Early events in alpha-synuclein aggregation

(1)

Early Events in α-Synuclein Aggregation

Burst analysis and Fluorescence Correlation Spectroscopy on a

dual color fluorescence microscope

Author: J.D. Vliegenthart Studentnumber: 0934968

Study: Physics

Organisation: Leiden University Masterthesis

(2)

i

ABSTRACT

Α-synuclein is a small 14 kDa protein, which is believed to play a role in the Parkinson’s disease. The aggregation process of α-synuclein could be the reason for the death of the neural cells in the human brain during the disease. In order to understand this aggregation process, we need to set up a method for measuring the small aggregates of α-synuclein. This report will show the results of burst analysis, Fluorescence Correlation Spectroscopy in solution and in gel. Using the last method, we can find a size dependence on the diffusion coefficient of -1.64. FCS in gel could be performed for finding the early events in α-synuclein aggregation (monomer, dimer and tetramer)

(5)

2

1 INTRODUCTION

Parkinson’s disease is one of the most common neurodegenerative movement disorders [1]. This movement disorder is caused by the deaths of nerve cells in the human brain[2]. During the disease, Lewy bodies are found in the cytoplasm of surviving nigral neurons in the brain of the patients[1]. These Lewy bodies are spherical protein inclusions[1]. They contain fibrillar aggregates of α-synuclein and are found in the surviving nigral neurons in the brain of the patients[1].

There is evidence that suggest that the forming of those fibrillary aggregates of α-synuclein will cause the death of the nerve cells[1]. Normally α-synuclein is found in the human brain[3]. It is a 14 kDa [1], 140-amino-acid-long protein and the function of α-synuclein is still unclear[4]. For the role in Parkinson’s disease, the aggregation process of α-synuclein has to be understand.

This aggregation process can be described by a simple kinetic model of fibrillation. The conversion of an unfolded monomer (Nu) to partially folded intermediates can lead to aggregates, oligomers or

fibrils[5]. The process depends on the experimental conditions[5]. For example, by slowing down fibril formation, the formation of oligomers will be increased[5].

The goal of the research is to set up a method for measuring the early events in the α-synuclein aggregation process. This can be done by labeling α-synuclein molecules with fluorescent dyes. In a dual color fluorescence microscope, the labeled molecules could be placed and the single molecules could be measured. The measured size of these single molecules will give information about the number of monomers in the oligomer. So we can get information about the aggregation process. This report describes the experiments done to image the early events in α-synuclein aggregation. The next section starts by discussing the useful techniques for detecting single α-synuclein molecules.

(6)

3

2 THEORY

Different methods and physical phenomena are used for setting up a method to detect the early events in α-synuclein aggregation. This section shows the relevant theory to explain how they can be used for observing the early events in α-synuclein aggregation.

2.1 F

LUORESCENCE

One of the difficulties in observing single molecules is to reduce the background as much as possible. Surrounding molecules influences the detected signal and increase the background noise [6]. The measured intensity must be dominated by the single molecule and not by the surrounding

molecules, otherwise it will be impossible to focus on the single molecule. A method that reduces the influence of the surrounding molecules is to label the single molecule with a fluorescent dye

molecule.

Fluorescence is a physical process of a molecule. During this process, the molecules absorbs a photon and uses its energy to excite an electron into a higher excited state. Vibrational relaxation causes the electron to relax into the lowest vibrational level of the excited state, from which it decays to the vibrational level of the ground state [6]. A photon is released during the relaxation of the electron from the lowest vibrational state of the excited state to the vibrational levels of the ground state. The wavelength of the photon must be in a specific energy range, dependent on the molecule. Every molecule has its own fluorescence spectrum [6]. The same holds for the emitted photon. The wavelength of this photon also depend on the molecule. The emitted photon that is released during the relaxation has a lower wavelength than the absorbed photon, due to the energy loss caused by the vibrational relaxation [6]. Because of this, a laser which emits photons at a specific wavelength can only excite the molecules that can absorb photons with this wavelengths. The emitted photons of this excited molecules can be observed. The surrounding molecules cannot be excited by this laser and therefor they do not emit photons. The obtained signal then only consists of the photons

emitted by the excited molecules.

Figure 1: A Jablonski Energy Diagram of the fluorescence process. Excitation occurs from the ground state to a vibrational state of the excited states. Due to internal conversion and vibrational relaxation the molecule returns to the relaxed excited state. From here the molecule can return to its ground state through fluorescence, non-radiative relaxation, quenching or through intersystem crossing via its excited triplet state [7].

(7)

4 The Jablnski Energy Diagram of the fluorescence process is shown in figure 1. This diagram shows that it is also possible for a molecule to return to its ground state through non-radiative relaxation, quenching and intersystem crossing. These phenomena limit the fluorescence properties of the molecule. The rate constant of the different relaxation phenomena are given by

• 𝑘_𝑟𝑆_{rate constant for radiative deactivation 𝑆}

1→ 𝑆0 emission of fluorescence

• 𝑘_𝑖𝑐𝑆_{rate constant for non-radiative relaxation and quenching decay by internal conversion} 𝑆1→ 𝑆0

• 𝑘_𝑖𝑠𝑐𝑇 _{rate constant for intersystem crossing}

The overall non-radiative rate constant is denoted by 𝑘𝑛𝑟𝑆 = 𝑘𝑖𝑐𝑆 + 𝑘𝑖𝑠𝑐𝑇 [8]. The lifetime a molecule will stay in the excited state 𝜏𝑆 is given by [8]

𝜏𝑆= 1

𝑘𝑟𝑆+𝑘𝑛𝑟𝑆 1

The fluorescent molecule is more unstable when it stays longer in its excited state. A good fluorescent molecule thus has a short lifetime. Another important quantity that determines the quality of the fluorescent molecule is the fluorescence quantum yield ΦF [8]. The fluorescence quantum yield is the

fraction of excited molecules that returns to the ground state S0 by emission of fluorescent photons

[8]: 𝛷𝐹= 𝑘𝑟𝑆 𝑘𝑟𝑆+𝑘𝑛𝑟𝑆 = 𝑘𝑟 𝑆_𝜏 𝑆 2

The lifetime and quantum yield are important quantities in fluorescent experiments. The fluorescent dye molecules, used in the experiments, must have a low lifetime of the excited state and a high quantum yield to be detected.

2.2 F

LUORESCENT

R

ESONANCE

E

NERGY

T

RANSFER

By using two fluorescent dye molecules with different excitation wavelengths, the energy of the absorption of a photon from one molecule could also be transferred to the other molecule [9]. This process is called Fluorescence Resonance Energy Transfer (FRET) and is important in many biophysical researches [9].

FRET is a physical process in which non-radiative energy transfer plays a role [9]. An excited molecule (the donor) transfers its energy to another molecule (the acceptor), so the excited molecule returns to its ground state (without releasing a photon) and the acceptor will become excited [9]. The acceptor may then emit a photon, by which it returns to its ground state [9]. FRET operates over short distances it can be used to determine if molecules are bound to each other and to measure the distance between donor and acceptor in the range of up to a few nanometers [9].

The rate of the energy transfer 𝑘𝑇 is given by [9]: 𝑘𝑇 = ( 1 𝜏𝐷) × ( 𝑅0 𝑅) 6 3

With 𝜏𝐷 the fluorescence lifetime of the donor, 𝑅 the distance between the donor and acceptor and 𝑅0 the critical transfer distance at which the energy transfer rate is equal to the fluorescence decay rate of the donor [9]. 𝑅0 is given by [10]

𝑅06= 8,785× 10−5 𝜅 2_𝛷

𝐷𝐽

(8)

5 where 𝜅2 is a factor which depends on the relative orientation of transition dipole moment which has usually a-well defined direction in the molecular frame, 𝛷𝐷 the quantum yield of the donor, 𝑛 the index of refraction and 𝐽 the overlap integral between the donor and acceptor [10].

The FRET efficiency (E) is a quantitative measure of the number of quanta that are transferred from the donor to the acceptor. E is defined as [9]:

𝐸 = 1

1+(𝑅 𝑅0)6

5

Note that the energy transfer efficiency is 50% when R = R0. By determining the FRET efficiency, the

distance between the donor and acceptor can be calculated [9].

