. 7745-7748.

35. Hunenberger, P.H., A.E. Mark, and W.F. Vangunsteren, Computational
approaches to study protein unfolding - hen egg white lysozyme as a case study.
Proteins-Structure Function and Genetics, 1995. 21(3): p. 196-213.

36. Williams, M.A., J.M. Thornton, and J.M. Goodfellow, Modelling protein
unfolding: hen egg-white lysozyme. Protein Engineering, 1997. 10(8): p. 895-903.

80

37. Pepys, M.B., et al., Human lysozyme gene mutations cause hereditary systemic
amyloidosis. Nature, 1993. 362(6420): p. 553-557.

38. Kazmirski, S.L. and V. Daggett, Non-native interactions in protein folding
intermediates: Molecular dynamics simulations of hen lysozyme. Journal of
Molecular Biology, 1998. 284(3): p. 793-806.

39. Smith, L.J., et al., Molecular dynamics simulations of peptide fragments from hen
lysozyme: Insight into non-native protein conformations. Journal of Molecular
Biology, 1998. 280(4): p. 703-719.

40. Chen, L.L., K.O. Hodgson, and S. Doniach, A lysozyme folding intermediate
revealed by solution X-ray scattering. Journal of Molecular Biology, 1996.
261(5): p. 658-671.

41. Zhou, R.H., et al., Destruction of long-range interactions by a single mutation in
lysozyme. Proceedings of the National Academy of Sciences of the United States
of America, 2007. 104(14): p. 5824-5829.

42. Dobson, C.M., Protein folding solid evidence for molten globules. Current
Biology, 1994. 4(7): p. 636-640.

43. Lee, P.K.J. and Jirgenso.B, Conformational transitions of immunoglobulin
fragment FC(T) incited by alkyl sulfates of various hydrophobic chain lengths.
Biochimica Et Biophysica Acta, 1971. 229(3): p. 631-&.

44. D. van der Spoel, E.L., B. Hess, A. R. van Buuren, E. Apol, P. J. Meulenhoff,,
A.L.T.M.S. D. P. Tieleman, K. A. Feenstra, R. van Drunen and H. J. C., and
Berendsen, Gromacs User Manual version 3.3, www.gromacs.org. 2005.

45. Bekker, H., Berendsen, H. J. C., Dijkstra, E. J., Achterop, S., van Drunen, R., van
der Spoel, D., Sijbers, A., Keegstra, H., Reitsma, B., Renardus, M. K. R.,
Gromacs: A parallel computer for molecular dynamics simulations. In Physics
Computing 92 (Singapore, 1993).de Groot, R. A., Nadrchal, J., eds. . World
Scientific.

46. Berendsen, H.J.C., D. Vanderspoel, and R. Vandrunen, Gromacs - A message
passing parallel molecular dynamics implementation. Computer Physics
Communications, 1995. 91(1-3): p. 43-56.

47. Lindahl, E., B. Hess, and D. van der Spoel, GROMACS 3.0: a package for
molecular simulation and trajectory analysis. Journal of Molecular Modeling,
2001. 7(8): p. 306-317.

81

48. Van der Spoel, D., et al., GROMACS: Fast, flexible, and free. Journal of
Computational Chemistry, 2005. 26(16): p. 1701-1718.

49. van Gunsteren, W.F., Billeter, S.R., Eising, A.A., Hunenberger, P.H., Kruger, P.,
Mark, A.E., Scott, W.R.P., Tironi, I.G., Biomolecular Simulation: The Gromos96
Manual and User Guide. Hochschulverlag AG an der ETH, Zurich., 1996.

50. Neria, E., S. Fischer, and M. Karplus, Simulation of activation free energies in
molecular systems. Journal of Chemical Physics, 1996. 105(5): p. 1902-1921.

51. Jorgensen, W.L., et al., Comparison of simple potential functions for simulating
liquid water. Journal of Chemical Physics, 1983. 79(2): p. 926-935.

52. Bockmann, M., et al., Atomistic force field for azobenzene compounds adapted for
QM/MM simulations with applications to liquids and liquid crystals. Journal of
Chemical Theory and Computation, 2007. 3(5): p. 1789-1802.

53. Diamond, R., Real-space refinement of structure of hen egg-white lysozyme.
Journal of Molecular Biology, 1974. 82(3): p. 371-&.

54. Wang, S.C. and C.T. Lee, Enhanced enzymatic activity through photoreversible
conformational changes. Biochemistry, 2007. 46(50): p. 14557-14566.

55. Lu, D.N., et al., Molecular simulation of surfactant-assisted protein refolding.
Journal of Chemical Physics, 2005. 122(13).

56. Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, W.F., DiNola, A., Haak, J.R.,
In: Pullman, B. (Ed.) Intermolecular Forces. Reidel Dordrecht, The
Netherlands, 1981.

57. Tironi, I.G., et al., A generalized reaction field method for molecular dynamics
simulations. Journal of Chemical Physics, 1995. 102(13): p. 5451-5459.

58. Schulz, R., et al., Scaling of Multimillion-Atom Biological Molecular Dynamics
Simulation on a Petascale Supercomputer. Journal of Chemical Theory and
Computation, 2009. 5(10): p. 2798-2808.

59. Berendsen, H.J.C., et al., Molecular dynamics with coupling to an external bath.
Journal of Chemical Physics, 1984. 81(8): p. 3684-3690.

60. Silberstein, M., et al., Identification of substrate binding sites in enzymes by
computational solvent mapping. Journal of Molecular Biology, 2003. 332(5): p.
1095-1113.

82

61. Adcock, S.A. and J.A. McCammon, Molecular dynamics: Survey of methods for
simulating the activity of proteins. Chemical Reviews, 2006. 106(5): p. 1589-
1615.

62. Housaindokht, M.R., M.R. Bozorgmehr, and M. Bahrololoom, Analysis of ligand
binding to proteins using molecular dynamics simulations. Journal of Theoretical
Biology, 2008. 254(2): p. 294-300.

63. Deshpande, A., S. Nimsadkar, and S.C. Mande, Effect of alcohols on protein
hydration: crystallographic analysis of hen egg-white lysozyme in the presence of
alcohols. Acta Crystallographica Section D-Biological Crystallography, 2005. 61:
p. 1005-1008.

64. Liepinsh, E. and G. Otting, Organic solvents identify specific ligand binding sites
on protein surfaces. Nature Biotechnology, 1997. 15(3): p. 264-268.

65. Pike, A.C.W. and K.R. Acharya, A structural basis for the interation of urea with
lysozyme. Protein Science, 1994. 3(4): p. 706-710.

66. Salem, M., Y. Mauguen, and T. Prange, On the edge of the denaturation process:
Application of X-ray diffraction to barnase and lysozyme cross-linked crystals
with denaturants in molar concentrations. Biochimica Et Biophysica Acta-
Proteins and Proteomics, 2006. 1764(5): p. 903-912.

67. Mande, S.C. and M.E. Sobhia, Structural characterization of protein-denaturant
interactions: crystal structures of hen egg-white lysozyme in complex with DMSO
and guanidinium chloride. Protein Engineering, 2000. 13(2): p. 133-141.

68. Haliloglu, T. and I. Bahar, Structure-based analysis of protein dynamics:
Comparison of theoretical results for hen lysozyme with X-ray diffraction and
NMR relaxation data. Proteins-Structure Function and Genetics, 1999. 37(4): p.
654-667.

69. Kabsch, W. and C. Sander, Dictionary of protein secondary structure - pattern -
recognition of hydrogen-bonded and geometrical features. Biopolymers, 1983.
22(12): p. 2577-2637.

70. Yang, A.S., B. Hitz, and B. Honig, Free energy determinants of secondary
structure formation .3. beta-Turns and their role in protein folding. Journal of
Molecular Biology, 1996. 259(4): p. 873-882.

71. Moraitakis, G. and J.M. Goodfellow, Simulations of human lysozyme: Probing the
conformations triggering amyloidosis. Biophysical Journal, 2003. 84(4): p. 2149-
2158.
83

72. Hooke, S.D., S.E. Radford, and C.M. Dobson, Refolding of human lysozyme - a
comparison with the structurally homologous hen lysozyme. Biochemistry, 1994.
33(19): p. 5867-5876.

73. MorozovaRoche, L.A., et al., Structural characterisation and comparison of the
native and A-states of equine lysozyme. Journal of Molecular Biology, 1997.
268(5): p. 903-921.

74. Radford, S.E. and C.M. Dobson, Insights into protein-folding using physical
techniques - studies of lysozyme and alpha-lactalbumin. Philosophical
Transactions of the Royal Society of London Series B-Biological Sciences, 1995.
348(1323): p. 17-25.

75. Morozova, L.A., et al., Structural basis of the stability of a lysozyme molten
globule. Nature Structural Biology, 1995. 2(10): p. 871-875.

76. Schwalbe, H., et al., Structural and dynamical properties of a denatured protein.
Heteronuclear 3D NMR experiments and theoretical simulations of lysozyme in 8
M urea. Biochemistry, 1997. 36(29): p. 8977-8991.

77. Mirarefi, P. and C.T. Lee, Photo-induced unfolding and inactivation of bovine
carbonic anhydrase in the presence of a photoresponsive surfactant. Biochimica
Et Biophysica Acta-Proteins and Proteomics, 2010. 1804(1): p. 106-114.

78. Arai, M. and K. Kuwajima, Role of the molten globule state in protein folding, in
Advances in Protein Chemistry; Protein folding mechansims. 2000, Academic
Press Inc., 525 B Street, Suite 1900, San Diego, CA, 92101-4495, USA Harcourt
Place, 32 Jamestown Road, London, NW1 7BY, UK. p. 209-282.

79. Hamada, D., et al., Salt-induced formation of the molten globule state of
cytochrome-C studied by isothermal titration calorimetry. Proceedings of the
National Academy of Sciences of the United States of America, 1994. 91(22): p.
10325-10329.

80. Ichiye, T., et al., Structure and internal mobility of proteins - a molecular
dynamics study of hen egg-white lysozyme. Biopolymers, 1986. 25(10): p. 1909-
1937.

81. Sinha, S.K. and S. Bandyopadhyay, Differential flexibility of the secondary
structures of lysozyme and the structure and ordering of surrounding water
molecules. Journal of Chemical Physics, 2011. 134(11).

84

Chapter 3: Massively Parallel Simulations of Early Unfolding
α Domain of Lysozyme Induced by Surfactants, azoTABs

We performed ~27 μs molecular dynamic simulations of lysozyme molecules and
surfactants in the explicit solution of pH = 7.0 at 300 K. The results showed that at low
(1:10) lysozyme : surfactant ratios (either all trans or all cis), azoTABs unfold the α
domain and the β domain of lysozyme molecules concurrently; especially, the α domain
encounters apparent unfolding while keeping the β-sheets relatively intact with the
indication of the fast increase of the RMSD value of the α domain along with the greater
increase of the α-domain gyration radius than those of the β domain. The secondary
structure analysis showed that α-helix structure is lost more than β-sheet during the
simulation course. Helix D is fast denatured by azoTABs, and then helix A and helix C
are subsequently partially unfolded; helix B maintains most of its structure. On the other
hand, about 5%~10% of β sheet turns to coil. In the α-β domain interface, the total
number of contacts and hydrogen bonds do not decrease compared with those of the
native state, implying that the α domain is not leaving the β domain apart in response to
the not-dramatically-increased domain distances. The surfactants, azoTABs, reduce more
native contacts, total contacts, and number of hydrogen bonds but apparent increase of
solvent accessible surface area (SASA) in the α domain than those of the β domain.
Disordered helix packing induced by azoTABs is investigated and the residence time of
azoTAB toward lysozyme reveals the behavior of azoTAB on main chains and chains of
85

lysozyme. We have demonstrated that the surfactants, azoTAB, can early unfold the α
domains of lysozyme molecules at the room temperature, as seen in the previous SANS
data combined with FT-IR analysis, while the lysozymes have the major unfolding β
domain under most other denaturing conditions.

3.1 Introduction
Hen egg-white (HEW) lysozyme (Figure 3.1) is a 129-residue protein with two
domains (α and β). The α domain includes residues 1 to 39 and residues 88 to 129 while
the β domain has residues 40 to 87. The active site cleft is between the two domains and
there are a total of four disulfide bonds within the lysozyme molecule, where two of them
are situated in the intra-α-domain (C6-C127 and C30-C115), one is in the intra-β-domain
(C64-C80) and the other (C76-C94) is bonding α and β domains together. The α-domain
is basically composed of helices, four α-helices (helices A-D) and one C-terminal 310
helix. The β-domain is mainly a three-stranded anti-parallel β sheet, followed by a 310
helix, and an irregular loop. The HEW lysozyme is the most well-studied protein whose
intermediates are delicately characterized so that a good understanding about its
hierarchical folding pathway has been achieved [1].
86

Figure 3.1 : Crystal structure of hen egg white (HEW) lysozyme adopted from 6LYZ.pdb.
The α domain is colored in red and it has the α-helices A, B, C, D plus the C-terminal 310
helix. The β domain is colored in green with a three-stranded β-sheet and a 310 helix.
Each disulfide bonds are shown as a pair of two disulfide atoms (yellow spheres). There
are two disulfide bonds in the α domain, one in the β domain, the other connecting α and
β domain. And, six Trp residues (28, 62, 63, 108, 111, 123) are indicated. VMD program
[2] is adopted.

Hen lysozyme folds in multiple pathways [3]. Structural studies of folding
intermediates from both theory and experiment reveal that there are approximately 20%
of the lysozyme molecules, as the I αβ intermediates, forming the α and β domains at the
same time through a ‘fast track’ [4], then quickly folding into its native state. On the free
energy surfaces [5], those 20% molecules follow along the monotonically-decreasing
curve toward the native states with the hydrogen-bonding network fast built inside the
entire lysozyme molecules. On the other hand, there are near 10% molecules fold very
slowly due to their likely involvement of the proline isomerism. The remaining molecules
(70%), the majority ones, are folding back to the native structures through a ‘slow track’
87

[4], where there is a minimum on the free-energy landscape [5] and those trapped
intermediates, I α, have only the persistent helix structure in the α domain resulting from
the early formation of the hydrogen bonds in the α domain. Along with the ‘slow track’,
the helix formation of the α domain precedes the β-sheet formation of the β domain. The
final step, also the rate-limiting step, in folding process is to correctly dock the residues
from one domain into the other and form the α-β interdomain interface. It is particularly
important to insert residues Leu 55 and Ile 56 of the β-domain into a hydrophobic slot in
the α domain [6]. All-atom molecular dynamics simulations have been developed to
study the folding pathways of proteins by offering the intermediate structures at atomic
resolution. Some simulation work about unfolding of lysozyme has indicated that the
observed intermediates are with an unfolded β-domain but a structured α-domain [7-9] in
agreement with those intermediates found on the dominant slow-track pathway. The β-
domain has shown less stable than the α-domain in lysozyme [10]. Under 500 K of
unfolding simulations, hen egg white lysozyme is thermal denatured rendering loss of
the β-domain before the α-domain unfolds [11]. Moreover, the separate simulations of the
α-domain peptide and β-domain peptide showed that the β-sheet structure disappeared
earlier than the α-helix at high temperature [12]. It was investigated that the lysozyme
molecules were unfolded in the urea solution. Through X-ray scattering, the overall
increase of the gyration radius of the denatured lysozyme induced by urea was monitored.
The mean increase of the β-domain gyration radius is about 0.5 nm while there is around
0.2 nm increase in α-domain. This suggested that the β-domain unfolds much faster than
the α-domain [13]. The MD simulations have the same results and support the same
88

conclusion for the lysozyme-urea solution [12, 14]. In contrast to the large population of
the intermediates (70%) trapped in a free-energy minimum on the slow track, another
20% of the intermediates are expected to be accumulated along the fast track of folding
[4]. The key to stabilize the α-domain is to maintain its hydrophobic contacts between
residues; nevertheless, side chain-side chain hydrogen bonding and salt bridges together
with both hydrophobic and polar contacts all play significant roles to resist the possibly
dynamic fluctuations leading the β-domain to unfold. Some other unfolding simulations
are showing relatively persistent β-sheets. A modeling procedure, which introduces more
water molecules penetrating into the core of lysozyme, led the protein into the unfolding
pathway similar to its fast track on free-energy diagram [5]. Subsequently, the helices
fluctuate rapidly and further the compactness of the α domain is lost; nevertheless, the
native β-sheet is essentially intact [9]. With force field GROMOS87, two simulation
works performed by Mark and van Gunsteren et al. [7-8] are under high temperature,
high pressure, force controlled, or kinetic energy controlled. Both of the simulations have
lysozyme with persistent β-sheets but unstructured α-helices.
As chemical denaturants, like urea and GdmCl [15-16], unfold proteins,
surfactants are also used to control the unfolding extent of proteins. The surfactants,
azoTAB, have been proven to stabilize the partially-folded states, the intermediates, of
lysozyme in solution [17]. It carries a head, the trimethyl ammonium, and a big
photoresponsive tail, two phenol rings. With different light illumination, the relative
position of the two phenol rings can be adjusted, which results in two different
structures/forms of the azoTAB, trans (visible light) and cis (UV-light). Because the two
89

forms of azoTAB, trans and cis, have different extent of interaction with lysozyme under
low concentration of azoTAB (<12mM), it was demonstrated that the unfolding event of
lysozyme can be controlled by light illumination. Basically, trans-azoTAB induces much
more unfolding conformation changes of lysozyme than cis-azoTAB at such low
surfactant concentration. As a result, the unfolding mostly happens in the α-domain of the
lysozyme, which makes the apparent swelling of its α-domain leaving a relatively intact
structure of its β-domain through the observation via SANS combined with the analysis
of the protein secondary structure via FT-IR spectroscopy. These induced intermediates
by azoTAB show separate characteristics than those well-studied lysozyme intermediates
that have an intact folded α domain but unfolded β domain.
In the present study, we conducted massively parallel molecular dynamic
simulations for ~27 µs at 300K and found that either trans or cis azoTAB partially unfold
the α domain as well as the β domains of lysozyme molecules. With conformational
analysis of protein intermediates, the extent of the unfolded α-helix is more than that of
the β-sheets. This corresponds to the experimental observation, where lysozyme unfolds
even under low concentration (<12mM) of azoTAB and more significant unfolding
happens in the α domain while the β domain shows relatively reserved. Our further study
indicates that azoTABs disturb helix packing dramatically and residence time of
azoTABs toward lysozyme shows the behaviors of the azoTAB on different regions of
lysozyme.

90

3.2 Simulation Details
The Gromacs 3.3.3 package [18-22] is used to perform our simulation. The
GROMOS 45a3 force field [23] is chosen for protein lysozyme and azoTAB along with
the TIP3P [24-25] water model. The whole set of reparametrization to the
photoswitchable azo functional unit is from Böckmann’s previous work [26] and CH3 and
CH2 are typical ‘atom type’ in Gromos force field.

