Computer modeling of human islet amyloid polypeptide

PDF

Download

Open document

Flip pages

Contact Us

Copy asset link

Request this asset

Transcript (if available)

Content
Computer modeling of human islet amyloid polypeptide

By

Bianfei Xuan

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(PHARMACEUTICAL SCIENCES)
August 2016

Copyright 2016 Bianfei Xuan

2

TABLE OF CONTENTS
ACKNOWLEDGEMENT
..............................................................................................................
3

ABSTRACT
...................................................................................................................................
4

CHAPTER 1 INTRODUCTION
....................................................................................................
5

1. Introduction
.........................................................................................................................
5

1.1 Misfolding of proteins
................................................................................................
5

1.2 Amyloid fibrils
...........................................................................................................
5

1.3 Alpha-synuclein
.........................................................................................................
7

1.4 Amyloid-beta
.............................................................................................................
8

1.5 Human Islet Amyloid Polypeptide (hIAPP) Fibrils
...................................................
9

1.6 Overview of the Chapters
........................................................................................
12

CHAPTER 2 COMPUTATIONAL METHODS
.........................................................................
13

1. Introduction
.......................................................................................................................
13

2. Methods
.............................................................................................................................
13

2.1 MFIBRIL
.................................................................................................................
15

2.2 VMD
........................................................................................................................
16

2.3 NAMD
.....................................................................................................................
26

CHAPTER 3 HUMAN ISLET AMYLOID POLYPEPTIDE FIBRIL STRUCTURE
................
29

1. Introduction
.......................................................................................................................
29

2. Methods
.............................................................................................................................
29

3. Results
...............................................................................................................................
29

4. Discussion
.........................................................................................................................
43

REFERENCES
.............................................................................................................................
44

3

ACKNOWLEDGEMENT
I would like to express my great appreciation to my supervisor Dr. Ian S. Haworth for his
support and guidance. Without him, this thesis would not have been possible.
I sincerely appreciate Dr. Ralf Langen, Dr. Clay C. Wang for serving on my master
committee.
Thanks to the members of Dr. Ian Haworth’s laboratory, Asma S. El-Magboub and Dab A.
Brill, for their help and friendship.
Finally, I am deeply grateful to my family and my friends for their love, support and
encouragement in my life.

4

ABSTRACT
It is believed that a variety of human diseases are related to the misfolding of proteins or
peptides. Amyloid fibrils are highly organized aggregates with a cross-β structure and are
generated by misfolding of proteins. These aggregates are also related to disease. For instance,
human islet amyloid polypeptide (hIAPP), α-synuclein, and amyloid-β are associated with Type
II Diabetes, Parkinson’s disease and Alzheimer’s disease when they become misfolded,
respectively. However, these amyloid fibrils are neither soluble nor crystallizable, and so it is
hard to acquire fibril structure information through direct experimental approaches. Therefore,
analysis and determination of the fibril structure, and formation process remain as compelling
questions.
The objectives of this thesis are to investigate the fibril structure of hIAPP by computer
modeling. The fibril model with the most reasonable potential interaction interfaces is
determined, and then the resulting model is used to explain experimental data from election
paramagnetic resonance (EPR) and electron microscopy (EM).
This thesis mainly focuses on producing structures of hIAPP fibrils consistent with
experimental data. Numerous potential models of the hIAPP fibril were built using MFIBRIL
and then refined by equilibration utilizing simulations in the NAMD program. The refined
models were evaluated by comparing their minimized energy. Through analysis of several
favorable models, we identified suitable values of selected parameters, and further narrowed
down the range values of some parameters to find a more exact tendency of the energy changes
related to continuous values. The results suggest that x=0, y=6.5 and z=-1.5 (MFIBRIL) are one
of the most favorable sets of values, but other combinations can also contribute to favorable
structures.

5

CHAPTER 1 INTRODUCTION
1. Introduction
1.1 Misfolding of proteins
Some prevalent human degenerative diseases, such as Type II Diabetes, Parkinson’s disease
and Alzheimer’s disease, are linked to misfolding and aggregation of proteins (Gonzalez and
Soto, 2011). Diseases caused by misfolding of proteins or peptides into β-sheet aggregated
structures are referred to as protein misfolding diseases or conformational diseases (Gonzalez
and Soto, 2011). Misfolding indicates that the proteins cannot achieve their normal 3-
dimensional conformation to function appropriately, but are folded into a strange conformation
and thus lose their original function. The conformation of misfolded proteins goes from the
formation of oligomers that result from intermolecular interactions, to production of
protofilaments and bigger fibrils, and eventually accumulation as amyloid deposits (Soto, 2003;
Soto and Estrada, 2008).
1.2 Amyloid fibrils
Among the misfolded proteins, the majority of misfolding leads to formation of amyloid
fibrils; that is, highly organized fibrillar aggregates with a cross-β structure and other
characteristics (Chiti and Dobson, 2006; Gonzalez and Soto, 2011). Amyloid fibrils form
insoluble fibrous protein aggregates, including both intracellular amyloid-like inclusions and
extracellular deposits (Li et al., 2013). There are many proteins or peptides that form amyloid
deposits. These peptides have various amino acid sequences, but the fibrils have several common
features, such as being long, straight and resistant to protease digestion (Alvarado et al., 2000).
These common features of amyloid fibrils result from their secondary, tertiary and quaternary
structures (Ow and Dunstan, 2014). Due to distinct cross-β arrangement, each amyloid fibril

6

from different proteins has individual specific tertiary structure, but is likely to form fibrils with
similar diameters (Peter and Lansbury, 1999).
Nelson and Eisenberg proposed three types of models, refolding, natively disordered and
gain-of-interaction models (Figure 1.1), to illustrate conversion from soluble proteins to amyloid
fibrils (Nelson and Eisenberg, 2006). The refolding model is the most prevalent one. In this
model, a native protein first unfolds into an intermediate before forming an amyloid fibril
(Nelson and Eisenberg, 2006). If a protein is natively disordered and folds to amyloid fibrils
without the intermediate state, it fits the natively disordered model (Nelson and Eisenberg, 2006).
The gain-of-interaction model is the most complicated one, because this model needs one
amyloid protein monomer to interact with another protein monomer to reach the amyloid-like
state (Nelson and Eisenberg, 2006).
Prion protein, islet amyloid polypeptide, serum amyloid A, α-synuclein, and amyloid-β are
examples of amyloid fibrils, and normally each fibril will contribute to one specific disease. For
example, prion protein leads to Prion disease, amyloid-β causes Alzheimer’s disease and
Huntingtin results in Huntington’s disease (Gonzalez and Soto, 2011). α-synuclein, amyloid-β,
and hIAPP are three common amyloid fibrils, and the following sections are focused on these
proteins.

7

Figure 1.1 A representation showing three models of conversion from native proteins to amyloid fibrils.
The refolding model demonstrates the process in which a native protein first unfolds to an intermediate
and then forms an amyloid fibril. The natively disordered model requires that native proteins have little
secondary structure, and the gain-of-interaction model shows the interaction with other proteins to form
amyloid fibrils. (Nelson and Eisenberg, 2006)
1.3 Alpha-synuclein
α-synuclein is a 14 kDa insoluble protein that is the major component of Lewy bodies. α-
synuclein is expressed in brain at presynaptic terminals in a large quantity (Iwai et al., 1995), but
also exists in biological fluids such as blood and plasma (El-Agnaf et al., 2006). Moreover, α-
synuclein comprises more than 1% of the total proteins in soluble cytosolic brain fractions (Iwai
et al., 1995). Therefore, α-synuclein plays a crucial role in neural function, including modulating
the stability of the neuronal membrane and membrane trafficking (Braak et al., 2003).
α-synuclein contains 140 amino acids and might have a greater possibility of abnormal
aggregation due to posttranslational modification for all sites (Levitan et al., 2011). α-synuclein
can be divided into two parts: N-terminal and C-terminal. The N-terminal part, starting from