2.2.1 Optical Microscope

Until now, we have found a way to observe single molecule with a high signal to noise ratio and a way to detect the distance between two fluorescent molecules. The detection of these fluorescent particles can be done by an optical microscope [6]. An optical microscope can be seen as a simple lens system for magnifying small objects [6]. It exists of two parts, the objective (short focal length, a few mm), creates an image of the object, and the eye-piece, which can provide further magnification [6]. Maximizing the resolution of the microscope is critical for obtaining reliable measurements [6]. The Abbe-Rayleigh criterion states that the smallest resolvable distance 𝑑𝑚𝑖𝑛 between two point sources depends on the wavelenght 𝜆 and the numerical aperture 𝑁𝐴 = 𝑛 × sin 𝛼, with 𝑛 the index of refraction and 𝛼 half the maximal angle under which the objective lens collects light from the object, by [6]:

𝑑𝑚𝑖𝑛 = 1.22× 𝜆

2𝑁𝐴 6

For minimizing the spatial resolution, the numerical aperture should be as large as possible [6]. Another reason to maximize the numerical aperture is to maximize the collection efficiency. The fraction of light collected for an isotropic light source is [6]:

Figure 2: The FRET efficiency graph of two fluorescent molecules attached to the same protein. The FRET efficiency depends on the distance between the two fluorescent molecules. [11]

(9)

6 Ω 4𝜋= 1 2(1 − cos 𝛼) = 1 2(1 − √1 − ( 𝑁𝐴 𝑛) 2 ) 7

A confocal microscope images only one point of the sample into the photodetector [6]. The light coming from the objective is focussed on a pinhole [6]. This pinhole only passes through the light coming from the focus point of the objective [6]. This method is very useful to detect single molecules [6]. The laser light will only excites the focus with high efficiency and the detector only detects the fluorescence light coming from molecules in this confocal volume [6].

For the alingment of the set up, knowing the value of the eccentricity of the confocal volume is important. The expectation value of the eccentricity is found by starting with the intensity of a gaussian beam (TEM00) directed along the z axis [12]

2 2 2 2 2 2( )/ 0 z x y w

I



I e

   8 where

w

is the 2

e radius in the xy direction and  is the eccentricity. The eccentricity is the resolution in vertical (z) direction divided by the resolution in the xy direction (eccentricity) [12]:

𝜅 = 𝑟𝑧

𝑟_𝑥𝑦 9

For the resolution in the horizontal xy plane, we use Rayleigh limit [12]: 0.61 xy r NA   10

The resolution in the z direction is given by [12] 2 2 z n r NA   11

Combining (12), (13) and the definition of  gives [12]

2.33

z xy

r

n

r

NA





12

Putting in the

n 

1.3

for water and

NA 

0.6

for a water immersion objective, gives a lower limit of

5.05 



[12]. The optical microscope has to approximate this eccentricity when the alignment of the microscope is optimal.

For detecting fluorescent molecules, a laser light has to be focused on the confocal volume [6]. When a fluorescent molecule comes into the confocal volume, it can absorb a photon and emits a photon on a different wavelength [6]. By filtering out the excitation wavelength (laser light) the signal of the fluorescent molecule can be detected. During the time a molecule is in the confocal volume it can absorb and emit photons [6]. This will give rise to photon bursts. These bursts are determined by the length of time the molecule spends in the confocal volume, the laser intensity, the concentration of the sample and the number of dye molecules in the confocal volume [13].

(10)

7

P

HOTON

S

TATISTICS

A typical time trace obtained from a dual color fluorescence microscope looks like figure 3. A laser beam will excite an electron of a fluorescent dye to a higher energy state. A short time after, this electron falls back into its ground state and emits a photon. These photons will be detected by a photon counter, which consists of a very sensitive light detector such as an avalanche photodiode (APD) connected to an electronic counter. From this process, we can make the following assumptions [14]:

1. The probability of the emission of a photon in a short time interval ∆𝑡 is proportional to the intensity 𝐼, the area 𝐴 illuminated, and the time interval ∆𝑡.

2. If ∆𝑡 is sufficiently small, the probability of emitting two photons is negligibly small. 3. Photoemission events registered in different time intervals are statistically independent of

each other.

From these assumption the probability to observe 𝑛 photons at a time 𝑡 is given by 𝑃𝑛(𝑡) = 𝑛̅𝑛 𝑛!𝑒 −𝑛̅ ₁₃ With 𝑛̅ = 𝜉𝐼𝑡 = 𝐶𝑡 14

Where I is the intensity of the signal and 𝜉 is proportional to the area illuminated, and is equal to the emission probability per unit time per unit intensity. This shows that we obtain a Poisson distribution when 𝐼(𝑡) is constant.

2.3 B

URST

A

NALYSIS

Single molecules are detected by using an optical microscope, which collects the fluorescent light emitted by the dye molecules. In this project, dye molecules are labeled to α-synuclein molecules to measure the early events in α-synuclein aggregation. The signal coming from the fluorescent molecules are analyzed to find the important properties of the α-synuclein molecules. The sample that is placed in the microscope has to obtain a low concentration of fluorescent molecules for observing the single molecules moving in and out of the confocal volume.

A time trace of an experiment with an average of less than one molecule in the confocal volume looks like figure 4. Every burst consists of a number of photons (burst intensity) and last for a time ∆𝑡 (burst duration) [13]. The burst intensity and the burst duration give information about the diffusion coefficient and size of the molecule in the confocal volume [13]. The burst intensity depends on the number of dye molecules in the confocal volume, the burst duration depends on the diffusion

Figure 3 A time trace of a dual color fluorescence experiment. When a molecule comes in the confocal volume it will be excited by the laser light and starts emitting photons. These photons will be detected by an APD. The bursts in this time trace come from the different molecules in the confocal volume. ∆𝑡 is the time between photons coming from the same molecule. ∆𝑇 is the time between two molecules coming in the confocal volume.

Burst 1

∆𝑡 ∆𝑇

(11)

8 coefficient of the fluorescent molecule and the time between the burst ∆𝑇 depends on the

concentration of the sample. A lower concentration will give longer times between the bursts ∆𝑇 [13]. This helps to determine the properties of the molecules more accurately.

For example, a time trace of an α-synuclein sample labeled with a fluorescent dye molecule has many individually bursts. By selecting and analyzing these burst one can find the average burst duration and photons/burst. These quantities will help to identify the size and diffusion coefficient of the molecules in the sample.

2.3.1 Fluorescence Correlation Spectroscopy

Another way to analyze an aged α-synuclein sample is by using Fluorescence Correlation Spectroscopy (FCS). In a typical experiment a laser beam excites a solution of fluorescently labeled molecules in the confocal volume. FCS is a technique, which focusses on the spontaneous intensity fluctuations caused by the minute deviations of a small system from thermal equilibrium.

The trajectory of the molecules is described by the mean squared displacement < 𝑟2> , which describes the area that the particle covers in a certain time [15]. The mean squared displacement is given by [15]:

< 𝑟2> = 6 ∙ 𝐷 ∙ 𝜏 15

With 𝜏 the time the molecule is diffusing and 𝐷 the diffusion coefficient, given by [15] 𝐷 = 𝑘𝐵𝑇

6𝜋∙𝜂∙𝑅ℎ 16

Where 𝑘𝐵 = 1.3806504 · 10−23 𝐽/𝐾 is Boltzmann’s constant, 𝑇 is the absolute temperature and η is the solutions viscosity [15]. So the different molecules have a different trajectory, as shown in figure 5 [15]. When we focus the objective on the center of the sample, we have a fluctuating number of molecules in the confocal volume [15]. When the emitted intensity of the molecules in the confocal volume is measured, changes in fluorescence intensity can be seen, due to the changing number of molecules in confocal volume [15]. The time trace of the intensity fluctuations can be described as [15]

𝐼(𝑡) = < 𝐼 > + 𝛿𝐼(𝑡) 17

Figure 4 The detection area and the number of particle graph of a fluorescent sample with the diffusion particles in blue and red [15].

