Paper1: Improving protein expression, stability and function with ProteinMPNN

Problem and solution

• problem

Natural proteins are highly optimized for function but are hard to produce at a scale suitable for biotechnical applications because of their poor solubility, thermostability and expression in heterologous systems

• Solution

Set up a strategy to apply ProteinMPNN to natural proteins to improve stability, solubility and function

General workflow

1. Choose design space

   1. Fix AA identitties of residues in the protein that are near the ligand

   2. Fix AA ID of residues that are highly conserved in MSA

   3. Optional backbone redesign

2. Supply the 3D backbone and fixed residues to proteinMPNN that will generate new backbone sequences

3. Predict structure with AlphFold2

   1. filer by pLDDT and C(alpha) RMSD to input structure

pLDDT

predicted Local Distance Difference Test

= estimates its confidence in the predicted local structure of each residue independent of any known experimental structure

How certain am I about the relative postition of residue i connected to j if I assume j is in the right postion

scores range from 0-100 where above 85 is good

RMSD

Average distance between two different conformations of the same or similar biomolecules after they’ve been structurally alligned

Design of Myoglobin variants with increased stability

1. Fix AA IDs of residues near the Heme ligand

2. Backbone redesign (optional)

   1. redesign of poorly ordered/flexible regions to increase stability

   2. RosettaFold impainting= remove flexible region and find ideal backbone to reconnect them

3. apply ProteinMPNN

4. Use AlphaFold single sequence prediction

Results of Myoglobin variants

1. All exhibited higher pLDDT and lower RMSD than the native

   1. >85, <1A vs 50.6, 7.5A

2. Solubility

   1. 13/20 had a higher soluble protein yield

3. Fucntionality

   1. all exhibted similar heme binding spectra

4. Thermostability

   1. All had a higher melting point

   2. all preserve heme binding at higher temperatures

TEV protease

Cleaves a specific recognition sequence

• can be used to remove a tag after purification

Design of TEV protease variants with improved stability and activity

1. Fix AA IDs of residues:

   1. active site only

   2. active site + 30% of the most conserved residues in the TEV family

   3. active site + 50%

   4. active site + 70%

2. Apply ProteinMPNN (gives 4 sets of sequences)

3. Structure prediction with AlphaFold2

4. Experimental analysis

TEV protease variants: results

1. Using a fluorescent gel where the cleaved substrate becomes visible

   1. Active site residues only: no turnover

   2. 50%: had the highest turnover

      1. some had higher activity than the native

      2. Higher k_cat could maybe be attributed to a higher fraction of the protein being in a catalytically competent state

2. The most active ones

   1. had a higher Tm

   2. preserved catalytic activity at higher temperatures

Is the stability of the catalytic conformation involved in the activity enhancement?

Investigate by mutating some residues in the catalytic site and then running 10 µs MD simulations to see its effect on overall protein dynamics.

1. MD showed that over all proteins the loops were more rigid for the redesigned variants

   1. more rigid proteins often bind substrates better because it loses less conformational entropy upon binding

   2. The best performing one showed the most rigid version of a redesigned region

   3. eventhough this region wasn’t directly at the active site, it still affects catalysis

2. The variants showed overall less catalytically competent conformations of the Cis-His diad thcompared to the native

   1. this was the least significant for the best performing variant explaining its higher relative k_cat