8

residue 1 to residue 103, plays a role in lipid binding, which is based on the ability of
amphipathic apolipoprotein helical motifs to bind to phospholipids and form a helical
conformation (Eliezer et al., 2001). The N-terminal region is believed to have a hydrophobic
region (non-amyloid-β component) that can result in formation of amyloid fibrils (Eliezer et al.,
2001). This hydrophobic region allows α-synuclein to change from a random coil to a β-sheet
structure that is a prerequisite for formation of amyloid fibrils (Eliezer et al., 2001). The C-
terminal part (residues 103-140) is unfolded and negatively charged because of many acidic
residues within the sequence (Ritchie and Thomas, 2012), and is related to regulating interaction
between α-synuclein and other proteins (Eliezer et al., 2001). However, there is a C-terminal
truncated form of α-synuclein that has the capability of forming aggregates faster than the
normal full-length protein in vitro (Ritchie and Thomas, 2012). Therefore, the C-terminal region
could also contribute to Lewy body formation.
1.4 Amyloid-beta
Amyloid-β, a small protein (36-43 amino acids) seen in mammals, is expressed from a gene
located in chromosome 21 in humans (Selkoe, 1994). Under physiological conditions, amyloid-β
will first form cytotoxic intermediates that can lead to neuronal disruptions, such as inhibiting
long-term potentiation, losing glutamate receptors and disrupting calcium ion uptake (Hamley,
2012), which is also the cause for some symptoms found in the Alzheimer’s disease (Paola et al.,
2000), and later form the fibrils (Hamilton-Brown et al., 2008).
Amyloid-β is regarded as a natively unfolded protein on account of the little secondary
structure shown in its original conformation (Uversky, 2002). When misfolding happens,
amyloid-β first forms an α-helical rich intermediate via folding of the native random-coil rich

9

conformation, later goes to a β sheet rich amyloid monomer and lastly self-assembles into fibrils
(Kimov and Thirumalai, 2003).
Amyloid-β originates from the membrane-bound amyloid precursor protein, which has two
possible cleavage sites in human cells (Hamley, 2012; Kopan and Ilagen, 2004). At the first site,
α-secretase plays a role in cleavage to produce the peptides α-amyloid precursor proteins and
C83 (Kopan and Ilagen, 2004). The second site is acted on by β-secretase to produce β-amyloid
precursor proteins and C99 (Kopan and Ilagen, 2004). After cleavage by α-secretase or β-
secretase, γ-secretase, a large multiprotein complex, works on both possible amyloids to produce
peptide P3 for αAPP and Aβ1-40 or Aβ1-42 for βAPPs (Selkoe, 1994; Kopan and Ilagen, 2004).
1.5 Human Islet Amyloid Polypeptide (hIAPP) Fibrils
IAPP is a 37-residue peptide that is found in all mammals and shares similar sequences in
different species. Human IAPP (hIAPP) is co-produced and co-secreted with insulin and the ratio
of IAPP: insulin is five times higher in Type II Diabetes than in healthy individuals: 1:100
compared to 1:20 (Caillon et al., 2015). Therefore, hIAPP is important in Type II Diabetes.
Initially, hIAPP was named insulinoma amyloid peptide (Westermark et al., 1986), then
diabetes-associated peptide (Cooper et al., 1987), and eventually islet amyloid polypeptide
(Westermark et al., 1987). It is believed that amyloid deposits of hIAPP lead to the death of
insulin islet β-cells and thus result in the development of Type II Diabetes (Hoppener et al.,
2000).
hIAPP appears in various states (monomer, oligomer and fibril) and all of these states show
distinct structures (Caillon et al., 2015). In solution, circular dichroism data indicated that the
monomeric hIAPP is natively unfolded, expect for a rigid ring structure formed by the disulfide

10

bridge between Cys2 and Cys7 (Caillon et al., 2015). The second state oligomer is unstable and
persists for a shorter time compare to other states, therefore its structure information is limited.
High-resolution microscopy (electron and atomic-force) and spectroscopy techniques (NMR) are
two common techniques used for detecting oligomeric species. However, NMR has a
disadvantage that it generally lacks the required time resolution to get a snapshot of oligomers
(Caillon et al., 2015).
Fibrils of hIAPP are stable and their structures have been more comprehensively described.
Electron microscopy (EM) revealed two major forms: ribbon-like fibrils and left-handed twisted
fibrils (Goldsbury et al., 1977). The ribbon-like fibrils are formed when oligomers associate
with each other laterally in long and striated ribbon-like strands (Caillon et al., 2015). The left-
handed twisted fibrils exist because some oligomers organize themselves as helical fibrils of
variable width, and thus present periodical twists (Caillon et al., 2015).
In this study, our target is the left-handed twisted fibrils. According to the EM observation, a
twisted hIAPP fibril consists of 1-4 protofilaments with a 250 or 500 Å pitch (Goldsbury et al.
1977). Twisted fibrils formed by 2 protofilaments are most commonly observed, therefore this
model is used in our study. The β-sheet structures were revealed by Fourier transform infrared
spectroscopy and circular dichroism (Higham et al., 2000; Kayed et al., 1999);), and X-ray and
electron diffraction data showed that β-strands are perpendicular to fibril axis and the hydrogen
bonding spacing between successive β-strands is about 4.7 Å (Sumner Makin and Sperpell,
2004). Depending on the electron paramagnetic resonance spectroscopy, the arrangement of β-
strands is parallel and in-register (in register: the same residues in different molecules stack on
top of each other) and the N-terminal region is slightly less ordered (Jayasinghe and Langen,
2004).

11

The functions and properties of hIAPP are directly related to its residues, and previous
studies have identified the roles of several residues. A great many fragments of hIAPP have the
capacity of self-assembling into amyloid or non-amyloid oligomeric aggregates (Ilitchev et al.,
2016). Yoon et al. set up a method called CSSP2 for predicting contact-dependent secondary
structure propensity, and the predictions indicated that the IAPP region residues 13-29 showed
the highest β-sheet propensity (Yoon et al., 2007). For the region 13-29, segment 20-29 is
regarded as the central amyloidogenic module of the polypeptide (Moriarty and Raleigh, 1999).
Early work regarded the segment 20-29 as a region for amyloidogenicity, but later researchers
found this segment should be considered as a key contributor to hIAPP aggregation (Buchanan et
al., 2013; Westermark et al., 1990; Tenidis, et al., 2000). Moreover, several studies indicated that
fragment 10-19 is crucial for IAPP fibrillogenesis (Tracz et al., 2004; Scrocchi et al., 2003;
Gilead and Gazit, 2008). Specially, the ionization state of histidine at position 18 was found to
have the capacity of substantially influencing the rate of assembly and the morphology of the
hIAPP amyloid fibrils (Abedini and Raleigh, 2005).
The N-terminal, which contains disulfide and plays an important role in in vivo function, is
also an interesting segment for study, and may help with in vitro aggregation (Ilitchev et al.,
2016). There are conflicting reports on the role of the disulfide bond. Cope et al. claimed that the
segment 1-8 of hIAPP could lead to non-amyloid aggregation into large macroscopic structures,
and the removal of the disulfide bond clearly prohibits the aggregation process (Cope et al.,
2013). However, Ilitchev et al. recently published a paper that held an opposite view, indicating
that cleavage of the disulfide bond would increase the possibility of aggregation, and its presence
may inhibit the self-aggregation of the segment because of a reduction of inter-peptide hydrogen
bonding (Ilitchev et al., 2016).

12

As for interfacial residue interactions, Zhao et al. modeled three distinct interfaces formed
by the C-terminal β-sheet and C-terminal β-sheet (CC), N-terminal β-sheet and N-terminal β-
sheet (NN), and C-terminal β-sheet and N-terminal β-sheet (CN) to explore their intermolecular
interactions. The CC interface was dominated by polar interactions, the NN interface mainly
depended on the hydrophobic interactions, and the CN interface combined the previous two
interactions (Zhao et al., 2011).
1.6 Overview of the Chapters
This thesis is separated into two chapters, in addition to the introduction (Chapter 1).
Chapter 2 describes computational methods for the applications of the MFIBRIL, VMD and
NAMD programs in detail. We mainly focus on the basic fibril model named model_22 to
identify the most favorable interaction positions. MFIBRIL is a program established in our lab,
which can generate fibril models with more than one protofilament. Model_22 is the most
reasonable basic model selected from the previous study, and further detailed models are built
based on model_22, and then refined by equilibration via molecular dynamics simulations in the
NAMD software.
For Chapter 3, various fibril structures were built by the means of changing selected
parameters (x, y, z, pitch and n) related to fibril structures, and refined using the computational
methods in Chapter 2. In this chapter, energy is the predominant standard used to identify
favorable models. According to the energy, several favorable models were selected and these can
be further examined using other programs in future work.

13

CHAPTER 2 COMPUTATIONAL METHODS
1. Introduction
In this chapter, the method of computational calculation is illustrated in detail, including the
applications of the MFIBRIL (Li et al., 2012), VMD and NAMD (Phillips et al., 2005) programs.
This introduction of methods contains three main parts: building of hIAPP fibril structures by
MFIBRIL and energy minimization of fibril models by VMD and NAMD. Model_22 was
selected as the original basic model, and MIBRIL was used to build more detailed fibril model
with different potential interaction positions. VMD and NMAD were utilized to analyze the
energy of the models, including electrostatic and van der Waals interaction (Humphrey et al.,
1996), to get the minimized energy for each model.
2. Methods
In a previous study, four basic fibril models, model_11, model_12, model_22 and model_LL,
were constructed by MFIBRIL (Li et al., 2013). These shared some common features. Each basic
model has two identical protofilaments that consist of 35 peptides, and the obvious difference
among the models is the interaction interface. The two protofilaments (named protofilament A
and protofilament B for illustration) interact with each other in four ways: the first (N-terminal)
β-strands of protofilament A and that of protofilament B (model_11), the first β-strand in
protofilament A and the second (C-terminal) β-strand in protofilament B (model_12), the second
β-strands of protofilament A and B (model_22), and the inter-strand (loop) regions of two
peptides (model_LL) (Figure 2.1) (Li et al., 2013).