(12)

9 Where 𝐼(𝑡) is the total intensity over time, < 𝐼 > the average intensity and 𝛿𝐼(𝑡) the fluctuations in the intensity [15]. For measuring how fast the signal is fluctuating, we calculate the autocorrelation function [15]:

𝑔(𝜏) = <𝛿𝐼(𝑡)∙𝛿𝐼(𝑡+𝜏)>𝑡

<𝐼(𝑡)>_𝑡2 18

Where <. . . >𝑡 is the time average over the time variable t, so we can write [15] < 𝐼(𝑡) >𝑡=

1

𝑇∫ 𝐼(𝑡)𝑑𝑡 𝑇

1 19

By calculating those integrals and inserting the diffusion equation, we get [15] 𝑔(𝜏) =1 𝑁∙ (1 + 4𝐷𝜏 𝑤_𝑥𝑦2 ) −1 ∙ (1 +4𝐷𝜏 𝑧₀2) −1/2 20

Where 𝑤𝑥𝑦 is the width in the xy-plane and 𝑧0 the length of the measurement volume. The intensity fluctuations of the sample are not only dependent on the diffusion of the molecules, but also on the fluctuations of the in the fluorescence process, like a triplet state [15]. These triplet dynamics lead to an extra term in the autocorrelation function [15]

𝑔(𝑡𝑟𝑖𝑝𝑙𝑒𝑡)(𝜏) =

(1−𝐹+𝐹𝑒−

𝑡 𝜏𝐵)

1−𝐹 ∙ 𝑔(𝜏) 21

With F the fraction of molecules related to the blinking reaction described by the blinking time 𝜏𝐵. So FCS can be used for calculating local concentrations, mobility coefficients or characteristic rate constants of fluorescently labeled biomolecules [16]. A higher concentration will give more fluorescence fluctuations, so the FCS traces have a lower amplitude and the plotted data will not be reliable [16].

Figure 5 The diffusing molecules and the fluorescence processes which cause the fluctuations of the top graph. The autocorrelation of these fluctuations shown in the down graph. [16]

(13)

10

2.4 G

EL

E

LECTROPHORESIS

FCS only gives the average values of the mobility coefficients. A homogenous sample gives the average value of the fluorescent labeled molecules. However, when a sample consists of molecules with different sizes the results from FCS does not give a reliable value of the mobility coefficients. A way to separate the molecules based on their size is by using Gel Electrophoresis.

Gel electrophoresis is performed in an electrophoretic chamber, with a cathode at one end and an anode on the other side. Before inserting the proteins, they will be treated with a detergent, so they will be negatively charged. After this the proteins, will be loaded into the gel. By applying an electric field to the gel the proteins will be moving towards the cathode. This movement is influenced by the porous gel, in a way that the bigger molecules will move slower than the smaller molecules. After a period of time, the molecules with the same size will form a band. Every band will then contain molecules of one particular size.

FCS traces could be taken from these bands. The results show the diffusion coefficient and the concentration of the molecules in the band. The size of the molecules could be determined from these quantities.

2.5 D

IFFUSION

In this project the diffusion of a protein plays a crucial role in measuring the size of the molecules in the sample. Bigger molecules will have a slower diffusion, therefor the diffusion coefficient found in experiments can tell what the size is of the molecules [17]. In order to achieve this, we need to understand the diffusion in solution and in gel. I will start this discussion with the diffusion of a spherical bead and then extend this to the diffusion of a protein.

2.5.1 The Rouse Model

A spherical bead of mass m and radius a in a solution experiences a force in opposite direction to its velocity [17]. This force is of size −𝜉𝒗, with ξ the friction constant [17]. The bead will also experience random forces from the continuous bombardment of the surrounding solvent molecules [17]. Therefor the equations of motion can be written as [17]

𝑑𝒓

𝑑𝑡= 𝒗 22

Figure 6: A schematic overview of the gel electrophoresis process, where the sample is loaded into the gel. By applying an electric field to the gel, the proteins will be moving towards the cathode, so different band will be formed.

(14)

11 𝑑𝒗

𝑑𝑡 = −𝝃𝒗 + 𝑭 23

Where the friction constant is given by [17] 𝜉 =6𝜋𝜂𝑠𝑎

𝑚 24

With 𝜂𝑠 the viscosity of the solvent. The solution of the equations of motion can be written as [17] 𝒗(𝑡) = 𝒗0𝑒−𝜉𝑡+ ∫ 𝑑𝜏 𝑒−𝜉(𝑡−𝜏)𝑭(𝑡)

𝑡

0 25

And integrating this equation will give a formula of the displacement of the bead [17] 𝒓(𝑡) = 𝒓0+ 𝒗0 𝜉 (1 − 𝑒 −𝜉𝑡_{) + ∫ 𝑑𝜏 ∫ 𝑑𝜏}𝜏 ′_𝑒−𝜉(𝜏−𝜏′)_𝑭(𝜏′₎ 0 𝑡 0 26

From here we can write the mean square displacement as [17] 〈(𝒓(𝑡) − 𝒓0)2〉𝒗0 = ( 𝒗0 𝜉) 2 (1 − 𝑒−𝜉𝑡₎2₊3𝑘𝐵𝑇 𝑚𝜉2 (2𝜉𝑡 − 3 + 4𝑒−𝜉𝑡− 𝑒−2𝜉𝑡) 27 For large t, this equation will become [17]

〈(𝒓(𝑡) − 𝒓0)2〉 = 6𝑘𝐵𝑇

𝑚𝜉 𝑡 28

Using 〈(𝒓(𝑡) − 𝒓0)2〉 = 6𝐷𝑡 will give the Einstein equation [17] 𝐷 =𝑘𝐵𝑇

Ϛ 29

With 𝜉 =_𝑚Ϛ. In the case of a protein, a single bead is not enough to describe its diffusion process. According to Rouse, a protein can be seen as a Gaussian chain, consisting of N+1 beads, connected by N springs of strength 𝑘 = 3𝑘𝐵𝑇/𝑏2 [17]. We can assume that each bead in the chain will experience the same friction and the diffusion coefficient is independent of the position of the bead [17]. This model is called the Rouse chain. The mean square displacement of the center of mass can be calculated from [17]

𝑿0(𝑡) = 𝑿0(0) + ∫ 𝑑𝜏 𝑭0(𝜏) 𝑡

0 30

So that the mean square displacement is given by [17] 𝑔𝐶𝑀 = 〈(𝑿0(𝑡) − 𝑿0(0))2〉 = 〈∫ 𝑑𝜏 𝑡 0 ∫ 𝑑𝜏′𝑭0(𝜏) ∙ 𝑭0(𝜏′) 𝑡 0 〉 = 6𝐷 𝑁+1𝑡 ≡ 6𝐷𝐺𝑡 31

From there we can write the diffusion coefficient of the center of mass as [17] 𝐷𝐺 =

𝑘𝐵𝑇

(𝑁+1)Ϛ 32

So the diffusion coefficient of the protein in solution depends on 𝑁−1 [17]. When a big molecule moves through water, it only comes to the resistance of small water molecules. In gel, it happens in a different way. Gel consists of bigger molecules (gel fibers)[18]. So when a protein is moving in the gel, it cannot move as freely as in water. A protein spans several pores in the gel[18]. The gel fibers restrict the lateral movement, so it can be seen as the proteins are moving in a tube[18]. In this way, we can replace the obstacles by a fictitious tube[18]. The short length scale dynamics of the chain

(15)

12 inside its tube being unimportant in this model, the chain is simply replaced by a “primitive” chain with rigid links moving along the tube axis[18]. The primitive chain has N segments, one per gel pore occupied by the polymer chain[18].

2.5.2 Reptation model

The reptation model implies the diffusion of molecules in the gel electrophoresis process[18]. Gel electrophoresis is a process for separating different molecules by their size applying electric field to a gel, the molecules start moving through it. The bigger molecule will face the higher resistance after a curtain time the molecules separate in bands, the smaller bands at the bottom and the bigger band at the top of the gel. The reptation model could be used to calculate the diffusion coefficient.

Figure 7: Reptation theory scheme: A) In gel the DNA spans several points. B) Because these points (gel fibers) restrict the movement of the DNA, it can be seen as this DNA molecule is moving in a tube. C) Inside the tube, the short length scale dynamics are not important. So the DNA chain is simply replaced by a ¨primitive¨ chain[18].

