🧬 Directed Evolution & Protein Engineering — Full Theoretical Summary

This lecture is about how we generate diversity in proteins (mutations) and then select the best variants to improve properties like activity, stability, or specificity.

🔁 1. Why Do We Need Many Mutations?

A single mutation → only gives:

• Either original amino acid
• Or one changed amino acid

➡️ That’s very limited diversity

👉 For directed evolution, we need:

• Huge libraries (millions of variants)
• Many mutations at multiple positions

🧩 2. Cassette Mutagenesis (KEY CONCEPT)

📌 Definition: Replace a small region of a gene with a synthetic DNA fragment (cassette) containing randomized codons.

🧠 How it works:

• You choose a region (e.g., active site)
• Replace it with a sequence like:

  NNK NNK NNK ...
  

Each codon is randomized.

🔤 What do N and K mean?

You were close, but let’s fix it precisely:

👉 So:

• Yes: N = any nucleotide ✅
• Correction: K = G or T (50/50), not “50% G only” ❌

🧬 Why NNK?

Genetic code facts:

• Total codons = 64
• Encode:
  • 61 amino acids
  • 3 stop codons

❓ Why not NNN?

NNN → includes ALL codons:

• Includes 3 stop codons ❌ (bad for proteins)

✅ NNK advantages:

• Encodes all 20 amino acids
• Reduces stop codons from 3 → 1 only

✔ Your statement:

“NNK: 20 aa and reduce stop codon?”

👉 Correct ✅

🔍 Third position (K)

You asked:

“K → G and T only? correct?”

👉 Yes, exactly correct ✅

That means:

• Codons ending in A or C are excluded
• Reduces unwanted codons

🔢 Library size

If you randomize:

• 6 positions → 20⁶ ≈ 64 million variants

✔ This matches the lecture

💧 3. Hydrophobic Environment in Active Sites

You asked:

“Is it to remove water because water is active?”

👉 Yes — but let’s clarify deeper:

🧠 Why exclude water?

Water:

• Is chemically reactive
• Can:
  • Hydrolyze substrates
  • Compete with enzyme reactions
  • Cause unwanted side reactions

➡️ In active sites:

• Enzymes often create a hydrophobic pocket
• This:
  • Stabilizes substrate binding
  • Prevents side reactions

⚠️ Mutation implication

If mutations introduce:

• Polar residues → water enters
• Active site becomes less controlled

👉 Result:

• Reduced specificity
• Side reactions
• Lower efficiency

🎯 4. NNK vs NTN (Smart Library Design)

NNK → Full diversity

• All 20 amino acids
• Large library

NTN → Biased selection

👉 You asked:

“What is NTN?”

NTN:

• Middle base = T only
• Selects specific amino acids

From genetic code: ➡️ Gives mainly:

• Hydrophobic amino acids
  • Phe, Leu, Ile, Met, Val

🧠 Purpose:

Instead of randomizing everything: 👉 You bias toward useful chemistry

✔ Example:

• Active site → often hydrophobic
• So you restrict library to hydrophobic residues

⚠️ Limitation of NNK / Error-prone PCR

You asked:

“2 mutations in same codon?”

👉 Correct issue:

• Error-prone PCR:
  • Mostly causes single nucleotide changes
  • Hard to jump between very different amino acids

Example:

• Phe → Gly requires multiple base changes ❌ Hard with random mutation

🧬 5. Trinucleotide Synthesis (VERY IMPORTANT)

🧠 What is it?

Instead of adding:

• 1 nucleotide at a time

👉 Add:

• 3 nucleotides at once (one codon)

🎯 Purpose:

Full control over:

• Which amino acids appear
• Their ratios

Example from lecture:

• 67% Phenylalanine
• 3% Glycine
• Rest cysteine

👉 You design the library composition exactly

✅ Advantages:

• No stop codons
• No unwanted amino acids
• Precise control

❌ Disadvantage:

• Very expensive (~100× cost)

🧬 6. DNA Shuffling (Recombination Evolution)

🧠 Concept:

Combine existing genes to create new ones.

🔁 Workflow:

1. Take similar genes (e.g., lipases from different organisms)
2. Cut into fragments (DNase)
3. Mix fragments
4. Reassemble via PCR (no primers)
5. Amplify full-length genes

💡 Result:

👉 Chimeric genes

• Hybrid of multiple parents

🧠 Why powerful?