The starting lysozyme structure is the x-ray crystal structure (6LYZ.pdb [27]).
One lysozyme is first centered in the box of 6nm * 6nm * 6nm and 10 azoTABs (either
all trans or all cis structures) are randomly put around the lysozyme. If any azoTAB
molecule overlaps with other azoTAB molecules or lysozyme, it will be replaced by
another randomly distributed azoTAB. The experimental system [17, 28] under low
concentration (<12 mM) of azoTABs, the surfactant-protein complex has the
approximate 12 – 25 bounded azoTAB to one lysozyme [17]. And, the optimal molar
ratio of CTAB to protein was 10 to obtain good refolding yield and avoid the lysozyme
aggregates [29]. Hence, we enclose one lysozyme and 10 azoTAB in each simulation
box. Next, add solvent molecules as the distance between any atom of the solvent
molecule and any atom of the solute molecule(s) is greater than the sum of the Van der
Waals radii of both atoms or solvent molecules are removed from this box [18]. Eighteen
CL- counter ions are then added to neutralize the pH=7 solution. In total, 21 simulation
91

boxes are created and each is with 6nm*6nm*6nm dimension; TIP3P water model is
applied. All systems are listed in 0. Ten boxes / simulation runs are created with one
lysozyme and 10 trans azoTABs immersed in water (system: T1~T10) and another ten
boxes / simulation runs are created with one lysozyme and 10 cis azoTABs immersed in
water (system: C1~C10). In the beginning, the 10 azoTABs are randomly-distributed
around the lysozyme located in the center of the box. One additional simulation run, for
comparison, is made to mimic pure lysozyme solution (system: P) where one lysozyme is
put in the center of the box.
System Dimension
of box
No. of
water molecules
Azotab
Structure
No.
of azoTAB
Simulation time (ns)
P 6.16 6396 none 0 1014
T1 6.15 6318 all trans 10 1382
T2 6.15 6316 all trans 10 1231
T3 6.15 6303 all trans 10 1301
T4 6.15 6298 all trans 10 1284
T5 6.15 6325 all trans 10 1364
T6 6.15 6305 all trans 10 1247
T7 6.15 6299 all trans 10 1260
T8 6.14 6296 all trans 10 1298
T9 6.15 6300 all trans 10 1285
T10 6.15 6302 all trans 10 1334
C1 6.14 6296 all cis 10 1307
C2 6.14 6307 all cis 10 1316
C3 6.15 6313 all cis 10 1250
C4 6.15 6314 all cis 10 1334
C5 6.16 6306 all cis 10 1325
C6 6.14 6287 all cis 10 1315
C7 6.14 6308 all cis 10 1327
C8 6.15 6305 all cis 10 1233
C9 6.14 6278 all cis 10 1248
C10 6.14 6297 all cis 10 1208

Table 3.1: Summary of simulation systems. Dimension of box is average after
equilibrium.
The energy minimization procedures are performed in the beginning and the
steepest descent algorithm is adopted to minimize the energy of each system. Next, the
two-stage procedure is carried out to equilibrate each system. In the first stage, the
solvents are allowed to move freely while keeping the atoms of the proteins fixed to its
92

original positions; therefore, only the solvent molecules are equilibrated. In the second
stage, all atoms are free moving and the whole system is equilibrated for 5ns. Following
is the NPT (i.e. isothermal and isobaric) simulation at 300 K and 1 atm with time step 2
fs. LINCS algorithm [30] was employed to fix chemical bonds. For the long-range
electrostatic interactions, we use Reaction-Field (RF) method [31] (the cutoff is 14
angstroms) because the multimillion-atom simulations on the petascale supercomputers
exists good agreement for calculating the long range electrostatics in water with method
of RF and PME [32] but RF computation is not expensive. For the van der Waals
interaction, 14 angstroms cutoff is applied. The Brendsen coupling algorithm is used [33]
to maintain a constant temperature and pressure during simulations. Periodic boundary
conditions are employed throughout our simulations. We calculate the root-mean-square
deviation (RMSD), the radius of gyration (Rg), solvent-accessible surface area (SASA),
and counts of hydrogen bonds with Gromacs analysis programs.
In this work, we conduct massively parallel computing with 16 nodes for each
simulation systems on USC high performance computing center. As listed in 0, the
simulation time after equilibrium for each trans or cis system is about 1300ns (1.3µs) and
there is 260 ns (0.26 µs) simulation time for pure lysozyme in water system. Totally,
approximate 30 µs simulation was performed in our study. We have taken structural
analysis of intermediates per 0.5 ns, collecting for 25994 lysozyme intermediates induced
by 10 trans-azoTABs and 25746 lysozyme intermediates induced by10 cis-azoTABs
while 2028 lysozyme conformations in water without azoTABs are gathered. All
statistics of this report are based on these lysozyme intermediates.
93

3.3 Results and Discussion
3.3.1 Relative Unfolding Effects on the α / β Domains
Figure 3.2 is the time evolution of induced protein structures by the surfactants
over time for ten trans (a)~(e) systems (T1~T10) and ten cis (f)~(j) systems, printed in
blue dots while the time evolution of pure lysozyme in water (P) is shown in red dots.
Each dot on the plot represents lysozyme intermediates; the induced ones are in blue, the
pure lysozyme in water colored in red. There are 25994 blue dots and 524 red dots. We
define:
ΔRMSD = RMSD(α) – RMSD(β).
where RMSD(α) is the root mean square deviation of the α domain based on crystal
structure and RMSD(β) represents the β domain.
The increase difference radius of gyration of α and β domain is defined as
ΔRg (%) = ΔRg_α(%) – ΔRg_β(%).
ΔRg_y(%) = (Rg of the y domain of lysozyme intermediate – Rg of the y domain
of crystal structure) / (Rg of the y domain in crystal structure) *100%
94

Figure 3.2 : Time evolution of the relative unfolding effects of α / β domains (ΔRg (%)
and ΔRMSD(nm) ). Time evolution of the number of hydrogen bonds are displayed in (c)
and (h) and that of the total contacts are displayed (d) and (i) between α – β domain
interface. The α – β domain distance with time is shown on (e) and (j). Blue cross in
(a)~(e) stands for the 25994 intermediates induced by 10 trans-azoTABs. Blue cross in
(f)~(j) stands for the 25746 intermediates induced by 10 cis-azoTABs. The red cross
represents the 2028 lysozyme intermediates in water without azoTABs.
95

It has been considered that the β domain is more easily unfolded (unstable) than
the α domain [10] and easily perturbed by water molecules [34]. Our simulation, the pure
lysozyme in water (P system), has the corresponding results. Among the 524 lysozyme
conformations of pure lysozyme in water, many have shown both negative increase
difference of α-β domain in ΔRMSD(nm) and negative increase difference of α-β domain
in ΔRg(%), which means that RMSD (ΔRg) of β domain is larger than RMSD (ΔRg) of α
domain and therefore in our simulation, water molecules disturb the β domain more
easily than the α domain. Adding surfactants azoTABs in solution, the α domain is early
denatured more likely than the β domain due to the apparent positive difference in both
ΔRMSD(nm) and ΔRg(%) from most of induced intermediates by surfactantsm, azoTAB,
which is consistent with the experimental observation from SANS and FTIR [17].
Induced by the azo-TABs, the Rg increase of the α domain is up to 20% gain than the
increase of the β domain. The increase RMSD of the α domain could be 0.6nm gain than
that of the β domain. The time evolutionary of positive increase difference in ΔRg and
ΔRMSD indicates that azo-TABs unfold the α domain of lysozymes earlier than the β
domain, the span of the increase difference in ΔRg and ΔRMSD become larger, showing
the bigger positive increase difference in ΔRg and ΔRMSD; that is, more and more
partially unfolding α domain than the β domain by the surfactants, azo-TABs.
In Figure 3.2 (c), (h), the pure lysozyme in water has the number of hydrogen
bonds between the α-β domain interface is 7.9±2.4 while the induced lysozyme
intermediates by trans-azoTABs have 7.9±2.9 comparing 5 hydrogen bonds in crystal
structures. Some of the induced intermediates have more than 20 hydrogen bonds, some
96

have no hydrogen bond, and most of the rest have 8±3 hydrogen bonds between interface.
Figure 3.2 (d), (i) show the total contacts between the α-β domain interface. There are 16
contacts between α-β domain interface in crystal structure. For the pure lysozyme in
water, the interface contact number is 14.9±2.5 and for the intermediates induced by
trans-surfactants is 14.4±5.3. Therefore, one contact reduced in average after perturbed
by water with the range 8-24 contacts; however, trans-azoTABs make obvious contacts
difference in the domain interface; the maximum contact number can be up to 40 and the
minimum contacts can be less than 5. Figure 3.2 (e), (j) show the distance between the α-
β domains. The domain distance is defined as distance between two centers of mass of
α/β domains. It is 1.76 nm in the crystal structure and 1.57±0.08 nm of the pure lysozyme
in water but the induced structure has the domain distance 1.7±0.1. The domain distance
has slight increase made by addition of the trans-azoTABs. In overall view, azoTAB does
not have the dramatic effect on the interface of the two domains but convincing effects on
the α domain than the β domain.
ΔRg_α(%) versus ΔRg_β(%) is plotted in Figure 3.3. Almost all of the
intermediate conformations have shown ΔRg_α(%) > 0, which range over 0%~25%
induced by both trans-azoTABs and cis-azoTABs. On the other hand, more than 80% of
intermediates experience ΔRg_β(%) > 0. The increase percentage of β domain
(ΔRg_β(%)) has the range -5%~15% by trans-azoTABs, -4%~25% by cis-azoTABs.
Apparently, most of blue dots (lysozyme intermediates) are below the diagonal line (red);
that is, ΔRg_α(%) > ΔRg_β(%) and therefore the increase Rg of the α domain is more
than the increase Rg of the β domain corresponding to the experimental result, which
97

reveals that the α domain is more puffy than the β domain. For comparison, we conduct
the simulation which only a lysozyme molecule is immersed in the water bath and both of
its ΔRg_α(%) > 0 and ΔRg_β(%) > 0, shown as green dots in Figure 3.2. It is found that
green dots are along the diagonal line mostly but some are above the diagonal line
implying that the water molecules tend to disturb the β domain than the α domain. The
lysozyme molecule in water without azoTABs has the increase range of Rg for the α
domain is 0%~10%; for the β domain is 0%~14%.

Figure 3.3 : The α and β domains of lysozyme intermediates. Increase difference of
protein intermediates ΔRg (%) - (a),(b) in the α and β domain. RMSD_α (nm) and
RMSD_β (nm) are shown in (c) and (d). Blue cross in (a) and (b) stands for the 25994
intermediates induced by 10 trans-azoTABs. Blue cross in (c) and (d) stands for the
25746 intermediates induced by 10 cis-azoTABs. The green cross represents the 2028
lysozyme intermediates in water without azoTABs. Red lines are simply the diagonals
where ΔRg_ α (%) = ΔRg_ β (%) and RMSD_ α (nm) = RMSD_ β (nm).
RMSD of the α (β) domain ranges over 0.1~1.1 (0.1~0.7) for the intermediate
conformation induced by trans-azoTABs and ranges over 0.1~1.2 (0.1~0.9) induced by
cis-azoTABs. RMSD of these induced intermediates are recognized as blue dots and
98

more than 90% of blue dots are under the diagonal line, discovering that the α domain is
more disordered than the β domain found in the 90% of the induced intermediates. The
pure lysozyme in water has the RMSD of the α domain ranging 0.3~0.6 and RMSD of
the β domain ranging 0.2~0.65. Since the green dots are mainly along the diagonal line,
the disordering extent of the α domain and the β domain is almost the same, perturbed by
the water molecules.

Figure 3.4 : The normalized distributions are based on the 25994 intermediates induced
by 10 trans-azoTABs, the 25746 intermediates induced by 10 cis-azoTABs and the 2028
lysozyme intermediates in water without azoTABs. The black squares represent the
crystal structure. The curves in the right are showing the α domains and the curves in the
left are showing the β domains.
We consider that there exists native contacts between two different residues i, j if
the distance between the Cα atoms of the two residues, i, j, is within 0.6nm. In crystal
structure, there are 224 native/total contacts in the α domain and 125 native/total contacts
in the β domain. The pure lysozyme immersed in water has the consistent total contacts
(α domain:210±15 ; β domain:120±10 ) while existed native contacts are174±4 in the α
99

domain ; and 101±3 in the β domain. By adding trans-azoTABs, the normalized
distribution of total contacts(%) and native contacts(%) (shown in Figure 3.4 (a) and (b) )
in the α domain have a big shift to the left with 190±30 total contacts , 160±25 native
contacts. And, there are 115±15 total contacts and 90±6 native contacts in the β domain;
therefore, the normalized distribution of the contacts in the β domain shows some shift to
the left as compared with that of the pure lysozyme in water. Similar curve shift is found
from the lysozyme intermediates induced by cis-azoTABs. Obviously, the surfactants,
azoTABs (either trans or cis), reduce total contacts as well as native contacts in the α
domain while the total contacts of the β domain remains approximate the same but some
decrease of native contacts compared with the pure lysozyme in water. Number of
hydrogen bonds has good agreement with number of contacts within lysozyme. There are
49(35) hydrogen bonds in the α(β) domain of the crystal structure of lysozyme. For the
pure lysozyme immersed in water, there are 50±4 hydrogen bonds in the α domain and
31±3 hydrogen bonds in the β domain. In Figure 3.4 (c), with perturbing by the
surfactants (either trans or cis), the normalized distribution of the α domain shift to the
left apparently; however, a slight shift to the left for the β domain, concluding that the
surfactants make more reduce of the hydrogen bonds in the α domain than in the β
domain.
We calculate solvent accessible surface area (SASA) of lysozyme with 0.14 nm
radius of solvent probe. The solvent accessible surface area of lysozyme in crystal
structure (6LYZ) is 90.20 nm
2
(α: 59 nm
2
; β:32 nm
2
) and SASA of pure lysozyme in
water is around 90~95 nm
2
(α:61 nm
2
; β:35 nm
2
). SASA of the unfolding lysozyme
100

molecules induced by azoTAB, is around 120~140 nm
2
for all simulation cases (α:
60~100nm
2
; β:32~60 nm
2
). In the Figure 3.4 (d), the α domain-SASA of the induced
intermediate by trans-azoTABs could be up to 110 nm
2
and the β domain-SASA of the
induced intermediate by trans-azoTABs could be 65 nm
2
. Much more increase SASA in
the α domain than in the β domain because the normalized distribution of the α domain
moves to the right more clearly than the β domain. The trans-azoTABs have almost the
same effect on SASA as the cis-azoTABs.

3.3.2 Loose Helix Packing
The orientation of each helix vectors is defined and shown in the Figure 3.5. In
the lysozyme crystal structure (6LYZ.pdb), the angles between helix A and helix B (θ AB)
is 126.3˚, the angle between helix A and helix C (θ AC) is 74.7˚, the angle between helix A
and helix D (θAD) is 118.2˚, the angle between helix B and helix C (θBC) is 103.1˚, the
angle between helix B and helix D (θBD) is 112.2˚, the angle between helix C and helix D
(θCD) is 74.6˚. The pure lysozyme in water results in the distribution of helix orientations
(see Figure 3.5), in which θ AB = 128.0˚±6.5˚; θ AC = 71.3˚±8.6˚; θAD = 127.3˚±12.7˚; θBC =
118.2˚±6.2˚; θBD = 102.9˚±10.0˚; θCD = 77.3˚±7.5˚. The approximate 10˚ increase is
found from θ AD, θBC, and θBD, showing some arrangement between helices. With addition
of surfactants (either trans-azoTABs or cis-azoTABs), the distribution of each helix
angles become broad (see Figure 3.5), indicating that the increased helix motion results in
the extended distribution of helix angles upon the perturbation of surfactants. Therefore,
101

the protein molecule explore its conformation space by extensive helix motion within the
α domain due to existence of the surrounding surfactants. The analysis of helix angle
distribution provides unpacked structural information of the α-helix domain as well as the
relative position of helices. Among these surfactants-induced protein intermediates, about
~35% of which have their helix arrangements within the ranges of helix angles sets from
the pure lysozyme in water, appearing that the surfactants, azoTABs, have made the
obvious changes of helix packing on the lysozyme molecules. Secondary structure
analysis (Table 3.2) shows that the helix D lost most of its helix structures and this
corresponds to the large motion of helix D with large degrees of freedom, leading to the
wide ranges of helix angles, (see Figure 3.5 (b), (c), (f): θ AD = 40˚~180˚, θBD = 60˚~160˚
, θCD =40˚~150˚) , and therefore the helix angles of θ AD, θBD, θCD tend to be flat
distributed. The angle between helix A and helix B is most narrow distributed (θ AB =
90˚~150˚) and the angle between helix B and helix C is also narrow distributed (θBC =
70˚~140˚) possibly because that helix B is embedded inside the α domain, not only
keeping its helix structure intact (Table 3.2) but also having few degrees of freedom to
move by itself. Helix A and helix C are disturbed and unfolded by surfactants as shown
in the secondary structure analysis (Table 3.2) and so the distribution of helix angle (θ AC
= 40˚~120˚ mostly) of the induced intermediates is wider than that of the pure lysozyme
in water (θ AC = 50˚~90˚). Some intermediates induced by cis-azoTAB are showing either
very high θ AC = 120˚~140˚ or very low θ AC = 0˚~40˚ due to apparent secondary structure
loss of helix A and helix C.

102

Figure 3.5 : The normalized distributions are based on the 25994 intermediates induced
by 10 trans-azoTABs, the 25746 intermediates induced by 10 cis-azoTABs and the 2028
lysozyme intermediates in water without azoTABs. The black squares represent the
crystal structure. The curves in the right are showing the α domains and the curves in the
left are showing the β domains. The widely spread of the helix angle distribution implies
the large degrees of motions of the helices, making the loose packing of helices in the α
domain.

3.3.3 Analysis of Secondary Structures
The secondary structures are determined with the DSSP program [35]. The
average percentage of secondary structure calculation overall trajectories (13µs) of
10trans system (T1~T10) / (13µs) of cis systems (C1~C10) is shown in Table 2.3. One
intermediate conformation per 0.5 ns was taken to be averaged over the whole
trajectories. The secondary structure conformations are considered as helix, strand, β-turn
as well as coil and the percentage average of each structure was calculated over each
specific regions. For comparison, the average percentage of secondary structure
calculation was taken over the 1014 ns of the pure lysozyme in water (system: P). Most
103

helix D of lysozyme are lost in the simulation of both trans and cis systems. But, helices
A, C, and B have apparent helical structures even perturbed by the surfactants. Helix B is
embedded inside the α-domain core and therefore there is less chance for azoTAB to
reach it. The calculation of the average secondary structure reveals that helix B keeps its
high percentage of helix structure after long time perturbation of surfactants. Although
helix C is exposed on the surface of the lysozyme, both trans and cis azoTAB have little
interaction with it and so it is less perturbed and well maintains its helix structure. Helix
A is also on the surface of the lysozyme, but more helix loss of helix A than helix C. The
β domain is composed of three β-strands (β-sheet). The β sheet I is located on the front
position of the active cleft and thus a little few percentages of strand structure are lost;
nevertheless, β sheets II and III appeared more stable. With existence of trans azoTABs,
lysozyme molecules keep more helical structure in the α domain and more strand
structures in the β domain than the existence of cis azoTABs. On the other hand, in both
α and β domain, trans azoTAB induces more β-turn structures than cis while cis azoTAB
induces more coil structure than trans. During folding course, β-turns occur earlier than
the formation of helices and sheets [36], and thus the overall higher β-turns induced by
trans-azoTAB indicates the relative instability before trans azoTABs unfolds hen
lysozyme. Although cis unfold lysozyme more than trans based on our simulation results
(more coil found in the cis systems), the fact that more β-turns induced by trans could
foresee more extent of unfolding induced by trans with extension of simulation time.
Loop structure moves from strand to coil. 310 helix lost more helix structure and have
more coil structure perturbed by both trans and cis azoTABs. Overall, as a result of
104

interaction with azoTAB, lysozyme is swollen primarily in the α domain, while the β-
sheets remain relatively intact. Hamill et al. [17] conducted both SANS and FT-IR
experiment of lysozyme in the presence of <12 mM azoTAB upon exposure of either
visible (trans : cis = 75 : 25) or UV (trans : cis = 10 : 90) revealed the loss of the α-helix
and increase of the unordered structure, which could correspond to the coil structures as
the results from the simulation. However, the β structures maintain relatively stable and
unchanged. The helix-to-unordered transition indicates that under the very low
concentration of azoTAB, the very little amount of these photo-controlled surfactants can
swell the α domain of lysozyme while keeping β domain relatively intact. Our present
simulation work supports these phenomena observed from the experimental work.
Helix (%) Strand /β-Sheet (%) β-Turn (%) Coil (%)
Regions residue Pure Trans Cis Pure Trans Cis Pure Trans Cis Pure Trans Cis
HelixA (4-15) 84 75 78 2 8 3 6 6 6 8 11 10
HelixB (24-36 ） 65 67 61 18 13 16 11 10 7 6 10 17
HelixC (89-99 ） 93 85 84 3 1 3 1 8 6 2 6 6
HelixD (108-115) 64 33 27 5 19 14 12 23 15 18 25 44
Beta I (41-45) 0 0 0 64 55 59 9 5 3 26 40 38
Beta II (50-53) 0 0 0 75 76 75 2 3 0 23 22 25
Beta III (58-60) 0 1 6 67 64 55 30 27 26 2 8 13
Loop (61-78) 0 2 4 55 42 44 22 26 20 22 31 33
3-10 (80-84) 80 65 54 3 7 13 16 16 13 2 11 20
3-10 (120-124) 69 42 45 7 14 25 19 20 10 5 24 21

Table 3.2: Average percentages of secondary structures contents of lysozyme
intermediates overall simulation course, ~27 ns.

Figure 3.6 displays the time evolution of the protein secondary structure
calculated from the trajectories of the T5 system, in which one lysozyme and ten trans-
azoTAB are immersed in water over 1300 ns. Overall view, there are more disruptions on
105

the α domain (residues 1 to 35; 85 to 129) than on the β domain (residues 36 to 84) by the
end of the simulation time, 1300 ns. In the beginning, 310 helix (120-124) disappeared in
200 ns and followed by helix D, which unfolds and reforms within 1000 ns but totally
unfolds after that. Helix A is perturbed earlier and more than helix C for either trans or
cis than helix C. The helical structure of helix A is totally lost in 1100 ns but 50% of
helical structure of helix C remains at the 1300
th
ns. Helix B is quite intact during the
simulation course because it is situated in the core of the α domain and hence has low
chance toward either trans or cis azoTAB. It is found that β-sheet I (41-45) was gone
around the 400
th
ns in T5 system but still exists through the entire simulation course
under other simulation systems. β-sheet II (50-53) and β-sheet III (58-60) are relatively
stable. All other trajectories derived from either trans or cis systems share the similar
results of the secondary structure deformation as discussed above although the lyozyme
in each system behaves differently and unfolds at different timing.
106

Figure 3.6 : Time evolution of the protein secondary structure in T5 system with DSSP
program [35]. On the right hand side, we indicates location of helix A, helix B, helix C,
helix D, the two 310 helices and the β sheet.