The following derivation is derived from the book ¨The Theory of Polymer Dynamics¨[19]. With the assumption of the properties of the polymer are represented by the Rouse model, consisting of 𝑁 segments with bond length (𝑏) and friction constant (𝜉) [19]. The dynamics can be visualized by the tube model [19]. Figure 7 shows a polymer moving through a primitive chain [19]. In short time scale, the polymer is wriggling around the primitive path [19]. On a longer time scale, the primitive path changes and so it creates and destroys the ends of the primitive path [19]. The mathematical treatment of this model is complicated, because of the short time wriggling of the polymer [19]. We are focused on the large-scale motion of the polymer, so the details of the wriggling motion are irrelevant [19]. Neglecting the short time movement in the calculation and taking the following assumptions [19].

(16)

13 • The primitive chain has constant contour length (L)

• The primitive chain can move back and forth only along itself with a certain diffusion constant (𝐷𝑐)

• The correlation of the tangent vectors 𝑢(𝑠, 𝑡) and 𝑢(𝑠′, 𝑡) decreases quickly with |𝑠 − 𝑠′_| The first assumption states that the fluctuations of the contour length could be neglected, the second states that the motion of the primitive chain is reptation and the third states that the conformation of the primitive chain becomes Gaussian [19]. The diffusion constant 𝐷𝐺 of the centre of mass is given as [19]

𝐷𝐺 = 𝑘𝐵𝑇𝑎2

3𝑁2_𝜉𝑏2 33

With 𝑇 the temperature, 𝑘𝐵 Boltz constant, 𝑎 the step length of the primitive chain, N the number of beads in the polymer

So for the FCS results we can expect a 𝑁−2 dependence on the diffusion coefficient. So when we plot the diffusion coefficient against the α-synuclein aggregates size, we can expect a -2 slope in a log-log plot of the size dependence.

Using the techniques discussed in this section helps us to observe single molecules and detect the properties of these single molecules. This project is focused on α-synuclein aggregation. To find the early events in α-synuclein aggregation, α-synuclein is labeled with fluorescent dyes. The size of the aggregates in an α-synuclein sample could be analyzed by using Burst Analysis, FCS and FRET. The next section discusses the materials and methods used for performing Burst Analysis, FCS and FRET experiments.

(17)

14

3 MATERIALS & METHODS

3.1 G

LASS

C

LEANING

P

ROTOCOL

The coverslips (nr. 1.5, thickness 0.16 – 0.19 mm, from Gerhard Menzel GmbH, Germany) were sonicated for 30 minutes first in acetone and then in 4 M NaOH solution. In between these steps, the coverslips were rinsed three times in water (de-ionized MilliQ water, 30 min in Sonicator) and sonicated in water (de-ionized MilliQ water, 30 min in Sonicator). The glasses were gently dried using nitrogen flow. Where mentioned cleaned coverslips were treated with 100 μL 0.5% BSA to prevent sticking of the protein.

3.2 L

ABELING

P

ROCEDURE

The α-synuclein mutant A140C was incubated overnight at 4o_{C with 2-fold molar access of the Alexa}

488 and Alexa 647 separately, which gave A140C-Alexa 488, A140C-Alexa 647. The free dyes and possible aggregates were removed by size exclusion chromatography. The samples were stored at a temperature of 193.15 K.

3.3 D

YES

For fluorescent measurements, α-Synuclein is labeled with fluorescent dyes. For this purpose, we have used three different dyes: Alexa 488, Alexa 647 and Atto 655.

3.3.1.1 Alexa 488

The absorption maximum is at 495 nm and the emission maximum is at 519 nm [20]. The lifetime of Alexa 488 is 4.1 ns [21]. The extinction coefficient is 73,000 cm−1_M−1_{[20] and the fluorescence}

quantum yield is 0.92 [21]. The diffusion constant is 4.3×10−6 cm2_s−1_{(at 25}o_{C, in water) [22]}

3.3.1.2 Alexa 647

The absorption maximum of Alexa 647 is 651 nm and the emission maximum is at 667 nm [20]. The lifetime of Alexa 647 is 1.0 ns [21]. The extinction coefficient is 270,000 cm−1_M−1 _{[20] and the}

quantum yield is 0.33 [21]. The diffusion constant is 3.3×10−6 cm2s−1 [23].

3.3.1.3 Atto 655

The absorption max of Atto 655 is 663 nm and the emission max is 684 nm [24]. The lifetime of Atto 655 is 1.8 ns [24]. The extinction coefficient is 125,000 cm-1_M-1_{and the quantum yield is 0.30 [24].}

Diffusion constant is 4.26 ∙ 10−6 𝑐𝑚2𝑠−1 (25o_{C, in water)}_[23].

3.4 S

ET

U

P

The experiments are performed using a single molecule dual color fluorescence microscope. In this microscope, the sample is excited using two lasers, performing on different wavelengths. Therefor a molecule will absorb a photon of one of these lasers and emit a photon corresponding to its emission wavelength. A crucial part of these microscope is to separate the fluorescent light coming from the different molecules. In the next part I will explain the excitation path and the emission path

separately. The excitation path will combine the two lasers and focus the beam on the same spot in the sample. And the emission path will filter the fluorescent light from the laser light, separate the fluorescent light coming from the different molecules and focus its light on the detectors.

(18)

15

3.4.1.1 Excitation path

Two pulsed lasers, 488 nm and 635 nm diode lasers (Power Technology), are used for exciting the sample. Filtering the laser light is done by a 488 nm and 635 nm band pass filter. A dichroic mirror overlaps the laser light of the two lasers, 635 nm laser light is passing through the dichroic mirror and 488 nm laser light is reflected by it. The combined bundle is focused on the input end of an optical single-mode fiber, so the output end of the single-mode optical fiber emits a clean gaussian beam profile (TEM00). This gaussian beam is passed through a collimetry lens. The objective (60x, NA=1.2

water imersion or 100x, NA=1.4 oil imersion, Olympus) focusses the beam on the sample.

3.4.1.2 Emmision path

The epifluorescent light, collected by the objective, is filtered by a notch filter. The epifluorescent light is focussed on a 50 µm pinhole. The light coming from the pinhole was passed trough a dichroic mirror for separating the 488 nm and 635 nm fluorescent light. These beams were focussed on two single-photon avalance photodiodes (Perkin Elmer Inc., USA). The signal from the diodes was read out using a TimeHarp200 counting board (PicoQuant GmbH, Berlin, Germany). The anaylisis of the signal was done by the SimphoTime software (PicoQuant, GmbH, Berlin, Germany).

The rep rate lasers were adjusted to 40 MHz and (unless stated elsewhere) on 20 µW laser intensity, measured on the objective. The transmission of the optical fiber was 16% and 10% for the 488 nm

Figure 8 First part of the excitation path, with 1) two lasers (488 nm and 635 nm), 2) 488 nm filter, 3) 633 nm filter, 4) two mirrors, 5) a dichroic mirror, 6) a gray filter and 7) the fiber coupler.

Figure 9 Second part of the excitation path, with 1) the fiber decoupler, 2) a mirror and 3) the scanning stage.

Figure 10 Emission path, with 1) scanning stand, 2) a mirror, 3) a notch filter, 4) a lens (focal length 16 cm), 5) a pinhole, 6) a lens (20 cm), 7) a dichroic mirror, 8) two lenses (focal length 4 cm) and 9) two APDs (for blue and red fluorescent light). 1 1 2 ₃ 2 2 3 2 3 4 2 4 2 5 2 6 2 7 2 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 1 9 1

(19)

16 and 635 nm laser beams, respectively, and the transmission of the pinhole was 20% for the 488 nm laser light. The signal was processed by a TimeHarp200 counting board (Picoquant Gmbh, Berlin, Germany). The Symphotime software is used for data processing. Especially for calculating and plotting the lifetime, plotting the Fluorescence Correlation Spectroscopy (FCS) traces and analyzing time traces.

Figure 11 FCS Trace, with the lifetime fit in order to reduce the after pulsing, and the FCS trace.