Unlike random mutation:

• Uses mutations already tested by evolution

✔ More likely:

• Proper folding
• Stability
• Function

🧪 7. Error-Prone PCR

🧠 What it does:

Introduce random mutations during PCR

How?

Modify conditions:

• Low fidelity polymerase
• High Mg²⁺
• Imbalanced nucleotides

Result:

• Many mutated copies
• Random mutations across gene

🧬 8. Chimeric Genes

👉 Created by DNA shuffling

• Contain parts from different parent genes
• Combine beneficial mutations

🧪 9. Library → Screening vs Selection

🧠 Problem:

Huge libraries:

• Millions of variants

👉 You must find the one good mutant

🔍 Screening:

• Test each variant individually
• Very slow

Example:

• Frances Arnold screened 20,000 clones

⚡ Selection:

• Only functional variants survive
• Much faster

🧫 Array approach:

You asked:

“Split library into arrays?”

👉 Yes:

• Each variant → separate well/spot
• Express protein
• Test function

🔥 10. Directed Evolution Example (Subtilisin E)

🧬 Protein:

• Subtilisin E = protease

🎯 Goal:

Increase thermal stability (half-life at 65°C)

🔁 Process:

1. Error-prone PCR → mutants
2. Screen → pick best
3. DNA shuffling → combine mutations
4. Repeat cycles

🧠 Result:

• 6 rounds
• 200× increase in half-life

⏱️ Half-life meaning:

👉 Time for:

• 50% of enzyme activity to be lost

✔ Higher half-life = more stable enzyme

⚙️ 11. Thermostability & Mutagenesis

🧠 Concept:

Mutations can:

• Increase rigidity
• Improve folding
• Reduce unfolding

Strategy:

• Mutate key residues
• Test stability
• Combine beneficial mutations

🧠 12. Protein Engineering Strategy (Big Picture)

Pipeline:

1. Generate diversity:
   • Error-prone PCR
   • Cassette mutagenesis
   • DNA shuffling
2. Create library
3. Screen/select
4. Repeat cycles

🧬 13. DNA-Encoded Libraries & Selection

Concept:

• Each protein linked to DNA tag
• Enables tracking

Selection cycle:

1. Bind to target
2. Wash away weak binders
3. Amplify DNA of strong binders

⚠️ 14. Major Bottleneck: Screening

You asked:

“Problem of screening?”

👉 Biggest issue:

• Libraries = millions
• Screening = slow, expensive

Solutions:

• Automation (robots)
• Selection systems (phage display, etc.)

🧠 Final Key Takeaways

• NNK = efficient randomization with minimal stop codons
• NTN = bias toward specific chemical properties
• Trinucleotide synthesis = precise control
• DNA shuffling = recombine natural mutations
• Directed evolution = iterative improvement
• Main challenge = screening large libraries

🧩 1. Cassette Mutagenesis (deeper clarification)

You already got the core idea, but one subtle point:

👉 You can introduce:

• 1 mutation
• 2, 3, or more mutations in the SAME oligo

This is important because:

• Unlike error-prone PCR (random),
• Cassette mutagenesis is designed mutation

✔ Meaning: You can precisely control which positions are mutated simultaneously

🧬 2. “64 different codons?” (clarification)

👉 Yes, exactly:

• 4 nucleotides (A, T, G, C)
• 3 positions per codon

→ 4³ = 64 codons

Split into:

• 61 → amino acids
• 3 → stop codons

✔ Your understanding was correct

🔬 3. “Chemical properties” (what does it mean here?)

This refers to grouping amino acids by behavior, not identity.

Main categories:

🧠 Why important in mutagenesis?

Instead of randomizing blindly: 👉 You choose mutations that preserve function

Example:

• Active site → hydrophobic → use NTN instead of NNK

🧪 4. “Protease with saquinavir bound”

This is a structure-based design example

👉 Saquinavir = inhibitor of HIV protease

🧠 What it illustrates:

• You look at:
  • Enzyme structure
  • Bound ligand (drug)

👉 Then identify:

• Which residues interact with substrate/inhibitor

🎯 Purpose in lecture:

• Shows how to target mutagenesis to functional regions
• Instead of mutating whole protein

🔁 5. DNA Shuffling (clarifying your confusion)

You asked:

“4 genes? lipase? cleave DNA?”

