Notes 20100721 Aron Broom's PhD Thesis Defense

From SnOwy - Ed's Wiki Notebook

Contents 1 Rational Protein Design; Carbohydrate Binding 2 In Vivo Screening 3 In Silico Screening 4 Rational Design 5 Rational Design, Why? 6 Vs in Vivo screening 7 Question 8 Protein Binding Carbohydrates 9 Why Carbs? 10 Approaches; possible methods 11 Protein stability 12 Questions 13 Protein changes 14 Computational approach 15 Detail scaling 16 Model (prototype) selection 17 3RAB revisited 18 Initial work 19 Rational design 20 Question about Glycomics Gateway

Rational Protein Design; Carbohydrate Binding

1. rational design, in silico screening
2. protein binding (carbohydrates)
3. stability
4. approach and methods
5. the 3RAB model revisited

In Vivo Screening

• a method that's often used
• if we're for instance changing binding specificity; increasing catalytic function
  • we create a library
  • use a kind of faulty PCR so we end up with many many mutants
  • use a screening approach -- perhaps a fluorescently tagged molecule
• can we get specificity for a new carbohydrate?
• colonies that can bind it have such a changed specificity.

In Silico Screening

• nothing physical is done
• no PCR is actually done
• randomly change sequence on the computer
  • homology modeling
  • model of initial model, make changes; able to use something even more sophisticated than homology modeling
• get the best results
• best results are then screened in vivo

Rational Design

• rather than screening, take initial gene; model interaction
• improve the binding at a site; we have some idea of what forces we need at a local area
  • engineering approach
• we need specific distances, electrostatics.
• do this once
• express this
  • possible to incorporate in silico screening into this

Rational Design, Why?

• a general solution
  • applicable for any problems that can be modelled
• requires only computational resources
• returns only a putatively good subset of answers
• has considerable room for improvement
• fails too often
• possible to improve: this is the major appeal

Vs in Vivo screening

• requires either many volunteers or robots
• high throughput required
• in practice, cannot be improved upon

Question

• how are we sure that we haven't eliminated something valuable in the search space with rational design?
• how do you select your starting protein? - was discussed later.

Protein Binding Carbohydrates

• universal binding
• that is, all proteins bind to something
• binding can be described with classical (Newtonian) forces rather than quantum ones
  • actually does end up using some quantum description; is some kind of approximation
• if we're just modeling carbohydrates and proteins, the Newtonian frame suffices
• drugs etc., all rely on binding

Why Carbs?

• ubiquitous
• many many cellular functions
• medicinal use -- microbes have carbohydrate shells
• cancer cells: profile of glycoproteins changes
• is a way of targetting cancer cells

Approaches; possible methods

• sequence analysis: motifs
• geometry: known crystals, energy minimalization
• molecular dynamics: using forces

Protein stability

• ΔG
  • thermodynamic
  • kinetic
• therapeutic stability
  • solubility
  • protease resistance (half-life)
• want to retain function for a while without being able to aggregate

Questions

• if it stays too long, we'll trigger an immune response...

Protein changes

• hydrophobic core packing -- tighter → stable
• surface sidechains: acid/base; long/short; movement, rotation
• disulfides -- more → stable?
• loop length -- short → stable
• alpha/beta propensity -- fewer beta → reduced aggregation probability

Computational approach

• exponential (combinatorial really)
• many scenarios -- many inaccurate answers
• few scenarios -- few accurate answers

Detail scaling

• least detailed method on largest dataset
• perform methods and filter
• highest detailed method on smallest dataset

Model (prototype) selection

• ideal model: has carbohydrate binding sites
• hydrophobic core
• variable length loops
• can accommodate disulfides

3RAB revisited

• contains three carbohydrate binding sites
  • multivalence
  • feedback: possible to create a large carbohydrate tether protein as well to bind all three sites
• trefoil is a superfold: many diverse sequences may conform to this fold
• NMR indicates that the protein is stably folded in aqueous solution
  • inferred from multiple peaks in the NMR
  • null hypothesis: unfolded -- would have been inferred from a single peak
• the structure that's been crystallized is co-crystallized with a non-carbohydrate small molecule
  • this small molecule is similar because of it's carbon backbone and hydroxyls.

Initial work

• Structure
• Stability
  • melting point of 94°C

Rational design

• in silico to in vivo
• HIS-tagged proteins
  • purification with nickel-affinity chromotography
• biochemical characterization
• carbohydrate binding specificity and affinity
  • Glycomics gateway -- microarray screening
  • a bunch of physiologically relevant sugars are rapidly screened against your protein
  • the protein must be tagged with something that fluoresces, such as GFP
• protein and protein-ligand structure
  • crystallization

Question about Glycomics Gateway

• do they use a branch and bound method?
• what kinds of sugars has the 3RAB shown binding propensity for so far?