3.5 A

LIGNMENT

C

HECK

3.5.1 Atto 655 Intensity Measurement

Atto 655 intensity measurement was done for checking the alignment of the set up. It was performed using the 635 nm diode laser with laser intensity of 20 μW measured on the objective, at 40 MHz. The Atto 655 solution was diluted to 10 nM using a 10 mM TRIS buffer of pH 7.4. A drop of 100 μL sample was placed on a clean coverslip. The laser light was focused using the water immersion objective, 60x, NA=1.2. The objective was focused 20 μm in solution. The intensity was measured using the Oscilloscope function of the SimphoTime software.

The intensity of 10 nM Atto 655 with a laser power of 20 μW (635 nm laser) is expected to be around 250,000 photon counts/second [25]. 10 nM ATTO 655 intensity was 300,000 photon counts/s. On a previous set-up, 250,000 photon counts/second was obtained. So the intensity of 10 nM Atto 655 is good.

3.5.2 Lifetime measurements Alexa 488 and Atto 655

The second check for the set-up performance is done by measuring the lifetime of the known sample. The lifetime measurements were performed using the 488 nm diode laser or the 635 nm diode laser, depending on the sample (Alexa 488 or Atto655, respectively). The laser intensity was 20 µW on the objective, and at 40 MHz. The Alexa 488 and Atto 655 samples were diluted to a 10 nM concentration, using a 10 mM TRIS buffer of pH 7.4. A drop of 100 μL sample was placed on a clean coverslip. The laser light was focused using the water immersion objective, 60x, NA=1.2. The

objective was focused 20 μm in solution. The lifetime was measured using the point scan function of the SimphoTime software and analyzed with the FLCS function of the SimphoTime software.

Another set up performance check is performed by measuring the lifetimes of Alexa 488 and Alexa 647. From theory, the values of the lifetimes of Alexa 488 and Alexa 647 are expected to be 4.1 ns and 1.9 ns for Alexa 488 and Alexa 647, respectively. By fitting the lifetime graphs, a lifetime of 4.11 ns and 1.87 ns for Alexa 488 and Atto 655, respectively, were measured. So it is as expected.

(20)

17 3.5.3 Nanobead Resolution

To check the resolution, the nanobead sample was measured. This was done with a laser power (488 nm diode laser) around 1 µW, adjusted so the fluorescence intensity was around 200000 counts/s, at a frequency of 40 MHz. The 100 µL sample (10-5_{M TransFluoSpheres carboxylate-modified, 0.04um,}

633/720) was spin coated on a clean coverslip. The laser light was focused using the oil immersion objective, 100x, NA=1.4. The objective was focused on the surface of the coverslip (highest intensity). The images were made using the Area scan function of the SimphoTime program. Those images were exported to a burst distribution analysis program for measuring the average resolution of the

measured molecules.

The resolution of the nanobeads should be 291 nm, when the alignment is optimized. By analyzing the area scans in the burst distribution analysis program, a mean value for the x- and y-radius should be obtained. The average radius was 272±0.10 nm and 260±0.09 nm for the x- and y-radius, respectively.

Table 1 The mean and SD values of the resolution of the nanobeads.

X-radius Y-radius

Mean 272 260

SD 0.10 0.09

3.6 B

URST

A

NALYSIS

P

ROGRAM

In order to perform burst analysis on the time traces collected by symphotime, we use a burst analysis program. This program will select the bursts and calculate all the important quantities, like burst duration, photons/burst and FRET efficiency.

Burst Analysis is performed by using Alternating Laser Excitation (ALEX). This is a technique were the sample is excited by alternating between the two lasers in the set up. The two pulsed lasers send every 25 ns a laser pulse towards the sample, where one laser sends a pulse at 12.5 ns after the other laser has send a pulse towards the sample. The fluorescent light is detected by two detectors, which detect the fluorescent light after the sample is excited by the 488 nm laser and 633 nm laser, respectively. The individual burst in the signal are analyzed to obtain the burst duration and the number of photons per burst. From these properties the diffusion coefficient and size of the molecules in the sample could be obtained.

(21)

18 There are 4 different photons detected during the experiments: red detected/red excited (RemRex),

blue detected/red excited (BemRex), red detected/blue excited (RemBex) and blue detected/blue excited

(BemBex). The red excited photons are the photons detected within the first 12 ns of the life time

graph, the blue excited photons are the photons detected within the second 12 ns of the life time graph, the red detected photons are the photons detected at the red APD and the blue detected photons are the photons detected at the blue APD. The BemRex photons are coming from background

noise, because a red excited molecule cannot transfer energy to a blue fluorescence molecule. These four different photons are used in the calculation of the FRET efficiency. For every burst the FRET efficiency 𝐸 is calculated by the formula

𝐸 = 𝑅𝐵

𝑅𝐵+𝐵𝐵 34

In this way, a graph is formed as shown in figure 14. In the top graph the FRET efficiency is shown. This graph shows a FRET efficiency of 0.28. In the graph on the side, the stoichiometry is shown. The stoichiometry is calculated by the blue excited photons, divided by the total photons. So the

stoichiometry 𝑆 is calculated by

𝑆 = 𝑅𝐵+𝐵𝐵

𝑅𝐵+𝐵𝐵+𝑅𝑅 35

In the perfect measurement, the stoichiometry has to be 0.5. In this case, there are as many blue excited photons as red excited photons. Also, an average of the burst duration and photons/burst is calculated in this program.

(22)

19

Figure 14 The burst analysis calculations, with the FRET efficiency and stoichiometry graph of the selected bursts.

Using this program, we can calculate the FRET efficiency, the burst duration and the photons/burst. With a sample of fluorescently labeled α-synuclein, in theory one can compare the intensity of the burst, the FRET efficiency and the burst duration of the sample and can predict what kind of aggregate is in the sample. Prior to start, the program has to be tested. This is done by placing an artificial time trace in the burst analysis program, so we can predict what the outcome has to be.

3.7 DNA

BURST ANALYSIS

A single DNA string was labeled with Alexa 488 and Atto 655, separated by 10 nm. The samples were diluted using 10 mM TRIS buffer pH 7.4. For the determination of the best concentration, a labeled molecule concentration of 5 nM, 2.5 nM, 1 nM, 0.5 nM, 0.25 nM and 0.1 nM were used, excited with a laser intensity of 20 µW. For the determination of the optimal laser intensity, a laser intensity of 5 µW, 10 µW, 15 µW, 20 µW, 25 µW and 30 µW were tested, with a labeled molecule concentration of 0.25 nM. 100 µL of labeled sample was placed on a clean coverslip. The objective was focused 20 µm into solution. The time traces were taken using Symphotime and were placed in the burst analysis program. The bursts of the time trace were selected using an InterphotonTimeThreshold and a Photons/burstThreshold of 0.1 ms and 10 photons, respectively.

3.8 Α-S

YNUCLEIN BURST ANALYSIS

The α-synuclein sample was diluted using 10 mM TRIS buffer pH 7.4. A labeled molecule

concentration of 0.25 nM was used. The sample was excited using a laser intensity of 20 µW. The measurements were performed with samples consisting of A140C Alexa 488, A140C Atto 655 or a mixture of both. 100 µL of labeled sample was placed on a clean coverslip. The objective was focused 20 µm into solution. The time traces were taken using Symphotime and were placed in the burst analysis program. The bursts of the time trace were selected using an Interphoton Time Threshold of 0.1 ms and a minimum number of photons per burst of 10 photons per burst.

3.9 Gel electrophoresis

SDS gel electrophoresis was performed with 12.5% SDS gel, 200 V, 400 mA, 50 min, cracking buffer without β-mercaptoethanol and no boiling of the samples. The experiment was done to separate the aged samples by size.

(23)

20

3.10 F

LUORESCENCE

C

ORRELATION

S

PECTROSCOPY

The FCS experiments in solution are performed using pH 7.4 TRIS buffer. The samples were freshly prepared before each measurement. The volume of sample placed on the coverslip was 100 µL. The concentration of labeled protein and Atto 655 was 1 nM. The objective was focused 20 µm into solution to avoid the detection of proteins sticking to the surface, unless stated otherwise. The time traces were taken for 5 minutes.

