dynamics of filament growth

Upload: tarun-meena

Post on 14-Apr-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/30/2019 Dynamics of filament Growth

    1/39

    Dual Degree Project Presentation

    Dynamics of filament growthUnder the supervision of

    Prof.Mandar Inamdar

    Neeraj KookadaDepartment of Civil Engineering

    IIT Bombay

    May 2, 2013

  • 7/30/2019 Dynamics of filament Growth

    2/39

    Flow of Presentation

    Introduction Objectives

    Stochastic Simulations of Chemical Reactions

    Gillespie Algorithm

    Degradation Reaction

    Degradation and Production Reaction

    Diffusion using Compartment-based approach Force generation by filament polymerization

    Introduction

    2-State Model

    Theoretical Stall force for a single filament

    Stall force for a two filament model

    Pseudo code Simulation Results and Discussions

    Conclusion and Future work

    References

  • 7/30/2019 Dynamics of filament Growth

    3/39

    Introduction Optimization problems in dynamic, stochastic environments are an increasingly

    important part of biological systems.

    Simulation methods can provide solutions for complex partial differential

    equations and problems which involve uncertainty about models to draw

    conclusions about the related variables.

    The growth dynamics of fibres like microtubules and actin filaments play an

    important role in cellular biology. These fibres generate forces that play a role in cellular motility processes such as

    the motion of chromosomes during mitosis.

    Recent experimental advances have allowed to determine the dynamics of growth

    of cytoskeleton proteins with a single-molecule precision at different conditions.

    These experiments suggest that the complex biochemical transitions and

    intermediate states in the cytoskeleton proteins influence their dynamic properties

    and functions and thus the control the force that a cell can produce.

  • 7/30/2019 Dynamics of filament Growth

    4/39

    Objectives To develop stochastic models for processes like degradation, production and

    diffusion which mimic the interactions between different components in molecularbiology

    To apply the Gillespie algorithm as it turns out to be a more efficient model

    reducing the computational time when compared with Monte Carlo simulations to

    predict the time evaluation of processes.

    To develop a two-state model for a single actin filament which assumes thepolymerization and depolymerisation, as well as switching of states, where the

    filament is restrained to grow against a wall (movable) and the resulting force

    generated is studied

    The maximum force generated by the filaments stall force and its relations with

    the velocity of the wall and length fluctuations of the filament is plotted using

    Monte Carlo Simulations. The one filament model is extended to a two filament model and numerically it is

    derived that the stall force for N filaments is N times the stall force for a single

    filament.

  • 7/30/2019 Dynamics of filament Growth

    5/39

    Gillespie Algorithm

    Introduction

    Gillespie Algorithm generates a statistically correct trajectory (possible solution) of

    a stochastic equation

    Gillespie algorithm allows a discrete and stochastic simulation of a system with fewreactants because every reaction is explicitly simulated.

    A Gillespie realization represents a random walk that exactly represents the

    distribution of the Master equation.

    The physical basis of the algorithm is the collision of molecules within a reaction

    vessel (well mixed). All reactions within the Gillespie framework must involve at most two molecules.

    Reactions involving three molecules are assumed to be extremely rare and are

    modeled as a sequence of binary reactions.

  • 7/30/2019 Dynamics of filament Growth

    6/39

    Flowchart for Gillespie Algorithm

    Initialize Start Initialize System time to zero

    Event list

    Create event list of all possible transitions

    Calculate the time to-event outcomes

    Time step

    If events available, perform next event, otherwise stop

    Advance time

    Terminate

    Re-compute the event list

    Time reached

    Stop

  • 7/30/2019 Dynamics of filament Growth

    7/39

  • 7/30/2019 Dynamics of filament Growth

    8/39

    Algorithm

    1. Choose a small time step t. We compute the number of molecules A(t) at times t = i t, i =

    1, 2, 3, . . . , as follows. Starting with t = 0 and A(0) = n0, we perform two steps at time t,

    2. Generate a random number r uniformly distributed in the interval (0, 1).

    3. If r < A(t)kt, then put A(tt) = A(t)1; otherwise, A(tt) = A(t).

    4. Continue with step (1) for time t +t.

    5. Generate a random number r uniformly distributed in the interval (0,1)6. is computed from eq (1.3)

    7. Compute the number of molecules at time t by A (t) = A (t)-1.

    8. Continue with step (4).

    since r is a random number uniformly distributed in the interval (0, 1), the probability that

    r < A(t)kt is equal to A(t)kt.