14

Figure 2.1 Diagrams show four fibril models. Two peptides are in the same layer in starting structures of
fibril model_11 (A), model_12 (B), model_22 (C) and model_LL (D). Residues 12-19 and 31-36 are
displayed as β-strands (Li et al., 2013). The blue peptide and green one show protofilament A and
protofilament B, separately. This figure was generated by Yiyu Li in her Ph.D. thesis.

In the earlier study, model_22 was thought to be the most promising basic model because of
its appropriate interaction between two peptides and the match with experimental data (aromatic
ring-to-ring distance) (Li et al., 2013). Therefore, we selected model_22 as the original basic

15

model for fibril construction, and a lot more models derived from model_22 were made using
MFIBRIL. These derived models were constructed by adjustment of the separation distance and
the shift between the two interacting peptides, and selection of 250-Å and 500-Å pitches. Then,
energy analysis by VMD and NAMD was used to identify more favorable fibril models from all
the derived models. To better illustrate the whole process, an example is used to go over the
process. The whole process consists of three primary parts using MFIBRIL, VMD and NAMD.
2.1 MFIBRIL
In MFIBRIL, model_22 is used as the original model and several parameters are changed to
adjust the interaction interfaces. Figure 2.2 (A) shows the terminal window for LINUX system
and basic commands to construct hIAPP fibrils. Command pwd indicates the current location,
and command cd is used to change to the target directory. MFIBRIL.bat is the major function for
the MFIBIRL program, and Figure 2.2 (B) demonstrates all the related parameters for
constructing fibril structures.
All the meanings of parameters in MFIBRIL can be found in MFIBRIL_Manual.docx. In
previous studies, Li et al. assumed that hIAPP protofilaments might be limited to staggers of
about 2 and 4 peptides and pitches should be within the range from 250 Å to 500 Å (Li et al.,
2013). Fibrils with 2 interacting peptides are the most common model, and with the pre-test with
possible parameters, we decided to concentrate on three parameters: shiftc, pitch and nm. Shiftc
is an important parameter, containing three sub-parameters x, y, and z, for our study to change
the coordinate peptide center related to the origin (0, 0, -1 for model_22). Pitch is used to set the
distance between repeated peptides of hIAPP. In the pitch section, if the number is larger, the
hIAPP fibrils will become straighter. For example, Figure 2.2 (D) shows two similar structures
only with a pitch difference, and the structure with pitch 500 is obviously straighter than the

16

structure with pitch 250. The last parameter nm refers to the number of layers in the model, and
we have primarily constructed four kinds of layers: 1-layer, 5-layer, 35-layer and 100-layer.
After setting correct values for the above parameters and running the last command, a new
fibril called x0y7z-1p250L5.pdb (Figure 2.2 (E)) is constructed. Its name contains the coordinate,
pitch and number of layers. PyMOL software can be used to display its structure (Figure 2.2 (F)).
2. 2 VMD
The above pdb file containing the structure information does not work directly in the VMD
software. Therefore, a code for reformatting the original pdb file is required. Figure 2.3
demonstrates the codes needed for this process. All that is needed is typing the right names of
fibrils in the codes to get the corresponding revised versions. Figure 2.3 (A) is prepared for
hIAPP fibrils with less than 35 layers, and Figure 2.3 (B) shows the codes needed for hIAPP
fibrils with longer layers, starting from 35 layers to 100 layers.
A

17

B

C

D

Pitch 500

Pitch 250

18

E

F

Figure 2.2 Fibril construction process by using MFIBRIL program. A. Commands for opening MFIBRIL
B. MFIBRIL command C. Meanings of some key parameters in the MFIBRIL_Manual.docx D.
Illustration of fibrils with different pitches. E. Integral commands for fibril building F. Structure display
in the PyMOL

19

A

B

20

Figure 2.3 Python codes for documents (.pdf) formatting. A Codes for structures less than 35 layers
B Codes for structures more than 35 layers (included)

Regarding the difference between the original version (Figure 2.4 (A)) and revised version
(Figure 2.4 (B)), there is an extra line, either filled with ‘A’ or ‘B’, in the revised version. In
Figure 2.4 (C), the extra line was changed from ‘A’ to ‘B’, and then ‘B’ was changed back to ‘A’,
demonstrating the function of these codes. If the original version is directly applied in the VMD
and NAMD programs, it could not work normally owing to failure to distinguish two stacks of
peptides. For each layer of the hIAPP fibrils, there is two peptides interacting with each other,
and ‘A’ and ‘B’ are used to mark two peptides. Each peptide of the model has 25 amino acids,
containing 352 atoms, so when the atom number reaches 352, the first peptide has all its atoms,
and ‘B’ should be used to mark the second peptide. Each layer (one cycle) means 50 amino acids,
so when one cycle is finished, a new cycle should start to distinguish the following peptides. In
this way, the revised version could separate all the peptides in two stacks and help VMD and
NAMD to understand the actual structure of the input fibrils.

21

A
B
C
D

Figure 2.4 Difference between the original version and revised version of documents of hIAPP fibrils A
Original version (.pdf) of hIAPP fibrils B Revised version (.pdf) for the same hIAPP fibril C Revised
version (.pdf) of the same hIAPP fibril - (‘A’ was changed to ‘B’) D Revised version (.pdf) of the same
hIAPP fibril - (‘B’ was changed to ‘A’)

22

Running VMD is a prerequisite for further NAMD running by creating all the required
structure files. Figure 2.5 (A) demonstrates the primary windows for the VMD program: VMD
Main window and VMD OpenGL Display window. The VMD Main window contains all the
functions for VMD, and OpenGL Display window is just for displaying the structures.
The first step is to locate the directory containing the required structures (folder
‘minimization-y’: revised structures have been copied to this folder) by entering codes (‘cd’
means change directory) in VMD TkConsole window (Extensions è VMD TkConsole). This
step is not necessary in this study, but it can save some time for finding the required structures.
Then, Molecule File Browser window (Figure 2.5 (B)) is used for inputting the required
structures into VMD (File è New Molecule è Molecule File Brower: Browse to find the target
structure file è Load). If the first step is missed, then browsing the target folder will require
more time. Figure 2.5 (C) showed the name, atoms and other information of the input file in the
VMD Main window, and the corresponding structure is displayed in the OpenGL Display
window.
The following steps are key operations for obtaining the required files (e.g. .psf file) for
NAMD running. Auto PSF builder (Figure 2.5 (D): Extensions è Modeling è Auto PSF
builder) needs four steps to produce new files.
Step one (Input and Output Files) is clicking the button ‘Load input files’ to load files, and
take care of the ‘Output basename’ at the top to name the output files. Step Two (Selections to
include in PSF/PDB) is selecting the required kind of PSF/PDB files, either protein or nucleic
acid, but in this case, only protein exists in the input file, so just keeping the choice ‘Everything’
and then click the button ‘Guess and spilt chains using current selections’ is fine. At this time, a

23

dialogue ‘Do you want to specify the location of the original coordinate file?’ appears and the
answer should be ‘No’.
Step three (Segments identified) is the reason why revised versions of the original pdb files
are needed. VMD does not have the ability to identify two stacks of fibrils separately, so a pre-
marked input file is necessary. After clicking the button ‘ Create chains’, step four (Patches) is
just simply clicking the button ‘Apply patches and finish PSF/PDB’. Because no patches were
automatically assigned to the input molecule, VMD will send a message to claim that adding
patches and regenerating these files are still available. After this message, a message is sent
informing the user that the structure is complete and in new files. In the VMD Main window
(Figure 2.5 (E)), a second line shows the new structure has been built. Also, if you browse the
target folder, you could find three new files (Figure 2.5 (E)) lying behind the input files.
A

24

B

25

C

D

26

E

Figure 2.5 Integral procedures for VMD operation A Primary windows for VMD. B Input molecule file
in VMD. C A diagram showing the information of inputted structures. D Process for building PSF files. E
Diagrams indicating three new files have been built.