Under high temperature (500K or 700K) simulation, the unfolding behavior of the
lysozyme has shown different events [11-12, 37]. The β-sheet starts to disappear earlier
than the α-helix. Helix C is lost rapid at 700K within 500 ps [12] while the significant
residual helicity in helix D and 310 helix (120-124) exist after that. Found in the
conformation of ~80% of the refolding molecules [38-39], the amide hydrogen atoms of
α domain are protected before those of β domain. Other 20%, the amide hydrogen atoms
in both domains were at the equally protected rate. The further experimental studies [40-
43] suggest that the significant hydrogen exchange protection for amides within helix A,
107

B and D in the absence of helix C. The hen lysozyme denatured in 8M urea has the
earlier unfolding of β domain than α domain, which is observed from both the
experiments [13, 44] and the computer simulations [14]. The helix C is first denatured by
urea along other helices and hence the ensemble of the denatured lysozyme molecules
only contains the native helical structures of helix A, B, D and 310; however, without
helix C [44]. Through the detailed non-native contact analysis [12], it is considered that
the contacts between α domain and β domain can stabilize the helix C; particularly, the
burial of hydrophobic groups (either main chain or side chain) are importantly involved
in this process. As shown in Figure 3.2, total contact counts as well as hydrogen bond
counts in the α-β domain interface does not decrease and the α-β domain distance does
not dramatically increase over all trajectories long in the systems T1 ~ T10 and C1~C10
as compared with the lysozyme structure in the water without surfactants, which could
explain that our helix C maintain most of its helix structure because of being protected by
the β domain.

3.3.4 Residence Time
We define that the distance between any atom of the protein and any atom of
surfactants is within 0.4 nm (reference) where the surfactant is close enough to the
protein, residing on the protein as the case. Since the surfactant could be near enough to a
residue both its main chain (including backbone atoms: Cα , N, C and carbonyl oxygens)
and its sidechain (all non-main chain atoms). The α domain contains the residues 1 to 35
108

and 85 to 129. The β domain has the residues 36 to 84. The residence time of the 200
surfactants (100 trans and 100 cis) on the protein are summarized from twenty systems,
see Table 3.3. For each system, the simulation duration is about 1300 ns and we took the
snapshots to capture the behavior of all the surfactants per 0.5 ns. Once any atom of the
surfactant and any atom of protein is within 0.4 nm in one snapshot, we count the
residence time for the surfactant is 0.5 ns. If the consecutive snapshot also concludes that
at least one atom of the surfactant is within 0.4 nm with the protein molecules, we rise the
residence time to 1 ns, and so on. The counting distribution of the residence time, ) (t C
y
,
based on the y domain or the whole protein at which the surfactants maintain t ns close to
the protein within 0.4 nm, and the counting is the sum up all the simulation work for 10
trans systems and 10 cis systems.
The average residence time (ns) of the surfactants on the y domain of the protein
molecules:




y
y
dt t C
dt t t C
t
) (
) (

: ) (t C The counting distribution of the residence time where 5 . 0  t .
The percentage of the residence time where the surfactants are on the y domain of the
protein is as follows:
109

Residence
 



systems ts surfac
y
time simulation total
dt t t C
10 tan 10
) (
(%) for separate trans or cis
For each system (either trans or cis), the simulation time is up to 1300ns, shown in
Table 3.3. Overall, we have ten trans systems and ten cis systems. There are ten enclosed
ten surfactants (azoTABs, either all trans or all cis) in each system.
time simulation total
systems ts surfac
 
10 tan 10
~ 130,000 ns for separate trans or cis

Lysozyme Trans Cis
Main Chain Side Chain Main Chain Side Chain
(A) (B) (C) (D) (a) (b) (c) (d)
α domain 38.4% 8.0 ns 35.2% 12.2 ns 37.1% 8.8 ns 33.2% 17.7 ns
β domain 22.0% 5.0 ns 24.2% 8.0 ns 23.5% 6.1 ns 25.4% 10.9 ns
Whole protein 46.0% 13.6 ns 39.0% 34.0 ns 46.4% 15.7 ns 39.5% 56.2 ns

Table 3.3 : (A), (C), (a), (c) are the Residence (%) of which the distance between any
atom on mainchain (sidechain) of the α domain/the β domain/the protein and any atom of
surfactants is within 0.4 nm. (B), (D), (b), (d) are average residence time (ns) while the
distance between any atom on mainchain (sidechain) of the α domain/the β domain/the
protein and any atom of surfactants is within 0.4 nm.

In Table 3.3, overall view, the Residence (%) of the surfactants on the whole
protein is less than 50% for all calculations (46.0%; 39.0%; 46.4%; 39.5%), showing that
the surfactants is far away from the protein 0.4 nm during the simulation courses. The
Residence (%) of surfactant towards the main chains of the protein molecules is larger
110

than the side chains of the protein (46.0%>39.0%; 46.4>39.5%) but the average time of
residing on side chains is longer than the main chains (13.6ns<34.0ns; 15.7ns<56.2ns),
bring the information that azoTABs could binding on the side chains (function groups) of
some specific residues for a bit long; on the other hand, there are 516 atoms constituting
the main chain but 805 atoms forming the side chains. Further instigation we have
discovered that the average residence time per residue is around 2.0 ns (either trans or
cis), much less than 13.6ns, 34.0ns, 15.7ns and 56.2ns, revealing that the surfactants
spend most of time searching on the surface of the protein.
The average time for trans-surfactants is less than the cis-surfactants. (8.0 ns < 8.8
ns; 5.0 ns < 6.1 ns; 12.2 ns < 17.7ns; 8.0ns<10.9ns) That corresponds to the higher
mobility of trans-surfactants than the cis-surfactants. [17, 26] Surfactants spend higher
Residence (%) of simulation time as well as the average residence time on the α domain
than the β domain. ( 38.4%>22.0%; 8.0ns>5.0ns; 35.2%>24.2%; 12.2ns >8.0ns;
37.1%>23.5%; 8.8ns>6.1ns; 33.2%>25.4%; 17.7ns>10.9ns ); this potentially results in
the puffy α domain as observed in the experimental work [17]. In the α domain,
surfactants spend more Residence (%) closer with any atom of the main chain of the α
domain than the side chain of the α domain (38.4%>35.2%; 37.1%>33.2%); however, the
average residence time on the main of the α domain is much less the side chain of the α
domain. (8.0ns<12.2ns; 8.8ns<17.7ns) This implies that the surfactants have higher
affinity on the side chains on the α domain than the main chains on the β domain. In the β
domain, surfactants spend more Residence (%) closer with any atom of the side chain of
the β domain than the main chain of the β domain(22.0%<24.2%; 23.5%<25.4%);
111

therefore, the average residence time is higher in the side chains of the β domain
(5.0ns<8.0ns; 6.1ns<10.9ns). Cis-azoTABs spend more Residence (%) on the β domain
(either main chains or side chains) than the Trans-azoTABs (22.0%<23.5% ;
24.2%<25.4%). Trans-azoTABs spend more Residence (%) on the α domain (either main
chains or side chains) than the Cis-azoTABs (38.4%>37.1%; 35.2%>33.2%). From Table
3.3, column (B); 8.0ns+5.0ns=13.0ns~13.6 ns (for main chain). For column (b),
8.8ns+6.1ns=14.9ns~15.7 ns (for main chain). For column (D),
12.2ns+8.0ns=20.2ns<34.0 ns (for side chain). For column (d),
17.7ns+10.9ns=28.6ns<56.2 ns (for side chain). Obviously, the surfactants are crossing
on the boundaries (cleft and hinge) of the α domain and the β domain. Due to the large
difference between 20.2ns and 34.0 ns (28.6ns and 56.2 ns), the surfactants resides on the
side chains of boundaries between the α domain and the β domain, implying that the side
chains of the residues on the cleft or hinge are not embedded so well after perturbing by
the surfactants, corresponding to the experimental work which concludes the hinge is
potentially puffy [17].

3.4 Conclusions
Our MD simulations illustrate that a little amount of surfactants, azoTAB (either
trans or cis), can unfold the lysozyme at room temperature, where the α domain and the β
domain unfolds concurrently though β-domain is showing relatively intact. This
corresponds to the results of the previous unfolding experiment of lysozyme under low
112

concentration of azoTAB (<12 mM) upon either visible light or UV light exposure and
the induced-intermediates, I αβ, represent the loss of helical structures of the swollen,
unfolding α-domain but the β domain adopts relatively few conformation changes. The
well-known I α intermediates display folded α domain without a structured β domain and
hence the surfactants-induced-intermediates, I αβ, are showing separate structures than the
I α intermediates. Along with the consistent domain distance, the increase number of total
contacts and hydrogen bonds between α-β domain interface during the simulation course
suggest the protected-intact structure of β domain as well as the stable helix C. It is also
found that the surfactants disorder the helix packing and make the greater increase of
solvent accessible surface area (SASA) but father decrease of contacts in the α domain
than the β domain. Our simulation work has revealed that the ability of azoTAB to unfold
the solid 3D structure of the α domain is making different residence time with main
chains and side chains of the α, β domains. Further, the unfolding pathway of lysozyme
where α and β domain unfold concurrently can be tracked through characterizing these
surfactant-unfolded intermediates.

113

3.5 References

1. Dobson, C.M., P.A. Evans, and S.E. Radford, Understanding how proteins fold -
the lysozyme story so far. Trends in Biochemical Sciences, 1994. 19(1): p. 31-37.

2. Humphrey, W., A. Dalke, and K. Schulten, VMD: Visual molecular dynamics.
Journal of Molecular Graphics, 1996. 14(1): p. 33-&.

3. Radford, S.E., C.M. Dobson, and P.A. Evans, The folding of hen lysozyme
involves partially structured intermediates and multiple pathways. Nature, 1992.
358(6384): p. 302-307.

4. Matagne, A., S.E. Radford, and C.M. Dobson, Fast and slow tracks in lysozyme
folding: Insight into the role of domains in the folding process. Journal of
Molecular Biology, 1997. 267(5): p. 1068-1074.

5. Dinner, A.R., et al., Understanding protein folding via free-energy surfaces from
theory and experiment. Trends in Biochemical Sciences, 2000. 25(7): p. 331-339.

6. Gilquin, B., C. Guilbert, and D. Perahia, Unfolding of hen egg lysozyme by
molecular dynamics simulations at 300K: Insight into the role of the interdomain
interface. Proteins-Structure Function and Genetics, 2000. 41(1): p. 58-74.

7. Mark, A.E. and W.F. Vangunsteren, Simulation of the thermal denaturation of
hen egg white lysozyme - Trapping the molten globule state. Biochemistry, 1992.
31(34): p. 7745-7748.

8. Hunenberger, P.H., A.E. Mark, and W.F. Vangunsteren, Computational
approaches to study protein unfolding - hen egg white lysozyme as a case study.
Proteins-Structure Function and Genetics, 1995. 21(3): p. 196-213.

9. Williams, M.A., J.M. Thornton, and J.M. Goodfellow, Modelling protein
unfolding: hen egg-white lysozyme. Protein Engineering, 1997. 10(8): p. 895-903.

10. Pepys, M.B., et al., Human lysozyme gene mutations cause hereditary systemic
amyloidosis. Nature, 1993. 362(6420): p. 553-557.

11. Kazmirski, S.L. and V. Daggett, Non-native interactions in protein folding
intermediates: Molecular dynamics simulations of hen lysozyme. Journal of
Molecular Biology, 1998. 284(3): p. 793-806.

114

12. Smith, L.J., et al., Molecular dynamics simulations of peptide fragments from hen
lysozyme: Insight into non-native protein conformations. Journal of Molecular
Biology, 1998. 280(4): p. 703-719.

13. Chen, L.L., K.O. Hodgson, and S. Doniach, A lysozyme folding intermediate
revealed by solution X-ray scattering. Journal of Molecular Biology, 1996.
261(5): p. 658-671.

14. Zhou, R.H., et al., Destruction of long-range interactions by a single mutation in
lysozyme. Proceedings of the National Academy of Sciences of the United States
of America, 2007. 104(14): p. 5824-5829.

15. Matthews, C.R., Pathways of protein folding. Annual Review of Biochemistry,
1993. 62: p. 653-683.

16. Dobson, C.M., Protein folding solid evidence for molten globules. Current
Biology, 1994. 4(7): p. 636-640.

17. Hamill, A.C., S.C. Wang, and C.T. Lee, Probing lysozyme conformation with
light reveals a new folding intermediate. Biochemistry, 2005. 44(46): p. 15139-
15149.

18. D. van der Spoel, E.L., B. Hess, A. R. van Buuren, E. Apol, P. J. Meulenhoff,,
A.L.T.M.S. D. P. Tieleman, K. A. Feenstra, R. van Drunen and H. J. C., and
Berendsen, Gromacs User Manual version 3.3, www.gromacs.org. 2005.

19. Bekker, H., Berendsen, H. J. C., Dijkstra, E. J., Achterop, S., van Drunen, R., van
der Spoel, D., Sijbers, A., Keegstra, H., Reitsma, B., Renardus, M. K. R.,
Gromacs: A parallel computer for molecular dynamics simulations. In Physics
Computing 92 (Singapore, 1993).de Groot, R. A., Nadrchal, J., eds. . World
Scientific.

20. Berendsen, H.J.C., D. Vanderspoel, and R. Vandrunen, Gromacs - A message
passing parallel molecular dynamics implementation. Computer Physics
Communications, 1995. 91(1-3): p. 43-56.

21. Lindahl, E., B. Hess, and D. van der Spoel, GROMACS 3.0: a package for
molecular simulation and trajectory analysis. Journal of Molecular Modeling,
2001. 7(8): p. 306-317.

22. Van der Spoel, D., et al., GROMACS: Fast, flexible, and free. Journal of
Computational Chemistry, 2005. 26(16): p. 1701-1718.

115

23. van Gunsteren, W.F., Billeter, S.R., Eising, A.A., Hunenberger, P.H., Kruger, P.,
Mark, A.E., Scott, W.R.P., Tironi, I.G., Biomolecular Simulation: The Gromos96
Manual and User Guide. Hochschulverlag AG an der ETH, Zurich., 1996.

24. Neria, E., S. Fischer, and M. Karplus, Simulation of activation free energies in
molecular systems. Journal of Chemical Physics, 1996. 105(5): p. 1902-1921.

25. Jorgensen, W.L., et al., Comparison of simple potential functions for simulating
liquid water. Journal of Chemical Physics, 1983. 79(2): p. 926-935.

26. Bockmann, M., et al., Atomistic force field for azobenzene compounds adapted for
QM/MM simulations with applications to liquids and liquid crystals. Journal of
Chemical Theory and Computation, 2007. 3(5): p. 1789-1802.

27. Diamond, R., Real-space refinement of structure of hen egg-white lysozyme.
Journal of Molecular Biology, 1974. 82(3): p. 371-&.

28. Wang, S.C. and C.T. Lee, Enhanced enzymatic activity through photoreversible
conformational changes. Biochemistry, 2007. 46(50): p. 14557-14566.

29. Lu, D.N., et al., Molecular simulation of surfactant-assisted protein refolding.
Journal of Chemical Physics, 2005. 122(13).

30. Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, W.F., DiNola, A., Haak, J.R.,
In: Pullman, B. (Ed.) Intermolecular Forces. Reidel Dordrecht, The
Netherlands, 1981.

31. Tironi, I.G., et al., A generalized reaction field method for molecular dynamics
simulations. Journal of Chemical Physics, 1995. 102(13): p. 5451-5459.

32. Schulz, R., et al., Scaling of Multimillion-Atom Biological Molecular Dynamics
Simulation on a Petascale Supercomputer. Journal of Chemical Theory and
Computation, 2009. 5(10): p. 2798-2808.

33. Berendsen, H.J.C., et al., Molecular dynamics with coupling to an external bath.
Journal of Chemical Physics, 1984. 81(8): p. 3684-3690.

34. Sinha, S.K. and S. Bandyopadhyay, Differential flexibility of the secondary
structures of lysozyme and the structure and ordering of surrounding water
molecules. Journal of Chemical Physics, 2011. 134(11).

35. Kabsch, W. and C. Sander, Dictionary of protein secondary structure - pattern -
recognition of hydrogen-bonded and geometrical features. Biopolymers, 1983.
22(12): p. 2577-2637.
116

36. Yang, A.S., B. Hitz, and B. Honig, Free energy determinants of secondary
structure formation .3. beta-Turns and their role in protein folding. Journal of
Molecular Biology, 1996. 259(4): p. 873-882.

37. Moraitakis, G. and J.M. Goodfellow, Simulations of human lysozyme: Probing the
conformations triggering amyloidosis. Biophysical Journal, 2003. 84(4): p. 2149-
2158.

38. Miranker, A., et al., Demonstration by NMR of folding domains in lysozyme.
Nature, 1991. 349(6310): p. 633-636.

39. Miranker, A., et al., Detection of transient protein folding populations by mass
spectrometry. Science, 1993. 262(5135): p. 896-900.

40. Hooke, S.D., S.E. Radford, and C.M. Dobson, Refolding of human lysozyme - a
comparison with the structurally homologous hen lysozyme. Biochemistry, 1994.
33(19): p. 5867-5876.

41. MorozovaRoche, L.A., et al., Structural characterisation and comparison of the
native and A-states of equine lysozyme. Journal of Molecular Biology, 1997.
268(5): p. 903-921.

42. Radford, S.E. and C.M. Dobson, Insights into protein-folding using physical
techniques - studies of lysozyme and alpha-lactalbumin. Philosophical
Transactions of the Royal Society of London Series B-Biological Sciences, 1995.
348(1323): p. 17-25.

43. Morozova, L.A., et al., Structural basis of the stability of a lysozyme molten
globule. Nature Structural Biology, 1995. 2(10): p. 871-875.

44. Schwalbe, H., et al., Structural and dynamical properties of a denatured protein.
Heteronuclear 3D NMR experiments and theoretical simulations of lysozyme in 8
M urea. Biochemistry, 1997. 36(29): p. 8977-8991.

117

Chapter 4: Ribonuclease A – structure and dynamics

Previous experiment has proven that photosensitive surfactants, azoTABs, can
stabilize partially unfolded intermediates of ribonuclease A (RNase A) with techniques of
small-angle neutron scattering (SANS) and test of FTIR. The surfactants have two
different structures, a trans planar form under visible and a cis bent form under UV light,
unfolding the proteins primarily in the α-helical region. In this study, we perform the 25
µs molecular dynamic simulations. Because each protein molecule has a certain number
of binding sites available to the surfactants and RNase A is a relative small protein, we
have the protein/surfactant ratio 1:10, enclosing one protein and ten surrounding
azoTABs (either all trans or all cis) in the simulation box. Twenty simulation boxes (ten
for trans, ten for cis) are created, each of which has different initial configurations. The
secondary structure analysis of proteins from simulations reveals that the α-helix unfolds
earlier than the β-sheet.

4.1 Introduction
Proteins are the three-dimensional folded structures, performing vast biological
functions within living organisms. The structures of proteins determine how the proteins
functions; therefore, proteins are not static and they constantly rearrange their
conformational structures during the course of functioning [1-2]. The environmental
perturbations (e.g. temperature, pressure and pH) can lead to changes of the protein
118

structures as well as its dynamics, further causing deviation of its functions. A wide
variety of techniques are introduced to detect protein dynamics including static
crystallography [3], neutron spectroscopy [2, 4], single-molecule measurements [5], and
computer simulations such as molecular dynamics simulations [6-8]. In the past decades,
molecular dynamics simulations has been largely used to probe the protein structures and
protein dynamics over a range of time scale (ns ~ µs) and length scale (nm), providing
atomic resolutions on the events of protein folding/unfolding.
Bovine pancreatic ribonuclease A (RNase A) is a small monomeric enzyme [9],
consisting of 124 amino acid residues and ~13.7 kDa molecular weight. The pancreatic
RNase A family catalyzes the hydrolysis of phosphodiester linkage of single-strand RNA
and further decomposes the RNA polymer chain. It is one of the proteins being
determined its amino acid sequence as well as its three dimensional structure and being
an important protein model in the 20
th
century [10-12]. The shape of the RNase A
molecule is like a kidney bean and divided into two halves with a deep cleft for binding
the RNA substrate. The first half consists of a predominantly N-terminal α helices
(residues 1-60; α1: residues 3-13; α2: residues 24-34; α3: residues 50-60) and the second
half consists of a predominantly C-terminal β sheet (residues 65-124; β1:61-74; β2:79-
104; β3:105-124). The secondary structure of RNase A is 23% α-helix, 46% β-sheet, 21%
β-turn and 10% random coil. RNase A molecule is tightly linked by four disulfide bonds
(Cys26-Cys84, Cys58-110, Cys40-95 and Cys65-72), which stabilize the protein 3-D
structure and significantly restraint conformational changes. The disulfide bond Cys26-
Cys84 is linking an α-helix (α2) and a β-sheet (β2) and Cys58-Cys110 is connecting α3-
119

β1. The other two disulfide bonds simply link the loop segments; either one can be
reduced without affecting the native structure under physiological conditions.