The experiments in gel are performed by cutting a band from the gel, made by the Gel

electrophoresis method described above. The gel was placed on a cleaned coverslip, incubated with 100 µL of miliQ, for moving the gel to the center of the coverslip. The objective was focused 20 µm into solution to avoid the detection of proteins sticking to the surface.

The results of both experiments were obtained and analyzed using the SimphoTime program. The data obtained was fitted to the following equation:

𝐺(𝑡) = 1 𝑁(1 + 4𝐷𝑡 𝑤₀2) −1 ∗ (1 + 4𝐷𝑡 𝑘2_𝑤 02) −1 2 ∗ (1−𝐹+𝐹𝑒− 𝑡 𝜏𝐵) 1−𝐹 36 With the eccentricity (k) the lateral 1/e2_{-radius 𝑤}

0, the axial 1/e2-radius 𝑧0, 𝜏𝐵 the decay time of blinking reactions, and F the fraction of molecules related to blinking, 𝑁 the number of molecules, 𝑡 the time and 𝐷 the diffusion coefficient.

(24)

21

4 RESULTS

4.1 B

URST

A

NALYSIS

The sizes of the molecules in a sample can be determined using time traces from the dual color fluorescence microscope. At a low enough concentration, when only a single photon is in the confocal volume, the detected burst duration will depend on the sizes of the molecules. A bigger molecule will have a lower diffusion coefficient and will spend more time in the confocal volume, which gives a longer burst duration. The aggregation process can also be followed by looking at the burst intensities. When every molecule is labeled with a dye, a dimer will give twice the burst

intensity of a monomer, because the dimer has two dyes labeled to it. The burst analysis experiments are discussed in the following part, starting with testing the working of the burst analysis program. 4.1.1 Artificial time trace measurements

As discussed in the introduction, the probability of detecting the nth_{photon at a time t later is given}

by the Poisson distribution: 𝑃𝑛(𝑡) = 𝑛̅𝑛 𝑛!𝑒 −𝑛̅ ₃₇ With 𝑛̅ = 𝜉𝐼𝑡 = 𝐶𝑡 38

Where I is the intensity of the signal and 𝜉 is proportional to the area illuminated, and is equal to the emission probability per unit time per unit intensity. The burst analysis program can be tested using artificial time traces with these statistics. The FRET efficiency and stoichiometry could be varied by changing the distribution of the RemRex, RemBex, BemRex and BemBex photons in the artificial time trace.

The calculated FRET efficiency and Stoichiometry of the artificial time trace can be compared to the expectation values, so the calculations can be checked.

The values of RemRex, RemBex, BemRex and BemBex in the artificial time trace are shown on the left side of

the Table 2.1. In the middle of the table are the expected values of the FRET efficiency and the stoichiometry. And on the right part of the table the measured values of the mean FRET efficiency and mean Stoichiometry.

The FRET efficiency is calculated by equation 34. It depends on the BemBex and RemBex photons. The

FRET efficiency is 0 when RemBex is zero and BemBex is non-zero, 1 when RemBex is non-zero and BemBex is Table 2 The constants, expected values and the results of the artificial time trace burst analysis

BemRex RemRex BemBex RemBex Eexpected Sexpected Emean Esd Smean Ssd

0 100 0 0 0 0 NaN NaN NaN NaN

0 50 50 0 0 0.5 0 0 0.39 0.11 0 50 25 25 0.5 0.5 0.5 0.2 0.39 0.11 0 50 0 50 1 0.5 1 0 0.39 0.11 0 0 100 0 0 1 0 0 1 0 0 0 75 25 0.25 1 0.25 0.12 1 0 0 0 50 50 0.50 1 0.5 0.14 1 0 0 0 25 75 0.75 1 0.75 0.12 1 0 0 0 0 100 1 1 1 0 1 0

(25)

22 zero and 0.5 when RemBex equals BemBex. The expectation values of the time traces are shown in table

2. The calculated FRET efficiencies of these time traces show similar results. Only when BemBex and

RemBex are both zero the calculations fail, because the denominator of the FRET efficiency equals

zero. In all the other cases the calculated FRET efficiency is equal to the expectation value.

The Stoichiometry is calculated by equation 35. It depends on the RemRex, RemBex, and BemBex photons.

It is given by the ratio between RemBex + BemBex and RemBex + BemBex + RemRex. The stoichiometry is 0

when there are only RemRex photons in the time trace, it is 1 when there are no RemRex photons and it

is 0.5 when the excitation photons are equally divided. These expectation values are shown in table…. The calculated stoichiometry is comparable to the expectation values. It is shown in table 2 that the stoichiometry is a little bit off for the time traces with a stoichiometry of 0.5. The calculation fails for the first time trace because of the error in the calculation of the FRET efficiency.

4.1.2 DNA Burst Analysis

In the previous part, it is shown that the calculation in the burst analysis program produce similar results as the expected values. It is now important to find the best conditions for performing the α-synuclein burst analysis experiments. So we need to ask ourselves the questions: What is the ideal laser intensity? And what is the best concentration to perform the experiments?

We can find answers to these questions by using a stable molecule with a known FRET efficiency. Α-synuclein cannot be used for this, because the molecules can react to each other and will give different values for the FRET efficiency. For finding the right conditions we need to use a sample which will give in theory the same FRET efficiency. Therefore, we used a sample with Alexa 488 and Atto 655 labeled to a string of DNA, so that the dyes are separated by 10 nm. This string of DNA will then give a FRET efficiency of 0.5 when it is measured in the confocal volume. The stoichiometry will also be 0.5, because every string of DNA is labeled with both dyes.

We can vary the concentration and laser intensity to find the best conditions for performing the α-synuclein burst analysis experiments. Observing individual bursts is important to perform the calculations in the burst analysis program. Overlapping burst will increase the variance of the FRET efficiency. Therefor we need to observe one molecule in the confocal volume. The concentration has to be low enough to fulfill this criterion, but a low concentration means less burst per second and will increase the duration of the experiment. The best concentration will be a concentration low enough so the number of overlapping bursts can be neglected, but high enough to observe lots of burst in a period of 10 minutes.

A high laser intensity will cause the fluorescent molecules to emit a photon more frequently. Therefor a high laser intensity will give a higher burst intensity. This will improve the accuracy of the calculations, because the distribution between the different photons (explained in the previous part) is more accurate for more detected photons. A disadvantage of a high laser intensity is that the chance of bleaching is increased during the experiment.

First the best concentration for the experiment was found by changing the concentration of the DNA sample. The experiments are performed as described in section 2.2.7. The samples consisted of concentrations in the range of 5 nM, 2.5 nM, 1 nM, 0.5 nM, 0.25 nM and 0.1 nM. The time traces were obtained by placing the samples in the set up and were used in the burst analysis program. The lifetime graphs were used to select the blue and red excited photons, so the time traces of the blue and red excited photons could be obtained as shown in figure 15.

It is clear to see that by decreasing the concentration, the measured photon intensity is decreasing. This is because there are too many molecules in the confocal volume. All these molecules will emit

(26)

23 photons in the same time interval and the burst of the individual molecules cannot be seen. This is the case in 5 nM and 2.5 nM. At these concentrations, there always at least one fluorescent molecule in the confocal volume, because the intensity never reaches zero. At a concentration of 1 nM the first individual bursts are observed, but also in this time trace the intensity almost never reaches zero. The first time trace with clearly individual burst is the time trace of 0.5 nM. As can be seen the time trace reaches zero when no molecule is in the confocal volume and it gives peaks when a molecule went through the confocal volume, but also in this time trace a lot of bursts are overlapping. The time trace of a sample with a DNA concentration of 0.25 nM shows more individual burst coming from a single molecule. The time trace gives most of the time an intensity of 0, so most of the time there is no molecule in the confocal volume. This is exactly what we are looking for. A sample of 0.1 nM DNA gives even better time traces. The distances between the bursts are big, so they can be selected accurately by the burst analysis program. The disadvantage of this concentration is that there is a low number of bursts per second. Therefor the duration of the experiment needs to be extended in order to get accurate calculations of the FRET efficiency and the Stoichiometry.