2. 3 NAMD
In order to get the refined structures of built fibril models, NAMD is used for refining
structures by energy minimization. For using the NAMD software, some files are required to
enable its running. Figure 2.6 (A) shows some of the required files: autopsf.psf, autopsf.pdb,

27

parameter file (par_all27_prot_na_prm1.txt). Autopsf.psf file, which includes the structure
information for the input file, together with autopsf. pdb file, which contains the initial structure
coordinates, are the integral files. Curc-1.conf is an important configuration file, but dealing with
this file simply requires changing five names: the autopsf.psf, the autopsf.pdb, outputname,
restartname and dcd file. The latter three files (Figure 2.6 (C)) and log file (whose name is
assigned in Figure 2.6 (B)) are four results produced by the NAMD run.
A

28

B

C

Figure 2.6 NAMD run A Content of curc.conf file B code for NAMD running C Three outputs for
NAMD running

The NAMD running procedure can be divided into two steps. The first step is adjusting the
five names mentioned above, according to the name of the input file. The other step is running
the code shown in Figure 2.6 (B). Then, the log file is in the folder ‘minimization-y’ and the
other three files (min.coor, min.dcd, min.vel) can be found in the folder ‘output’, as designed in
the curc.conf file. The log file contains the minimized energy of each fibril model, and the
min.coor file is the minimized structure of the corresponding original fibril model.

29

CHAPTER 3 HUMAN ISLET AMYLOID POLYPEPTIDE FIBRIL
STRUCTURE
1. Introduction
In this chapter, MFIBRIL, VMD and NAMD were combined together to investigate the
hIAPP fibril structures using the methods in Chapter 2. Many detailed fibril models with
different potential interaction positions were built using MFIBRIL, and then the energies of these
models were analyzed after minimization using VMD and NAMD. These results were shown in
tables and figures for illustration and analysis.
2. Methods
All the detailed parameters have been described in Chapter 2. Several parameters (x, y, z,
pitch and n) were selected to build different fibril structures. The values are shown in the Results
section.
3. Results
Tables 3.1 to 3.4 give the primary results of this study. These results identify the impact of
different parameters on fibril structures by comparing the energy for refined structures. These
four figures share three common elements: the value of x is 0, the values of y ranges from -10 to
10 (integer), and all have 5, 35, 100 layers. Keeping x at 0 demonstrates the influence for y, z
and pitches. x refers to the position inside or outside of the plane, and every single peptide in the
model is not plain, so it is hard to identify the impact of x at the very beginning. y refers to the
positions up and down the plane; therefore a wide range of models can be tested from peptide A
at the top to peptide B at the top (-10 and 10, which indicate that the two peptides almost have no
interaction in the center). z refers to the distance between two stacks of peptides. Theoretically, if
two stacks get closer, they will have more possibilities to interact with each other. However, if

30

two stacks are too close, they will overlap with each other and thus fail to have low energy.
In Table 3.1, for example, there are 13 columns, containing x value, y value, z value, pitch
value, 5-layer, energy of 5-layer, 35-layer, energy of 35-layer, energy of 35-layer/energy of 5-
layer, 100-layer, energy of 100-layer, energy of 100-layer/energy of 5-layer and energy of 100-
layer/energy of 35-layer. The meaning of the ratios is as follows.
Column 9, Column12 and Column 13 have the same function. If we treat 5-layer as a single
unit and assume the energy of 35-layer is 5 times of that of 5-layer, then the ratio between them
should be about 7. If the value is close to 7, we could assume the 5-layer structure is a good
sample to estimate the energy of structures with more layers. Comparing all four Column 9
values, fibrils with pitch 250 have higher values than 7, while values of pitch 500 are around 7 or
below 7. Structures with 100 layers were tested to ensure the model is long enough to give
reasonable results. If we assume 35 layers as a single unit for estimation, the values of Column
13 are close to 2.85, as shown in Table 3.1, Therefore, the 35-layer fibril is suitable for
estimating the energy of structures with larger layers in this parameter setting.
Table 3.1 Energies for (x = 0, y (-10 to 10), z = -1, pitch 250) refined fibrils with 5, 35, 100 layers
X
Y
Z
Pitch
Layer
Energy
Layer
Energy
35/5
Layer
Energy
100/5

100/3
5

0
-‐10
-‐1
250
5
-‐1902

35
-‐13735

7.22

100
-‐39274

20.65
2.86

0
-‐9
-‐1
250
5
-‐1893

35
-‐14249

7.53

100
-‐40352

21.31
2.83

0
-‐8
-‐1
250
5
-‐1983

35
-‐14493

7.31

100
-‐40105

20.23
2.77

0
-‐7
-‐1
250
5
-‐1874

35
-‐14448

7.71

100
-‐40992

21.88
2.84

0
-‐6
-‐1
250
5
-‐1945

35
-‐14561

7.49

100
-‐41633

21.41
2.86

0
-‐5
-‐1
250
5
-‐2116

35
-‐15132

7.15

100
-‐45565

21.53
3.01

0
-‐4
-‐1
250
5
-‐1940

35
-‐14978

7.72

100
-‐40548

20.90
2.71

0
-‐3
-‐1
250
5
-‐1876

35
-‐14603

7.78

100
-‐39602

21.11
2.71

0
-‐2
-‐1
250
5
-‐1969

35
-‐14849

7.54

100
-‐40449

20.55
2.72

0
-‐1
-‐1
250
5
-‐1923

35
-‐13770

7.16

100
-‐38718

20.14
2.81

0
0
-‐1
250
5
-‐2011

35
-‐15076

7.50

100
-‐43426

21.60
2.88

0
1
-‐1
250
5
-‐2272

35
-‐16354

7.20

100
-‐45718

20.12
2.80

0
2
-‐1
250
5
-‐2185

35
-‐16684

7.64

100
-‐45855

20.99
2.75

31

0
3
-‐1
250
5
-‐2147

35
-‐17152

7.99

100
-‐49291

22.96
2.87

0
4
-‐1
250
5
-‐2184

35
-‐17202

7.88

100
-‐49245

22.55
2.86

0
5
-‐1
250
5
-‐2215

35
-‐16866

7.61

100
-‐47484

21.43
2.82

0
6
-‐1
250
5
-‐2287

35
-‐17820

7.79

100
-‐50867

22.24
2.85

0
7
-‐1
250
5
-‐2205

35
-‐20407

9.26

100
-‐56403

25.58
2.76

0
8
-‐1
250
5
-‐2208

35
-‐17860

8.09

100
-‐49946

22.62
2.80

0
9
-‐1
250
5
-‐2039

35
-‐20159

9.89

100
-‐43420

21.29
2.15

0
10
-‐1
250
5
-‐2003

35
-‐19151

9.56

100
-‐54829

27.37
2.86

Table 3.2 Energies for (x = 0, y (-10 to 10), z = -2, pitch 250) refined fibrils with 5, 35, 100 layers
X
Y
Z
Pitch
Layer
Energy
Layer
Energy
35/5
Layer
Energy
100/5

100/3
5

0
-‐10
-‐2
250
5
-‐1886

35
-‐13131

6.96

100
-‐37215

19.73

2.83

0
-‐9
-‐2
250
5
-‐2023

35
-‐14107

6.97

100
-‐39242

19.39

2.78

0
-‐8
-‐2
250
5
-‐1975

35
-‐14476

7.33

100
-‐41320

20.92

2.85

0
-‐7
-‐2
250
5
-‐1961

35
-‐14092

7.18

100
-‐40421

20.61

2.87

0
-‐6
-‐2
250
5
-‐1832

35
-‐13406

7.32

100
-‐38449

20.99

2.87

0
-‐5
-‐2
250
5
-‐1933

35
-‐13424

6.95

100
-‐38394

19.87

2.86

0
-‐4
-‐2
250
5
-‐1812

35
-‐13247

7.31

100
-‐39003

21.52

2.94

0
-‐3
-‐2
250
5
-‐1823

35
-‐13228

7.26

100
-‐37828

20.75

2.86

0
-‐2
-‐2
250
5
-‐2032

35
-‐14728

7.25

100
-‐43789

21.55

2.97

0
-‐1
-‐2
250
5
-‐1977

35
-‐16063

8.12

100
-‐45596

23.06

2.84

0
0
-‐2
250
5
-‐2054

35
-‐16049

7.81

100
-‐45874

22.33

2.86

0
1
-‐2
250
5
-‐2152

35
-‐17323

8.05

100
-‐50415

23.43

2.91

0
2
-‐2
250
5
-‐2132

35
-‐16253

7.62

100
-‐50511

23.70

3.11

0
3
-‐2
250
5
-‐2111

35
-‐17875

8.47

100
-‐49750

23.57

2.78

0
4
-‐2
250
5
-‐2115

35
-‐17872

8.45

100
-‐52436

24.80

2.93

0
5
-‐2
250
5
-‐2174

35
-‐18016

8.29

100
-‐54596

25.11

3.03

0
6
-‐2
250
5
-‐2234

35
-‐20177

9.03

100
-‐56272

25.19

2.79

0
7
-‐2
250
5
-‐2186

35
-‐19931

9.12

100
-‐58695

26.85

2.94

0
8
-‐2
250
5
-‐2190

35
-‐20153

9.20

100
-‐47675

21.77

2.37

0
9
-‐2
250
5
-‐1931

35
-‐18107

9.38

100
-‐42619

22.08

2.35

0
10
-‐2
250
5
-‐2017

35
-‐19560

9.70

100
-‐53569

26.56

2.74

Table 3.3 Energies for (x = 0, y (-10 to 10), z = -1, pitch 500) refined fibrils with 5, 35, 100 layers
X
Y
Z
Pitch
Layer
Energy
Layer
Energy
35/5
Layer
Energy
100/5

