yingting0803 by nWBA2KPB

VIEWS: 7 PAGES: 36

									     Multiscale Modeling of Phosphorylation and
     Inhibition of the Epidermal Growth Factor
     Receptor Tyrosine Kinase: Linking Somatic
         Mutations to Differential Signaling

                                                Yingting Liu
                             Advisor: Dr. Ravi Radhakrishnan
                               Department of Bioengineering
                                  University of Pennsylvania




University of Pennsylvania             Department of Bioengineering
                                  Outline


   Backgrounds


   Hypothesis and Specific aims


   Experimental design and preliminary results




     University of Pennsylvania               Department of Bioengineering
 ErbB Family Receptors and the Signaling Pathways
                             Yarden and Sliwkowski, nature reviews, 2001




University of Pennsylvania                    Department of Bioengineering
Tyrosine Phosphorylation and Receptor Inhibition




Zhang and Kuriyan,Cell, 2006


 University of Pennsylvania        Department of Bioengineering
                EGFR Kinase Domain Mutations




Choi and Lemmon, Oncogene, 2007   Carey and Sliwkowski, Cancer Res, 2006
                                      Zhang and Kuriyan,Cell, 2006

   University of Pennsylvania                 Department of Bioengineering
                        Hypothesis and Methods


We hypothesize that mutants in the EGFR              MD simulation for protein kinase

kinase domain will alter the kinase-inhibitor,
kinase-substrate interactions, and the           Multiple conformation molecular docking
catalytic reaction efficiency of the turn-over
of different EGFR substrates by affecting
the properties of EGFRTK active site,                  MD simulation for complex

therefore lead to differential characteristics
in the downstream signaling in pathways             Structural and energetic analysis
mediated by EGFR.
We propose to employ multiscale                      Inhibition     MD simulation for
computational methods based on molecular                            EGFRTK-ATP-MG-Peptide
docking, molecular dynamics (MD), and                               complex

quantum mechanics molecular mechanics
(QM/MM) simulations to test this
hypothesis.                                       QM/MM calculation on catalysis




     University of Pennsylvania                          Department of Bioengineering
                             Specific Aims



Aim1. Developing empirical force-field parameters for small molecule
inhibitors for use in in-silico docking and molecular dynamics simulations.

Aim2. Exploring the conformational and free energy landscape for
wildtype and L834R mutant EGFR kinase complexed with small molecule
inhibitors and peptide substrates.

Aim3. Modeling the catalytic mechanism and activity of the EGFR
tyrosine kinase.




University of Pennsylvania                         Department of Bioengineering
                             Specific Aims


Aim1. Developing empirical force-field parameters for small molecule
  inhibitors for use in in-silico docking and molecular dynamics
  simulations.

Aim2. Exploring the conformational and free energy landscape for
  wildtype and L834R mutant EGFR kinase complexed with small
  molecule inhibitors and peptide substrates.

Aim3. Modeling the catalytic mechanism and activity of the EGFR
  tyrosine kinase.




University of Pennsylvania                    Department of Bioengineering
             MD Simulation and CHARMM Potential Energy

 Molecular Dynamic (MD) simulations:


    Essential part is the potential energy function. V


 CHARMM potential energy:


V (r )                   2              2               2
              K ( b - b )   K (S - S )   K (  -  )           K (1  cos( n - ))
                 b      0       UB     0              0               
           bonds            UB             angles           dihedrals
                                                      R       12
                                                                     R         
                                                                                     6
                                                      min ij                          qi q j
                                                                 -          
                                                                        min ij
                     K     ( j - j )2                                             
                                          nonbond  rij              r        
                         imp        0             ij
             impropers                                                                 erij
                                                                    ij           



       University of Pennsylvania                                       Department of Bioengineering
                                        Erlotinib Parameterization (1)
 Define new atom types and initiate the parameter set.
 Optimize the structure using ab-initio methods and obtain
 equilibrium constants.
 Obtain partial charges of each atom using CHELPG
 (CHarges from ELectrostatic Potentials using a Grid based method)                                            .
 Get Van der Waals constants (                                       and  ) from existing
                                                               Rmin ij           ij


 CHARMM parameters.
 Guess the force field constants based on those assigned for
 similar structure in existing CHARMM parameters.
          HA

                                                               V (r )                   2              2               2
                                                                             K ( b - b )   K (S - S )   K (  -  )           K (1  cos( n - ))
                                                                                                                                     
          CC3
                                                                                b      0       UB     0               0
                                                                          bonds            UB             angles           dihedrals
          CC3

