Week 1 HW: Principles and Practices

Researchers — anyone working on developing the crops. Farmers — anyone working on growing the crops. Government — anyone working on lawmaking associated with the crops. Business — any business-side component of funding or selling the crops. Elders — indigenous or other wisdom teachers.

My focuses to best mitigate risk on this project would be threefold:

1. Establish a great brand that creates great products. Nothing builds institutional trust like authentic, direct communication and consistent top-tier execution. The success next generation of transformed crops will depend on the end-consumer brands’ ability to communicate to the audience better than that haters will. The only true solution to this problem is direct communication between scientists and consumers via a company that works across the range from research to end-product deployment.
2. Get involved in lobbying from day one with a focus on the comparatively under-invested arena of executive branch lobbying. If we don’t want to get regulated out of existence, scientists must be in conversation with lawmakers as early and as often as possible to develop smart regulations and proper incentives for the synthetic biology economy.
3. Make deep relationships with the farmers growing our plants. No one knows the land or the plants like the ones tending it. If we want to do this right we must befriend our growers and learn from their daily experience brining our seeds to fruition.

Week 1 HW: Principles and Practices

Section A

1. How many amino acid molecules are in 500g of meat?

500 g meat at ~20% protein = 100 g protein

Avg amino acid ≈ 100 g/mol

100 g ÷ 100 g/mol = 1 mole of amino acids

1 mole = 6.022 × 10²³ molecules

2. Why do we eat cow but not become cow?

Cow give us protein. Protein broken down into building blocks. Human make building blocks into human protein. Identity does not transfer, just blocks.

3) Why are there only 20 natural amino acids?

There are actually more than 20, but 20 is the standard. This was frozen early on in natural development because it is chemically sufficient, and added translations opens up higher possibilities of catastrophic error. Enough is enough.

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

Yes. Thousands exist.

β-sheet breaker polar amino acid: A proline-like backbone constraint plus a polar, H-bonding side chain that disrupts inter-strand packing.

Tunable active-site histidine analog: An imidazole variant with a shifted pKa/metal affinity to dial down catalytic rates in GAME-pathway enzymes.

5) Where did amino acids come from before life?

The most plausible routes are Miller-Urey type chemistry occurring in natural boiling bots, meteorites transferring them to Earth (maybe they grew on the meteorites from cosmic chemistry), or possible synthesis along hydrothermal vents. Essentially, the conditions to create them exist on Earth in suboptimal, but sufficient ways to create amino acids before life perfected the process.

6) If you build an α-helix from D-amino acids, what handedness?

Left-handed.

7) Why are most molecular helices right-handed?

Because biological polymers are made of L-amino acids making right-handed helices the lowest-energy solution given L-chirality.

8) Why do β-sheets tend to aggregate?

Because their geometry allows for easy and infinitely extensible exposed bonding. Elongation is energetically downhill once nucleated.

9) Why do many amyloid diseases form β-sheets?

Because they are the thermodynamic ground state ****for many unfolded peptides and their extended chains can hydrogen bond indefinitely.

Section B

1. Briefly describe the protein you selected and why you selected it.

   UGT74F2 is a UDP-dependent glycosyltransferase: it transfers glucose from UDP-glucose–type donors onto small-molecule acceptors (here, salicylic acid), changing solubility, localization, and bioactivity. I picked a plant UGT because I am interested in engineering edibility in tomato leaves and Solanaceae toxicity pathways repeatedly use UGT-like chemistry to decorate small molecules.

2. Identify the amino acid sequence of your protein.

   • How long is it? What is the most frequent amino acid? 458 amino acids long. Leucine (L) followed closely by Glycine (G) and Serine (S).
   • How many protein sequence homologs are there for your protein? If restricted to Arabidopsis only there are ~100+ UGT paralogs, Solanaceae (tomato/potato) there are dozens of clear homologs, in all plants there are thousands.
   • Does your protein belong to any protein family? Yes, Family 1 UDP-glycosyltransferases (UGTs).

3.  Identify the structure page of your protein in RCSB

   • When was the structure solved? Is it a good quality structure? Good quality structure is the one with good resolution. Smaller the better (Resolution: 2.70 Å) Yes it was solved in 2017 at very good quality: ~2.0 A. **
   • Are there any other molecules in the solved structure apart from protein? Yes, the 5U6N structure was solved with ligands bound.
   • Does your protein belong to any structure classification family? Yes, the Glycosyltransferase GT-B fold family.

4. Open the structure of your protein in any 3D molecule visualization software:

   In PyMOL, I loaded 5U6N and visualized UGT74F2 as cartoon/ribbon to inspect fold architecture, then displayed ligands and the surrounding residues in ball-and-stick. Coloring by secondary structure shows a GT-B glycosyltransferase fold with substantial β-sheet cores and many α-helices wrapping them (overall, more helices than sheets). Coloring by residue type shows hydrophobic residues concentrated in the domain cores and polar/charged residues enriched on the surface and lining the ligand-binding cleft. A surface representation reveals a prominent inter-domain groove (“hole”) that forms the binding pocket; residues within ~4 Å of the ligands define the catalytic pocket around the bound UDP and salicylic acid.

Section C

Section D

I did not have other students to work on this with so I built some agents to work on it with. Here is our solution:

I propose to engineer the MS2 lysis protein L to increase its stability and reduce its dependence on the host chaperone DnaJ. L is a 75-amino-acid membrane-associated lysis protein whose activity depends on proper insertion, oligomerization, and interaction with DnaJ in E. coli. Because it is small and highly sensitive to sequence changes, it is well suited for computational mutagenesis.

To improve stability, I will use protein language models to generate and score single and combinatorial mutations. Variants will be filtered using physicochemical metrics such as net charge, isoelectric point, hydrophobic moment, predicted transmembrane helix propensity, and aggregation risk. AlphaFold will then be used to evaluate predicted structural confidence and overall fold integrity. Variants that maintain structure while improving membrane insertion or helical stability will be prioritized.

To reduce DnaJ dependence and potentially increase effective toxicity, I will focus on modifying the basic N-terminal region implicated in chaperone interaction. AlphaFold-Multimer will be used heuristically to compare predicted L–DnaJ interactions between wild type and mutants. Relative changes in interface contacts and oligomerization tendencies will guide ranking.

The rationale is that protein language models efficiently explore sequence space while preserving protein-like constraints, and structure prediction tools enable rapid filtering for destabilizing or disruptive mutations. However, predictions for small membrane peptides may be unreliable, and chaperone interactions are highly context dependent. All computational outputs will therefore be treated as ranking heuristics prior to experimental validation.

The pipeline is: generate mutations with a protein language model, filter by physicochemical properties, evaluate structure and interfaces with AlphaFold, rank candidates for stability or reduced DnaJ dependence, and select a small panel for experimental testing.

Week 1 Lab: Pipetting

Individual Final Project

Group Final Project