100/3
5

0
-‐10
-‐1
500
5
-‐1882

35
-‐12952

6.88

100
-‐36916

19.62

2.85

0
-‐9
-‐1
500
5
-‐1965

35
-‐13218

6.73

100
-‐37224

18.94

2.82

0
-‐8
-‐1
500
5
-‐2010

35
-‐13070

6.50

100
-‐36237

18.03

2.77

0
-‐7
-‐1
500
5
-‐1881

35
-‐13202

7.02

100
-‐37071

19.71

2.81

0
-‐6
-‐1
500
5
-‐1962

35
-‐13344

6.80

100
-‐37398

19.06

2.80

32

0
-‐5
-‐1
500
5
-‐2053

35
-‐14044

6.84

100
-‐40934

19.94

2.91

0
-‐4
-‐1
500
5
-‐1907

35
-‐12796

6.71

100
-‐30048

15.76

2.35

0
-‐3
-‐1
500
5
-‐1830

35
-‐12730

6.96

100
-‐33461

18.28

2.63

0
-‐2
-‐1
500
5
-‐2012

35
-‐14230

7.07

100
-‐34850

17.32

2.45

0
-‐1
-‐1
500
5
-‐1875

35
-‐12502

6.67

100
-‐35581

18.97

2.85

0
0
-‐1
500
5
-‐1895

35
-‐12510

6.60

100
-‐33915

17.90

2.71

0
1
-‐1
500
5
-‐1820

35
-‐12460

6.85

100
-‐32033

17.60

2.57

0
2
-‐1
500
5
-‐1704

35
-‐12254

7.19

100
-‐30797

18.08

2.51

0
3
-‐1
500
5
-‐1817

35
-‐13240

7.29

100
-‐35241

19.40

2.66

0
4
-‐1
500
5
-‐1982

35
-‐13401

6.76

100
-‐38145

19.25

2.85

0
5
-‐1
500
5
-‐1990

35
-‐13853

6.96

100
-‐32986

16.58

2.38

0
6
-‐1
500
5
-‐2018

35
-‐14276

7.07

100
-‐36548

18.11

2.56

0
7
-‐1
500
5
-‐2064

35
-‐14791

7.17

100
-‐41261

19.99

2.79

0
8
-‐1
500
5
-‐1899

35
-‐14630

7.70

100
-‐34532

18.18

2.36

0
9
-‐1
500
5
-‐1916

35
-‐14238

7.43

100
-‐34433

17.97

2.42

0
10
-‐1
500
5
-‐1802

35
-‐15613

8.67

100
-‐28992

16.09

1.86

Table 3.4 Energies for (x = 0, y (-10 to 10), z = -1, pitch 500) refined fibrils with 5, 35, 100 layers
X
Y
Z
Pitch
Layer
Energy
Layer
Energy
35/5
Layer
Energy
100/5

100/3
5

0
-‐10
-‐2
500
5
-‐1993

35
-‐12996

6.52

100
-‐35875

18.00

2.76

0
-‐9
-‐2
500
5
-‐2013

35
-‐13127

6.52

100
-‐36991

18.38

2.82

0
-‐8
-‐2
500
5
-‐1958

35
-‐13044

6.66

100
-‐36278

18.53

2.78

0
-‐7
-‐2
500
5
-‐1907

35
-‐13065

6.85

100
-‐37527

19.68

2.87

0
-‐6
-‐2
500
5
-‐1867

35
-‐13788

7.39

100
-‐39540

21.18

2.87

0
-‐5
-‐2
500
5
-‐2042

35
-‐14607

7.15

100
-‐41205

20.18

2.82

0
-‐4
-‐2
500
5
-‐1885

35
-‐13286

7.05

100
-‐36822

19.53

2.77

0
-‐3
-‐2
500
5
-‐1885

35
-‐13335

7.07

100
-‐36361

19.29

2.73

0
-‐2
-‐2
500
5
-‐1846

35
-‐13071

7.08

100
-‐36514

19.78

2.79

0
-‐1
-‐2
500
5
-‐1909

35
-‐12217

6.40

100
-‐32997

17.28

2.70

0
0
-‐2
500
5
-‐1649

35
-‐12704

7.70

100
-‐35503

21.53

2.79

0
1
-‐2
500
5
-‐1772

35
-‐12797

7.22

100
-‐34146

19.27

2.67

0
2
-‐2
500
5
-‐1794

35
-‐11589

6.46

100
-‐32003

17.84

2.76

0
3
-‐2
500
5
-‐1941

35
-‐12826

6.61

100
-‐33991

17.51

2.65

0
4
-‐2
500
5
-‐1847

35
-‐12353

6.69

100
-‐33514

18.15

2.71

0
5
-‐2
500
5
-‐2019

35
-‐13790

6.83

100
-‐38305

18.97

2.78

0
6
-‐2
500
5
-‐1967

35
-‐13901

7.07

100
-‐36036

18.32

2.59

0
7
-‐2
500
5
-‐2014

35
-‐14622

7.26

100
-‐37153

18.45

2.54

0
8
-‐2
500
5
-‐2036

35
-‐13472

6.62

100
-‐34074

16.73

2.53

0
9
-‐2
500
5
-‐1869

35
-‐14752

7.89

100
-‐30037

16.07

2.04

0
10
-‐2
500
5
-‐1907

35
-‐14267

7.48

100
-‐31940

16.75

2.24

To get a clear understanding of the importance of various parameters, these four tables are

33

recombined into three tables in terms of layers in Tables 3.5, 3.6 and 3.7. Corresponding figures
are shown for illustration. For 5 layers, structures with y ranging from -10 to 0 (left side) have
similar energy, while structures with y ranging from 0 to 10 (right side) have distinct energy. It is
also clear that the two 250-pitch groups have lower energy than the 500-pitch groups. The same
phenomena happen with 35 and 100 layers. However, there are still several 250-pitch individual
structures showing energy similar to 500-pitch groups in 5 layers. But the energy trends are clear
in all three diagrams, and further analyses was focused on structures on the right side of the
figures.
Looking at the values of 250-pitch groups on the right side in the three diagrams, the peak of
the lowest energy is more and more clear as the number of layers increases. When the values of y
are in the range of 0 to 5, the energy goes down to a lower level. When y is in the range 6 to 8,
the energy reaches the lowest level, and then becomes higher as y increases. Therefore, we
structure with lowest energy has a 250 pitch and y ranging from 6 to 8.
Table 3.8 demonstrated the efforts to identify an ideal y value based on the original structure
with lowest energy in the existing structures. The y values we interested are 6.5, 7 and 7.5. Also,
we want to get some ideas about the function of x, but x will not have a wide range because of
the close distance between layers, so x could be -1, 0 and 1. As for z value, we picked up -1.5
and -2.5, and -1.5 is supposed to give some interesting results. In this condition, 18 structures are
built to pick the ideal combination of x, y and z values.