                                                                                                                 R       12
                                                                                                                                R         
                                                                                                                                                6
                                                                                                                 min ij                          qi q j
PH        CA          HP

                                                                                                                            -          
                                                                                                                                   min ij
                                                                                        ( j - j )2 
     CA         CA
                                                                                K                                                             
                                                                                                     nonbond  rij              r        
                                                                                    imp        0             ij
                                                                        impropers                                                                 erij
                                                                                                                               ij       
     CA         CA    AQ1   H
PH        CA          N            HP
                                                    HA
                                                          HA
                                                               HA
                                                                     HA
                                                                          HA
                                                                                                                                                 
                                               S
          HP   Q2AC             CA             O         T2C        T3C
           Q2AN             CAQ3        CAQ4        C          O            HA
                                                    T2         S
                                                                     HA
                                                               S
                C           CAQ3        CAQ4       T2C    T2   O            HA
                AQ1
          PH          N            C        O             C         T3C
                      AQ1          A        S
                                                   AH          HA
                                                                           HA
                                   HP                     HA




          University of Pennsylvania                                                                         Department of Bioengineering
                                        Erlotinib Parameterization (2)
 Refine Partial charges manually.
                                                                                                   H33


                                                                                                   O3
                                                                                            N1H           H32
                                                                                       N1          H8
                                                               O2
                                               H21                   H22               C6          C8
                                                                                 N3          C7           C9


                                                                                 C19         C18          C13
                                                                     H19               N2          C17

          HA
                                                                                       H11          H17
          CC3

          CC3
                                                                                       O1
PH        CA          HP
     CA         CA
                                                                                H12
     CA         CA    AQ1   H
PH        CA          N            HP                     HA          HA
                                                    HA          HA         HA
                                               S
          HP   Q2AC             CA             O         T2C         T3C
           Q2AN             CAQ3        CAQ4        C           O          HA
                                                    T2          S
                                                                      HA
                                                                S
                C           CAQ3        CAQ4       T2C    T2    O          HA
                AQ1
          PH          N            C        O             C          T3C
                      AQ1          A        S
                                                   AH           HA
                                                                           HA
                                   HP                     HA




          University of Pennsylvania                                                                Department of Bioengineering
                                        Erlotinib Parameterization (3)
Refine dihedral parameters to reproduce ab initio dihedral energy surface.
     Using genetic algorithm to automatically minimize the merit function:
                                        NGRID
                                      i 1
                                                    ( DiC  DiG )2                                         Dihedral potential energy surface
                                                                                                     1.5



     NGRID is the number of potential values




                                                                                 Energy (Kcal/mol)
                                          G
     calculated in the surface. DiC and Di                                                            1


     are potential values from CHARMM and
     GAUSSIAN.                                                                                       0.5

          HA

          CC3

                                                                                                      0
          CC3                                                                                          0      100        200         300       400
PH        CA          HP                                                                                        Dihedral (degree)
     CA         CA

     CA         CA    AQ1   H
PH        CA          N            HP                       HA         HA
                                                      HA         HA         HA
                                                S
          HP   Q2AC             CA              O          T2C        T3C
           Q2AN             CAQ3        CAQ4          C          O          HA
                                                      T2         S
                                                                       HA
                                                                 S
                C           CAQ3        CAQ4         T2C    T2   O          HA
                AQ1
          PH          N            C        O               C         T3C
                      AQ1          A        S
                                                     AH          HA
                                                                            HA
                                   HP                       HA




          University of Pennsylvania                                                                       Department of Bioengineering
                                             Erlotinib Parameterization (4)
Refine force constants to reproduce vibrational eigenvalues and eigenvectors.
     Using genetic algorithm to automatically minimum the merit function:

                       3 N 6
                                                                                                       1
                             (        i
                                               C
                                               i            G
                                                              j max
                                                                    2
                                                                       )               i 
                                                                                              max j (  ic   G )
                          i 1                                                                               j
                                             3N  6

                             ,  ;i ,  j : the ith frequency and eigenvector from CHARMM and GAUSSIAN
                              C           C         G         G
                              i          i
                                                c        G                                          c G
          HA
                            Project              into { }, and find the index jmax which maxmum    .
                                               i         j                                         i   j
          CC3


                                                C    G           c G
                            In the ideal case,          and     
          CC3