Figure 4.1 : The picture of Ribonuclease A molecule. Three helices are shown in purple
and from left to right, they are helix A, helix B, and helix C. The β sheets are printed in
yellow. The α-layer is in front but the β-layer is in the back. We keep the same
orientation of the protein in this chapter, including the protein shape reconstruction from
SANS data. VMD program is used for the drawing of the protein molecule [13].
RNase A is a basic protein (pI=9.63) and is positively charged at neutral pH. The
three important residues for catalysis are His12, His119 and Lys 41. The major
hydrophobic core is composed of α1, α3, and β3. The minor hydrophobic core contains
α2 and β2. Increasing temperature, α2 becomes destabilized rendering the loose of the
packing α1 in the major hydrophobic csUpon the existence of the denaturants, RNase A
unfolds with some increase of the gyration radius, keeping disulfide bonds intact;
therefore, the protein does not run into a random coil. The common used denaturant is
sodium dodecyl sulfate (SDS) [14-15].
120

In this decade, azobenzene-based surfactants “azoTABs” have been used to
reversibly photo-control protein structure and protein activity due to the light sensitive
properties of the surfactants. The azobenzene group of the surfactant undergoes a
photoisomerization from trans form upon visible light exposure (436 nm) to cis form
upon UV light exposure (365 nm). Experiment work has shown that azoTABs own
higher binding affinity on the α-helical segments of several proteins (lysozyme, BSA, and
α-chymotrypsin) than the β-sheets [16-19]. Under visible light, lysozyme is
swollen/unfolding near the hinge region (where the α domain and the β domain meets)
opposite to the active site. This allows for a flexible hinge, making enhanced domain
motion and superactivity of lysozyme (eight fold higher than the native state). Further,
the study of the molecular dynamics simulations has investigated the effects of azoTABs
on the induced structures of lysozyme, showing apparent swollen state on the α domain
of lysozyme and the loose type of hinge which corresponds to the experimental work.
The MD simulations offer details of an unfolding lysozyme induced by the azoTABs. It
is helix D to quickly unfold, followed by unfolding of helix A or helix C while the
embedded helix B is showing relatively intact. On the other hand, the β sheets have been
reserved mostly even though they are also perturbed by the surfactants.
In the present study, we conduct the µs molecular dynamics simulations on the
system of RNase A and azoTABs to better understand the internal dynamics, the
secondary structures, the tertiary structures of the induced protein intermediates. Every
simulation is based on one protein in the center of the simulation box as a start and ten
121

azoTABs (either all trans or all cis) randomly distributed around the protein, resulting in
RNase A molecules to partially unfold.

4.2 Simulation details
The starting protein structure is from the x-ray crystal (1RBX.pdb [20] ). The
protein is solvated in the center of the water box of size 6nm * 6nm * 6nm and 10 azotabs
(either all trans or all cis structures) are randomly distributed around the protein
molecule. There are about 6300 water molecules in the box. 14 (number) CL- counter
ions are added to neutralize the positively charged protein molecules and surfactants. The
GROMOS 45a3 force field [21] and TIP3P [22-23] water models are used for the
simulation. TIP3P represents transferable intermolecular potential, three position model.
The Gromacs 3.3.3 package [24-28] is employed to perform our simulation work. The
whole set of the photoswitchable azo functional unit is reparameterized from
Böckmann’s previous work [29] (see trans and cis structures in Figure 2.2) and CH3 and
CH2 are typical ‘atom type’ in Gromos force field.
The experimental work has reveals that protein molecules of ribonuclease A are
unfold even under the very low concentration of surfactants (<10 mM). According to the
previous SANS data, the surfactant-protein complex has the rough estimate of 15~25
binding azoTAB on one protein [30] derived from its effective molecular weight. Hence,
we enclose one ribonuclease A molecule and 10 azo-TAB in each simulation box to study
122

the unfolding event of ribonuclease A induced by the azo-TABs. All surfactants are
surrounding the protein in random as a start. Avoiding any overlaps between azoTABs or
azoTAB on the protein, a new azoTAB will replace any existed azoTAB which overlaps
with protein or other surfactants. While adding solvent molecules, the distance between
any atom of the solvent molecule and any atom of the solute molecule(s) should be
greater than the sum of the Van der Waals radii of both atoms (reference). In this
simulation study, we create 21 simulation boxes (6nm*6nm* 6nm, Table 4.1). Ten boxes
are for simulation runs (T1~T10, Table 4.1) of one initially centered protein surrounded
by ten randomly distributed trans azoTABs. On the other hand, another ten boxes are
simulation runs (C1~C10, Table 4.1) of one centered protein and ten cis azoTABs
randomly spread in the box, but not overlap the protein. Each simulation run has different
initial configurations, where either the ten trans-azoTABs or ten cis-azoTABs are present
in different positions as well as locations in the system at the beginning. For comparison,
pure protein solution was created (system: P, see Table 4.1), where one protein molecule
is centered in the box of the size 6nm*6nm*6nm and there are no surfactants around the
protein.
The energy minimization procedures are performed before running the MD
simulations and the steepest descent algorithm is applied to minimize the energy of each
system. Following is the two-stage procedure to equilibrate each system. In the first
stage, the solvents are allowed to move freely while keeping the atoms of the proteins
fixed to its original positions; therefore, only the solvent molecules are equilibrated. In
the second stage, all atoms are free moving and the whole system is equilibrated. Then,
123

we perform the NPT (i.e. isothermal and isobaric) simulation at 300 K and 1 atm with
time step 2 fs. LINCS algorithm [31] was employed to fix the all of chemical bonds.
Reaction-Field (RF) method [32] is used for the long-range electrostatic interactions as
the cutoff is 14 angstroms. Although Particle-Mesh Ewald (PME) is widely applied on
calculation of the electrostatic interactions in the infinite periodic system and PMD
provides more accurate results than RF, there is much more computational cost of PME
than RF. The multimillion-atom simulations on the petascale supercomputers exists good
agreement for calculating the long range electrostatics in water with method of RF and
PME [33]. In our 20 simulation systems, each represents separate RibonucleaseA–
azoTAB solution, and it should take long (high computational cost) to unfold protein
molecules under quite low concentration of surfactants at room temperature; therefore,
we have RF method to deal with electrostatic interactions; moreover, the chosen
GROMOS force fields are developed using RF. For the van der Waals interaction, 14
angstroms cutoff is applied. To maintain a constant temperature and pressure during
simulations, we use the Brendsen coupling algorithm [34]. Throughout our simulations,
periodic boundary conditions are employed. With Gromacs analysis programs, we
calculate the root-mean-square deviation (RMSD), the radius of gyration (Rg), solvent-
accessible surface area (SASA), and counts of hydrogen bonds.

124

Figure 4.2 : Molecular structures of the surfactants, azoTAB. Trans structure is on the left
and the right is the cis structure. As Gromos force field is applied, alkyl group CH3 and
CH2 are treated as typical atoms. The VMD program is adopted to demonstrate the
molecular structure of the surfactants [13].

4.3 Results and Discussion
4.3.1 Dynamic Affinity and Binding Sites
The interaction between proteins and small molecules / ligands is essential in drug
design [35-38]. Current methods to investigate protein-ligand interaction include
calorimetric techniques [39-42] and analysis of spectroscopic structure images [40, 43-
44], for example, SANS, NMR, AFM and Raman spectroscopy. The increased
development of supercomputers makes it possible to simulate the interaction of
macromolecules, such as proteins, and the ligands [45-49]. To predict protein–ligand
binding sites, docking, organic solvent mapping [50], and calculation of free energy of
binding [51] are three different algorithm. Docking allows low degrees of rotational
freedom of ligands and treats proteins as rigid bodies, only few conformation changes.
Organic solvent mapping considers small organic molecules as probes rolling over the
protein surface trying to fit the probes into the binding sites but it could result in false
125

prediction due to the simple assumption of sphere shapes of ligands. Moreover, the
binding free energy calculation (TI, FEP) is high computational cost [26]. Then, the
technique uses molecular dynamics simulations to determine the ligand binding sites
[52], which involves the atomic information and provide full degrees of freedom for not
only protein but also ligands. In this work, we also conducted the MD simulations to find
the binding sites of the Ribonuclease A for surfactants, azo-TABs, which are supposed to
collide more frequently with binding sites of proteins. Simulation is longer than 1 µs for
each of the protein-surfactants system and further recognized the binding sites of azo-
TABs around Ribonuclease A based on the dynamic affinity [52]. The more collisions
between proteins and surfactants, the higher dynamics affinity. The effective collision is
defined as the distance within 0.4 nm or less from any atom of the protein residue to any
atom of azo-TAB through the simulation trajectories. The definition of the dynamic
affinity is n n P
i i
/  [52], where
i
n indicates the number of collisions between the ith
residue and the surfactants, azoTABs. N M n /  is the average number of collisions for
a single residue,



N
i
i
n M
1
is the total number of collisions, N is the number of the
residues of the protein. There are 124 residues for Ribonuclease A ( 124  N ). The
dynamic binding affinity (
i
P ) is the average overall simulation paths of the ten
simulations for either all trans (~13 ns) or all cis (~13 ns) system (also summing up over
either 100 trans-azoTABs or 100 cis-azoTABs). The binding affinity is shown in (Figure
4.3; Table 4.2); red color represents high binding affinity (Pi > 1.0); blue represents low
binding affinity (Pi <0.7); and yellow is moderate affinity (0.7< Pi <1.0).
126

In Figure 4.3, it shows that both trans and cis azotab have high binding affinity on
helix C, and β3 but have low binding affinity on helix B and β2. Otherwise, helix A and
β1 have moderate binding affinity toward both trans and cis azotabs. Generally azoTABs
prefer to bind to hydrophobic site. Residue Lysine (LYL) carry positive charge of side
chains and are considered as non-hydrophobic residues. Most of LYL
(1,7,31,37,41,61,66,91,98,104) have the quite low dynamic affinity (<0.5) toward either
trans or cis azoTAB. Residue LYL61 results in the lowest binding affinity, 0.058 (trans)
and 0.038 (cis) among all the protein residues. For residues 74, 75, and 85, 88, 108 the
trans and cis azotabs have the very low binding affinity of less than 0.3. Residue 61, 74
and 75 are situated on the β1 and residue 85 and 88 are on the β2; residue 108 is on the
β3. Among the residues in β3, almost all the residues are owning polar side-chains,
rendering low binding affinity toward surfactants (trans or cis). On the other hand, in α3,
there are so many hydrophobic residues, such as Ala, Leu, Val, and Cys, involving high
interaction with surfactants (either trans or cis). There are three catalytically active
residues, His12, Lys 41 and His 119 in the Ribonuclease A. His 12 and His 119 residues
have mediate binding affinity toward surfactants but Lys 41 has low binding affinity.
Moreover, either trans or cis azotabs have frequent interaction with the residues, Phe8,
Glu9, Met13, Asp14, Ser15, His48, Glu49, Leu51, Glu55, Ser80,Glu101, Val116 (Table
4.2). In Figure 4.3, dynamic affinity >0 for all residues, that implies that azotabs can
penetrate into the ribonuclease A and resides in grooves and cavities of molecule of
ribonuclease A. Of the 20 natural amino acids, Ribonuclease A possesses 19 of them,
excluding tryptophan (Trp). The phenol rings of azoTABs have high-frequent interaction
127

with Trp groups of lysozyme molecules. In Figure 4.3, the bottom part of the protein
molecule (residues 14 ~ 23 and residues 45 ~50) is colored in red (high binding affinity)
for both trans and cis azotabs but the helix B (on the left side) is colored in blue (low
binding affinity). The high binding affinity in the bottom of the protein, where is close to
the catalytically active sites, might explain that azotabs (both trans and cis) increase the
activity of ribonuclease A [53-55]. The structural difference between trans and cis affects
the significantly different binding affinity toward some residues (Ser15, Ser21, Leu51,
Ala52, Gln55, Ala56, Cys58, Tyr73, Pro114, Tyr115, Val118).
Three of the enzymatic subsites (B1, B2, and B3) [56-58] interact with the bases
of a bound substrate. B1 (Thr45, Asp83, Phe120, Ser123), B2 (Asn71, Gln69, Glu111),
B3(Lys1). The B1 subsite appears to bind only pyrimidine bases. In contrast, the B2 and
B3 subsites bind all bases, but B2 has a preference for an adenine base and B3 has a
preference for a purine base. Three other enzymic subsites (P0, P1, and P2) interact with
the phosphoryl groups of a bound substrate. P0(Lys66), P1(active sites: Gln11, His12,
Lys41, His119, Asp121), P2(Lys7, Arg10). The main interaction with RNA and DNA
occurs through three phosphate binding sites, P0, P1 and P2, located in the active binding
pocket of RNAse [57-58].The main phosphate binding site P1 is the active site while P0
and P2 are considered non-catalytic. Overall, the surfactants, azoTABs, do not
specifically bind on the active sites of the Ribonuclease A molecules, also, the low
binding affinity toward the subsites (P0, P1, P2, B1, B3). The binding of the surfactants is
mainly the adsorption effects toward non-active sites of the protein, and therefore, it is
the unspecific binding between the surfactants and Ribonuclease A. [59-62].
128

System Dimension
of box
No. of
water molecules
Azotab
Structure
No. of
azoTAB
Simulation time (ns)
P 6.15 6531 none 0 425
T1 6.14 6317 all trans 10 1494
T2 6.15 6324 all trans 10 1481
T3 6.14 6312 all trans 10 1351
T4 6.14 6313 all trans 10 1469
T5 6.14 6322 all trans 10 1467
T6 6.15 6329 all trans 10 1397
T7 6.14 6315 all trans 10 1367
T8 6.14 6316 all trans 10 1474
T9 6.14 6319 all trans 10 1382
T10 6.14 6316 all trans 10 1372
C1 6.14 6325 all cis 10 1424
C2 6.14 6313 all cis 10 1424
C3 6.14 6318 all cis 10 1468
C4 6.14 6325 all cis 10 1483
C5 6.15 6328 all cis 10 1370
C6 6.14 6309 all cis 10 1347
C7 6.14 6315 all cis 10 1461
C8 6.14 6314 all cis 10 1460
C9 6.14 6307 all cis 10 1446
C10 6.15 6334 all cis 10 1380

Table 4.1: Summary of 21 simulation system for RNase; ten trans-azoTABs system
(T1~T10); ten cis-azoTABs (C1~C10) system; one pure protein in water (P). Dimension
of box is average after equilibrium.

Figure 4.3 : Average Dynamic Affinity of the surfactants, azoTABs, on the protein
Ribonuclease A all over ten simulation system across whole simulation duration. The
129

black line stands for the the dynamic affinity of trans-azoTABs toward RNase while the
red line is the dynamic affinity for cis-azoTABs on RNase. Distribution of the secondary
structures of crystal RNase is along the axis of the residue number. The moderate
dynamic affinity is defined between 0.7~1.0, in where two horizontal dashed lines are
shown in light blue.
Res. No. Trans Cis Res. No. Trans Cis Res. No. Trans Cis
1 0.43 0.51 43 0.72 0.79 85 0.24 0.23
2 1.13 1.28 44 0.84 0.67 86 0.25 0.40
3 1.12 1.49 45 0.48 1.07 87 0.34 0.28
4 1.30 1.52 46 0.46 0.95 88 0.17 0.14
5 0.93 1.15 47 0.99 1.32 89 0.30 0.24
6 0.70 1.17 48 3.33 3.11 90 0.33 0.35
7 0.67 0.59 49 3.25 2.96 91 0.34 0.40
8 2.89 2.62 50 2.38 1.84 92 0.47 0.45
9 2.44 2.61 51 3.70 2.81 93 0.49 0.55
10 0.63 0.90 52 1.66 0.51 94 0.62 0.68
11 0.65 0.72 53 1.06 0.66 95 0.66 0.83
12 0.93 1.42 54 1.09 0.77 96 0.33 0.56
13 3.21 2.85 55 3.57 2.19 97 0.61 0.84
14 2.53 2.30 56 1.41 0.65 98 0.39 0.34
15 3.11 2.14 57 0.37 0.29 99 0.89 0.98
16 1.68 1.41 58 1.55 0.81 100 0.44 0.43
17 1.53 1.97 59 1.30 0.78 101 2.27 2.49
18 1.53 2.10 60 0.52 0.17 102 0.83 0.97
19 1.42 1.87 61 0.06 0.04 103 1.35 1.58
20 0.98 1.14 62 0.76 0.75 104 0.18 0.38
21 0.66 1.12 63 0.56 0.60 105 0.10 0.42
22 0.71 0.70 64 0.47 0.69 106 0.18 0.36
23 0.62 0.79 65 0.63 0.90 107 0.33 0.54
24 0.53 0.73 66 0.54 0.77 108 0.24 0.12
25 1.17 1.30 67 1.34 1.16 109 0.43 0.50
26 0.20 0.38 68 1.28 1.31 110 0.85 0.73
27 0.68 0.63 69 1.76 2.01 111 1.77 2.02
28 0.24 0.39 70 0.89 1.02 112 1.36 1.28
29 0.25 0.69 71 0.95 0.90 113 1.84 1.38
30 0.69 0.66 72 0.66 0.52 114 2.61 1.74
31 0.41 0.44 73 1.52 0.90 115 3.31 1.51
32 0.35 0.38 74 0.30 0.25 116 2.99 2.74
33 0.92 0.68 75 0.15 0.07 117 1.68 1.77
34 0.62 0.63 76 0.52 0.18 118 1.22 2.40
35 0.57 0.89 77 0.52 0.49 119 0.83 1.35
36 0.51 0.75 78 0.83 0.95 120 0.88 0.94
37 0.54 0.60 79 0.60 0.64 121 0.92 0.69
38 0.53 0.69 80 2.61 2.31 122 0.40 0.83
39 0.44 0.44 81 0.80 1.18 123 0.62 0.55
40 0.52 0.57 82 0.97 1.37 124 0.34 0.70
41 0.36 0.41 83 0.32 0.47
42 0.36 0.51 84 0.18 0.43

Table 4.2 : List of the average dynamic affinity, in which trans or cis azoTABs interact
with the protein molecules, RNase, at different extent.
130

(a)

(b)

Figure 4.4 : (a) Binding Affinity of trans-azoTAB toward Ribonuclease A. (b) Binding
Affinity of cis-azoTAB toward Ribonuclease A. Residues with high binding affinity, Pi >
1.0, are colored in red; low binding affinity, Pi <0.7 in blue; and moderate binding
affinity, 0.7< Pi <1.0, in yellow. All plots are made with VMD Program [13].