34

Table 3.5 Energies of 5-layer fibrils

X
Y
Z-‐1p250L5
Z-‐2p250L5
Z-‐1p500L5
Z-‐2p500L5

0
-‐10
-‐1902

-‐1886

-‐1882

-‐1993

0
-‐9
-‐1893

-‐2023

-‐1965

-‐2013

0
-‐8
-‐1983

-‐1975

-‐2010

-‐1958

0
-‐7
-‐1874

-‐1961

-‐1881

-‐1907

0
-‐6
-‐1945

-‐1832

-‐1962

-‐1867

0
-‐5
-‐2116

-‐1933

-‐2053

-‐2042

0
-‐4
-‐1940

-‐1812

-‐1907

-‐1885

0
-‐3
-‐1876

-‐1823

-‐1830

-‐1885

0
-‐2
-‐1969

-‐2032

-‐2012

-‐1846

0
-‐1
-‐1923

-‐1977

-‐1875

-‐1909

0
0
-‐2011

-‐2054

-‐1895

-‐1649

0
1
-‐2272

-‐2152

-‐1820

-‐1772

0
2
-‐2185

-‐2132

-‐1704

-‐1794

0
3
-‐2147

-‐2111

-‐1817

-‐1941

0
4
-‐2184

-‐2115

-‐1982

-‐1847

0
5
-‐2215

-‐2174

-‐1990

-‐2019

0
6
-‐2287

-‐2234

-‐2018

-‐1967

0
7
-‐2205

-‐2186

-‐2064

-‐2014

0
8
-‐2208

-‐2190

-‐1899

-‐2036

0
9
-‐2039

-‐1931

-‐1916

-‐1869

0
10
-‐2003

-‐2017

-‐1802

-‐1907

-‐2400.0000

-‐2300.0000

-‐2200.0000

-‐2100.0000

-‐2000.0000

-‐1900.0000

-‐1800.0000

-‐1700.0000

-‐1600.0000

-‐1500.0000

-‐10
-‐5
0
5
10

z-‐1p250L5

z-‐2p250L5

z-‐1p500L5

z-‐2p500L5

35

Table 3.6 Energies of 35-layer fibrils

-‐22000.0000

-‐20000.0000

-‐18000.0000

-‐16000.0000

-‐14000.0000

-‐12000.0000

-‐10000.0000

-‐10
-‐5
0
5
10

z-‐1p250L35

z-‐2p250L35

z-‐1p500L35

z-‐2p500L35

X
Y
Z-‐1p250L35
Z-‐2p250L35
Z-‐1p500L35
Z-‐2p500L35

0
-‐10
-‐13735

-‐13131

-‐12952

-‐12996

0
-‐9
-‐14249

-‐14107

-‐13218

-‐13127

0
-‐8
-‐14493

-‐14476

-‐13070

-‐13044

0
-‐7
-‐14448

-‐14092

-‐13202

-‐13065

0
-‐6
-‐14561

-‐13406

-‐13344

-‐13788

0
-‐5
-‐15132

-‐13424

-‐14044

-‐14607

0
-‐4
-‐14978

-‐13247

-‐12796

-‐13286

0
-‐3
-‐14603

-‐13228

-‐12730

-‐13335

0
-‐2
-‐14849

-‐14728

-‐14230

-‐13071

0
-‐1
-‐13770

-‐16063

-‐12502

-‐12217

0
0
-‐15076

-‐16049

-‐12510

-‐12704

0
1
-‐16354

-‐17323

-‐12460

-‐12797

0
2
-‐16684

-‐16253

-‐12254

-‐11589

0
3
-‐17152

-‐17875

-‐13240

-‐12826

0
4
-‐17202

-‐17872

-‐13401

-‐12353

0
5
-‐16866

-‐18016

-‐13853

-‐13790

0
6
-‐17820

-‐20177

-‐14276

-‐13901

0
7
-‐20407

-‐19931

-‐14791

-‐14622

0
8
-‐17860

-‐20153

-‐14630

-‐13472

0
9
-‐20159

-‐18107

-‐14238

-‐14752

0
10
-‐19151

-‐19560

-‐15613

-‐14267

36

Table 3.7 Energies of 100-layer fibrils

X
Y
Z-‐1p250L100
Z-‐2p250L100
Z-‐1p500L100
Z-‐2p500L100

0
-‐10
-‐39274
-‐37215
-‐36916
-‐35875

0
-‐9
-‐40352
-‐39242
-‐37224
-‐36991

0
-‐8
-‐40105
-‐41320
-‐36237
-‐36278

0
-‐7
-‐40992
-‐40421
-‐37071
-‐37527

0
-‐6
-‐41633
-‐38449
-‐37398
-‐39540

0
-‐5
-‐45565
-‐38394
-‐40934
-‐41205

0
-‐4
-‐40548
-‐39003
-‐30048
-‐36822

0
-‐3
-‐39602
-‐37828
-‐33461
-‐36361

0
-‐2
-‐40449
-‐43789
-‐34850
-‐36514

0
-‐1
-‐38718
-‐45596
-‐35581
-‐32997

0
0
-‐43426
-‐45874
-‐33915
-‐35503

0
1
-‐45718
-‐50415
-‐32033
-‐34146

0
2
-‐45855
-‐50511
-‐30797
-‐32003

0
3
-‐49291
-‐49750
-‐35241
-‐33991

0
4
-‐49245
-‐52436
-‐38145
-‐33514

0
5
-‐47484
-‐54596
-‐32986
-‐38305

0
6
-‐50867
-‐56272
-‐36548
-‐36036

0
7
-‐56403
-‐58695
-‐41261
-‐37153

0
8
-‐49946
-‐47675
-‐34532
-‐34074

0
9
-‐43420
-‐42619
-‐34433
-‐30037

0
10
-‐54829
-‐53569
-‐28992
-‐31940

-‐60000

-‐55000

-‐50000

-‐45000

-‐40000

-‐35000

-‐30000

-‐25000

-‐10
-‐5
0
5
10

z-‐1p250L100

z-‐2p250L100

z-‐1p500L100

z-‐2p500L100

37

Table 3.8 and Figure 3.1 shows the energies of 18 further structures. If these are divided into
2 groups by the difference of z, it is clear that the group with z of -1.5 has much lower energy
and -2.5 is not better than -1.5. When analyzing the influence of the x value, it is hard to decide
the ideal values, but the diagrams give some ideas. Even though the structures with the lowest
two energies have x is 1 or -1, the x = 0 group exhibits the lowest average energy among the
three groups (points in the diagram are concentrated in the center, and the average energy for
x=1, x=0, x=-1 are -51665, -55509 and -52842). Therefore, 0 may be a better choice to identify
the overall structures, and more modifications may produce better structures by changing the x
value. As for the y value, 6.5 may be a better option, but it is not very clear and further tests are
needed.
Table 3.8 Energies of 100-layer fibrils with intermediate x, y and z values

X
Y
Z
P250L100

1
6.5
-‐1.5
-‐61633

1
6.5
-‐2.5
-‐44090

1
7
-‐1.5
-‐59650

1
7
-‐2.5
-‐48134

1
7.5
-‐1.5
-‐59633

1
7.5
-‐2.5
-‐36848

0
6.5
-‐1.5
-‐59979

0
6.5
-‐2.5
-‐48612

0
7
-‐1.5
-‐58186

0
7
-‐2.5
-‐53080

0
7.5
-‐1.5
-‐59003

0
7.5
-‐2.5
-‐54194

-‐1
6.5
-‐1.5
-‐61622

-‐1
6.5
-‐2.5
-‐55429

-‐1
7
-‐1.5
-‐59653

-‐1
7
-‐2.5
-‐38351

-‐1
7.5
-‐1.5
-‐59593

-‐1
7.5
-‐2.5
-‐42403

38

Figure 3.1 Graphs drawn to determine the ideal y value in the fibrils in Table 3.8
-‐65000

-‐60000

-‐55000

-‐50000

-‐45000

-‐40000

-‐35000

6.5
(Z-‐1.5)

6.5
(Z-‐2.5)

7
(Z-‐1.5)

7
(Z-‐2.5)

7.5（Z-‐1.5）
7.5（Z-‐2.5）
P250L100

-‐65000

-‐60000

-‐55000

-‐50000

-‐45000

-‐40000

-‐35000

6.5
(Z-‐1.5)

6.5
(Z-‐2.5)

7
(Z-‐1.5)

7
(Z-‐2.5)

7.5
(Z-‐1.5)

7.5
(Z-‐2.5)

P250L100

-‐65000

-‐60000

-‐55000

-‐50000

-‐45000

-‐40000

-‐35000

6.5
(Z-‐1.5)

6.5
(Z-‐2.5)

7
(Z-‐1.5)

7
(Z-‐2.5)

7.5
(Z-‐1.5)

7.5
(Z-‐2.5)

P250L100

39

The results in Table 3.8 and Figure 3.1 narrow down the range of ideal values for x, y and z.
Nevertheless, there are still some possible values that require analysis. Therefore, we added 17
new structures for further calculation. In previous experiments, the range for x was quite narrow,
and x is a parameter that is hard to explore. We assume x values should not be changed too much
due to the high possibility of overlapping between two stacks, which could be verified by several
extreme high energy of structures with x=1 or -1 compared with that of x=0. However, changing
the x values is likely to give some good structures as well, which means -1 and 1 did not reach
the maximum distance between two stacks. As a result, -2 and 2 were chosen for testing values
for x.
Table 3.9 Energies for fibrils with x=2 or -2

