Week 5: Protein Design Part II

Week 5: Protein Design Part II

Part A: SOD1 Binder Peptide Design

Background

Superoxide dismutase 1, or SOD1, is an enzyme that helps protect cells from oxidative stress by converting superoxide radicals into hydrogen peroxide and oxygen. SOD1 normally folds into a stable structure and forms a homodimer. It also binds metal cofactors, which are important for its activity.

1. PepMLM Binder Generation

I used the mutant SOD1 sequence as the target and generated short peptide candidates. I also kept the known binder sequence as a reference:

FLYRWLPSRRGG

The known binder has a mix of aromatic residues and positively charged residues. This seemed important because aromatic residues can help make surface contacts, while arginine and lysine can form electrostatic interactions or hydrogen bonds with exposed residues on SOD1.

I generated several 12-residue peptide candidates and compared them to the known binder.

I wanted to avoid choosing a peptide only because it looked hydrophobic and sticky. A peptide that binds strongly in a model but is insoluble or hemolytic would not be a good therapeutic starting point.

2. AlphaFold3 Complex Screening

I screened the peptides by looking at their predicted complexes with A4V SOD1. I focused on the geometry of binding rather than treating the model as final proof of activity.

The main criteria were:

• whether the peptide localized near the N-terminal region
• whether it made a compact surface contact
• whether it avoided unrealistic insertion into the folded protein
• whether the predicted interface looked more specific than diffuse

Structural screening summary

P2 had the strongest-looking structural interaction, but I did not automatically choose it as the final peptide because it also looked more hydrophobic. P4 looked slightly less aggressive but more balanced.

The main trend was that peptides with aromatic residues and positive charge tended to look better. This matched the known binder style.

3. PeptiVerse Property Screening

Next, I compared the peptides using therapeutic-style peptide properties. I looked at predicted binding affinity, solubility, hemolysis risk, charge, and molecular weight.

Property screening results

P2 had the strongest predicted binding, but P4 had the best overall profile. P4 had good binding, better solubility, and lower hemolysis risk. I chose P4 as the best candidate to advance.

Selected peptide

FQYRWLKSGRGS

I chose this peptide because it was not just the strongest binder. It had the best balance between binding and peptide-like properties. For a therapeutic peptide, that balance matters more than maximizing one score.

4. moPPIt Optimization

For the optimization step, I used P4 as the starting peptide. My goal was to improve the peptide slightly while keeping the same overall design logic.

Starting sequence:

FQYRWLKSGRGS

Design goals:

• preserve aromatic residues for binding
• keep moderate positive charge
• improve solubility if possible
• keep hemolysis risk low
• avoid making the sequence too hydrophobic
• keep the length around 12 amino acids

Optimized candidates

Optimized property comparison

The best optimized peptide depends on what we prioritize. If the goal is maximum binding, O2 is attractive. If the goal is peptide safety and solubility, O4 is attractive. I chose O1 because it kept binding close to the original while slightly improving solubility and hemolysis risk.

Final optimized peptide

FQYRWLKSGRGT

This was my final SOD1 binder candidate. It keeps the aromatic/basic pattern that seemed useful for SOD1 binding, while avoiding the more hydrophobic profile of P2 and P3.

Part C: Final Project: L-Protein Mutants

Background

Phage lysis proteins are important because they help release newly produced phage particles from infected bacteria. For MS2, the L protein is involved in lysis of E. coli. Since lysis is the core function, I did not want to mutate the membrane-associated part too aggressively.

The L protein sequence used was:

METRFPQQSQQTPASTNRRRPFKHEDYPCRRQQRSSTLYVLIFLAIFLSKFTNQLLLSLLEAVIRTVTTLQQLLT

The hydrophobic region beginning near YVLIFLAIFL... looks membrane-associated, so I focused my mutations mostly before that region.

Design Strategy

I used these rules for choosing mutations:

1. Avoid the predicted transmembrane region.
2. Prefer mutations in the soluble N-terminal region.
3. Avoid making the protein more hydrophobic.
4. Use mostly conservative substitutions.
5. Add polarity or charge when it might improve solubility.
6. Avoid disrupting residues that may be important for lysis.
7. Do not mutate too many residues at once.

The main idea was to improve folding or stability without destroying the biological function.

Proposed Mutants

Each mutant contains three substitutions, mostly in the soluble region.

Mutant 1: Q8E, T12S, A14S

This mutant adds one acidic residue and two small polar substitutions. Q8E changes glutamine to glutamate, adding negative charge. T12S is conservative because threonine and serine are similar. A14S adds a small polar side chain.

I liked this mutant because it changes the soluble region without touching the membrane-associated region.

Expected benefit:

• improved solubility
• low risk of disrupting membrane function
• moderate change to local charge

Main risk:

• the added charge could affect local interaction behavior

Mutant 2: F5Y, P6A, H23Q

This mutant changes the early N-terminal region more strongly. F5Y is a conservative aromatic substitution, but tyrosine adds a polar hydroxyl group. P6A removes a proline, which could reduce backbone rigidity. H23Q removes a pH-sensitive histidine and replaces it with glutamine.

Expected benefit:

• slightly more polar N-terminus
• less rigid local backbone
• reduced pH sensitivity near position 23

Main risk:

• removing proline could disrupt a local structural feature

Mutant 3: Q9E, S11T, K22R

This mutant is relatively conservative. Q9E adds a negative charge, S11T is a small polar-to-polar change, and K22R preserves positive charge.

K22R is useful because lysine and arginine are both positively charged, but arginine can make stronger hydrogen-bonding or salt-bridge interactions.

Expected benefit:

• preserves basic character
• adds solubility through Q9E
• avoids the membrane region

Main risk:

• charge redistribution could change an interaction site

Mutant 4: P6S, A14T, E24D

This mutant increases polar character while staying fairly close to the original sequence. P6S replaces proline with serine, A14T adds a hydroxyl group, and E24D keeps an acidic residue but shortens the side chain.

Expected benefit:

• improved polar character
• possible improvement in folding flexibility
• keeps acidic character at residue 24

Main risk:

• P6S may make the local region too flexible

Mutant 5: Q10N, T13S, H23N

This is the least aggressive design. Q10N keeps amide chemistry but shortens the side chain. T13S is conservative. H23N removes the histidine imidazole and replaces it with a polar amide.

Expected benefit:

• low disruption risk
• improved polar character
• reduced pH sensitivity

Main risk:

• changes may be too small to produce a meaningful improvement

Mutant Ranking

I ranked the mutants by balancing stability, solubility, and risk to lysis function.

If I had to pick one mutant to test first, I would choose M1.

Selected mutant:

M1: Q8E, T12S, A14S

I chose M1 because it improves polarity in the soluble region without making the protein more hydrophobic or changing the membrane-associated region.

How I Would Test the Mutants

A good L-protein mutant should improve folding or stability without reducing lysis activity. Stability alone is not enough because the biological function has to be preserved.

I would evaluate the mutants using:

• predicted folding confidence
• preservation of the hydrophobic membrane-associated region
• lack of major structural disruption
• solubility of the N-terminal region
• preservation of lysis activity in bacteria

Experimentally, the key test would be whether the mutant still lyses E. coli efficiently. If a mutant folds better but does not lyse cells, it would not be useful.