Week 4 HW: Protein Design Part 1

South American Rattlesnakes (Crotalus durissus terrificus) with Crotamine protein

Part A: Conceptual Questions

1. How many molecules of amino acids do you take with a piece of 500 grams of meat? (on average an amino acid is ~100 Daltons)

• I intake approx. 5 * 10²³ Daltons of amino acids when ingesting 500 grams of meat. This is based off results indicating I ingest approx. 10²¹ Daltons of amino acids when ingesting 1 gram of meat

2. Why do humans eat beef but do not become a cow, eat fish but do not become fish?

• Humans eat beef but don’t become cattle, and eat fish but don’t become fish because genetic information from the lifeform being ingested isn’t transferred wholesale. Much of the genetic material being eaten is broken down during digestion and more importantly a human beings’ cells follow instructions derived from their DNA. Human beings’ cells utilize amino acids from the lifeform being ingestion, but perform this utilization according to specific genetic instructions. The lifeform being ingested and its amino acids are the raw materials the cell uses for various means.

3. Why are there only 20 natural amino acids?

• There are several broad reasons why there are only 20 standard natural amino acids. The first reason is that early on in the history of evolution, this group of amino acids became more or less ’locked in’, meaning that once the basic relationship between three letter codons and these 20 standard natural amino acids became widely distributed across the kingdom of life, it becamde too risky/dangerous from an evolutionary standpoint to alter this core set. Another reason is that the group of 20 gives enough range in structure and chemistry to build a large chunk of what evolution or directed evolution might desire. The other reasons seem to amount to various types of evolutionary trade-offs. Adding more than 20 amino acids to this standard set would add additional, potentially unwanted complexity, while decreasing the number of amino acids in the set might lead to issues with a lack of uniqueness with amino acids side chain sharing, which would in turn limit the functional flexibility of amino acids to do things like fold precisely.

4. Can you make other non-natural amino acids? Design some new amino acids.

• Yes you can. My attempts to design some new amino acids usng SwissSideChain and the Cryo-EM structure of Receptor Tyrosine Kinase ROS1 PDB file in PyMol (open-source) are shown below:

Attempt at creating a non-natural amino acid residue mutation of Tyrosine Kinase ROS1 using cyclohexanecarboxylic acid

Attempt at creating a non-natural amino acid residue mutation of Tyrosine Kinase ROS1 using cyclopropanecarboxylic acid

5. Where did amino acids come from before enzymes that make them, and before life started?

• Amino acids come from metabolic molecules within the cell. These metabolic molecules consist of carbon atom chemical backbone, inorganic nitrogen, and enzyme-facilitated chemical reactions. Before life as we know it started, amino acids originated from abiotic (not from living organisms) chemical reactions on Earth before the emergence of life as we know it. The chemical reactions occurred in the atmosphere, hydrothermal and oceanic vents, and via meteorite and comet (i.e., extraterrestrial) delivery

6. If you make an α-helix using D-amino acids, what handedness (right or left) would you expect?

• If I made an a-helix using D-amino acids, I’d expect left handedness because the majority of alpha helices are built from L-amino acids, which exhibit right handedness.

7. Can you discover additional helices in proteins?

• Yes additional helices can be discovered in proteins. These can include non-alpha helices and even π-helices. These can be discovered via x-ray crystallography, structure databases, or with the aid of AI prediction tools.

8. Why are most molecular helices right-handed?

• Most molecular helices are right-handed because L-amino acides force the helix backbones to conform to right-handedness. Selection pressures also favors this form of handedness in the helix for hydrogen atom bonding and overall functionality

9. Why do β-sheets tend to aggregate?

• β-sheets tend to aggregate because the edges of β-sheets have exposed nitrogen and hydrogen (donor) atomic pairs and carbon-oxygen pairs (acceptor) atoms. What happens is that when sheets get in close proximity these atomic pairs snap together, almost like a zipper.

  • What is the driving force for β-sheet aggregation?
    • The driving force for β-sheet aggregation is the attraction of the donor and acceptor atomic pairs

  Supporting prompts for this section listed below:

Part B: Protein Analysis and Visualization

• I selected the crotamine protein found in the South American rattlesnake, the protein’s abilities to penetrate cells as a toxin allows it to serve as a template for antivenom and potentially targeted therapies.