    We use k = 0.1sec-1 and t = 0.005 sec. A(0) = 20, Time = 30s.

    For the given input values, A(t)kt (0, 0.01).

  • 7/30/2019 Dynamics of filament Growth

    9/39

    Result

    Results of ten realizations of SSA (a2)(c2)(solid lines; different colours show different

    realizations) and stochastic mean plotted by the dashed line

    Since the function A(t)

    has only integer values

    {0, 1, 2, . . . , 20}, someof the computed

    curves A(t) partially

    overlap.

    It would be very

    unlikely that there

    would be two

    realizations giving

    exactly the same

    results.

    Details of realization

    A(t) depend on thesequence of random

    numbers)

    Average values give a

    reproducible

    characteristic of the

    system

  • 7/30/2019 Dynamics of filament Growth

    10/39

    Degradation & Production Reactions

    Consider the chemical reactions

    The first reaction describes the degradation of chemical A with the rate constant k1. It was

    already studied previously as reaction (1.1). We couple it with the second reaction which

    represents the production of chemical A with the rate constant k2. This reaction is similarto a random walk carried by a particle in the following case.

    The particle in the given figure has three mechanisms r1,r2 and r3 possible at different

    possibilities of p1,p2 and p3 respectively, where r1,r2 and r3 are set of mutually exclusive

    events.

    (2.4)

  • 7/30/2019 Dynamics of filament Growth

    11/39

    Algorithm

    (starting with A(0) = n0 at time t = 0):

    1. Generate two random numbers r1, r2 uniformly distributed in (0, 1).

    2. Compute 0 = A(t)k1 + k2.

    3. Compute the time when the next chemical reaction takes place as t where from

    (1.3)

    4. = 1/0 ln(1/r1)

    5. Compute the number of molecules at time t by

    A(t ) =A(t) 1 if r2 < k2/ 0;

    = A(t) 1 if r2 k2/ 0.

  • 7/30/2019 Dynamics of filament Growth

    12/39

    Graphs

    frequency over 10000 simulationsDistance v/s Time

  • 7/30/2019 Dynamics of filament Growth

    13/39

    Diffusion using Compartment-based

    approachDiffusion is the random migration of molecules (or small particles) arising from motion due tothermal energy. As shown by Einstein, the kinetic energy of a molecule (e.g. protein) isproportional to the absolute temperature. In particular, the protein molecule has a non-zeroinstantaneous speed at, for example, room temperature or at the temperature of the humanbody. A typical protein molecule is immersed in the aqueous medium of a living cell.Consequently, it cannot travel too far before it bumps into other molecules (e.g. water molecules)in the solution. As a result, the trajectory of the molecule is not straight but it executes a randomwalk.

    Consider diffusion of 1000 molecules

    .

    We simulate directly the time evolution of 40 compartments. Considering a computational

    domain [0,L] and dividing it into K = 40 compartments of length h = L/K = 25 m. Denote thenumber of molecules of chemical species A in the i-th compartment [(i 1)h, ih) by Ai, i = 1, . . .

    ,K. The rate of diffusion is assumed as d. Applying the Gillespie SSA provides a correct model of

    diffusion provided that the rate constant d is chosen as d = D/h2 where D is the diffusion constant

    and h is the compartment length.

  • 7/30/2019 Dynamics of filament Growth

    14/39

    AlgorithmGenerate two random numbers r1, r2 uniformly distributed in (0, 1).

    1. Compute propensity functions of reactions by i= A

    i(t)d for i = 1, 2, . . . ,K. Compute

    +

    2. Compute the time at which the next chemical reaction takes place as

    3. If < = , then for j = (1,2,,k-1),

    <

    =

    =

    4. The number of molecules at time t is given by:

    Aj(t ) = Aj (t)-1

    Aj+1(t ) = Aj+1 (t) + 1

    Ai(t ) = A

    i(t) for I j, I j 1

  • 7/30/2019 Dynamics of filament Growth

    15/39

    5. If

    = , then for j = (2,3,,k-1),

    +

    = <

    +

    =

    =

    =

    6. The number of molecules at time t is given by:

    Aj(t ) = Aj (t)-1

    Aj-1(t ) = Aj-1 (t) + 1

    Ai(t ) = Ai(t) for I j, I j 1

  • 7/30/2019 Dynamics of filament Growth

    16/39

    We simulate 1000 molecules starting from position 0.4 mm in the interval [0,L]for L = 1 mm. We use K = 40.