👉 Correct idea, but here is the clean workflow:

🧬 Step-by-step:

1. Take similar genes (e.g., lipases from different organisms)
2. Cut DNA randomly (DNase)
3. Mix fragments
4. Let them reassemble (PCR without primers)
5. Amplify full genes

🧠 Key idea:

• You are NOT just mutating
• You are recombining existing mutations

❗ Important correction:

👉 It is NOT:

• “one gene → cleave → random mutation”

👉 It IS:

• multiple related genes → mix → recombine

🧬 6. “Introduce mutation with error-prone PCR”

Already covered partly, but missing mechanism detail:

🧠 What actually changes:

DNA polymerase:

• Normally → high fidelity
• Here → forced to make mistakes

How?

• Increase Mg²⁺
• Add Mn²⁺
• Unbalanced dNTPs
• Use low-fidelity polymerase

Result:

• Random substitutions
• Occasionally insertions/deletions

🧬 7. Protein Engineering Strategy (deeper clarification)

You asked:

“strategy of the library?”

🧠 Two main strategies:

1. Random (exploration)

• Error-prone PCR
• NNK libraries

✔ Pros:

• Large diversity

❌ Cons:

• Huge screening effort

2. Rational / semi-rational (guided)

• NTN
• Trinucleotide synthesis
• Target active site

✔ Pros:

• Smaller, smarter libraries

🧪 8. DNA-Encoded Library & Selection Cycle

You asked:

“split into arrays? what does it mean?”

🧠 Two approaches:

A. Arrays (screening)

• Each mutant → separate well
• Measure activity individually

✔ Very controlled ❌ Slow

B. Selection (more important)

• Link protein to DNA tag
• Apply selection pressure

Example:

• Binding to target
• Survival in condition

👉 Only best variants survive

🔁 Selection cycle:

1. Generate library
2. Apply selection
3. Amplify survivors
4. Repeat

🧬 9. Protein Design (concept)

Different from evolution:

🧠 Two approaches:

Directed evolution

• Start with natural protein
• Improve it

Protein design

• Build protein from scratch or redesign

Uses:

• Structure knowledge
• Computational tools

🔥 10. Subtilisin E (deeper meaning)

You asked:

“increase half-life?”

🧠 Interpretation:

• Enzyme loses activity over time (especially at high temp)

👉 Half-life = time until:

• 50% activity remains

🎯 Increasing half-life means:

• More stable protein
• More resistant to heat denaturation

🔁 11. “Get many mutants → screen them?”

Yes, but important nuance:

🧠 Reality:

• You generate thousands to millions
• But only test a fraction

Workflow:

1. Generate mutants
2. Express in bacteria
3. Measure function
4. Select best

⚠️ 12. Problem of Screening (expanded)

You asked for deeper explanation:

🧠 Core problem:

• Library size grows exponentially

Example:

• 6 positions → 64 million variants

❌ Issues:

• Time-consuming
• Expensive
• Labor-intensive

🔧 Solutions:

• Automation (robots)
• Selection methods (phage display, etc.)
• Smarter libraries (NTN, trinucleotide)

🧬 13. “Two mutations in same codon?” (clarification)

👉 Important limitation:

• Error-prone PCR:
  • Mostly single nucleotide changes

Problem:

Some amino acid changes require:

• 2 or 3 nucleotide changes

👉 Hard to achieve randomly

Solution:

• Cassette mutagenesis
• Trinucleotide synthesis

🧠 Final Missing Insights

✔ Directed evolution is iterative

• Mutate → select → repeat

✔ Combining methods is common

Example:

• Error-prone PCR → find good mutations
• DNA shuffling → combine them

✔ Most mutations are useless

• Only very few are beneficial

✅ Summary of what was clarified

You now have:

• Exact meaning of NNK, NTN, K
• Why hydrophobic environments matter
• Full DNA shuffling workflow
• Difference between:
  • Random vs guided libraries
• Meaning of:
  • Half-life
  • Screening bottleneck
• Role of:
  • trinucleotide synthesis
  • chemical properties
  • structure-based design (saquinavir example)