4.3.2 Analysis of Secondary Structures
Helix (%) Strand /β-Sheet (%) β-Turn (%) Coil (%)
Regions residue Pure Trans Cis Pure Trans Cis Pure Trans Cis Pure Trans Cis
HelixA (3-13) 64 26 35 14 28 22 11 20 13 11 26 30
HelixB (24-34 ） 85 75 48 9 3 18 2 9 8 4 13 26
HelixC (50-60 ） 71 45 62 8 10 7 6 12 10 15 33 21
β I (61-74) 0 0 0 85 74 77 0 1 0 15 25 23
β II (79-104) 0 0 0 81 69 76 3 5 4 16 26 20
β III (105-124) 0 0 0 78 57 60 2 11 12 20 32 28

Table 4.3 : Average percentages of secondary structures contents of Ribonuclease A
intermediates induced by the surfactants during all of simulation courses collecting from
every simulation system.
131

We determined the secondary structures with the DSSP program [63]. The
average percentage of secondary structure is calculated overall the simulations of 10trans
(system:T1~T10) /cis systems (system: C1~C10), as shown in Table 4.3. Every 0.5 ns,
one snapshot was taken over full trajectories. The secondary structure conformations are
defined as helix, strand, β-turn as well as coil and for each specific regions, the
percentage average of each secondary structure conformation was calculated. Moreover,
we try to understand the pure water effects on the Ribonuclease A and therefore, the
simulation of the pure Ribonuclease A in water (system: P) without any of surfactants,
azoTABs is done for comparison. The duration of this simulation is around 425 ns and
the calculation of the percentage of secondary structure was taken average overall the
snapshots per 0.5 ns of the Ribonuclease A molecules perturbed by water molecules.
With either trans or cis azo-TABs, most helix A lost most of its helical structures and
gain ~30% coil structures as well as ~25% of β-sheet corresponding to the observation of
the high binding affinity of both trans and cis azotab toward helix A. On the other hand,
helices B has low binding affinity toward trans and cis structures of surfactants and
therefore there reserves lots of helical structures in the regions of helix B. Trans-azoTAB
has moderate interaction with helix C while cis-azoTAB has low bind affinity on helix C.
There are 45% of helical structures of helix C left after perturbing by trans-azoTAB and
62% of helical structures left of helix C due to cis-azoTABs. Both lost of helical
structures of helix B region and helix C region lead to coil structures (~25%). The
secondary structural analysis reveals that the β-1 has kept best of its β-sheet structures
(>70%), resulting from low binding affinity towards surfactants (either trans or cis), as
132

shown in Figure 4.3. The β-2 region also has low interaction frequency with both trans
and cis azo-TABs and so its β-sheet structures are mostly intact (~70%). Although Figure
4.3 shows that high binding affinity towards azo-TABs in the β-3 region, there are still
~60% of β-sheet structures left. For the β-1, the β-2, and the β-3 regions, both trans and
cis surfactants induce ~25% coil structures but 0% α-helix structures. The trans-azoTABs
perturbs the β-sheet structures more than the cis-azoTABs do, rendering more loss of the
β-sheet but more gain coil structures with surrounding trans-azoTABs. In Table 4.2,
trans-azoTABs induce more β-turns than cis-azoTABs and with extension of simulation
time, we foresee more extent of unfolding induced by trans-azoTABs as shown in [30,
64]. Overall view, the average percentages of secondary structures contents of
Ribonuclease A intermediates during full simulation courses (Table 4.2) reveal that the
surfactants, azo-TABs, mainly interact with the α-helices of molecules of Ribonuclease
resulting in the obvious unfolding of the α-helices while the β-sheets remain relatively
intact. The experimental work based on SANS and FT-IR [30, 64] discovered that in the
presence of <8 mM azoTAB upon exposure of either visible (trans : cis = 75 : 25) or UV
(trans : cis = 10 : 90), molecules of Ribonuclease A lost the α-helix but gain the
unordered structure, defined as the coil structures by DSSP [63], the same results
obtained from the simulation. On the other hand, the β structures show more stable than
the α-helices. The experimental work has indicated that under the very low concentration
of azo-TAB, the protein structures have encountered the helix-to-unordered transition as
seen in the simulation work, the very little amount of these photo-controlled surfactants
swell the α helices of Ribonuclease A while the β sheets are quite intact. Our current
133

simulation work corresponds to the experimental work, both of which support these
phenomena of the unfolding α-helices but the relatively reserved β-sheets.

Figure 4.5 : Time evolution of the protein secondary structure in C6 system with DSSP
program [63]. On the right hand side, we indicates location of helix A, helix B, helix C,
and the βI, βII, βIII sheets.

Figure 4.5 displays the time evolution of the protein secondary structure
calculated from the trajectories of the C6 system, in which one Ribonuclease A and ten
cis-azo-TAB are immersed in water over 1300 ns. Overall view, there are more
disruptions on the α helices (residues 3~13; 24~34; 50~60) than on the β sheets (residues
134

61~74; 79~104; 105~124) by the end of the simulation time, 1300 ns. In the beginning,
helix A (3-13) disappeared in 100 ns but after 1150ns, residues 3-8 reform helix
structures. Helix B (24-34) showed 40% fast-loss of helix structures (22-26) in the very
early simulation time (0~150ns); however, after 150ns, residues (22-26) reforms helix
structures but residues (27-34) totally lost its helix structures which are replaced by
strand / β sheets. In this simulation run (C6), helix C maintains almost intact during the
simulation time (1350ns). The average secondary structures of all trans systems shows
45% helical structures left of helix C and 62% left for all cis systems. At the end of
simulation time, 1350ns, helix B lost its total helical structures, which convert to the β
sheets instead. Helix A is perturbed earlier and more than helix C due to higher binding
affinity of azo-TAB for either trans or cis than helix C (Figure 4.3). The helical structure
of helix A is almost totally lost but more than 60% of helical structure of helix C remains
at the end of simulation course. Due to the average low binding affinity (Figure 4.3)
toward either trans or cis azoTABs, the average secondary structure of helix B reserve the
most helical structures than helix A and helix C (Table 4.3). The average reserved β-
sheets is up to 60% to 80% and β-sheet III lost most because of high binding affinity
which surfactants interact frequently on the protein. In C6 system, β-sheet II (79-104) and
β-sheet III (105-124) contains relatively low strand / β sheet structures corresponding to
the results of the average secondary structure analysis. In all other simulation systems,
the protein is disturbed variously by the surfactants, azoTABs from time to time and
hence the Ribonuclease A in each system does not share the same unfolding scenarios
135

within several microseconds but the secondary structure deformations share the similar
results derived from the trajectories data of the all atoms of the proteins.
4.3.3 Comparison with Experimental Results
(a) RNase : Crystal Structure

(b) pure RNase A in water

(c) C2 after 1300ns

(d)8.3 mM visible light (75%
trans-azoTAB)

Figure 4.6 : RNase: SANS images compared with P and C2 simulation results. (a) Crystal
Structure (1RBX) (b) SANS image of pure protein (RNase) in water. (c) Simulation
snapshot of protein in C2 system at 1300ns. (d) SANS image of protein (RNase) in 8.3
mM azo-TAB solution under visible light. Plots of simulation snapshot are obtained by
VMD program [13].
Ribonuclease molecules have been well-investigated with many different
techniques, not only by computational simulations [65-68] but also by the high-tech
experimental work, such as SANS, Small Angle Neutron Scattering, and FTIR, Fourier
Transform Infrared Spectroscopy. During 2006~2010, Hamill [30], Wang [69], and
Miraref [64] have continuing work on the unfolding event of Ribonuclease A with SANS,
resolving its tertiary structure of the protein, and FTIR, analyzing its secondary structure.
They discovered that the photo-sensitive surfactants, azoTABs, can manipulate the
folding/unfolding of the protein molecule, RNase, through the control under either UV or
136

Vis exposure. Their results of SANS images and FTIR data support the conclusion that
only the surfactant concentration renders the structural change on the RNase molecules;
however, increasing the concentration of the protein, RNase, does not have the direct
effects on the protein structures, which implies the limited binding sites of azoTABs on
the RNase moelcules.
Figure 4.6 are the images taken from the shape reconstruction of the SANS data.
The conformation of the pure RNase molecule in water is displayed in Figure 4.6 (b),
which is consistent with the X-ray crystal structure (PDB: 1RBX) in Figure 4.6 (a).
Therefore, we can trust the results from the SANS method. Figure 4.6 (d) shows the
SANS images which were from the experiment work [30, 64] where the RNase
molecules are immersed in the very low concentration of the surfactants (~8mM; 75%
trans-form, 25% cis-form) under the exposure of visual light. Figure 4.6 (d) indicates the
puffy parts located in the bottom of the protein, opposite from the active site. We cannot
exactly point out which series of residues unfold from the shape reconstruction images of
the SANS data but we proposed that the unfolding part of the protein are the helices
structures based on its opposite position from the active site. Also, from the previous
work [17], the hydrophobic regions of the surfactants play an main role to interact with
the α-helices of the protein and further unfold the α-helices.
137

Figure 4.7 : Time evolution of RMSD in T8 system. The three curves specifically
represent RMSD of whole protein (blue line), RMSD of α-layer (red line) and RMSD of
β-layer (green line). The RMSD of α-layer (red line) is leading among the three and the
RMSD of β-layer is under the other two curves.

Figure 4.8 : Time evolution of radius of gyration in T8 system. We have calculated the
gyration radius for the α-layer (red line) and β-layer (green line) and the whole protein
(blue line). Due to the small portion of the α-layer within the protein, the gyration radius
138

of α-layer (red line) is lower than the other two curves but has the most increase than the
other two.

Figure 4.9 : Time evolution of number of contacts within the protein in T8 system. The
plot is showing total contacts (blue curve), native contacts (red curve) and non-native
contacts (green curve). Total contacts remain almost ~300 because the decrease of the
native contacts (~200) is compensated by the increase of non-native contacts (~100).
After 800ns, the number of native contacts or non-native contacts persists in a range of
constants.
139

Figure 4.10 : Time evolution of number of native contacts within the protein in T8
system. The number of the total native contacts within the crystal structure of RNase
molecule are ~300; however, it decreases to ~200 after 800ns (blue curve), in which the
native contacts of the α-layer drops down more than that of the β-layer.

Figure 4.11 : Time evolution of distance between the two layers, α layer and β layer in T8
system. The distance decreases over most of time although it increase during 200~300 ns.

140

Figure 4.12 : Time evolution of the number of the hydrogen bonds within in the protein
(blue line), the α layer (red line), the β layer (green line) and the interface between the
two layer (purple line) in T8 system. Obviously, the number of the hydrogen bonds in the
β layer increases but that in the α-β layer interface decreases.

Figure 4.13 : The time evolution of solvent accessible surface area (SASA) in T8 system.
The blue line represents the SASA calculation of the protein and the purple line
represents the interface between the protein and 10 surfactants.
141

During the simulation trajectories, we collect all the coordinates of the
surfactants-induced intermediates and calculate the structural properties of these induced
protein intermediates, including root mean square deviation (RMSD), radius of gyration
(Rg), α-β layer distance, native contacts, total contacts, number of hydrogen bonds, and
solvent accessible surface area (SASA). Figure 4.7~Figure 4.13show the time evolution
of the structural properties of RNase molecules in the case of simulation, T8. Table 4.4
and Table 4.5 are the lists of the average structural properties of the RNase molecules in
all the simulation systems (P, T1~T10 and C1~C10) during the simulation time 1200 ns ~
1300 ns. The RMSD is the indication of similarity of the three-dimensional structure of
the induced intermediates by surfactants with the crystal structures. The higher the
RMSD, the less similarity of the induced intermediates with the crystal structures (RMSD
= 0). The RNase A molecule alone in water is perturbed by water and encountered slight
change of its 3D structure with a bit larger than zero RMSD value. While immersed in
the surfactant solution, RNase A molecules are disturbed by water molecules as well as
surfactants, making the RMSD values increase with time (Figure 4.7). The RMSD of the
α layer is a bigger number than the RMSD of the β layer with the existence of azoTABs,
either trans or cis, and the RMSD of the α layer is even larger with longer simulation time
due to more perturbation by azoTABs. The size of the protein can be measured by its
radius of gyration. For the crystal structure of RNase A molecule, the radius of gyration
is 1.43 nm (α layer:1.22nm; β layer:1.47nm). The radius of the gyration is increased by
interaction with surfactants, giving the enlarging size of both the α layer and the β layer,
but the α layer has the more increase than the β layer, corresponding to more increase of
142

RMSD of the α layer. In the T8 system, both RMSD and Rg keep increasing from
0ns~650ns as both the α layer and the β layer swell (Figure 4.7 and Figure 4.8). After 650
ns, the size of the RNase A molecule maintains approximately the same with steady Rg
of the α layer and the β layer and the stable structure of RNase A shows constant RMSD
values. Figure 4.11 displays the α-β layer distance, which is defined as the distance
between the center mass of the α layer and the center mass of the β layer, and the distance
decrease with time. However, the α-β layer distance appears increase in some other
simulations. We define that the contact exists if any two Cα atoms are within 0.6 nm
distance. The native contacts are the ones also found in the crystal structure of the protein
but the non-native contacts are not found in the crystal structure. The total contacts
between atom to atom within the protein molecules remains almost the same while the
native contacts of the protein decrease by ~30% accompanied with the increase of the
non-native contacts, as seen in Figure 4.9. Also, the α layer encounters more loss of its
native contacts than the β layer (Figure 4.10). The overall number of hydrogen bonds of
the whole protein decreases (Figure 4.12) and so the protein lost its secondary structures;
especially, the obvious loss of the hydrogen bonds is between the interface of the two
layers while the number of the hydrogen bonds in the β layer increases but some
decreases of the hydrogen bonds in the α layer. The time evolution of the solvent
accessible surface area of the RNase molecule appears 90~160 nm
2
(Figure 4.13) and the
interface between the protein and the ten surfactants is around 10 nm
2
. Therefore, the
average of the protein’s SASA is around 125 nm
2
with fluctuation / amplitude ~30 nm
2

143

since the ten azoTABs keep interacting with the RNase molecule and make the up and
down change of the SASA of the RNase molecule.

RMSD-
protein
(nm)

RMSD- α
layer
(nm)
RMSD-β
layer
(nm)
Radius of
gyration -
protein
(nm)
Radius of
gyration -
α layer
(nm)
Radius of
gyration -
β layer
(nm)
α-β layer
distance
(nm)

Protein –
azotabs-
interface
(nm
2
)
SASA-
hydrophobi
c
(nm
2
)

SASA-
hydrophilic
(nm
2
)

Crystal 0 0 0 1.43 1.22 1.47 0.93 None 37.11 30.87
P 0.28±0.03 0.20±0.02 0.25±0.02 1.45±0.03 1.25±0.03 1.50±0.03 0.95±0.03 None 45.55±4.21 39.60±3.22
T1 0.83±0.05 0.69±0.04 0.47±0.03 1.53±0.04 1.37±0.03 1.49±0.02 1.02±0.03 11.01±1.02 69.21±7.12 50.54±5.47
T2 0.62±0.03 0.59±0.02 0.40±0.02 1.50±0.02 1.34±0.02 1.54±0.03 0.80±0.04 13.76±0.97 69.00±7.42 52.60±6.21
T3 0.65±0.04 0.69±0.05 0.45±0.02 1.52±0.02 1.30±0.04 1.55±0.02 1.00±0.04 16.97±1.24 71.48±9.18 55.53±7.63
T4 0.66±0.06 0.70±0.04 0.50±0.08 1.52±0.04 1.38±0.03 1.54±0.06 0.83±0.05 13.40±1.33 70.58±8.25 55.12±6.91
T5 0.58±0.02 0.55±0.02 0.50±0.03 1.56±0.02 1.37±0.03 1.61±0.03 0.84±0.03 10.03±1.17 65.06±6.53 46.36±5.60
T6 0.58±0.02 0.61±0.02 0.49±0.01 1.54±0.02 1.37±0.02 1.45±0.02 1.21±0.03 15.42±1.13 68.58±7.21 52.01±5.82
T7 0.48±0.02 0.53±0.02 0.31±0.01 1.50±0.02 1.36±0.02 1.51±0.02 0.86±0.04 14.78±1.06 65.75±8.68 50.18±6.95
T8 0.92±0.02 0.99±0.02 0.62±0.04 1.53±0.03 1.40±0.02 1.56±0.04 0.70±0.04 11.43±1.05 70.27±7.54 51.45±6.30
T9 0.57±0.03 0.63±0.05 0.44±0.02 1.46±0.02 1.34±0.02 1.40±0.03 1.02±0.05 11.80±1.12 66.21±8.25 51.03±6.41
T10 0.53±0.01 0.64±0.01 0.36±0.02 1.46±0.02 1.30±0.02 1.44±0.02 1.01±0.02 15.38±1.08 66.23±7.26 49.19±6.07
C1 0.81±0.02 0.94±0.02 0.44±0.03 1.58±0.02 1.55±0.02 1.53±0.02 0.76±0.04 13.30±1.08 70.61±7.60 53.76±5.90
C2 0.64±0.02 0.71±0.03 0.46±0.02 1.50±0.02 1.35±0.02 1.50±0.02 0.90±0.03 20.88±1.23 72.02±9.12 54.58±7.61
C3 0.60±0.03 0.63±0.02 0.39±0.02 1.53±0.03 1.36±0.03 1.56±0.03 0.88±0.03 11.94±1.34 66.50±7.79 50.20±6.54
C4 0.72±0.03 0.90±0.04 0.40±0.02 1.56±0.02 1.42±0.02 1.54±0.02 0.95±0.03 18.67±1.16 73.42±6.88 56.06±5.67
C5 0.72±0.03 0.75±0.03 0.53±0.03 1.60±0.02 1.44±0.03 1.61±0.02 0.97±0.04 17.53±1.33 73.04±9.23 55.94±7.38
C6 0.64±0.03 0.69±0.03 0.44±0.04 1.58±0.02 1.34±0.02 1.62±0.03 1.06±0.05 15.61±1.18 68.71±7.76 51.54±6.39
C7 0.50±0.02 0.59±0.03 0.34±0.02 1.49±0.03 1.33±0.02 1.47±0.04 0.97±0.03 15.57±1.30 69.28±6.66 54.04±5.26
C8 0.58±0.07 0.64±0.09 0.48±0.05 1.52±0.03 1.43±0.03 1.45±0.03 0.98±0.04 15.82±1.25 70.51±7.66 53.42±6.11
C9 0.60±0.03 0.66±0.03 0.39±0.02 1.51±0.02 1.39±0.03 1.51±0.02 0.81±0.04 15.37±1.14 66.80±7.28 50.31±6.26
C10 0.48±0.03 0.48±0.04 0.43±0.03 1.53±0.02 1.32±0.03 1.59±0.02 0.86±0.05 15.53±1.20 69.60±8.75 54.53±6.87

Table 4.4: Average structural properties of the intermediates of Ribonuclease A. In the
system T1~T10 and C1 ~ C10, the time range for averaging is over protein trajectories
1,200ns – 1,300ns and we adopt one protein structure per 0.5 ns. The system, P, is
simulating pure protein solvated in water and its average protein structures are over all
simulation time. α-β layer distance is defined as the distance between two centers of mass
from α and β layers. All solvent access surface area is calculated using Gromacs program,
which employs the Lee and Richards algorithm with a probe radius of 0.14m. Transition
point is read from all plots. The reference structure of RMSD calculation is crystal
protein.

144

#Native
Contacts
-protein

#Native
Contacts-
α-layer

#Native
Contacts-
β-layer

#Native
Contacts-
α-β
interface
#Total
Contacts-
α-β
interface
#Hbond-
protein

#Hbond-
α-layer

#Hbond-
β-layer

#Hbond-
α-β
interface

Crystal 312 143 135 34 34 71 36 21 14
P 281±7 125±4 115±5 18±2 40±4 74.08±5.25 38.68±2.79 23.08±2.26 13.32±2.46
T1 223±8 98±5 109±4 16±2 38±5 70.91±7.22 28.02±3.92 29.51±3.67 13.38±2.76
T2 214±8 86±4 114±5 14±2 36±5 76.32±6.34 23.75±3.94 33.84±3.96 18.74±3.44
T3 210±10 90±6 112±5 9±2 24±4 67.10±6.16 28.14±3.91 26.72±3.82 12.23±2.66
T4 220±9 98±5 115±5 8±2 26±5 69.10±6.24 27.80±3.54 32.64±3.94 8.66±2.51
T5 228±7 105±4 105±4 18±3 50±5 82.76±6.10 29.04±3.07 31.57±3.84 22.14±3.06
T6 218±7 100±4 105±4 13±2 24±4 76.68±6.60 32.56±3.86 30.62±4.15 13.50±2.25
T7 231±8 100±5 117±5 13±2 44±5 88.25±6.51 29.70±3.89 35.19±3.44 23.36±2.63
T8 188±6 84±4 101±3 3±1 19±4 78.35±6.22 30.62±3.88 38.37±4.28 9.36±2.25
T9 220±8 98±5 110±4 12±2 32±4 83.62±6.47 28.44±3.70 39.15±3.88 16.02±2.61
T10 211±7 89±3 112±4 11±2 34±4 81.64±6.09 30.13±3.29 31.33±3.86 20.18±2.52
C1 199±7 81±4 106±4 12±2 39±5 76.64±6.41 26.10±3.53 34.51±4.21 16.02±2.53
C2 211±7 87±4 113±5 12±2 24±4 72.29±5.94 23.71±3.32 33.00±4.05 15.58±2.58
C3 225±9 99±5 114±4 12±3 27±5 76.65±5.83 31.35±3.91 34.25±3.61 11.05±2.53
C4 204±7 84±4 108±4 14±2 33±5 74.87±5.57 23.62±3.55 37.36±4.05 13.89±2.29
C5 214±9 91±5 109±4 14±3 32±7 68.79±6.40 27.24±3.82 31.18±3.97 10.36±2.44
C6 220±7 93±4 111±5 16±2 26±4 69.78±6.30 30.34±4.14 28.81±3.81 10.63±2.25
C7 217±6 93±4 112±4 13±2 29±4 74.47±5.41 24.46±3.46 34.45±4.17 15.56±2.19
C8 217±9 95±5 111±5 11±2 25±4 71.97±6.31 26.43±3.50 31.15±4.07 14.38±2.66
C9 223±8 97±5 114±5 13±2 29±4 73.82±5.80 26.59±3.46 31.03±3.82 16.20±2.34
C10 225±9 103±5 105±4 17±3 31±5 71.60±6.17 29.74±3.73 28.79±3.70 13.06±2.46

Table 4.5: Number of residue intra-molecular contacts and number of hydrogen bonds in
the intermediates of HEW lysozyme. Average is over 1,200ns-1,300ns for system
T1~T10 and C1~C10 while total simulation course for system P. Two residues are
considered to have contacts with each other when their Cα atoms are within 0.6 nm.
Hydrogen bonds are determined based on cutoff radius, 0.35 nm (acceptor – donor) and
cutoff angle, 30
o
(acceptor - donor - hydrogen).