    Since 0.4 mm is exactly a boundary between the 16th and 17th compartment,the initial condition is given by A16(0) = 500, A17(0) = 500 and Ai(0) = 0 fori 16, i 17.

    As D = 104 mm2 sec1, we have d = D/h2 = 0.16 sec1. The numbers Ai(t),

    i = 1, . . . ,K, at time t = 4 min, are plotted. For example, only one chemical

    reaction occurs per time step. Consequently, only two propensity functions change and need to be updated

    here N = 1000 is the total number of molecules in the simulation (this numberis conserved because there is no creation or degradation of the molecules inthe system).

    Hence, we need to re-compute only when there is a change in or ,i.e. whenever the boundary compartments were involved in the previousreaction

    0 1 k

  • 7/30/2019 Dynamics of filament Growth

    17/39

    Results

    Compartment-based approach

  • 7/30/2019 Dynamics of filament Growth

    18/39

    Force generation by filament

    polymerizationIntroduction

    Actin filaments and microtubules are key components of the cytoskeleton of

    eukaryotic cells.

    In order to understand the dynamical behaviour of single filaments such as actin or

    microtubules and the force they can generate, discrete stochastic models havebeen developed that incorporate at the molecular level the coupling of hydrolysis

    and polymerization.

    Single polymerizing filaments can generate forces in the pico-newton range, as has

    been demonstrated experimentally for microtubules (MTs). Such force generation

    mainly relies on the gain in chemical bonding energy upon monomer attachment. An opposing force slows down filament growth, which finally stops at the stall

    force representing the maximal polymerization force a filament can generate.

    Therefore, the stall force is the essential quantity to characterize polymerization

    forces.

  • 7/30/2019 Dynamics of filament Growth

    19/39

    2-state Model

    Assumptions

    Consider two rigid surfaces: one fixed where filaments are nucleated (nucleating

    wall) and one movable (barrier) whose position is defined to be the position of the

    filament furthest away from the nucleating wall .

    We do not model the internal structure of the filaments, and in particular we do

    not account for hydrolysis.

    After nucleation, the filaments grow or shrink by exchanging monomers with the

    surrounding pool of monomers, which acts as a reservoir.

    The filaments are coupled only through mechanical contact with the barrier.

    The filament consists of two species of monomers (type blue and type orange, as

    depicted in fig.(4.1). The filament is modelled as a two-state model depending onwhich type of monomer is in contact with the barrier. Only the filaments in contact

    feel the force exerted by the barrier on them, and as a result, this changes their

    polymerization rates as compared with free filaments.

  • 7/30/2019 Dynamics of filament Growth

    20/39

    2-state Model

    Assumptions

    We assume that a type blue and type orange monomers can be added to any free

    filament with a rate u0and removed with a rate w and w respectively. Similarly,

    the monomers can be added to the filament in contact with the barrier with a rate

    u(f) and removed with a rate w(f) and w(f) respectively. Assume that the barrierexerts a constant forcefon the filaments in contact

  • 7/30/2019 Dynamics of filament Growth

    21/39

    2-State Model

    Dynamics of a single filament

    For a filament that has exchanged work with the barrier through addition or loss of

    monomers, we use the following relation:

    00

    Where: u0 and w0 represent the polymerization and depolymerisation rates for monomers

    attaching to free filaments (not bounded by the barrier),

    u, w represent the polymerization and depolymerisation rates for monomers

    attaching to bounded ends of the filaments.

    Consider the switching of states (blue/orange = 1/2) with rates k12 and k21

    respectively.

  • 7/30/2019 Dynamics of filament Growth

    22/39

    For a single filament, the polymerization and depolymerisation rates are as

    follows:

    0 0 1

    where, the parameter [0, 1] depends on the location of the energy barrierwhich the wall has to overcome in moving one monomeric step.

    We have defined two states 1 and 2 for the filament. The state actually depends

    on the last molecule in the filament. We define the rate of conversion from state 1

    to 2 as k12 and vice versa. Let the probability of a filament of length l being in

    state 1 or 2 at time t be P1(l,t) and P2 (l,t) respectively. The probabilities obey the

    following :

  • 7/30/2019 Dynamics of filament Growth

    23/39

    1 ,

    1 1, + 11 + 1 , + 212 , + 1 + 12 1 ,

    2 ,

    2 1, + 22( + 1 , ) + 121(, ) + 2 + 21 2(,)

    (1())/212 121

    2 121 212()

    Summing over all l, we get

    For steady state, probabilities are independent of time.

