Modeling Proteins at an Oil Water Interface

1
Modeling Proteins at an Oil / Water Interface
Chemistry 699.08 Final Presentation
Patrick Brunelle December 13, 2001
2
Why care about oil / water interfaces?

• Proteins can have different biological effects,
  depending on their conformation.
• Some proteins have the ability of penetrating
  the membrane of cells, thus, participating in
  the cell functions.
• This can also be a problem (virus are a good
  example).

3
What are the theories available?

• Absorption of a polymer on a flat surface
  (Singer, 1948).
• Absorption of a flexible, high molecular weight,
  polymer on an impenetrable solid surface
  (Silbergberg, 1962).

• Extension to proteins
• Dickinsons Model
• (E. Dickinson and S.R. Euston, Adv. Colloid
  Interfaces Sci., 1992, 42, 89)
• Andersons Model
• (R.E. Anderson, V.S. Pande and C.J. Radke, J.
  Chem. Phys., 2000, 112, 9167)

4
Dickinsons Model - Purpose
Dickinson used his model to get an understanding
of the conformation of milk proteins at the
interface between an oil droplet and water.
For this study, beta-casein was chosen because of
its abundance in milk.
Beta-casein could be consider as a near random
chain of 209 amino acids.
5
Dickinsons Model - Basics
Monte-Carlo Algorithm is used on a tetrahedral
lattice, where each amino acid occupies on
lattice site. The bottom half of the lattice is
occupied by water molecules and the top half is
occupied by oil molecules.
But how to define the hamiltonian of the system???
6
Dickinsons Model - Basics
7
Dickinsons Model - Hamiltonian
The Hamiltonian is defined as follows

• There is no interaction between the solvent
  molecules.

• There is no interaction between two amino acids.

So, the Hamiltonian is the sum of the interaction
energies of the amino acids with the two
different type of solvents.
8
Dickinsons Model - Interaction Energies
Adsorption Energy (KT)
Adsorption Energy (KT)
E(np,oil) E(p,oil) E(c,oil) E(c-,oil)
0.5 -1.0 -10.0 -10.0
E(np,aq) E(p,aq) E(c,aq) E(c-,aq)
-1.0 0.0 5.0 5.0
Solvent Change (KT)
E(np,aq/oil) -E(np,oil/aq) E(p,aq/oil)
-E(p,oil/aq) E(c,aq/oil) -E(c,aq/oil) E(c-,a
q/oil) -E(c-,aq/oil)
-1.0 0.0 10.0 10.0
9
Dickinsons Model - Simulation

• The simulation is run for 5 x 106 steps to reach
  equilibration.
• And sampling is done at every 5 x 103 steps for
  a total of 20 x 106 steps.

• This model seems to agree with the CRISP
  procedure (neutron diffraction experiment). The
  CRISP procedure gave a maximum extension of the
  protein in aqueous phase of 12nm and the modeling
  shows 12nm.

10
Andersons Model - Purpose
Improve from Dickinsons model
11
Andersons Model - Basics

• The Hamiltonian is defined by using the matrix
  of Miyazawa and Jernigan.
• (Macromolecules, 1985, 18, 534)
• A octahedral lattice is used.

12
Andersons Model - Model Protein
FVHTGELYNAKTKGRIMQAESPRVLDS
The model peptide is 27 amino acids long and is
known to fold to one specific conformation. The
folding process is also very fast. (It folds and
unfolds in 3.24 x 107 Monte Carlo steps, in bulk
water)
13
Andersons Model - Hamiltonian
) 1 neighbouring amino acids ) 0 otherwise
Bij is the matrix element of amino acid i and j
from the matrix of Miyazawa and Jernigan.
BUT
The solvents are approximated water is taken
as histidine oil is taken as glycine
14
Andersons Model - Temperature Controlled
TT/Tm
The simulation is run at 0.94T which has been
shown to have 50 of the protein in the folded
state and 50 unfolded.
15
Andersons Model - Partition Function
(x) 1 for x0 (x) 0 otherwise
Each simulation is run for about 2 x 109 steps.
40 000 conformations are used.
16
Andersons Model - Bias Potential
"6.67
Biased free energy
Free Energy
17
Andersons Model - Enthalpy
18
Andersons Model - Entropy
TSt Et - F
Ss St - Sp
19
Andersons Model - Sample
20
Andersons Model - Results

• Based on the simulation, the entropy increases
  during the adsorption.
• However, the strongest driving force is the
  reduction of the unfavourable interaction between
  the oil and the water layer that is reduced by
  the presence of the protein.
• Anderson recommends the use of this model for
  short single domain proteins (50 to 70 amino
  acids).