PH
     CA
          CA
                CA
                      HP
                                                i    j max      i   j    ij
     CA         CA    AQ1    H
PH        CA          N             HP                            HA         HA
                                                         HA            HA         HA
                                                    S
          HP   Q2AC                CA               O         T2C           T3C
           Q2AN              CAQ3            CAQ4        C             O          HA
                                                         T2            S
                                                                             HA
                                                                       S
                C            CAQ3            CAQ4       T2C       T2   O          HA
          PH
                AQ1
                      N
                      AQ1
                                    C
                                    A
                                                 O
                                                 S
                                                                  C
                                                                       HA
                                                                            T3C        Vaiana, Computer Physics Communications, 2005.
                                                        AH                        HA
                                    HP                            HA




          University of Pennsylvania                                                                       Department of Bioengineering
                Erlotinib Parameterization (5)
                                    ------ Preliminary
                            results
    Water interactions   Interaction Energies         Distance (Å)
                         (Kcal/mol)
                         GAUSSIAN       CHARMM        GAUSSIAN        CHARMM
    N2…HOH                  -6.69           -6.61         2.13           1.91
    N3…HOH_2                -5.33           -5.3          2.32           2.01
    N1H…OHH_2               -6.52           -6.52        2.4415          2.63
     Dipole moment              GAUSSIAN                      CHARMM
    (Debye)                         4.868                         5.07

    Table 1 Water-mediated interactions and dipole moment for erlotinib. The
    ab-initio interaction energies are scaled by 1.16, and the distances should
    offset by –0.1 to –0.2 A. Experimental dipole moments are typically ~10 to
    20% larger than HF/6-31G*.




University of Pennsylvania                                   Department of Bioengineering
              Erlotinib Parameterization (6)
                                  ------ Preliminary
                          results
                                                 1.6


                                                 1.4


                                                 1.2


                                                  1




                             Energy (kcal/mol)
                                                 0.8


                                                 0.6


                                                 0.4


                                                 0.2


                                                  0
                                                   0   50      100     150         200     250   300   350
                                                                       Dihedral (degree)


     Frequency matching                                     Potential surface fitting




                                                             Genetic algorithm efficiency

University of Pennsylvania                                           Department of Bioengineering
                             Specific Aims

Aim1. Developing empirical force-field parameters for small molecule
inhibitors for use in in-silico docking and molecular dynamics simulations.

Aim2. Exploring the conformational and free energy landscape for
wildtype and L834R mutant EGFR kinase complexed with small molecule
inhibitors and peptide substrates.

Aim3. Modeling the catalytic mechanism and activity of the EGFR
tyrosine kinase.




University of Pennsylvania                         Department of Bioengineering
                          Methods: MD simulations

                                                  Molecular Dynamic (MD) protocol:
                                                  •Prepare protein conformation based
                                                  on available crystal structure or
                                                  homologies.
                                                  • Solvate the protein and neutralize
                                                  the systems by placing ions randomly.
                                                  • Minimize the solvated models
                                                  • Heat the system to 300 K
                                                  • Equilibrate at constant temperature
                                                  and constant pressure (300 K and 1
                                                  atm) for 200ps to stable the system.
                                                  • Run productive trajectory.
Solvated model for MD simulation of EGFRTK.
(Iceblue: sodium; yellow: chlorine; orange: protein; tan: water).

     University of Pennsylvania                               Department of Bioengineering
Methods: Multiple-Conformation Molecular Docking

• The idea of molecular docking: to generate a comprehensive set of
  conformations of the receptor-ligand complex and then to rank them
  according to their stability.
•    Single conformation docking: Ligand is flexibility, while receptors
    are usually treated as rigid during docking.
•    Multiple-conformation docking: An ensemble of 100 snapshots of
    the protein is collected from the equilibrated trajectory to perform
    molecular docking. The generated ligand conformations are clustered
    based on the relative RMSD and analyzed to explore the
    conformational and free energy landscape of the interaction between
    protein kinase and the ligands.
      The multiple-conformation docking jobs are submitted in parallel so
    that they will run simultaneously and then cluster the generated
    conformations upon completion of the docking runs using Fortran 90
    program.

University of Pennsylvania                        Department of Bioengineering
             Methods: Binding Free Energy Calculation
• AUTODOCK:
                     Aij Bij                    Cij Dij               qi q j
      G  GvdW   12  6   Ghbond  E (t )  12  6   Gelec 
                    
                 ij  Rij Rij                   R    Rij          ij  ( rij )rij
                                       ij        ij      
            Gtor Ntor  Gsol  SV j  S jVi   e
                                                        (  rij2 / 2 2 )
                                    i
                                  ij


• Molecular Mechanics Poisson-Boltzmann Surface Area (MMPBSA):
  G  Gcomplex - Greceptor - Gligand ;
 G  EMM  GPBSA - TSMM ;
 EMM  Ebond  Eangle  Etors  Eelec  Evdw ;
 GPBSA  Gsolvation  GPB  GSA .