    1 212 121 0212 121

  • 7/30/2019 Dynamics of filament Growth

    24/39

    The average position is given by

    , + ,

    =

    The velocity is given by

    (1) 1 ,

    +2 ,

    0

    1 1 , + ( 2)2 , 1 1 + ( 2)2

    In steady state

    (1) 1 21 + 2 1212 + 21

    Putting V(1) = 0 and u = ue-Fstall, we find the stall force for one filament system

  • 7/30/2019 Dynamics of filament Growth

    25/39

    Stall force expression for a single

    filament The stall force is defined as the value of the force applied on the barrier for which

    the velocity given by equation (4.7) vanishes.

    At stall force, the velocity can be assumed to be zero, which gives:

    ( ++ )

  • 7/30/2019 Dynamics of filament Growth

    26/39

    Stall force for a two filament model

    Dynamics of a two-filament system

    Here, two filaments have been modeled as a two state model with the rates of

    polymerization and depolymerization same as had been considered in the previouscase.

  • 7/30/2019 Dynamics of filament Growth

    27/39

    Pseudo Code : Stall force v/s length of filament

    function [ final_length ] = biofil( F )

    %UNTITLED Summary of this function goes here

    % Detailed explanation goes here

    iter_max=10000;

    k1=0.3; k2=0.15;

    u0=1;

    w1=0.2; w2=0.1; a1=50e3;

    b1=50.1e3;

    t_max=1; t=2;

    iter=1;

    stateA = 1; stateB=1; A=zeros(iter_max,2);

    while(t>t_max)

  • 7/30/2019 Dynamics of filament Growth

    28/39

    Pseudo Code

    if(a1>b1)

    s1=0;s2=1;

    elseif(b1>a1)

    s1=1;s2=0;

    else

    s1=u0*exp(-F)/(u0+u0*exp(-F));

    s2=u0*exp(-F)/(u0+u0*exp(-F));

    end

    %molecules entering

    r=rand*2;

    if(s2*u0*exp(-F)+s1*u0>r)

    a1=a1+1;

    end

    r=rand*2;

    if(s1*u0*exp(-F)+s2*u0>r)

    b1=b1+1;

    end

  • 7/30/2019 Dynamics of filament Growth

    29/39

    Pseudo Code

    %state change

    r=rand*2;

    if(stateA==1)

    if(k1>r)

    stateA=2;

    end

    else

    if(k2>r)

    stateA=1;

    end

    end

    r=rand*2;

    if(stateB==1)

    if(k1>r)

    stateB=2;

    end

  • 7/30/2019 Dynamics of filament Growth

    30/39

    Pseudo Code

    else

    if(k2>r)

    stateB=1;

    end

    end

    %molecules leaving

    r=rand*2;

    if(stateA==1)

    if(w1>r && a1>0)

    a1=a1-1;

    end

    else

    if(w2>r && a1>0)

    a1=a1-1;

    end

    end

  • 7/30/2019 Dynamics of filament Growth

    31/39

    Pseudo Code

    r=rand*2;

    if(stateB==1)

    if(w1>r && b1>0)

    b1=b1-1;

    end

    else

    if(w2>r && b1>0)

    b1=b1-1;

    end

    end

    %loop exit conditions

    if(iter>iter_max-1)

    t=0;

    end

    A(iter,1) = a1; A(iter,2) = b1;

  • 7/30/2019 Dynamics of filament Growth

    32/39

    Pseudo Code Stall Force v/s Velocity

    function [ output_args ] = velocity( )

    %UNTITLED Summary of this function goes here

    % Detailed explanation goes here

    v=zeros(23,1);

    i=1;

    f=linspace(0,4.4,23);

    for fs=0:0.2:4.4;

    v(i)=(biofil(fs)-50e3)/1000;

    i=i+1;

    end

    plot(f,v);

    end

  • 7/30/2019 Dynamics of filament Growth

    33/39

    Simulation Results and Discussions

    From the graph, it is evident that the position of the barrier isnt changing much with the

    iterations. The numerical value of the force used in the code is 4.23 units which is nearly

    equal to twice the stall force for a single element. Thus in absence of hydrolysis, the stall

    force of a system of N filaments is equal to N times the stall force for each filament

  • 7/30/2019 Dynamics of filament Growth

    34/39

    For forces below the stall force, the length of the filament goes on increasing with

    time.

    On the other hand, for forces above the stall force, the length of the filament goes on

    decreasing with time.