145

4.4 References
1. Johnson, L.N. and D.C. Phillips, Structure of some crystalline lysozyme-inhibitor
complexes determined by X-ray analysis at 6 A resolution. Nature, 1965.
206(4986): p. 761-&.

2. Daniel, R.M., et al., The role of dynamics in enzyme activity. Annual Review of
Biophysics and Biomolecular Structure, 2003. 32: p. 69-92.

3. Meinhold, L. and J.C. Smith, Protein dynamics from X-ray crystallography:
Anisotropic, global motion in diffuse scattering patterns. Proteins-Structure
Function and Bioinformatics, 2007. 66(4): p. 941-953.

4. Gabel, F., et al., Protein dynamics studied by neutron scattering. Quarterly
Reviews of Biophysics, 2002. 35(4): p. 327-367.

5. Chen, Y., et al., Probing single-molecule T4 lysozyme conformational dynamics
by intramolecular fluorescence energy transfer. Journal of Physical Chemistry B,
2003. 107(31): p. 7947-7956.

6. Hayward, S., A. Kitao, and H.J.C. Berendsen, Model-free methods of analyzing
domain motions in proteins from simulation: A comparison of normal mode
analysis and molecular dynamics simulation of lysozyme. Proteins-Structure
Function and Genetics, 1997. 27(3): p. 425-437.

7. Bu, Z.M., et al., Coupled protein domain motion in Taq polymerase revealed by
neutron spin-echo spectroscopy. Proceedings of the National Academy of
Sciences of the United States of America, 2005. 102(49): p. 17646-17651.

8. Arnold, G.E. and R.L. Ornstein, Protein hinge bending as seen in molecular
dynamics simulations of native and M6I mutant T4 lysozymes. Biopolymers,
1997. 41(5): p. 533-544.

9. Purcell, M., et al., Interactions of atrazine and 2,4-D with human serum albumin
studied by gel and capillary electrophoresis, and FTIR spectroscopy. Biochimica
Et Biophysica Acta-Protein Structure and Molecular Enzymology, 2001. 1548(1):
p. 129-138.

10. Raines, R.T., Ribonuclease A. Chemical Reviews, 1998. 98(3): p. 1045-1065.

11. Kuznetsova, L.L., et al., Artificial ribonucleases 5. Synthesis and ribonuclease
activity of tripeptides composed of amino acids involved in catalytic centers of
natural ribonucleases. Russian Chemical Bulletin, 2004. 53(2): p. 455-462.
146

12. Beloglazova, N.G., et al., Sequence-specific artificial ribonucleases. I. Bis-
imidazole-containing oligonucleotide conjugates prepared using precursor-based
strategy. Nucleic Acids Research, 2004. 32(13): p. 3887-3897.

13. Humphrey, W., A. Dalke, and K. Schulten, VMD: Visual molecular dynamics.
Journal of Molecular Graphics, 1996. 14(1): p. 33-&.

14. Otzen, D., Protein-surfactant interactions: A tale of many states. Biochimica Et
Biophysica Acta-Proteins and Proteomics, 2011. 1814(5): p. 562-591.

15. Andrade, M.I.P., et al., Enthalpy of interaction of ribonuclease-A and normal-
alkyl sulfates in aqueous-solution. Journal of the Chemical Society-Faraday
Transactions I, 1981. 77: p. 2939-2948.

16. Hamill, A.C., S.C. Wang, and C.T. Lee, Solution structure of an amyloid-forming
protein during photoinitiated hexamer-dodecamer transitions revealed through
small-angle neutron scattering. Biochemistry, 2007. 46(26): p. 7694-7705.

17. Hamill, A.C., S.C. Wang, and C.T. Lee, Probing lysozyme conformation with
light reveals a new folding intermediate. Biochemistry, 2005. 44(46): p. 15139-
15149.

18. Lee, C.T., K.A. Smith, and T.A. Hatton, Photocontrol of protein folding: The
interaction of photosensitive surfactants with bovine serum albumin.
Biochemistry, 2005. 44(2): p. 524-536.

19. Wang, S.C. and C.T. Lee, Protein secondary structure controlled with light and
photoresponsive surfactants. Journal of Physical Chemistry B, 2006. 110(32): p.
16117-16123.

20. Dunbar, J., et al., The effect of denaturants on protein structure. Protein Science,
1997. 6(8): p. 1727-1733.

21. van Gunsteren, W.F., Billeter, S.R., Eising, A.A., Hunenberger, P.H., Kruger, P.,
Mark, A.E., Scott, W.R.P., Tironi, I.G., Biomolecular Simulation: The Gromos96
Manual and User Guide. Hochschulverlag AG an der ETH, Zurich., 1996.

22. Neria, E., S. Fischer, and M. Karplus, Simulation of activation free energies in
molecular systems. Journal of Chemical Physics, 1996. 105(5): p. 1902-1921.

23. Jorgensen, W.L., et al., Comparison of simple potential functions for simulating
liquid water. Journal of Chemical Physics, 1983. 79(2): p. 926-935.

147

24. D. van der Spoel, E.L., B. Hess, A. R. van Buuren, E. Apol, P. J. Meulenhoff,,
A.L.T.M.S. D. P. Tieleman, K. A. Feenstra, R. van Drunen and H. J. C., and
Berendsen, Gromacs User Manual version 3.3, www.gromacs.org. 2005.

25. Bekker, H., Berendsen, H. J. C., Dijkstra, E. J., Achterop, S., van Drunen, R., van
der Spoel, D., Sijbers, A., Keegstra, H., Reitsma, B., Renardus, M. K. R.,
Gromacs: A parallel computer for molecular dynamics simulations. In Physics
Computing 92 (Singapore, 1993).de Groot, R. A., Nadrchal, J., eds. . World
Scientific.

26. Berendsen, H.J.C., D. Vanderspoel, and R. Vandrunen, Gromacs - A message
passing parallel molecular dynamics implementation. Computer Physics
Communications, 1995. 91(1-3): p. 43-56.

27. Lindahl, E., B. Hess, and D. van der Spoel, GROMACS 3.0: a package for
molecular simulation and trajectory analysis. Journal of Molecular Modeling,
2001. 7(8): p. 306-317.

28. Van der Spoel, D., et al., GROMACS: Fast, flexible, and free. Journal of
Computational Chemistry, 2005. 26(16): p. 1701-1718.

29. Bockmann, M., et al., Atomistic force field for azobenzene compounds adapted for
QM/MM simulations with applications to liquids and liquid crystals. Journal of
Chemical Theory and Computation, 2007. 3(5): p. 1789-1802.

30. Hamill, A.C., Photocontrol Of Protein Conformation Through The Use of
Photoresponsive Surfactants, Investigated by Small Angle Neutron Scattering.
2008.

31. Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, W.F., DiNola, A., Haak, J.R.,
In: Pullman, B. (Ed.) Intermolecular Forces. Reidel Dordrecht, The
Netherlands, 1981.

32. Tironi, I.G., et al., A generalized reaction field method for molecular dynamics
simulations. Journal of Chemical Physics, 1995. 102(13): p. 5451-5459.

33. Schulz, R., et al., Scaling of Multimillion-Atom Biological Molecular Dynamics
Simulation on a Petascale Supercomputer. Journal of Chemical Theory and
Computation, 2009. 5(10): p. 2798-2808.

34. Berendsen, H.J.C., et al., Molecular dynamics with coupling to an external bath.
Journal of Chemical Physics, 1984. 81(8): p. 3684-3690.

148

35. Caner, S., et al., The structure of the Mycobacterium smegmatis trehalose
synthase reveals an unusual active site configuration and acarbose-binding
mode(dagger). Glycobiology, 2013. 23(9): p. 1075-1083.

36. Sheng, J., J.H. Gan, and Z. Huang, Structure-based DNA-targeting strategies with
small molecule ligands for drug discovery. Medicinal Research Reviews, 2013.
33(5): p. 1119-1173.

37. Toth, G., et al., Site-specific basicities regulate molecular recognition in receptor
binding: in silico docking of thyroid hormones. European Biophysics Journal with
Biophysics Letters, 2013. 42(9): p. 721-730.

38. Tanneeru, K., I. Sahu, and L. Guruprasad, Ligand-based drug design for human
endothelin converting enzyme-1 inhibitors. Medicinal Chemistry Research, 2013.
22(9): p. 4401-4409.

39. Misra, P.P. and N. Kishore, Differential Modulation in Binding of Ketoprofen to
Bovine Serum Albumin in the Presence and Absence of Surfactants: Spectroscopic
and Calorimetric Insights. Chemical Biology & Drug Design, 2013. 82(1): p. 81-
98.

40. Divsalar, A., et al., Spectroscopic and Calorimetric Study of 2,2 '-Dibipyridin
Cu(II) Chloride Binding to Bovine beta-Lactoglobulin. Journal of Solution
Chemistry, 2013. 42(4): p. 705-715.

41. Quesada-Soriano, I., et al., Calorimetric Studies of Ligands Binding to
Glutathione S-Transferase from the Malarial Parasite Plasmodium falciparum.
Biochemistry, 2013. 52(11): p. 1980-1989.

42. Keswani, N., S. Choudhary, and N. Kishore, Quantitative aspects of recognition
of the antibiotic drug oxytetracycline by bovine serum albumin: Calorimetric and
spectroscopic studies. Journal of Chemical Thermodynamics, 2013. 58: p. 196-
205.

43. Bijari, N., et al., Spectroscopic study of interaction between osthole and human
serum albumin: Identification of possible binding site of the compound. Journal of
Luminescence, 2013. 143: p. 328-336.

44. Anacona, J. and J. Santaella, Synthesis, magnetic and spectroscopic studies of a
Schiff base derived from cephaclor and 1,2-diaminobenzene and its transition
metal complexes. Spectrochimica Acta Part a-Molecular and Biomolecular
Spectroscopy, 2013. 115: p. 800-804.

149

45. Blake, L. and M.E.S. Soliman, Bifunctional Anti-HIV/TB Inhibitors: Perspective
from In-Silico Design and Molecular Dynamics Simulations. Letters in Drug
Design & Discovery, 2013. 10(8): p. 706-712.

46. Kalyaanamoorthy, S. and Y.P.P. Chen, Ligand release mechanisms and channels
in histone deacetylases. Journal of Computational Chemistry, 2013. 34(26): p.
2270-2283.

47. Kaur, N., et al., Identification of Druggable Targets for Acinetobacter baumannii
Via Subtractive Genomics and Plausible Inhibitors for MurA and MurB. Applied
Biochemistry and Biotechnology, 2013. 171(2): p. 417-436.

48. Simanek, E.E., A.E. Enciso, and G.M. Pavan, Computational design principles
for the discovery of bioactive dendrimers: s -triazines and other examples. Expert
Opinion on Drug Discovery, 2013. 8(9): p. 1057-1069.

49. Tripathi, S.K., R. Muttineni, and S.K. Singh, Extra precision docking, free energy
calculation and molecular dynamics simulation studies of CDK2 inhibitors.
Journal of Theoretical Biology, 2013. 334: p. 87-100.

50. Silberstein, M., et al., Identification of substrate binding sites in enzymes by
computational solvent mapping. Journal of Molecular Biology, 2003. 332(5): p.
1095-1113.

51. Adcock, S.A. and J.A. McCammon, Molecular dynamics: Survey of methods for
simulating the activity of proteins. Chemical Reviews, 2006. 106(5): p. 1589-
1615.

52. Housaindokht, M.R., M.R. Bozorgmehr, and M. Bahrololoom, Analysis of ligand
binding to proteins using molecular dynamics simulations. Journal of Theoretical
Biology, 2008. 254(2): p. 294-300.

53. Joyner, J.C., et al., Antimicrobial metallopeptides with broad nuclease and
ribonuclease activity. Chemical Communications, 2013. 49(21): p. 2118-2120.

54. Liu, B., et al., Real time monitoring of junction ribonuclease activity of RNase H
using chimeric molecular beacons. Analyst, 2013. 138(11): p. 3238-3245.

55. Dayer, M.R., O. Ghayour, and M.S. Dayer, Mechanism of the Bell-Shaped Profile
of Ribonuclease A activity: Molecular Dynamic Approach. Protein Journal, 2012.
31(7): p. 573-579.

56. Delcardayre, S.B. and R.T. Raines, Structural determinants of enzymatic
processivity. Biochemistry, 1994. 33(20): p. 6031-6037.
150

57. Fisher, B.M., J.E. Grilley, and R.T. Raines, A new remote subsite in ribonuclease
A. Journal of Biological Chemistry, 1998. 273(51): p. 34134-34138.

58. Fisher, B.M., J.H. Ha, and R.T. Raines, Coulombic forces in protein-RNA
interactions: Binding and cleavage by ribonuclease A and variants at Lys7,
Arg10, and Lys66. Biochemistry, 1998. 37(35): p. 12121-12132.

59. Muchiri, J.M., et al., Efavirenz Stimulates HIV-1 Reverse Transcriptase RNase H
Activity by a Mechanism Involving Increased Substrate Binding and Secondary
Cleavage Activity. Biochemistry, 2013. 52(29): p. 4981-4990.

60. Barbas, A., et al., Rossmann-fold motifs can confer multiple functions to
metabolic enzymes: RNA binding and ribonuclease activity of a UDP-glucose
dehydrogenase. Biochemical and Biophysical Research Communications, 2013.
430(1): p. 218-224.

61. Leo, B., et al., The solution structure of the prototype foamy virus RNase H
domain indicates an important role of the basic loop in substrate binding.
Retrovirology, 2012. 9.

62. Zheng, X.H., et al., Metal and ligand binding to the HIV-RNase H active site are
remotely monitored by Ile556. Nucleic Acids Research, 2012. 40(20): p. 10543-
10553.

63. Kabsch, W. and C. Sander, Dictionary of protein secondary structure - pattern -
recognition of hydrogen-bonded and geometrical features. Biopolymers, 1983.
22(12): p. 2577-2637.

64. Mirarefi, P., Photocontrol of protein conformation and enzyme activity in the
presence of light-responsive surfactants.

65. Gu, H., et al., Experimental and computational analysis of the transition state for
ribonuclease A-catalyzed RNA 2 '-O-transphosphorylation. Proceedings of the
National Academy of Sciences of the United States of America, 2013. 110(32): p.
13002-13007.

66. Xia, Z., et al., RNA 3D Structure Prediction by Using a Coarse-Grained Model
and Experimental Data. Journal of Physical Chemistry B, 2013. 117(11): p. 3135-
3144.

67. Gumbart, J.C., B. Roux, and C. Chipot, Efficient Determination of Protein-
Protein Standard Binding Free Energies from First Principles. Journal of
Chemical Theory and Computation, 2013. 9(8): p. 3789-3798.

151

68. Camilloni, C., A. Cavalli, and M. Vendruscolo, Assessment of the Use of NMR
Chemical Shifts as Replica-Averaged Structural Restraints in Molecular
Dynamics Simulations to Characterize the Dynamics of Proteins. Journal of
Physical Chemistry B, 2013. 117(6): p. 1838-1843.

69. Wang, S.-C., Controlling the form-dynamics-function relationship of proteins
with light illumination. DIssertation_University of Southern California, 2008.

152

Chapter 5: Future Work

5.1 Enhanced enzymatic activity and selectivity of lysozyme through
surfactants-induced conformational changes
Lysozyme is an enzyme which has the ability to lyse certain Gram-positive
bacteria by hydrolyzing the β-linkage between N-acetylmuramic acid (NAM) and N-
acetylglucosamine (NAG) of the peptidoglycan layer in the bacterial cell wall. An 7-fold
increase in enzymatic activity of lysozyme induced by the surfactants, azoTAB, has been
tested under experimental work [1]. Small-angle neutron scattering (SANS)
measurements revealed that protein swelling occurs principally near the hinge region
connecting the α and β domains, and further neutron spin echo (NSE) measurements
detected a likely enhanced domain motions on the “nanometer” length scale over the
“nanosecond” time scale and increased flexibility within the protein.
Complementary to experiments, molecular dynamics (MD) simulations are a
widely used techniques to investigate collective motions in proteins. The relative
movement of the two domains of lysozyme molecule can cause large global
conformational changes which may permit access of substrate and generate an
appropriate environment for catalysis. A state-of-the-art approach to elucidate collective
motions from the protein dynamics (from MD trajectory data) is principal component
analysis (PCA). PCA is commonly used to extract the collective motions with the largest
contribution to the variance of the atomic fluctuations. Analysis of vibrational motions
153

and thermal fluctuational dynamics is another approach for studying structural, dynamic
and functional properties of proteins.

5.2 Microseconds simulation of unfolding α-lactalbumin Induced by
azoTABs
The structure of α-lactalbumin is well known and is composed of 123 amino acids
and 4 disulfide bridges. The molecular weight of α-lactalbumin is 14.2K Daltons. α-
lactalbumins (LAs) and c-type lysozymes (LYZs) have a 35–40% sequence homology
and share a common three dimensional (3D) fold (see Figure 5.1) but perform different
functions. These proteins are made up of two domains which are separated by a cleft. In
human α-LA, the residues 1–38 and 83–123 form the α-domain which predominantly
consists of the α-helices, A, B, C and D. The residues 39–82 constitute the β-domain
which consists of three β-strands and the extended loop region. The positions of the four
disulfide bonds are 6–120, 28–111, 61–77 and 73–91 in human α-LA.
In chapter2 and chapter3, the simulation of investigating the interaction between
lysozyme molecules and surfactants are described in detail. Table 5.1 lists the systems of
the microsecond simulations about the α-lactalbumin molecules and surfactants (either 10
trans or 10 cis). The simulation setup for the system of proteins, α-lactalbumin, and
surfactants is written in section 2.2. Figure 5.2 and Figure 5.3 are showing the binding
affinity of surfactants toward the α-lactalbumin molecules and the definition of the
binding affinity is shown at section 2.3.1. For the α lactalbumin molecules, both trans and
154

cis azo-TABs like close contacts with helix B, helix D and β sheets while helix C is the
least binding affinity among the other secondary structures and the former part of the
helix A (residues 5,6,7) also has low binding affinity with azoTABs. Figure 5.3 shows
that bottom part of the α-lactalbumin molecules has low binding affinity for azoTABs;
however, the bottom part of the lysozyme molecules has high binding affinity for
azoTAB (see Figure 2.4 (a), (b)). Previous experimental work [3] has proven that the
bottom part of the lysozyme is puffy after increasing the concentration of the surfactants
(~12mM); moreover, the enhanced enzymatic activity [1] is measured potentially because
of the improved flexibility of hinge of the lysozyme molecules (at the bottom of the cleft
in the α- and β-domain of the protein molecules) observed by the neutron spin echo
measurement [4]. The α-lactalbumin molecules are not bio-enzymes and has less binding
affinity to azoTABs on the bottom part of the proteins as observed from the simulations.
Otherwise, in c-type lysozyme, the most protected helix is helix B, followed by helix C,
A and D while in human α-LA, helix C is the most protected. Our simulation results have
shown that the azoTABs early unfold most of helix C in α-lactalbumin while keeping
most helix C structure in lysozyme (Table 2.3). α-Lactalbumin is a metallo-protein and
binds calcium, whereas only a few of the lysozymes bind calcium. Calcium is also bound
at a loop situated at the bottom of the cleft. The comparison of unfolding of α-
lactalbumin induced by azoTAB with unfolding of lysozyme induced by azoTAB can
have insights into the correlations of protein folding with their amino sequences.

155

(a)

(b)

Figure 5.1 (a) Crystal structure of α-lactalbumin (b) Crystal structure of hen egg white
lysozyme. The VMD program is used to show the protein secondary structures in New
Ribbons style [2].

System Dimension
of box
No. of
water molecules
Azotab
Structure
No.
of azoTAB
Simulation time (ns)
P none 0
T1 6.14 6425 all trans 10 1450
T2 6.14 6414 all trans 10 1253
T3 6.14 6433 all trans 10 1413
T4 6.14 6415 all trans 10 1343
T5 6.13 6425 all trans 10 1408
T6 6.13 6414 all trans 10 1449
T7 6.13 6416 all trans 10 1474
T8 6.13 6412 all trans 10 1476
T9 6.14 6437 all trans 10 1453
T10 6.13 6418 all trans 10 1450
C1 6.13 6407 all cis 10 1365
C2 6.14 6411 all cis 10 1442
C3 6.13 6415 all cis 10 1361
C4 6.13 6406 all cis 10 1342
C5 6.13 6411 all cis 10 1433
C6 6.13 6418 all cis 10 1405
C7 6.13 6415 all cis 10 1472
C8 6.13 6404 all cis 10 1477
C9 6.14 6424 all cis 10 1259
C10 6.14 6415 all cis 10 1388

Table 5.1 Summary of simulation systems (Alpha-Lactalbumin). Dimension of box is
average after equilibrium.
156

Figure 5.2 Binding Affinity of trans-azoTAB toward alpha-lactalbumin (black line).
Binding Affinity of cis-azoTAB toward alpha-lactalbumin (red line).
(a)

(b)

Figure 5.3 : (a) Binding Affinity of trans-azoTAB toward α-lactalbumin. (b) Binding
Affinity of cis-azoTAB toward α-lactalbumin. Residues with high binding affinity, Pi >
1.0, are colored in red; low binding affinity, Pi <0.7 in blue; and moderate binding
affinity, 0.7< Pi <1.0, in yellow. The two New Ribbons plots are constructed by VMD
program [2].
157

Figure 5.4 : Time evolution of the protein secondary structure in C5 system with DSSP
program [5-6].