Placing the time traces of the DNA sample with concentrations of 0.5 nM, 0.25 nM and 0.1 nM in the burst analysis program will give the graphs shown in figure 16. The expected FRET efficiency and Stoichiometry of a string of DNA with two dyes separated by 10 nm are 0.5 and 0.5 respectively. The calculated mean value of the stoichiometry in the samples with concentration of 0.5 nM, 0.25 nM and 0.1 nM are respectively 0.58, 0.54 and 0.52, which are comparable with the expectation value. The deviations can be coming from the different quantum yields of the dyes, the detection efficiency, reflection losses and the alignment of the set up. The expectation value of the FRET efficiency was 0.5. We observed a mean FRET efficiency of 0.23, 0.25 and 0.26 for a concentration of 0.5 nM, 0.25 nM and 0.1 nM, respectively. These values are the half of the expectation value. This can be coming from different formations of the DNA string or from photons coming from the background noise that influence the calculations.

Other quantities that are calculated are the number of burst, mean photons/burst and mean burst duration. These quantities can tell us if we have selected individual burst. If only individual bursts were selected, decreasing the concentration will not influence the mean photons/burst and the mean burst duration. Looking at the calculations of the DNA samples, we observe a mean

photons/burst of 113.93, 59.94 and 56.38 for 0.5 nM, 0.25 nM and 0.1 nM, respectively, and a burst duration of 6.01E-3, 1.37E-3 and 1.31E-3 for 0.5 nM, 0.25 nM and 0.1 nM, respectively. The values of the samples with 0.5 nM DNA concentration are different than the samples with DNA concentration of 0.25 nM and 0.1 nM. Therefor a sample of DNA with a concentration of 0.5 nM has too many overlapping bursts. The number of bursts in the 0.25 nM sample is 1247 and in the 0.1 nM sample is 331, but the results are comparable. Therefor the best concentration for performing the experiments is 0.25 nM.

(27)

24

5 nM 2.5 nM

1 nM 0.5 nM

0.25 nM 0.1 nM

Figure 15 The time traces obtained from samples of α-synuclein with various concentration (5 nM, 2.5 nM, 1 nM, 0.5 nM, 0.25 nM and 0.1 nM). The red and blue time traces were obtained based on the lifetime graph of the sample.

(28)

25 0.5 nM

0.25 nM

0.1 nM

Figure 16 The FRET efficiency and Stoichiometry calculations. In this figure, it can be seen that the FRET efficiency lies around 0.25 and the Stoichiometry around 0.6. The most of the bursts are observed at a concentration of 0.5 nM, but in the time trace of 0.5 nM (shown in figure 1), some background is also selected as noise.

(29)

26 The second question we wanted to answer was: What is the ideal laser intensity for performing the burst analysis experiment? Increasing the laser intensity will increase the number of photons/burst, but will also increase the chance of bleaching the fluorescent molecule. Therefor we need to find the lowest laser intensity at which the calculations will give the expected values. The ideal concentration was found to be 0.25 nM, so this was used for finding the ideal laser intensity. The laser intensity was increased from 5 μW to 30 μW in steps of 5 μW. The time traces of the 0.25 nM DNA sample at different laser intensities are shown in figure 17.

The probability of exciting a molecule in the confocal volume at a particular time will increase by increasing the laser intensity. Therefor the molecule will be excited more often at higher laser intensities than at lower intensities and will emit more photons. The bursts in the detected signal will consist of more photons and has a higher burst intensity at high laser intensity. So increasing the laser intensity will increase the burst intensity. Figure 17 shows the same dependence. The bursts in the time trace of 5 µW have bursts with 10 till 20 photons/burst, while the bursts in the time trace of 30 µW have bursts with 30 till 70 photons/burst. Selecting individual bursts can be performed more accurate for bursts with a high number of photons/burst. Background noise will influence the selection process of bursts with a low number of photons/burst. Therefor the calculation of the FRET efficiency and Stoichiometry will be more accurate for bursts with a high number of photons/burst. A disadvantage of increasing the laser intensity is that the contribution of the surrounding molecules to the background noise is increased. The time trace of 30 µW will almost never return to 0, because the background noise is too high. Another disadvantage of increasing the laser intensity is the increased probability of bleaching the fluorescent molecules in the sample. This cannot be observed in the time traces, but will influence the calculations. Therefor we need to find the lowest intensity for which the calculations give the expected values.

Placing the time traces in the burst analysis program will give results as shown in figure 18 and figure 19. The mean number of photons/burst is 10.57 photons/burst, 15.43 photons/burst, 20.61

photons/burst, 59.94 photons/burst, 53.81 photons/burst and 65.07 photons/burst for the time trace with a laser intensity of 5 µW, 10 µW, 15 µW, 20 µW, 25 µW and 30 µW, respectively. The mean number of photons/burst is increasing by increasing the laser intensity from 5 µW to 20 µW. The mean number of photons/burst seems to be constant at laser intensities above the 20 µW. This could be caused by the limitations of the burst analysis program or by using the wrong selection criteria for selecting individual bursts.

Like said before, we expect the calculations of the FRET efficiency and Stoichiometry to be more accurate for higher laser intensities. Figure 18 and Figure 19 show this behavior. The graphs of the FRET efficiency and Stoichiometry of 5 µW, 10 µW and 15 µW show a wide distribution of bursts with varying values for the FRET efficiency and Stoichiometry. The FRET efficiency and Stoichiometry distributions of the time traces with a laser intensity of 20 µW, 25 µW and 30 µW are similar to the distributions obtained by changing the concentration of the DNA sample. The FRET efficiency and Stoichiometry are given by 0.25 and 0.54, 0.25 and 0.56 and 0.23 and 0.63 for time traces with laser intensity of 20 µW, 25 µW and 30 µW, respectively. There is a peak at 1 which is increasing by

increasing laser intensity in the Stoichiometry distribution of the time traces with laser intensity of 20 µW, 25 µW and 30 µW. The origin of this peak is unknown and the effect of this peak is negligible. From this experiment, we can conclude that the calculations of the time traces with a laser intensity of 20 µW, 25 µW and 30 µW gave the expected results. To avoid bleaching the fluorescent molecules we will use a laser intensity of 20 µW in the α-synuclein burst analysis experiments.

(30)

27

5 µW 10 µW

15 µW 20 µW

25 µW 30 µW

Figure 17 The intensity dependence of 0.25 nM DNA sample. The number of photons/burst is increasing with increasing laser intensity as expected. At 25 and 30 uW the background noise from the other DNA molecules in the solution start increasing, so the intensity to get the nicest result is 20 μW.

(31)

28 5 µW

10 µW

15 µW

(32)

29 20 µW

25 µW

30 µW

(33)

30 4.1.3 Α-Synuclein burst analysis

An ideal laser intensity and concentration of 20 µW and 0.25 nM are found by performing DNA burst analysis. These values are also used in the α-synuclein experiments. We can follow the aggregation process of α-synuclein by labeling α-synuclein with Alexa 488 or with Atto 655. A monomer will consist of α-synuclein labeled with only one dye. These monomers will give bursts with a

stoichiometry of 1 when the molecule is labeled with Alexa 488 and it will give a stoichiometry of 0 when Atto 655 is labeled to it. A dimer can consist of two α-synuclein molecules, one labeled with Alexa 488 and the other labeled with Atto 655. These dimers will give bursts with a stoichiometry between 0.5 and a FRET efficiency higher than zero, because the molecules are close enough to make energy transfer possible. So we can define the bursts coming from a monomer or dimer based on the FRET efficiency and Stoichiometry graphs.

The oligomers could be observed by changing the labeling percentage. A FRET efficiency of 0.5 observed with a labeling percentage of 100% will probably be a dimer, but a FRET efficiency of 0.5 observed with a labeling percentage of 1% will probably be a higher oligomer. FRET is only observed when the dyes are close enough to make energy transfer possible. If 1% of the molecules is labeled with a dye, then the chance of observing a dimer with two dyes is 0.0001, when the sample is 100% labeled, the chance of observing a dimer with two dyes labeled to it is 1.0.