  • 7/30/2019 Dynamics of filament Growth

    35/39

    The mean velocity of the barrier becomes zero at stall force which from the monte carlo

    simulations comes out to be around 4.23

  • 7/30/2019 Dynamics of filament Growth

    36/39

    Conclusion and Future Work

    We presented SSAs for systems of chemical reactions and molecular diffusion. Thealgorithms for simulating systems of chemical reactions were based on the Gillespie

    algorithm.

    We presented a model of diffusion in this paper based on the chain of chemical

    reactions computing the time evolution of the numbers of molecules in compartments.

    Based on the models, we obtained the exact result for the stall force of polymerizing

    ensembles of rigid filaments with lateral interactions. The stall force is a linear functionof the number N of filaments in the ensemble and increases linearly with the lateral

    interaction.

    These results have been confirmed by simulations using the Gillespie algorithm.

    Our results are relevant for the interpretation of experimental data on the force-

    velocity relation in microtubule polymerization and in the polymerization of bundles of

    interacting actin filaments or microtubules.

    Future work would include extending the parameters like stall force, velocity for N-

    filament structures. The study and effects of hydrolysis on the polymerization and

    depolymerisation should be understood and a new exhaustive model should be

    prepared

  • 7/30/2019 Dynamics of filament Growth

    37/39

    References Erban, Radek, and S. Jonathan Chapman. "Stochastic reaction--diffusion processes." Physical Biology6 (2009): 046001.

    S. Andrews and D. Bray, Stochastic simulation of chemical reactions with spatial resolution and single molecule detail,Physical Biology 1 (2004), PP (137151)

    Li, Song. Numerical simulation of electrochemical adsorption process by kinetic Monte Carlo methods. Diss. Texas Tech

    University, 2010.

    Neal, RadfordM. "Monte Carlo implementation of Gaussian process models for Bayesian regression and classification." arXiv

    preprint physics/9701026 (1997).

    Ullah, M., Schmidt, H., Cho, K., & Wolkenhauer, O, Deterministic modelling and stochastic simulation of biochemical

    pathways using MATLAB. Systems biology, 153(2), 53.(2006)

    Scatena, Guido, Development of a stochastic simulator for biological systems based on the calculus of looping sequences.

    Diss. Masters thesis, University of Pisa, 2007.

    Bernstein, David. "Simulating mesoscopic reaction-diffusion systems using the Gillespie algorithm." Physical Review E71.4

    (2005): 041103.

    Sandmann, Werner. "Stochastic simulation of biochemical systems via discrete-time conversion." In Proceedings of the 2nd

    Conference on Foundations of Systems Biology in Engineering, pp. 267-272. 2007

    D. T. Gillespie. Approximate accelerated stochastic simulation of chemically reacting systems. The Journal of Chemical

    Physics, 115, issue 4:1716{1733,2001)

  • 7/30/2019 Dynamics of filament Growth

    38/39

    J.Krawczyk, J.Kierfeld. Stall force of polymerizing microtubules and filament bundles.EPL (93 28006 doi:10.1209/0295-

    5075/93/28006)

    Kostas Tsekouras, David Lacoste, Kirone Mallick and Jean-Francois Joanny. Condensation of actin filaments pushing against

    a barrier. New Journal of Physics 13 (2011) 103032 (16pp) 22011

    E. B. Stukalin and A. B. Kolomeisky. Polymerization dynamics of double-stranded biopolymers: Chemical kinetic approach.

    The Journal of Chemical Physics,122, 104903 (2005).

    G. S. van Doorn, C. Tanase, B. M. Mulder, and M. Dogterom . On the stall force for growing microtubules. Eur. Biophysics.

    Journal. 20, 2 (2000).

    T. L. Hill. Microfilament or microtubule assembly or disassembly against a force. Proc. Natl. Acad. Sci. USA 78, 5613 (1981).

    M. J. Footer, J. W. J. Kerssemakers, J. A. Theriot, and M. Dogterom . Direct measurement of force generation by actin

    filament polymerization using an optical trap. Proc. Natl. Acad. Sci. USA 104, 2181(2007)

    D. Vavylonis, Q. Yang, and B. OShaughnessy. Actin polymerization kinetics, cap structure and fluctuations. Proc. Natl. Acad.

    Sci. USA 102, 8543 (2005).

    Hill T. L. Free Energy Transduction and Biochemical Cycle Kinetics, (Springer) 1989

  • 7/30/2019 Dynamics of filament Growth

    39/39

    Thank You