[Mayo lecture summary] Part 2 Designing Protein Libraries

In this lecture, Prof. Mayo introduces about DBIS (Diversity Benefit applied to Interacting Sets of amino acids) and related success cases.

• Why Computational Protein Design is needed?

'Diversity of function' which is usually achieved by high level of mutation and 'Preservation of function' which is usually achieved by low level of mutation are anti-correlated with each other.

The solution to this problem is the usage of computational protein design. This is about a structure based method that allows for high mutation rate (diversity of function) with maintenance of stably folded struture (preservation of function).

• Result of Mutation
  

Typically lower mutation rates means higher preservation of functions.

Regarding this, the result of SCMF (self-consistent mean field) was counterintuitive. This lower mutation rate gave the lowest preservation of functions. (SCMF mutated two most mutation tolerant positions.)

Likewise, highest mutation rate gave the greatest preservation of functions.

Therefore, this design calculation is good since it shows where to put the mutations for given library size to select AA that most likely to stabilize the fold.

(Random mutation of course, gave the very low preservation of function as expected.)

• Making Diverse Function while Keeping Originally Desired Function

Although I couldn't understand all of this lecture, the main theme of this slide is this: this design calculation method succeeded making diverse functions while keeping the originally desired function (fluorescence). This success can be shown at the result by DBIS-ORBIT (DBIS-Optimization of Rotamers By Iterative Techniques).

Lower mutation rate (DBIS-ORBIT-4 degenerate codons at 4 positions, SCMF-32 degenerate codons at 2 positions) showed little diverse functions.

(Randomly mutated protein and error prone PCR showed no significant change from wild type of AA - no diverse function)

'Fold before Function' is a concept that computed stability is a good surrogate for function in library design, and Mayo team realized this.

• More Success Case
  

Prof. Mayo team introduced another success case. BFP was enhanced in terms of enhanced quantum yield, reduced pH sensitivity, much improved in photobleaching half-life.

source: http://www.youtube.com/watch?v=G76DINEwT0s&feature=channel

[Mayo lecture summary] Part 1 Protein Design by Computation

In this lecture, Prof. Mayo introduces basic concept of computational protein design and methods with exemplar processes.

• Protein Fold Prediction vs Protein Design

Protein fold prediction is about prediction of protein structure based on sequence while protein design is about prediction of protein sequence which has a desired function

• Methods of Computational Protein Design
  

Protein backbones typically come from PDB.
Due to conformational flexiblity, rotamer libraries are considered.

Building atom-based forcefields are important for accurate protein design.

With these backbone, rotamer, forcefields issues, combinatorial optimization algorithms are used for calculation.

Negative design is about a protein that prevents a certain function (ex. unwanted aggregation).

• Design Process Flow
  

1. Specify design objectives
2. Computationally design AA sequence using ORBIT
3. Assemble gene for new protein
4. Express gene in favorite host
5. Isolate and purify protein
6. Evaluate properties of designed protein
7. Iterate design

(these processes are similar to my QSAR experiences regarding that to accurately model some biological/chemical responses, iterative modification of modeling techniques is exploited)

• Sample Energy Calculation
  

To design a protein sequence which will bring a desired function, energy calculation is needed.

As a simple example (one body interaction such as the interaction between AA and backbone), many types of the energy between two AA are calculated such as van der waals force, electrostatic interaction (coulomb potential), hydrogen bond (direction specified), and solvation potential (penalize the exposure to hydrophobic surface area or polar surface area depending on design purpose).

Two body interaction is about the interaction between pairs (such as the interaction between the AAs).

Finally, the combinatorial energy is calculated as a following example, the combinatorial energy of Ser3 at position 1 and Ser2 at position 2 (-9.0) is a sum of 2 one body energy (-0.4, -3.1) and 1 two body energy (-5.5).

Among these combinatorial energies, the lowest energy is selected.

source: http://www.youtube.com/watch?v=a8CVm5S35ko