X
Y
Z
P250L100

-‐2
6.5
-‐1.5
-‐58178

-‐2
6.5
-‐2.5
-‐48436

-‐2
7
-‐1.5
-‐46002

-‐2
7
-‐2.5
-‐50666

-‐2
7.5
-‐1.5
-‐58435

-‐2
7.5
-‐2.5
-‐41135

2
6.5
-‐1.5
-‐58111

2
6.5
-‐2.5
-‐40568

2
7
-‐1.5
-‐41423

2
7
-‐2.5
-‐49504

2
7.5
-‐1.5
-‐58451

2
7.5
-‐2.5
-‐51587

40

Figure 3.2 Graphs drawn to determine the ideal x value in the fibrils in Table 3.9
Table 3.9 and Figure 3.2 shows 12 results for structures with x=2 or -2, with the range of y
and z the same as the setting in Table 3.8 and Figure 3.1. The average energy for x=-2 is -50475,
and that for x=2 is -49941. When comparing all the average energy for structures with five x
values, the conclusion that 0 is the best choice is clear. Moreover, if the absolute value of x = 0 is
larger, the energy of that structure will be larger. Therefore, -2 or 2 may not good be candidates
for the x parameter.
-‐65000

-‐60000

-‐55000

-‐50000

-‐45000

-‐40000

-‐35000

6.5
(Z-‐1.5)

6.5
(Z-‐2.5)

7
(Z-‐1.5)

7
(Z-‐2.5)

7.5
(Z-‐1.5)

7.5
(Z-‐2.5)

P250L100

-‐65000

-‐60000

-‐55000

-‐50000

-‐45000

-‐40000

-‐35000

6.5
(Z-‐1.5)

6.5
(Z-‐2.5)

7
(Z-‐1.5)

7
(Z-‐2.5)

7.5
(Z-‐1.5)

7.5
(Z-‐2.5)

P250L100

41

By analyzing the results in Table 3.8, 6.5 was selected as the most suitable y value. However,
for the z value, we did not include the combination of y=6.5 and z=-2. Therefore, 5 more
structures were tested for comparing z=-1.5 and z=-2. For these 5 structures, the range of x was -
2 to 2, and the values for z and y were fixed. Table 3.10 shows the calculated energy. These
results were interesting. When referring to the data in Figure 3.9, we assumed -2 or 2 are the
least favorable values. However, in this case, the top 3 favorable structures were -2, 0 and 2, and
structure with x=2 reached the lowest energy among these 5 models. For explaining this
phenomenon, further analysis of these specific structures is needed. However, the difference of
energy between structures with x=-1.5 and that of x=2 is still apparent.
Table 3.10 Energies for fibrils with x=(-2~2), y=6.5 and z=-2

X
Y
Z
P250L100

-‐2
6.5
-‐2
-‐54015

-‐1
6.5
-‐2
-‐44145

0
6.5
-‐2
-‐54811

1
6.5
-‐2
-‐49638

2
6.5
-‐2
-‐57209

To illustrate the best combination of x and z values, the previous data for y=6.5 was
rearranged. Table 3.11 shows the rearranged data in table and figure formats. When
-‐70000

-‐60000

-‐50000

-‐40000

-‐30000

-‐20000

-‐10000

0

-‐3
-‐2
-‐1
0
1
2
3

P250L100

42

y is equal to 6.5, the most favorable value for z is -1.5 because all the five structures showed the
lowest energy when compared to structures with same x values. The regularity of x values was
complicated in this case, and was too complicated to decide a favorable x value. However,
models with x=0 still exhibited the lowest average energy.
Table 3.11 Energies for fibrils with x=(-2~2), y=6.5 and z=(-1.5, -2 and -2.5)

X
Y
Z
P250L100

-‐2
6.5
-‐1.5
-‐58178

-‐2
6.5
-‐2
-‐54015

-‐2
6.5
-‐2.5
-‐48436

-‐1
6.5
-‐1.5
-‐61622

-‐1
6.5
-‐2
-‐44145

-‐1
6.5
-‐2.5
-‐55429

0
6.5
-‐1.5
-‐59979

0
6.5
-‐2
-‐54811

0
6.5
-‐2.5
-‐48612

1
6.5
-‐1.5
-‐61633

1
6.5
-‐2
-‐49638

1
6.5
-‐2.5
-‐44090

2
6.5
-‐1.5
-‐58111

2
6.5
-‐2
-‐57209

2
6.5
-‐2.5
-‐40568

-‐65000

-‐60000

-‐55000

-‐50000

-‐45000

-‐40000

-‐35000

-‐3
-‐2
-‐1
0
1
2
3

Z-‐1.5

Z-‐2

Z-‐2.5

43

4. Discussion
In this study, we built many hIAPP fibril structures to identify the most ideal parameters.
Among these parameters, the value of z is relatively easy to choose, and should be -1 or -1.5, and
-1.5 was best in most cases. The x value is the hardest one to choose. Setting x as 0 can just
ensure all the created structures have lowest energy in the general conditions. However, some
structures may just have a little overlap that affects the energy, and if the x changes a little, it can
give a significantly different result. Understanding the simulated structures well and more
simulations may help to omit these kinds of structures. The ideal value of y is identified in a
small range, probably from 6.5 to 7.5. This range is small enough, but smaller numbers are still
possible to use in the MFIBRIL program. In this study, 6.5 is the most favorable value for y.
However, there is still some possibility that y cannot be selected individually, but is dependent
on other parameters to choose the best combination.
Identifying a reasonable single unit for estimation of energies can save calculation time. A
35-layer is a good choice for estimation, but use of longer structures may more accurately
estimate the energy, but at the cost of a lot of time for analysis. Lastly, the selected favorable
models should be further examined and evaluated by PRONOX (a program developed in our
laboratory for measuring inter-spin label distances and residue mobilities) and WATGEN (a
program in our laboratory for modeling the water network at protein-peptide interfaces). This
will allow experimental results to be compared with computational modeling to see if modeling
can reproduce the actual structure of hIAPP.

44

REFERENCES
Abedini A and Raleigh DP. (2005). The role of His-18 in amyloid formation by human islet
amyloid polypeptide. Biochemistry. 44(49), 16284–16291.

Alvarado, R.M, Merkel, J.S., Regan, L. (2000). A systematic exploration of the influence of
protein stability on amyloid fibril formation in vitro. PNAS 97(16), 8979-8984.

Braak, H., Del Tredici, K., Rub, U., de Vos, R.A., Jansen Steur, E.N., Braak, E. (2003). Staging
of brain pathology related to sporadic Parkinson’s disease. Neurobiology of aging 24(2), 197-211.

Buchanan, L.E., Dunkelberger, E.B., Tran, H.Q., Cheng, P.N., Chiu, C.C.,
Cao, P., Raleigh, D.P., de Pablo, J.J., Nowick, J.S. and Zanni, M.T. (2013). Mechanism of IAPP
amyloid fibril formation involves an intermediate with a transient beta-sheet. Proc. Natl. Acad.
Sci. U. S. A. 110, 19285–19290.

Caillon, L., Hoffmann, A.R.F., Botz, A. and Khemtemourian, L. (2015). Molecular Structure,
membrane interactions, and toxicity of the islet amyloid polypeptide in type 2 diabetes mellitus.
Journal of Diabetes Research 2016, 1-13.

Chiti, F., and Dobson, C.M. (2006). Protein misfolding, functional amyloid, and human disease.
Annu Rev Biochem 75, 333-366.
Cooper, G. J. S., Willis, A. C., Clark, A., Turner, R. C., Sim, R. B. and K. B. M. Reid. (1987).
Purification and characterizationof a peptide from amyloid-rich pancreases of type 2
diabeticpatients. Proceedings of the National Academy of Sciences of the United States of
America 84(23), 8628–8632.
Cope, S.M., Shinde, S., Best, R.B., Ghirlanda, G. and Vaiana, S.M. (2013). Cyclic N-terminal
loop of amylin forms non-amyloid fibers. Biophys. J. 105, 1661–1669.
El-Agnaf, O.M., Salem, S.A., Paleologou, K.E., Curran, M.D., Gibson, M.J., Court, J.A.,
Schlossmacher, M.G., Allsop, D. (2006). Detection of oligomeric forms of alpha-synuclein
protein in human plasma as a potential biomarker for Parkinson’s disease. FASEB journal:
official publication of the Federation of American Societies for Experimental Biology 20(3),
419-425.
Eliezer, D., Kutluay, E., Bussell, R., Jr., Browne, G. (2001). Conformational properties of alpha-
synuclein in its free and lipid-associated states. Journal of molecular biology 307(4), 1061-1073.

Gilead, S. and Gazit, E. (2008). The role of the 14–20 domain of the islet amyloid polypeptide in
amyloid formation. Exp. Diabetes Res. 2008, 256954.

Goldsbury, C.S., Cooper, G.J., Goldie, K.N., Muller, S.A., Saafi, E.L., Gruijters, W.T., Misur,
M.P., Engel, A., Aebi, U., and Kistler, J. (1997). Polymorphic fibrillar assembly of human
amylin. J Struct Biol 119, 17-27.