Electrostatic solvation energy: Poisson-
Boltzmann equation.
Nonpolar term: depend on surface area.                                      Sitkoff and Honig 1993

    University of Pennsylvania                                              Department of Bioengineering
                      Kinase-Inhibitor Interactions
                                         ------ Proposed model
•   Motivation: Similar binding conformations presented in crystal structures but
    remarkably increase the binding affinities in L834R mutant systems.
    --- erlotinib (Carey and Sliwkowski, 2006), gefitinib and AEE788 (Yun and Eck
    2007)


•   Specific of Aim: using multiple-conformation molecular docking to obtain six
    top ranked complex conformations based on the approximate free energy from
    AUTODOCK and then perform MD based structural and energetic analysis
    (MMPBSA) for each conformations. Among the six, three conformations will be
    highlighted for analysis based on the more accurate binding free energy.


•   Possible reasons to test: unique interactions between L834R mutant kinase and
    inhibitors, subtle conformational differences, which is hard to be captured by
    crystallographic methods, effect of solvation, …



      University of Pennsylvania                       Department of Bioengineering
                    Kinase-Inhibitor Interactions
                       ------ Preliminary results and future work

  WT                                          L858R
                                    Crystal
                                    conf.




                                   Lowest energy conf.


Top ranked Erlotinib conformations in EGFR wildtype and mutant system.
 Use MD simulations to refine these structures with explicit solvent and resort the
structures using MMPBSA methods.
 Perform structural analysis for the refined conformations to explore the effect of
mutations on kinase-inhibitor interaction.
    University of Pennsylvania                          Department of Bioengineering
                     Kinase-Substrate interactions
                                        ------ Proposed model

•  Motivation: to predict the binding
modes for different substrates and test
the effect of mutation on kinase-
substrate interaction.


•   Substrates: Four seven-residue
sequences derived from the C-terminal
tail of EGFRTK (Y1068,Y1173,Y992
and Y1045).


•    Specific of Aim: perform the multiple-conformation molecular docking
    protocol followed by the MD based structural analysis and free energy
    calculation to predict the best binding modes and obtain the corresponding
    binding affinities, which can be correlated to Km values for each substrate.


     University of Pennsylvania                      Department of Bioengineering
                Kinase-Substrate interactions
                   ------ Preliminary results and future work

 L858R unphosphorylated EGFR
                                               2GS6
       Binding with 10687




University of Pennsylvania                    Department of Bioengineering
                Kinase-Subtrate interactions
                   ------ Preliminary results and future work



                               Approximate Binding
                Substrate       Energy(Kcal/mol)
                s
                             Y1068      Y1173      Y992
                 Wildtype    -5.42      -4.69       -4.7
                  L834R
                             -5.93      -3.78      -5.91
                  mutant




                                     Liu, Purvis and Radhakrishnan,2007

University of Pennsylvania                      Department of Bioengineering
                   Kinase-Substrate interactions
                      ------ Preliminary results and future work


           Free energy contributions of EGFRTK- peptide
           (VPEYINQ) binding from MMPBSA calculation.
           (Kcal/mol)
                 Internal energy          -139.7
                  Polar solvation                    140.5
                 onpolar solvation                    -6.4
             Total binding free energy                -5.6


 Future work: Use MD simulations to refine these structures with explicit
solvent and recalculate the binding free energy using MMPBSA methods.



   University of Pennsylvania                         Department of Bioengineering
                             Specific Aims


Aim1. Developing empirical force-field parameters for small molecule
inhibitors for use in in-silico docking and molecular dynamics simulations.

Aim2. Exploring the conformational and free energy landscape for
wildtype and L834R mutant EGFR kinase complexed with small molecule
inhibitors and peptide substrates.

Aim3. Modeling the catalytic mechanism and activity of the EGFR
tyrosine kinase.