So when a burst of 0.5 FRET efficiency is observed at a labeling percentage of 1%, the molecule in the confocal volume is probably an oligomer. The diffusion coefficient of the oligomer could be

measured by selecting the burst with a non-zero FRET efficiency. Then the diffusion coefficient could be calculated from the burst duration. The diffusion coefficient will also give an estimate of the size of the molecule. Bigger molecules will have a lower diffusion coefficient, because the volume of the sample that has to move in order to move the bigger molecule is much bigger.

In this experiment, we try to analyze an aged synuclein sample. The difference between an aged α-synuclein and a sample with α-α-synuclein monomers is that more bursts will have a non-zero FRET efficiency. Before placing an aged α-synuclein sample in the set up, we need to know what to expect for an synuclein sample. Therefor we started our synuclein experiments with samples of α-synuclein monomers labeled with only Atto 655 of samples of α-α-synuclein labeled with only Alexa 488. In these experiment, we expect to find an average FRET efficiency and Stoichiometry of 0 and 1 for α-synuclein labeled with only Alexa 488, respectively. We expect for α-synuclein labeled with only Atto 655 a Stoichiometry of 0 and FRET efficiency that is not defined, because the FRET efficiency is based on the Rex photons and not on the Bex photons. The photons coming from the α-synuclein

molecule labeled with Atto 655 will never be Rex.

The result of a sample of α-synuclein monomers labeled with Alexa 488 is shown in figure 20. The FRET efficiency gives a peak at 0 and the Stoichiometry gives a peak at 1, as expected. The result of a sample of α-synuclein monomers labeled with Atto 655 is shown in figure 21. The Stoichiometry gives 0 as we expected, but we observed a peak at 1 for the FRET efficiency. As said before the calculation of the FRET efficiency is not based on the Bex photons. So the peak at a FRET efficiency of

1 can come from the burst that are not properly selected. We need to combine the samples in order to compare the size of the FRET efficiency peak with the peak observed at 0 for Alexa 488.

(34)

31 Figure 20 The FRET efficiency and Stoichiometry graphs of a sample of α-synuclein labeled with Alexa 488.

Figure 21 The FRET efficiency and Stoichiometry graphs of a sample of α-synuclein labeled with Atto 655.

A sample of monomers where 0.5% is labeled with Alexa 488 and 0.5% is labeled with Atto 655 will give a Stoichiometry distribution with peaks at 0 and 1. The peak at 1 is coming from monomers labeled with Alexa 488 and the peak at 0 is coming from monomers labeled with Atto 655. The FRET efficiency is expected to be 0, because we only have monomers labeled with one dye, so energy transfer is not possible. Aging the sample will increase the number of oligomers in the sample. The monomers start to aggregate and bigger oligomers will be observed. These oligomers are labeled with more dyes and therefor energy transfer is possible. We expect that the number of bursts in which energy transfer is observed is increasing. The Stoichiometry will be 0.5 when a molecule is labeled with both dyes. This is more likely to observe in an aged sample. Therefor we expect an increase in bursts with a stoichiometry of 0.5.

The results of a sample with 1% labeled WT α-synuclein is shown in figure 22. This sample consists of monomers labeled with Alexa 488 or labeled with Atto 655. The Stoichiometry shows two peaks at 0 and 1. The sizes of the peaks are comparable, which means that there are as many photons coming from Alexa 488 as coming from Atto 655. We did not expect to observe bursts with a stoichiometry around 0.5, but the stoichiometry distribution shows that there are bursts with a stoichiometry around 0.5. It can be coming from overlapping bursts, when a monomer labeled with Atto 655 and Alexa 488 are observed at the same time in the confocal volume. The FRET efficiency distribution of 1% WT α-synuclein shows a peak at 0, as expected, but has a long tail ending at a FRET efficiency of 0.8. It is not clear where these bursts with a non-zero FRET efficiency are coming from. We saw a

(35)

32 peak at 1 in the FRET efficiency distribution of α-synuclein labeled with Atto 655, but this was a very sharp peak and this peak is not observed in the FRET efficiency distribution of 1% labeled WT α-synuclein. This tail can come from selecting background noise as a burst. This is very unlikely because we used the same selection criteria as in the α-synuclein samples with monomers labeled with only Alexa 488 or labeled only with Atto 655. There we did not observe this tail.

The results of the 10d aged 1% labeled WT α-synuclein sample are shown in figure 22. The Stoichiometry shows peaks at 0 and 1, as expected. We also expected to see an increase in bursts with a stoichiometry of 0.5, but we do not see this increase compared to the 0h WT 1% labeled α-synuclein sample stoichiometry distribution. The FRET efficiency distribution of 10d aged 1% labeled WT synuclein sample looks similar to the FRET efficiency distribution of the 0h WT 1% labeled α-synuclein sample. There is a peak at a FRET efficiency of 0 and a long tail, which ends at a FRET efficiency of 0.8. We expected an increase in bursts with a non-zero FRET efficiency, but this is not observed.

0h WT 1% labeled

10d WT 1% labeled

(36)

33

4.2 F

LUORESCENCE

C

ORRELATION

S

PECTROSCOPY

So the α-synuclein burst analysis experiments did not help us in understanding the aggregation process. The results of the sample of labeled monomers were similar to the 10 days aged sample. Therefor we could not determine the size of the molecules based on the stoichiometry and FRET efficiency. The sizes of the molecules could also be determined based on the diffusion coefficient of the molecules. The diffusion coefficient will influence the average burst intensity and the burst duration. So in theory we could determine the sizes of molecules based on the burst intensity and burst duration, but the burst selection criteria are limiting the accuracy of the burst intensity and burst duration. A more accurate technique to measure the diffusion coefficient of a sample is Fluorescence Correlation Spectroscopy.

Fluorescence Correlation Spectroscopy is a technique that uses the intensity fluctuations in order to measure the autocorrelation function given by

𝑔(𝜏) = 1 𝑁∙ (1 + 4𝐷𝜏 𝑤_𝑥𝑦2 ) −1 ∙ (1 +4𝐷𝜏 𝑧₀2) −1/2 39

With N the number of molecules in the confocal volume, D the diffusion coefficient, 𝜏 the time, 𝑤𝑥𝑦 is the width in the xy-plane and 𝑧0 the length of the measurement volume. The diffusion coefficient of the sample could be determined by the autocorrelation function when the eccentricity 𝑘 = 𝑧0

𝑤𝑥𝑦, width in the xy-plane and the length of the measurement volume are known. In the case of α-synuclein the diffusion coefficient of the sample will decrease by aging the sample. An aged sample contains bigger molecules, which will decrease the diffusion coefficient. So based on the diffusion coefficient we can estimate the sizes of the molecules.

4.2.1 Atto 655

Before we can measure the diffusion coefficient of an α-synuclein sample, we need a sample with a known diffusion coefficient for measuring the eccentricity, the width in the xy-plane and the length of the confocal volume. The eccentricity, the width in the xy-plane and the length of the confocal volume are quantities that are determined by the alignment of the set up. By comparing the dual color fluorescence setup with other setup in the lab, we can expect an eccentricity between 5 and 10 and a width in the xy-plane between 300 and 500 nm. From theory, we expected an eccentricity of 5.05 in an ideal setup.

The confocal volume measurements are performed as described in Methods and Materials. The FCS traces of samples with different concentrations of Atto 655 are shown in figure 23. As can be seen the starting point of the autocorrelation function increases by decreasing labeled molecule concentration. This is comparable with the expectations, because 𝑔(0) = 1

𝑁∝

1 𝐷.

Early events in alpha-synuclein aggregation

Early Events in α-Synuclein Aggregation

Burst analysis and Fluorescence Correlation Spectroscopy on a

dual color fluorescence microscope

CONTENTS

ABSTRACT

1 INTRODUCTION

2 THEORY

2.1 F

2.2 F

R

E

T

I



I e

w

2.33

r

n

r

NA







n 

1.3

NA 

0.6

5.05





P

S

2.3 B

A

2.4 G

E

2.5 D

3 MATERIALS & METHODS

3.1 G

C

P

3.2 L

P

3.3 D

3.4 S

U

3.5 A

C

3.6 B

A

P

3.7 DNA

3.8 Α-S

3.9 Gel electrophoresis

3.10 F

C

S

4 RESULTS

4.1 B

A

4.2 F

C

S