• The amino acid sequence of the crotamine protein protein is below ¹

  • AAF34911.1 crotamine [Crotalus durissus terrificus] MKILYLLFAFLFLAFLSEPGNAYKQCHKKGGHCFPKEKICLPPSSDFGKMDCRWRWKCCKKGSGK

  • This protein is 65 amino acids long. Its most frequent amino acid in this protein is lysine, which appears 11 times
  • There are 250 protein sequence homologs for the crotamine protein after running it in UniProt BLAST (ID: Q9PWF3)
  • Yes, this protein belongs to the crotamine-myotoxin family.

• Structural answers to this question are listed below:

  • The protein structure was discovered in 2005 ². It appears to be a mixed bag from a structural quality standpoint based on the NMR Structure Validation Report below: Crotamine protein RCSB NMR Structure Validation Report
  • No, there are no other molecules in the solved structure apart from the protein
  • It’s a defensin-like protein of the myotoxin family

• 3D visualization software (NGL Outputs)

  • “Cartoon”, “Ribbon”, and “Ball and Stick” combo
  • Protein coloring by secondary structure. I think it might have more sheets than helices, although I’m not sure
  • Protein by residue type. It seems like it has a good mix of hydrophobic vs hydrophilic residues
  • Protein surface visualization. It doesn’t seem like it has a lot of holes, although I’m not sure

All supporting prompts listed below:

Part C: Using ML-Based Protein Design Tools

C1. Protein Language Modeling

1. Deep Mutational Scans

   • See picture below: Crotamine Protein Deep Mutational Scan

   • There seem to be these really interesting columns that occur as one moves rightward past position 20 in the heatmap. There is a single almost uniform top-down color for these columns, which according to Gemini, indicate a special sensitivity. These are apparently disulfide bridges, which are quite important for holding the protein together (these are referred to as ‘anchors’)

2. Latent Space Analysis

   • See picture below Non-Crotamine tagged point cloud showing embedded proteins

   • Yes they do–there are approximations of similar proteins found throughout the cloud. Crotamine tagged point cloud showing embedded proteins and Crotamine’s location among embeddings

   • There are other proteins at the edge of the cloud that belong to either other viruses, or other organisms that might be found in similar locations/habitats to the South American rattlesnake (mice, chickens, humans), or other types of toxins or species carrying toxins. See picture below: Crotamine tagged point cloud showing embedded proteins and Crotamine’s location relative to its neighborhood

C2. Protein Folding

• Approach 1: It doesn’t look like the predicted coordinates match the original structure in the PDB much at all (see pictures below). Think this might be due to the fact that the original inputted structures in the PDB for this protein were apparently a bit of a mixed bag according to its NMR Structure Validation Report results (see previous section). Crotamine Protein PDB < > ESM Side-by-Side Structural Comparison (1)

  Crotamine Protein PDB < > ESM Side-by-Side Structural Comparison (2)

  Approach 2: Asked Gemini about this discrepancy and was recommended to input the mature sequence containing the last 42 residues. The result still visually seems to have some discrepancies compared with the original PDB visual upon first glance, but far less than the outputs under Approach 1 (see photo below) Crotamine Protein PDB < > ESM Side-by-Side Structural Comparison (3) – Post-Mature Sequence Input

• I tried the following mutations:

  • Cytesine Break: Per Gemini suggestion, replaced all the C residues to A (alanine). This notably decreased the predicted Local Difference Test (pLDDT) score of the outputed mutated sequence and the visual also seemed to indicate less structural integrity to the protein, implying less resilience to this mutation (see photo below) Crotamine Cytesine Break Mutation Results

  • Charge Swap: Per Gemini suggestion, replaced all the K (Lysine) and R (Arginine) residues to D (Aspartic Acid). This decreased the pLDDT score of the outputed mutated sequence a bit, but less than the Cytesine Break, indicating the protein was more comparably resilient to this mutation (see photo below) Crotamine Charge Swap Mutation Results

C3. Protein Generation

• Based on the heatmap below, there seems to be a difference between the predicted sequence probabilities here and the original heatmap generated in the ‘Deep Mutational Scans’ subsection. 48 of the 65 positions were changed and the sequence recovery rate was around 0.26, which didn’t seem all that promising Crotamine ProteinMPNN Results (1)
• Visually, this re-inserted ESMFold output also looks structurally different than the original protein structure (see photo below) Crotamine ProteinMPNN Results (2)

All supporting promopts for this section listed below:

Part D: Group Brainstorm on Bacteriophage Engineering

• My William & Mary Node Bacteriophage Engineering Group Brainstorm Google Doc. can be found here ³.