158

5.3 References
1. Wang, S.C. and C.T. Lee, Enhanced enzymatic activity through photoreversible
conformational changes. Biochemistry, 2007. 46(50): p. 14557-14566.

2. Humphrey, W., A. Dalke, and K. Schulten, VMD: Visual molecular dynamics.
Journal of Molecular Graphics, 1996. 14(1): p. 33-&.

3. Hamill, A.C., S.C. Wang, and C.T. Lee, Probing lysozyme conformation with
light reveals a new folding intermediate. Biochemistry, 2005. 44(46): p. 15139-
15149.

4. Wang, S.C., et al., Light-Controlled Protein Dynamics Observed with Neutron
Spin Echo Measurements. Biochemistry, 2011. 50(38): p. 8150-8162.

5. Kabsch, W. and C. Sander, Dictionary of protein secondary structure - pattern -
recognition of hydrogen-bonded and geometrical features. Biopolymers, 1983.
22(12): p. 2577-2637.

6. Joosten, R.P., et al., A series of PDB related databases for everyday needs.
Nucleic Acids Research, 2011. 39: p. D411-D419.

159

Bibliography
Acharya, K. R., J. S. Ren, D. I. Stuart, D. C. Phillips and R. E. Fenna (1991). Crystal
structure of human alpha-lactalbumin at 1.7 A resolution. Journal of Molecular Biology
221(2): 571-581.

Adcock, S. A. and J. A. McCammon (2006). Molecular dynamics: Survey of methods for
simulating the activity of proteins. Chemical Reviews 106(5): 1589-1615.

Alberts, B., Johnson, A., Lewis., J., et al. , (2002). Molecular Biology of the The Cell.
http://www.ncbi.nlm.nih.gov/books/NBK26830/.

Anacona, J. and J. Santaella (2013). Synthesis, magnetic and spectroscopic studies of a
Schiff base derived from cephaclor and 1,2-diaminobenzene and its transition metal
complexes. Spectrochimica Acta Part a-Molecular and Biomolecular Spectroscopy 115:
800-804.

Andrade, M. I. P., E. Boitard, M. A. Saghal, P. Manley, M. N. Jones and H. A. Skinner
(1981). Enthalpy of interaction of ribonuclease-A and normal-alkyl sulfates in aqueous-
solution. Journal of the Chemical Society-Faraday Transactions I 77: 2939-2948.

Anfinsen, C. B. (1973). Principles that govern folding of protein chains. Science
181(4096): 223-230.

Arai, M. and K. Kuwajima (2000). Role of the molten globule state in protein folding.
Advances in Protein Chemistry; Protein folding mechansims, Academic Press Inc., 525 B
Street, Suite 1900, San Diego, CA, 92101-4495, USA Harcourt Place, 32 Jamestown
Road, London, NW1 7BY, UK. 53: 209-282.

Arnold, G. E. and R. L. Ornstein (1997). Protein hinge bending as seen in molecular
dynamics simulations of native and M6I mutant T4 lysozymes. Biopolymers 41(5): 533-
544.

Bachmann, A., D. Segel and T. Kiefhaber (2002). Test for cooperativity in the early
kinetic intermediate in lysozyme folding. Biophysical Chemistry 96(2-3): 141-151.

Baldwin, R. L. (1995). The nature of protein-folding pathways - the classical versus the
new view. Journal of Biomolecular Nmr 5(2): 103-109.

Barbas, A., A. Popescu, C. Frazao, C. M. Arraiano and A. M. Fialho (2013). Rossmann-
fold motifs can confer multiple functions to metabolic enzymes: RNA binding and
ribonuclease activity of a UDP-glucose dehydrogenase. Biochemical and Biophysical
Research Communications 430(1): 218-224.
160

Baum, J., C. M. Dobson, P. A. Evans and C. Hanley (1989). Characterization of a partly
folded potein by NMR methods - studies on the molten globule state of guinea-pig alpha-
lactalbumin. Biochemistry 28(1): 7-13.

Bekker, H., Berendsen, H. J. C., Dijkstra, E. J., Achterop, S., van Drunen, R., van der
Spoel, D., Sijbers, A., Keegstra, H., Reitsma, B., Renardus, M. K. R. Gromacs: A parallel
computer for molecular dynamics simulations. In Physics Computing 92 (Singapore,
1993).de Groot, R. A., Nadrchal, J., eds. . World Scientific.

Beloglazova, N. G., M. M. Fabani, M. A. Zenkova, E. V. Bichenkova, N. N. Polushin, V.
V. Sil'nikov, K. T. Douglas and V. V. Vlassov (2004). Sequence-specific artificial
ribonucleases. I. Bis-imidazole-containing oligonucleotide conjugates prepared using
precursor-based strategy. Nucleic Acids Research 32(13): 3887-3897.

Berendsen, H. J. C., J. P. M. Postma, W. F. Vangunsteren, A. Dinola and J. R. Haak
(1984). Molecular dynamics with coupling to an external bath. Journal of Chemical
Physics 81(8): 3684-3690.

Berendsen, H. J. C., Postma, J.P.M., van Gunsteren, W.F., DiNola, A., Haak, J.R. (1981).
In: Pullman, B. (Ed.) Intermolecular Forces. Reidel Dordrecht, The Netherlands.

Berendsen, H. J. C., D. Vanderspoel and R. Vandrunen (1995). Gromacs - A message
passing parallel molecular dynamics implementation. Computer Physics Communications
91(1-3): 43-56.

Bijari, N., Y. Shokoohinia, M. R. Ashrafi-Kooshk, S. Ranjbar, S. Parvaneh, M. Moieni-
Arya and R. Khodarahmi (2013). Spectroscopic study of interaction between osthole and
human serum albumin: Identification of possible binding site of the compound. Journal
of Luminescence 143: 328-336.

Blake, C. C. F., D. F. Koenig, G. A. Mair, A. C. T. North, D. C. Phillips and V. R. Sarma
(1965). Structure of hen egg-white lysozyme - A 3-dimensional fourier synthesis at 2A
resolution. Nature 206(4986): 757-&.

Blake, L. and M. E. S. Soliman (2013). Bifunctional Anti-HIV/TB Inhibitors: Perspective
from In-Silico Design and Molecular Dynamics Simulations. Letters in Drug Design &
Discovery 10(8): 706-712.

Bockmann, M., C. Peter, L. Delle Site, N. L. Doltsinis, K. Kremer and D. Marx (2007).
Atomistic force field for azobenzene compounds adapted for QM/MM simulations with
applications to liquids and liquid crystals. Journal of Chemical Theory and Computation
3(5): 1789-1802.

161

Brooks, C. L. (1998). Simulations of protein folding and unfolding. Current Opinion in
Structural Biology 8(2): 222-226.

Bu, Z. M., R. Biehl, M. Monkenbusch, D. Richter and D. J. E. Callaway (2005). Coupled
protein domain motion in Taq polymerase revealed by neutron spin-echo spectroscopy.
Proceedings of the National Academy of Sciences of the United States of America
102(49): 17646-17651.

Buck, M., S. E. Radford and C. M. Dobson (1993). A partially folded state of hen egg
white lysozyme in trifluoroethanol - structural characterization and implications for
protein folding. Biochemistry 32(2): 669-678.

Callenberg, K. (2010), Illustration of solvent accessible surface in relation to the van der
Waals surface of an atom. http://en.wikipedia.org/wiki/File:Accessible_surface.svg.

Camilloni, C., A. Cavalli and M. Vendruscolo (2013). Assessment of the Use of NMR
Chemical Shifts as Replica-Averaged Structural Restraints in Molecular Dynamics
Simulations to Characterize the Dynamics of Proteins. Journal of Physical Chemistry B
117(6): 1838-1843.

Caner, S., N. Nguyen, A. Aguda, R. Zhang, Y. T. Pan, S. G. Withers and G. D. Brayer
(2013). The structure of the Mycobacterium smegmatis trehalose synthase reveals an
unusual active site configuration and acarbose-binding mode(dagger). Glycobiology
23(9): 1075-1083.

Chaffotte, A. F., Y. Guillou and M. E. Goldberg (1992). Kinetic resolution of peptide
bond and side chain far-UV circular dichroism during the folding of hen egg white
lysozyme. Biochemistry 31(40): 9694-9702.

Chan, H. S. and K. A. Dill (1994). Transition states and folding dynamics of proteins and
heteropolymers. Journal of Chemical Physics 100(12): 9238-9257.

Chen, L. L., K. O. Hodgson and S. Doniach (1996). A lysozyme folding intermediate
revealed by solution X-ray scattering. Journal of Molecular Biology 261(5): 658-671.

Chen, Y., D. H. Hu, E. R. Vorpagel and H. P. Lu (2003). Probing single-molecule T4
lysozyme conformational dynamics by intramolecular fluorescence energy transfer.
Journal of Physical Chemistry B 107(31): 7947-7956.

Creighton, T. E. (1990). Protein folding. Biochemical Journal 270(1): 1-16.

Creighton, T. E. and J. J. Ewbank (1994). Disulfide-rearranged molten globlue state of
alpha-lactalbumin. Biochemistry 33(6): 1534-1538.

162

D. van der Spoel, E. L., B. Hess, A. R. van Buuren, E. Apol, P. J. Meulenhoff,,
A.L.T.M.S. D. P. Tieleman, K. A. Feenstra, R. van Drunen and H. J. C., and Berendsen
(2005) Gromacs User Manual version 3.3, www.gromacs.org.

Daniel, R. M., R. V. Dunn, J. L. Finney and J. C. Smith (2003). The role of dynamics in
enzyme activity. Annual Review of Biophysics and Biomolecular Structure 32: 69-92.

Davidson, A. R. and R. T. Sauer (1994). Folded proteins occur frequently in libraries of
random amino-acid-sequences. Proceedings of the National Academy of Sciences of the
United States of America 91(6): 2146-2150.

Dayer, M. R., O. Ghayour and M. S. Dayer (2012). Mechanism of the Bell-Shaped
Profile of Ribonuclease A activity: Molecular Dynamic Approach. Protein Journal 31(7):
573-579.

Delcardayre, S. B. and R. T. Raines (1994). Structural determinants of enzymatic
processivity. Biochemistry 33(20): 6031-6037.

Demarest, S. J., J. A. Boice, R. Fairman and D. P. Raleigh (1999). Defining the core
structure of the alpha-lactalbumin molten globule state. Journal of Molecular Biology
294(1): 213-221.

Deshpande, A., S. Nimsadkar and S. C. Mande (2005). Effect of alcohols on protein
hydration: crystallographic analysis of hen egg-white lysozyme in the presence of
alcohols. Acta Crystallographica Section D-Biological Crystallography 61: 1005-1008.

Diamond, R. (1974). Real-space refinement of structure of hen egg-white lysozyme.
Journal of Molecular Biology 82(3): 371-&.

Dinner, A. R., A. Sali, L. J. Smith, C. M. Dobson and M. Karplus (2000). Understanding
protein folding via free-energy surfaces from theory and experiment. Trends in
Biochemical Sciences 25(7): 331-339.

Divsalar, A., A. A. Saboury, T. Haertle, L. Sawyer, H. Mansouri-Torshizi and L.
Barzegar (2013). Spectroscopic and Calorimetric Study of 2,2 '-Dibipyridin Cu(II)
Chloride Binding to Bovine beta-Lactoglobulin. Journal of Solution Chemistry 42(4):
705-715.

Dobson, C. M. (1991). Characterization of protein folding intermediates : Current
Opinion in Structural Biology 1991, 1:22-27. Current Opinion in Structural Biology 1(1):
22-27.

Dobson, C. M. (1994). Protein folding solid evidence for molten globules. Current
Biology 4(7): 636-640.
163

Dobson, C. M., P. A. Evans and S. E. Radford (1994). Understanding how proteins fold -
the lysozyme story so far. Trends in Biochemical Sciences 19(1): 31-37.

Dobson, C. M., A. Sali and M. Karplus (1998). Protein folding: A perspective from
theory and experiment. Angewandte Chemie-International Edition 37(7): 868-893.

Dolgikh, D. A., R. I. Gilmanshin, E. V. Brazhnikov, V. E. Bychkova, G. V. Semisotnov,
S. Y. Venyaminov and O. B. Ptitsyn (1981). Alpha-lactalbumin - compact state with
fluctuating tertiary structure. Febs Letters 136(2): 311-315.

Duan, Y. and P. A. Kollman (1998). Pathways to a protein folding intermediate observed
in a 1-microsecond simulation in aqueous solution. Science 282(5389): 740-744.

Dunbar, J., H. P. Yennawar, S. Banerjee, J. B. Luo and G. K. Farber (1997). The effect of
denaturants on protein structure. Protein Science 6(8): 1727-1733.

Evans, P. A. and S. E. Radford (1994). Probing the structure of folding intermediates.
Current Opinion in Structural Biology 4(1): 100-106.

Fersht, A. R. (1993). Protein folding and stability - the pathway of folding of barnase.
Febs Letters 325(1-2): 5-16.

Fersht, A. R. (1995). Mapping the structures of transition states and intermediates in
folding - delineation of pathways at high resolution. Philosophical Transactions of the
Royal Society of London Series B-Biological Sciences 348(1323): 11-15.

Fisher, B. M., J. E. Grilley and R. T. Raines (1998). A new remote subsite in
ribonuclease A. Journal of Biological Chemistry 273(51): 34134-34138.

Fisher, B. M., J. H. Ha and R. T. Raines (1998). Coulombic forces in protein-RNA
interactions: Binding and cleavage by ribonuclease A and variants at Lys7, Arg10, and
Lys66. Biochemistry 37(35): 12121-12132.

Gabel, F., D. Bicout, U. Lehnert, M. Tehei, M. Weik and G. Zaccai (2002). Protein
dynamics studied by neutron scattering. Quarterly Reviews of Biophysics 35(4): 327-367.

Gilquin, B., C. Guilbert and D. Perahia (2000). Unfolding of hen egg lysozyme by
molecular dynamics simulations at 300K: Insight into the role of the interdomain
interface. Proteins-Structure Function and Genetics 41(1): 58-74.

Ginsburg, A. and W. R. Carroll (1965). Some specific ion effects on the conformation
and thermal stability of Ribonuclease. Federation Proceedings 24(2P1): 412-&.

164

Gu, H., S. M. Zhang, K. Y. Wong, B. K. Radak, T. Dissanayake, D. L. Kellerman, Q.
Dai, M. Miyagi, V. E. Anderson, D. M. York, J. A. Piccirilli and M. E. Harris (2013).
Experimental and computational analysis of the transition state for ribonuclease A-
catalyzed RNA 2 '-O-transphosphorylation. Proceedings of the National Academy of
Sciences of the United States of America 110(32): 13002-13007.

Gumbart, J. C., B. Roux and C. Chipot (2013). Efficient Determination of Protein-Protein
Standard Binding Free Energies from First Principles. Journal of Chemical Theory and
Computation 9(8): 3789-3798.

Haezebrouck, P., W. Noppe, H. Vandael and I. Hanssens (1992). Hydrophobic
interaction of lysozyme and alpha-lactalbumin from equine milk whey. Biochimica Et
Biophysica Acta 1122(3): 305-310.

Haliloglu, T. and I. Bahar (1999). Structure-based analysis of protein dynamics:
Comparison of theoretical results for hen lysozyme with X-ray diffraction and NMR
relaxation data. Proteins-Structure Function and Genetics 37(4): 654-667.

Hamada, D., S. I. Kidokoro, H. Fukada, K. Takahashi and Y. Goto (1994). Salt-induced
formation of the molten globule state of cytochrome-C studied by isothermal titration
calorimetry. Proceedings of the National Academy of Sciences of the United States of
America 91(22): 10325-10329.

Hamill, A. C. (2008). Photocontrol Of Protein Conformation Through The Use of
Photoresponsive Surfactants, Investigated by Small Angle Neutron Scattering.

Hamill, A. C., S. C. Wang and C. T. Lee (2005). Probing lysozyme conformation with
light reveals a new folding intermediate. Biochemistry 44(46): 15139-15149.

Hamill, A. C., S. C. Wang and C. T. Lee (2007). Solution structure of an amyloid-
forming protein during photoinitiated hexamer-dodecamer transitions revealed through
small-angle neutron scattering. Biochemistry 46(26): 7694-7705.

Hayward, S., A. Kitao and H. J. C. Berendsen (1997). Model-free methods of analyzing
domain motions in proteins from simulation: A comparison of normal mode analysis and
molecular dynamics simulation of lysozyme. Proteins-Structure Function and Genetics
27(3): 425-437.

Hermans, J. and H. A. Scheraga (1961). Structural studies of Ribonuclease V. Reversible
change of configuration. Journal of the American Chemical Society 83(15): 3283-&.

Hooke, S. D., S. J. Eyles, A. Miranker, S. E. Radford, C. V. Robinson and C. M. Dobson
(1995). Cooperative elements in protein folding monitored by electrospray ionization
mass spectrometry. Journal of the American Chemical Society 117(28): 7548-7549.
165

Hooke, S. D., S. E. Radford and C. M. Dobson (1994). Refolding of human lysozyme - a
comparison with the structurally homologous hen lysozyme. Biochemistry 33(19): 5867-
5876.

Housaindokht, M. R., M. R. Bozorgmehr and M. Bahrololoom (2008). Analysis of ligand
binding to proteins using molecular dynamics simulations. Journal of Theoretical
Biology 254(2): 294-300.

Humphrey, W., A. Dalke and K. Schulten (1996). VMD: Visual molecular dynamics.
Journal of Molecular Graphics 14(1): 33-&.

Hunenberger, P. H., A. E. Mark and W. F. Vangunsteren (1995). Computational
approaches to study protein unfolding - hen egg white lysozyme as a case study.
Proteins-Structure Function and Genetics 21(3): 196-213.

Ichiye, T., B. D. Olafson, S. Swaminathan and M. Karplus (1986). Structure and internal
mobility of proteins - a molecular dynamics study of hen egg-white lysozyme.
Biopolymers 25(10): 1909-1937.

Ikeguchi, M., K. Kuwajima, M. Mitani and S. Sugai (1986). Evidence for identity
between the equilibrium unfolding intermediate and a transient folding intermediate - a
comparative study of the folding reactions of alpha-lactalbumin and lysozyme.
Biochemistry 25(22): 6965-6972.

Itzhaki, L. S., P. A. Evans, C. M. Dobson and S. E. Radford (1994). Tertiary interactions
in the folding pathway of hen lysozyme - kinetic studies using fluorescent probes.
Biochemistry 33(17): 5212-5220.

Johnson, L. N. and D. C. Phillips (1965). Structure of some crystalline lysozyme-
inhibitor complexes determined by X-ray analysis at 6 A resolution. Nature 206(4986):
761-&.

Joosten, R. P., T. Beek, E. Krieger, M. L. Hekkelman, R. W. W. Hooft, R. Schneider, C.
Sander and G. Vriend (2011). A series of PDB related databases for everyday needs.
Nucleic Acids Research 39: D411-D419.

Jorgensen, W. L., J. Chandrasekhar, J. D. Madura, R. W. Impey and M. L. Klein (1983).
Comparison of simple potential functions for simulating liquid water. Journal of
Chemical Physics 79(2): 926-935.

Joyner, J. C., W. F. Hodnick, A. S. Cowan, D. Tamuly, R. Boyd and J. A. Cowan (2013).
Antimicrobial metallopeptides with broad nuclease and ribonuclease activity. Chemical
Communications 49(21): 2118-2120.
166

Kabsch, W. and C. Sander (1983). Dictionary of protein secondary structure - pattern -
recognition of hydrogen-bonded and geometrical features. Biopolymers 22(12): 2577-
2637.

Kalyaanamoorthy, S. and Y. P. P. Chen (2013). Ligand release mechanisms and channels
in histone deacetylases. Journal of Computational Chemistry 34(26): 2270-2283.

Karplus, M. and A. Sali (1995). Theoretical studies of protein folding and unfolding.
Current Opinion in Structural Biology 5(1): 58-73.

Kato, S., M. Okamura, N. Shimamoto and H. Utiyama (1981). Spectral evidence for a
rapidly formed structural intermediate in the refolding kinetics of hen egg white
lysozyme. Biochemistry 20(5): 1080-1085.

Kato, S., N. Shimamoto and H. Utiyama (1982). Identification and characterization of the
direct folding process of hen egg white lysozyme. Biochemistry 21(1): 38-43.