University of Pennsylvania                         Department of Bioengineering
                           Catalytic Mechanism




•   In principle, the reaction mechanism can be either an associative or dissociative
    pathway.
•   pKa and nucleophile coefficient measurements support a dissociative transition
    state. (Kim and Cole, 1998)
•   QM/MM studies of cAMP agree with the dissociate mechanism. (Cheng and
    McCammon, 2005)
     University of Pennsylvania                          Department of Bioengineering
Proposed Catalytic Mechanism for EGFRTK based on cAMP

                                                                                                CH2
                                                                                                  Asp813
                                         O
                                                          -
                                                                   Mg2+
                         ATP                         O                                   O
                                 O           P                                                          O-
                     O                                             -
                                                                   O
                          P                                                         H                 H2
                   CH2                           O            P                              O        C
                                     -
                     O           O
                                                     O-            O                                Peptide
                                         Mg2+

                                                                                                     CH2
                                                                                                              Asp813
                                                 O                     Mg2+
                          ATP
                                                              O-                                O
                                         O           P                                                       O-
                         O                                             O-
                                 P                                                          H                 H2
                     CH2                                 O             P                            O         C
                                             -
                       O                 O                         O
                                                                            O                            Peptide
                                                              2+
                                                         Mg

                                                                                                         CH2
                                                                                                                  Asp813
                                                                               2+
                                                     O                    Mg
                              ATP
                                                                  O-                                O
                                             O           P                                                    O-
                             O                                                      O   -
                                     P                                                          H                 H2
                         CH2                                  O                                                   C
                                             O   -                                  P               O
                           O
                                                                           O                                 Peptide
                                                              Mg2+                  O-


   University of Pennsylvania                                                                           Department of Bioengineering
          Prepare the Enzyme-Substrate System



                                   Blue: 2GS6 bisubstrate;
                                   Pink: ATP conformation in
                                   2ITX;
                                   Yellow: proposed peptide
                                   conformation in aim 2;




University of Pennsylvania          Department of Bioengineering
                           QM/MM Calculation
Molecular Mechanics (MM): cannot account for the covalent transformations of
chemical bonds.
Quantum Mechanics (QM): limited system size due to computational complexity.
QM/MM: Treat atoms involved in chemical reaction with QM and others MM.



                             MM region        Link atoms



                  ATP
                                  QM region
                        MG
                                         PEPTIDE




     University of Pennsylvania                      Department of Bioengineering
                           Umbrella Sampling

• Umbrella sampling enables the calculation of the potential of mean force (free
energy density) along an a priori chosen set of reaction coordinates (or order
parameters), from which free energy changes are calculated by numerical
integration.




                           u(r )              u(r )  u(r )  W (r )

                                               W (r )  kw (r  r0 )    2


                            u( r )



    University of Pennsylvania                      Department of Bioengineering
        Free Energy Landscape Along Reaction Coordinates
                                                                                        CH2
                                                                                          Asp813
                                            O
                                                             -
                                                                     Mg2+
                              ATP
                          O
                                    O           P
                                                        O                        r2 O         O-
                                                                     O-
                               P                                                 H            H2
                        CH2                         O            P                      O     C
                                        -
                          O         O
                                                        O-           O                      Peptide
                                            Mg  2+                          r1
•   Umbrella sampling along two coordinates.
• 25 windows are sampled as a uniform 5×5 grid along
r1 and r2 .
• with each window harvesting a QM/MM MD
trajectory of 2 ps.
•free energy profile as a function of the coordinate will
be calculated using the weighted histogram analysis
method (WHAM).
•   Explore the effect of mutation on the reaction profile.
                                                                                                   Gregersen and York 2003

        University of Pennsylvania                                                          Department of Bioengineering
                       Summary and Significances

Effect of mutation on:
•   Kinase-Inhibitor interactions.
•   Kinase-Substrate interactions.
•   EGFR tyrosine kinase reaction profile.


Significances:
•   generate a rich amount of information concerning structural and dynamic
    properties of the system at atomic level.
•   help to further understand the mechanism of protein kinases inhibition
    and phosphorylation and therefore guide cancer therapy of protein kinase
    systems.




     University of Pennsylvania                       Department of Bioengineering
                             Thanks.




University of Pennsylvania             Department of Bioengineering
           Mutations increase kinase activities




                               Yun et al., (Eck) Cancer Cell (2007)




                                   Zhang et al., (Kuriyan) Cell (2006)

University of Pennsylvania                Department of Bioengineering
          Structural Studies of EGFRTK Active Site


             N-lobe
                         LYS721
                                            αC-helix
G-loop
                                  GLU738
                  ATP              ASP831      A-loop
MET769

                  ASP813


         C-loop                      peptide
                        C-lobe


  University of Pennsylvania                           Department of Bioengineering

								
To top