Gonzalez, I.M., and Soto, C. (2011). Misfolded protein aggregates: Mechanisms, structures and

45

potential for disease transmission. Seminars in Cell & Developmental Biology 22, 482-487.
Hamilton-Brown, P., Bekard, I., Ducker, W.A., Dunstan, D.E. (2008). How does shear affect
Abeta fibrillogenesis? J Phys Chem B 112, 16249–16252.

Hamley, I.W. (2012). The amyloid beta peptide: A chemist’s perspective. Role in Alzheimer’s
and fibrillization. Chem Rev 112, 5147–5192.

Higham, C.E., Jaikaran, E.T., Fraser, P.E., Gross, M., and Clark, A. (2000). Preparation of
synthetic human islet amyloid polypeptide (IAPP) in a stable conformation to enable study of
conversion to amyloid-like fibrils. FEBS Lett 470, 55-60.

H¨oppener, J. W. M., Ahr´en, B. and Lips, C. J. M. (2000). Islet amyloid and type 2 diabetes
mellitus,” The New England Journal of Medicine 343(6), 411–419.

Iwai, A., Masliah, E., Yoshimoto, M., Ge, N., Flanagan, L., de Silva, H.A., Kittel, A. and Saitoh,
T. (1995). The precursor protein of non-A beta component of Alzheimer’s disease amyloid is a
presynaptic protein of the central nervous system. Neuron 14(2), 467-475.

Ilitchev, A.I., Giammona, M.J., Do, T.D., Wong, A.G., Buratto, S.K., Shea, J.E., Raleigh, D.P.
and Bowers, M.T. (2016). Human islet amyloid polypeptide N-terminus fragment self-assembly:
Effect of conserved disulfide bond. J. Am. Soc. Mass Spectrom. 27, 1010-1018.

Jayasinghe, S.A., and Langen, R. (2004). Identifying structural features of fibrillar islet amyloid
polypeptide using site-directed spin labeling. J Biol Chem 279, 48420-48425.

Kayed, R., Bernhagen, J., Greenfield, N., Sweimeh, K., Brunner, H., Voelter, W., and
Kapurniotu, A. (1999). Conformational transitions of islet amyloid polypeptide (IAPP) in
amyloid formation in vitro. J Mol Biol 287, 781-796.

Kimov, D.K., Thirumalai, D. (2003). Dissecting the assembly of Aβ16–22 amyloid peptides into
antiparallel β sheets. Structure 11, 295–307.

Kopan, R., Ilagan, M.X.G. (2004). [gamma]-Secretase: proteasome of the membrane? Nat Rev
Mol Cell Biol 5, 499-504.

Levitan, K., Chereau, D., Cohen, S.I., Knowles, T.P., Dobson, C.M., Fink, A.L., Anderson, J.P.,
Goldstein, J.M., Millhauser, G.L. (2011). Conserved C-terminal charge exerts a profound
influence on the aggregation rate of alpha-synuclein. Journal of molecular biology 411(2), 329-
333.

Li, Y.Y. (2013). Algorithm development for modeling protein assembles. 1-149.

Moriarty, D. F. and Raleigh, D. P. (1999). Effects of sequential proline substitutions on amyloid
formation by human amylin 20–29. Biochemistry, 38, 1811–1818.

Nelson, R. and Eisenberg, D. (2006). Recent atomic models of amyloid fibril structure. Curr

46

Opin Struct Biol 16(2), 260-265.

Ow, S.Y. and Dunstan, D.E. (2014). A brief overview of amyloids and Alzheimer’s disease.
Protein Sci 23(10), 1315-1331.

Paola, D., Domenicotti, C., Nitti, M., Vitali, A., Borghi, R., Cottalasso, D., Zaccheo, D., Odetti,
P., Strocchi, P., Marinari, U.M., Tabaton, M., Pronzato, M.A. (2000). Oxidative stress induces
increase in intracellular amyloid β-protein production and selective activation of βI and βII PKCs
in NT2 cells. Biochem Biophys Res Commun 268, 642–646.

Peter, T., Lansbury, J. (1999). Evolution of amyloid: What normal protein folding may tell us
about fibrillogenesis and disease. Proc Natl Acad Sci USA 96, 3342–3344.

Ritchie, C.M. and Thomas, P.J. (2012). Alpha-synuclein truncation and disease. Health 4(11),
1167-1177.

Scrocchi, L. A., Ha, K., Chen, Y., Wu, L., Wang, F. and Fraser, P. E. (2003). Identification of
minimal peptide sequences in the (8–20) domain of human islet amyloid polypeptide involved in
fibrillogenesis. J. Struct. Biol. 141, 218–227.

Selkoe, D.J. (1994). Normal and abnormal biology of the β-amyloid precursor protein. Ann Rev
Neuro 17, 489–517.

Soto, C. (2003). Unfolding the role of protein misfolding in neurodegenerative diseases. Nature
Rev Neurosci 4, 49–60.

Soto, C., Estrada, L.D. (2008). Protein misfolding and neurodegeneration. Arch Neurol 65, 184-
189.

Sumner Makin, O., and Serpell, L.C. (2004). Structural characterisation of islet amyloid
polypeptide fibrils. J Mol Biol 335, 1279-1288.

Tenidis, K., Waldner, M., Bernhagen, J., Fischle, W., Bergmann, M.,Weber, M., Merkle, M.L.,
Voelter, W., Brunner, H., Kapurniotu, A. (2000). Identification of a penta- and hexapeptide of
islet amyloid polypeptide (IAPP) with amyloidogenic and cytotoxic properties. J. Mol. Biol. 295,
1055–1071.

Tracz, S. M., Abedini, A., Driscoll, M. and Raleigh, D. P. (2004). Role of aromatic interactions
in amyloid formation by peptides derived from human amylin. Biochemistry, 43, 15901–15908.

Uversky, V.N. (2002). Natively unfolded proteins: a point where biology waits for physics.
Protein Sci 11, 739–756.

Westermark, P., Engstrom, U., Johnson, K.H., Westermark, G.T. and Betsholtz, C. (1990). Islet
amyloid polypeptide - pinpointing amino-acidresidues linked to amyloid fibril formation. Proc.
Natl. Acad. Sci. U. S. A. 87, 5036–5040.

47

Westermark, P., Wernstedt, C., Wilander, E., Hayden, D.W., O’Brien, T.D. and K. H. Johnson.
(1987). Amyloid fibrils in human insulinoma and islets of Langerhans of the diabetic cat are
derived froma neuropeptide-like protein also present in normal islet cells. Proceedings of the
National Academy of Sciences of the United States of America 84(11), 3881–3885.

Westermark, P., Wernstedt, C., Wilander, E. and Sletten, K. (1986). A novel peptide in the
calcitonin gene related peptide family as an amyloid fibril protein in the endocrine pancreas.
Biochemical and Biophysical Research Communications 140(3), 827–831.

Yoon, S., Welsh, W. J., Jung, H. and Yoo, Y. D. (2007). CSSP2: an improved method for
predicting contact-dependent secondary structure propensity. Comput. Biol. Chem. 31, 373–377.

Zhao, J., Yu, X., Liang, G.Z. and Zheng, J. (2011). Structural polymorphism of human islet
amyloid polypeptide (hIAPP) oligomers highlight the importance of interfacial residue
interactions. Biomacromolecules. 12, 210-220.

Abstract (if available)

Abstract It is believed that a variety of human diseases are related to the misfolding of proteins or peptides. Amyloid fibrils are highly organized aggregates with a cross-β structure and are generated by misfolding of proteins. These aggregates are also related to disease. For instance, human islet amyloid polypeptide (hIAPP), α-synuclein, and amyloid-β are associated with Type II Diabetes, Parkinson’s disease and Alzheimer’s disease when they become misfolded, respectively. However, these amyloid fibrils are neither soluble nor crystallizable, and so it is hard to acquire fibril structure information through direct experimental approaches. Therefore, analysis and determination of the fibril structure, and formation process remain as compelling questions. ❧ The objectives of this thesis are to investigate the fibril structure of hIAPP by computer modeling. The fibril model with the most reasonable potential interaction interfaces is determined, and then the resulting model is used to explain experimental data from election paramagnetic resonance (EPR) and electron microscopy (EM). ❧ This thesis mainly focuses on producing structures of hIAPP fibrils consistent with experimental data. Numerous potential models of the hIAPP fibril were built using MFIBRIL and then refined by equilibration utilizing simulations in the NAMD program. The refined models were evaluated by comparing their minimized energy. Through analysis of several favorable models, we identified suitable values of selected parameters, and further narrowed down the range values of some parameters to find a more exact tendency of the energy changes related to continuous values. The results suggest that x=0, y=6.5 and z=−1.5 (MFIBRIL) are one of the most favorable sets of values, but other combinations can also contribute to favorable structures.

Linked assets