Kaur, N., M. Khokhar, V. Jain, P. V. Bharatam, R. Sandhir and R. Tewari (2013).
Identification of Druggable Targets for Acinetobacter baumannii Via Subtractive
Genomics and Plausible Inhibitors for MurA and MurB. Applied Biochemistry and
Biotechnology 171(2): 417-436.

Kazmirski, S. L. and V. Daggett (1998). Non-native interactions in protein folding
intermediates: Molecular dynamics simulations of hen lysozyme. Journal of Molecular
Biology 284(3): 793-806.

Keswani, N., S. Choudhary and N. Kishore (2013). Quantitative aspects of recognition of
the antibiotic drug oxytetracycline by bovine serum albumin: Calorimetric and
spectroscopic studies. Journal of Chemical Thermodynamics 58: 196-205.

Kiefhaber, T. (1995). Kinetic traps in lysozyme folding. Proceedings of the National
Academy of Sciences of the United States of America 92(20): 9029-9033.

Kim, P. S. and R. L. Baldwin (1990). Intermediates in the folding reactions of small
proteins. Annual Review of Biochemistry 59: 631-660.

Kronman, M. J., Cerankow.L and L. G. Holmes (1965). Inter and intramolecular
interactions of alpha-lactalbumin 3 special changes at acid PH. Biochemistry 4(3): 518-&.

Kuhlman, B., J. A. Boice, W. J. Wu, R. Fairman and D. P. Raleigh (1997). Calcium
binding peptides from alpha-lactalbumin: Implications for protein folding and stability.
Biochemistry 36(15): 4607-4615.

167

Kuwajima, K., Y. Hiraoka, M. Ikeguchi and S. Sugai (1985). Comparison of the transient
folding intermediates in lysozyme and alpha-lactalbumin. Biochemistry 24(4): 874-881.

Kuznetsova, L. L., N. S. Zhdan, M. A. Zenkova, V. V. Vlassov and V. N. Sil'nikov
(2004). Artificial ribonucleases 5. Synthesis and ribonuclease activity of tripeptides
composed of amino acids involved in catalytic centers of natural ribonucleases. Russian
Chemical Bulletin 53(2): 455-462.

LadyofHats (2008), Main protein structures levels.
http://en.wikipedia.org/wiki/File:Main_protein_structure_levels_en.svg.

Lee, B. and F. M. Richards (1971). Interpretation of protein structures - estimation of
static accessibility. Journal of Molecular Biology 55(3): 379-&.

Lee, C. T., K. A. Smith and T. A. Hatton (2005). Photocontrol of protein folding: The
interaction of photosensitive surfactants with bovine serum albumin. Biochemistry 44(2):
524-536.

Lee, P. K. J. and Jirgenso.B (1971). Conformational transitions of immunoglobulin
fragment FC(T) incited by alkyl sulfates of various hydrophobic chain lengths.
Biochimica Et Biophysica Acta 229(3): 631-&.

Leo, B., K. Schweimer, P. Rosch, M. J. Hartl and B. M. Wohrl (2012). The solution
structure of the prototype foamy virus RNase H domain indicates an important role of the
basic loop in substrate binding. Retrovirology 9.

Levintha.C How to fold graciously. In Proceedings of a Meeting Held at Allerton House,
Monticello, IL (Debrunner, P., Tsibris, J. C. M. & MuÈ nck, E., eds), pp. 22-24,
University of Illinois Press, Urbana., 1969.

Levintha.C (1968). Are there pathways for protein folding. Journal De Chimie Physique
Et De Physico-Chimie Biologique 65(1): 44-&.

Liepinsh, E. and G. Otting (1997). Organic solvents identify specific ligand binding sites
on protein surfaces. Nature Biotechnology 15(3): 264-268.

Lindahl, E., B. Hess and D. van der Spoel (2001). GROMACS 3.0: a package for
molecular simulation and trajectory analysis. Journal of Molecular Modeling 7(8): 306-
317.

Liu, B., D. Xiang, Y. Long and C. Y. Tong (2013). Real time monitoring of junction
ribonuclease activity of RNase H using chimeric molecular beacons. Analyst 138(11):
3238-3245.

168

Lu, D. N., Z. Liu, Z. X. Liu, M. L. Zhang and P. Ouyang (2005). Molecular simulation of
surfactant-assisted protein refolding. Journal of Chemical Physics 122(13).

Mande, S. C. and M. E. Sobhia (2000). Structural characterization of protein-denaturant
interactions: crystal structures of hen egg-white lysozyme in complex with DMSO and
guanidinium chloride. Protein Engineering 13(2): 133-141.

Mark, A. E. and W. F. Vangunsteren (1992). Simulation of the thermal denaturation of
hen egg white lysozyme - Trapping the molten globule state. Biochemistry 31(34): 7745-
7748.

Matagne, A., S. E. Radford and C. M. Dobson (1997). Fast and slow tracks in lysozyme
folding: Insight into the role of domains in the folding process. Journal of Molecular
Biology 267(5): 1068-1074.

Matthews, C. R. (1993). Pathways of protein folding. Annual Review of Biochemistry 62:
653-683.

Meinhold, L. and J. C. Smith (2007). Protein dynamics from X-ray crystallography:
Anisotropic, global motion in diffuse scattering patterns. Proteins-Structure Function and
Bioinformatics 66(4): 941-953.

Miranker, A., S. E. Radford, M. Karplus and C. M. Dobson (1991). Demonstration by
NMR of folding domains in lysozyme. Nature 349(6310): 633-636.

Miranker, A., C. V. Robinson, S. E. Radford, R. T. Aplin and C. M. Dobson (1993).
Detection of transient protein folding populations by mass spectrometry. Science
262(5135): 896-900.

Miranker, A. D. and C. M. Dobson (1996). Collapse and cooperativity in protein folding.
Current Opinion in Structural Biology 6(1): 31-42.

Mirarefi, P. Photocontrol of protein conformation and enzyme activity in the presence of
light-responsive surfactants.

Mirarefi, P. and C. T. Lee (2010). Photo-induced unfolding and inactivation of bovine
carbonic anhydrase in the presence of a photoresponsive surfactant. Biochimica Et
Biophysica Acta-Proteins and Proteomics 1804(1): 106-114.

Misra, P. P. and N. Kishore (2013). Differential Modulation in Binding of Ketoprofen to
Bovine Serum Albumin in the Presence and Absence of Surfactants: Spectroscopic and
Calorimetric Insights. Chemical Biology & Drug Design 82(1): 81-98.

169

Moraitakis, G. and J. M. Goodfellow (2003). Simulations of human lysozyme: Probing
the conformations triggering amyloidosis. Biophysical Journal 84(4): 2149-2158.

Morozova, L. A., D. T. Haynie, C. Aricomuendel, H. Vandael and C. M. Dobson (1995).
Structural basis of the stability of a lysozyme molten globule. Nature Structural Biology
2(10): 871-875.

MorozovaRoche, L. A., C. C. AricoMuendel, D. T. Haynie, V. I. Emelyanenko, H.
VanDael and C. M. Dobson (1997). Structural characterisation and comparison of the
native and A-states of equine lysozyme. Journal of Molecular Biology 268(5): 903-921.

Muchiri, J. M., D. G. Li, C. Dykes and R. A. Bambara (2013). Efavirenz Stimulates HIV-
1 Reverse Transcriptase RNase H Activity by a Mechanism Involving Increased
Substrate Binding and Secondary Cleavage Activity. Biochemistry 52(29): 4981-4990.

Neria, E., S. Fischer and M. Karplus (1996). Simulation of activation free energies in
molecular systems. Journal of Chemical Physics 105(5): 1902-1921.

Otzen, D. (2011). Protein-surfactant interactions: A tale of many states. Biochimica Et
Biophysica Acta-Proteins and Proteomics 1814(5): 562-591.

Peng, Z. Y., L. C. Wu and P. S. Kim (1995). Local structural preferences in the alpha-
lactalbumin molten globule. Biochemistry 34(10): 3248-3252.

Pepys, M. B., P. N. Hawkins, D. R. Booth, D. M. Vigushin, G. A. Tennent, A. K. Soutar,
N. Totty, O. Nguyen, C. C. F. Blake, C. J. Terry, T. G. Feest, A. M. Zalin and J. J. Hsuan
(1993). Human lysozyme gene mutations cause hereditary systemic amyloidosis. Nature
362(6420): 553-557.

Pike, A. C. W. and K. R. Acharya (1994). A structural basis for the interation of urea
with lysozyme. Protein Science 3(4): 706-710.

Purcell, M., J. F. Neault, H. Malonga, H. Arakawa, R. Carpentier and H. A. Tajmir-Riahi
(2001). Interactions of atrazine and 2,4-D with human serum albumin studied by gel and
capillary electrophoresis, and FTIR spectroscopy. Biochimica Et Biophysica Acta-Protein
Structure and Molecular Enzymology 1548(1): 129-138.

Quesada-Soriano, I., C. Baron, F. Garcia-Maroto, A. M. Aguilera and L. Garcia-Fuentes
(2013). Calorimetric Studies of Ligands Binding to Glutathione S-Transferase from the
Malarial Parasite Plasmodium falciparum. Biochemistry 52(11): 1980-1989.

Radford, S. E. and C. M. Dobson (1995). Insights into protein-folding using physical
techniques - studies of lysozyme and alpha-lactalbumin. Philosophical Transactions of
the Royal Society of London Series B-Biological Sciences 348(1323): 17-25.
170

Radford, S. E., C. M. Dobson and P. A. Evans (1992). The folding of hen lysozyme
involves partially structured intermediates and multiple pathways. Nature 358(6384):
302-307.

Raines, R. T. (1998). Ribonuclease A. Chemical Reviews 98(3): 1045-1065.

Raleigh, D. P., S. F. Betz and W. F. Degrado (1995). A de-novo designed protein mimics
the native-state of natural proteins. Journal of the American Chemical Society 117(28):
7558-7559.

Robertson, A. D. and R. L. Baldwin (1991). Hydrogen-exchange in thermally denatured
Ribonuclease A. Biochemistry 30(41): 9907-9914.

Roder, H. and W. Colon (1997). Kinetic role of early intermediates in protein folding.
Current Opinion in Structural Biology 7(1): 15-28.

Roder, H. a. G. A. E. (1994). in Mechanisms of Protein Folding (Pain, R. H., Ed.) pp 26-
54, Oxford University Press, New York.

Rooman, M., Y. Dehouck, J. M. Kwasigroch, C. Biot and D. Gilis (2002). What is
paradoxical about Levinthal paradox? Journal of Biomolecular Structure & Dynamics
20(3): 327-329.

Rothwarf, D. M. and H. A. Scheraga (1993). Regeneration of bovine pancreatic
ribonuclease - A steady state distribution. Biochemistry 32(10): 2671-2679.

Ryckaert, J. P., G. Ciccotti and H. J. C. Berendsen (1977). Numerical integration of
cartesian equations of motion of a system with constraints - Molecular dynamics of n-
alkanes. Journal of Computational Physics 23(3): 327-341.

Salem, M., Y. Mauguen and T. Prange (2006). On the edge of the denaturation process:
Application of X-ray diffraction to barnase and lysozyme cross-linked crystals with
denaturants in molar concentrations. Biochimica Et Biophysica Acta-Proteins and
Proteomics 1764(5): 903-912.

Sali, A., E. Shakhnovich and M. Karplus (1994). How does a protein fold. Nature
369(6477): 248-251.

Schulz, R., B. Lindner, L. Petridis and J. C. Smith (2009). Scaling of Multimillion-Atom
Biological Molecular Dynamics Simulation on a Petascale Supercomputer. Journal of
Chemical Theory and Computation 5(10): 2798-2808.

171

Schwalbe, H., K. M. Fiebig, M. Buck, J. A. Jones, S. B. Grimshaw, A. Spencer, S. J.
Glaser, L. J. Smith and C. M. Dobson (1997). Structural and dynamical properties of a
denatured protein. Heteronuclear 3D NMR experiments and theoretical simulations of
lysozyme in 8 M urea. Biochemistry 36(29): 8977-8991.

Sheng, J., J. H. Gan and Z. Huang (2013). Structure-based DNA-targeting strategies with
small molecule ligands for drug discovery. Medicinal Research Reviews 33(5): 1119-
1173.

Silberstein, M., S. Dennis, L. Brown, T. Kortvelyesi, K. Clodfelter and S. Vajda (2003).
Identification of substrate binding sites in enzymes by computational solvent mapping.
Journal of Molecular Biology 332(5): 1095-1113.

Simanek, E. E., A. E. Enciso and G. M. Pavan (2013). Computational design principles
for the discovery of bioactive dendrimers: s -triazines and other examples. Expert
Opinion on Drug Discovery 8(9): 1057-1069.

Sinha, S. K. and S. Bandyopadhyay (2011). Differential flexibility of the secondary
structures of lysozyme and the structure and ordering of surrounding water molecules.
Journal of Chemical Physics 134(11).

Smith, L. J., A. E. Mark, C. M. Dobson and W. F. van Gunsteren (1998). Molecular
dynamics simulations of peptide fragments from hen lysozyme: Insight into non-native
protein conformations. Journal of Molecular Biology 280(4): 703-719.

Sosnick, T. R., L. Mayne, R. Hiller and S. W. Englander (1994). The barriers in protein
folding. Nature Structural Biology 1(3): 149-156.

Tanford, C. (1968). Chemical basis for antibody diversity and specificity. Accounts of
Chemical Research 1(6): 161-&.

Tanneeru, K., I. Sahu and L. Guruprasad (2013). Ligand-based drug design for human
endothelin converting enzyme-1 inhibitors. Medicinal Chemistry Research 22(9): 4401-
4409.

Tironi, I. G., R. Sperb, P. E. Smith and W. F. Vangunsteren (1995). A generalized
reaction field method for molecular dynamics simulations. Journal of Chemical Physics
102(13): 5451-5459.

Toth, G., F. Baska, A. Schretner, A. Racz and B. Noszal (2013). Site-specific basicities
regulate molecular recognition in receptor binding: in silico docking of thyroid hormones.
European Biophysics Journal with Biophysics Letters 42(9): 721-730.

172

Tripathi, S. K., R. Muttineni and S. K. Singh (2013). Extra precision docking, free energy
calculation and molecular dynamics simulation studies of CDK2 inhibitors. Journal of
Theoretical Biology 334: 87-100.

Tsong, T. Y., R. P. Hearn, D. P. Wrathall and Sturteva.Jm (1970). A calorimetric study of
thermally induced conformational transitions of Ribonuclease A and certain of its
derivates. Biochemistry 9(13): 2662-&.

Van der Spoel, D., E. Lindahl, B. Hess, G. Groenhof, A. E. Mark and H. J. C. Berendsen
(2005). GROMACS: Fast, flexible, and free. Journal of Computational Chemistry
26(16): 1701-1718.

van Gunsteren, W. F., Billeter, S.R., Eising, A.A., Hunenberger, P.H., Kruger, P., Mark,
A.E., Scott, W.R.P., Tironi, I.G. (1996). Biomolecular Simulation: The Gromos96
Manual and User Guide. Hochschulverlag AG an der ETH, Zurich.

van Gunsteren, W. F. a. B. H. J. C. (1987). Gromos manual. BIOMOS, Biomolecular
Software, Laboratory of Physical Chemistry, University of Groningen, The Netherlands.

Vanderheeren, G., I. Hanssens, W. Meijberg and A. VanAerschot (1996).
Thermodynamic characterization of the partially unfolded state of Ca2+-loaded bovine
alpha-lactalbumin: Evidence that partial unfolding can precede Ca2+ release.
Biochemistry 35(51): 16753-16759.

Wang, S.-C. (2008). Controlling the form-dynamics-function relationship of proteins
with light illumination. Dissertation University of Southern California.

Wang, S. C. and C. T. Lee (2006). Protein secondary structure controlled with light and
photoresponsive surfactants. Journal of Physical Chemistry B 110(32): 16117-16123.

Wang, S. C. and C. T. Lee (2007). Enhanced enzymatic activity through photoreversible
conformational changes. Biochemistry 46(50): 14557-14566.

Wang, S. C., P. Mirarefi, A. Faraone and C. T. Lee (2011). Light-Controlled Protein
Dynamics Observed with Neutron Spin Echo Measurements. Biochemistry 50(38): 8150-
8162.

Wikipedia, http://en.wikipedia.org/wiki/Root_mean_square_fluctuation.

Williams, M. A., J. M. Thornton and J. M. Goodfellow (1997). Modelling protein
unfolding: hen egg-white lysozyme. Protein Engineering 10(8): 895-903.

Wu, L. C., Z. Y. Peng and P. S. Kim (1995). Bipartite structure of the alpha-lactalbumin
molten globule. Nature Structural Biology 2(4): 281-286.
173

Wu, L. C., B. A. Schulman, Z. Y. Peng and P. S. Kim (1996). Disulfide determinants of
calcium-induced packing in alpha-Lactalbumin. Biochemistry 35(3): 859-863.

Xia, Z., D. R. Bell, Y. Shi and P. Y. Ren (2013). RNA 3D Structure Prediction by Using
a Coarse-Grained Model and Experimental Data. Journal of Physical Chemistry B
117(11): 3135-3144.

Yang, A. S., B. Hitz and B. Honig (1996). Free energy determinants of secondary
structure formation .3. beta-Turns and their role in protein folding. Journal of Molecular
Biology 259(4): 873-882.

Zheng, X. H., G. A. Mueller, E. F. DeRose and R. E. London (2012). Metal and ligand
binding to the HIV-RNase H active site are remotely monitored by Ile556. Nucleic Acids
Research 40(20): 10543-10553.

Zhou, R. H., M. Eleftheriou, A. K. Royyuru and B. J. Berne (2007). Destruction of long-
range interactions by a single mutation in lysozyme. Proceedings of the National
Academy of Sciences of the United States of America 104(14): 5824-5829.

Zwanzig, R., A. Szabo and B. Bagchi (1992). Levinthals paradox. Proceedings of the
National Academy of Sciences of the United States of America 89(1): 20-22.

174

Appendices
Lysozyme with trans or cis azoTABs

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

Transition State

Abstract (if available)

Abstract How proteins fold and unfold has been a great focus for decades. Techniques of molecular dynamics simulations provide the atomic insight of protein folding/unfolding. Proteins solvated in water remain well at the native structures under room temperature. Being perturbed by a small amount of photoresponsive surfactants, azoTAB, at room temperature, protein molecules, such as lysozyme, ribonuclease A, and α‐lactalbumin, encounter the conformational changes and partially unfold, especially in the α domain. We conduct molecular dynamics simulation in microseconds and through analysis of the structural properties of protein intermediates as functions of time, we demonstrate that the surfactant‐unfolded intermediates of protein molecules, owning the unfolded α‐domain but the relatively intact β‐domain, although the hydrophobic interaction is higher in the α domain than the β domain. The increased internal dynamics of partially‐unfolded protein molecules induced by azoTABs is potentially contributed to the increase enzymatic activity of protein. Molecular dynamics simulation has offered supporting evidence to better understand experimental phenomena.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

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

PDF

Enhancement of biofuel enzyme activity and kinetics with azoTAB surfactants

PDF

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

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

The interaction of photo-responsive surfactants with biological macromolecules

PDF

Molecular dynamics studies of protein aggregation in unbounded and confined media

PDF

Controlling membrane protein folding with light illumination and catanionic surfactant systems

PDF

Controlling membrane protein folding using photoresponsive surfactant

PDF

Experimental study and atomic simulation of protein adsorption

PDF

Stability and folding rate of proteins and identification of their inhibitors

PDF

Photocontrol of DNA compaction using light-responsive surfactants

PDF

Ice: an impedance spectroscopy and atomistic simulation study

PDF

Nanomaterials under extreme environments: a study of structural and dynamic properties using reactive molecular dynamics simulations

PDF

Simulation and machine learning at exascale

PDF

Theoretical studies of lipid bilayer electroporation using molecular dynamics simulations

PDF

High throughput computational framework for synthesis and accelerated discovery of dielectric polymer materials using polarizable reactive molecular dynamics and graph neural networks

PDF

Shock-induced poration, cholesterol flip-flop and small interfering RNA transfection in a phospholipid membrane: multimillion atom, microsecond molecular dynamics simulations

PDF

Molecular-scale studies of mechanical phenomena at the interface between two solid surfaces: from high performance friction to superlubricity and flash heating

Asset Metadata

Creator Lin, Chih-Ying (author)

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

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Chemical Engineering

Publication Date 02/20/2014

Defense Date 12/10/2013

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

Tag folding,molecular dynamics,OAI-PMH Harvest,protein,simulations,surfactants,unfolding

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

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

Creator Email chihying@usc.edu,chihying2013@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-364713

Unique identifier UC11295813

Identifier etd-LinChihYin-2262.pdf (filename),usctheses-c3-364713 (legacy record id)

Legacy Identifier etd-LinChihYin-2262.pdf

Dmrecord 364713

Document Type Dissertation

Format application/pdf (imt)

Rights Lin, Chih-Ying

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