Lecture 5 — From Resistance to Next-Generation Design
1.27 million
deaths directly caused by antimicrobial resistance in 2019 — more than HIV/AIDS or malaria that year.
Projected: 10 million per year by 2050 if left unchecked.
You now have the tools to understand exactly how this happens at the atomic level — and how computational drug design is part of the response.
Resistance is evolution in real time
Step 1 — Random mutation
In millions of bacteria, random mutations occur constantly during DNA replication. Occasionally, one mutation changes an amino acid in a way that reduces drug binding.
Step 2 — Selection
When antibiotics are present, bacteria without the resistance mutation die. The one resistant bacterium survives, replicates, and within hours, millions of resistant bacteria exist.
Step 3 — Spread
Resistant bacteria spread between patients, hospitals, and countries. Some bacteria can pass resistance genes directly to others via horizontal gene transfer — without needing to reproduce.
Analogy: A stadium of bacteria. Spray an antibiotic. 999,999 die. One, by pure chance, had a protective mutation. That one survivor is now the ancestor of the entire population.
One amino acid change is enough
A drug works by binding tightly to a protein. Resistance often means the protein has changed — just enough to weaken drug binding, while still being able to do its original job.
The key tension in drug design
You want your drug to bind tightly to specific residues (high binding energy). But the more your drug depends on a small number of residues, the more vulnerable it is to a single mutation at one of those residues. This is the fundamental trade-off of targeted drug design.
Why this is hard to prevent
A drug target protein has hundreds of residues. Only a handful are hot spots for drug binding. But those same hot spots are often the positions where a single mutation can confer resistance — the drug relies on them, and so does the bacterium's escape strategy.
Connect to Project 6: You will take five clinically documented resistance mutations in TEM-1 β-Lactamase and calculate exactly how much each one weakens inhibitor binding.
β-Lactamase — antibiotic destroyer
β-Lactam antibiotics
The most widely prescribed class in the world — penicillins, cephalosporins, carbapenems. They block a bacterial enzyme needed to build the cell wall. Without a cell wall, bacteria cannot survive.
β-Lactamase (the resistance enzyme)
Bacteria produce β-Lactamase, which destroys β-Lactam antibiotics by breaking the β-Lactam ring. Once broken, the antibiotic is inactive.
The clinical response — inhibitor combinations:
Augmentin = Amoxicillin (antibiotic) + Clavulanic acid (β-Lactamase inhibitor). Bacteria have now evolved mutations in β-Lactamase that also weaken the binding of clavulanic acid.
Project 6 protein: TEM-1 β-Lactamase (PDB: 1ZG4) — the most clinically prevalent β-Lactamase.
Five mutations found in real patients
Mutation
Change
Mechanism of resistance
M69L
Met → Leu at 69
Alters local pocket geometry near the binding site
K73R
Lys → Arg at 73
Disrupts the active site catalytic mechanism
S130G
Ser → Gly at 130
Destroys a key H-bond between Ser -OH and inhibitor carbonyl
R244S
Arg → Ser at 244
Removes electrostatic interaction with inhibitor carboxylate
N276D
Asn → Asp at 276
Introduces a negative charge near the binding site
Apply Lecture 4 knowledge to S130G: Serine has an -OH group that forms hydrogen bonds. Glycine has no side chain at all. That hydrogen bond to the inhibitor is gone. This is exactly the kind of change you would predict to weaken binding in an alanine scan.
A good resistance mutation must
1. It must weaken drug binding
The mutation must reduce inhibitor binding energy enough that the drug can no longer block the enzyme at clinically achievable concentrations. A mutation that only weakens binding by 1 REU may not be enough to cause clinical resistance.
2. It must preserve enzyme function
The enzyme must still be able to destroy β-Lactam antibiotics. A mutation that eliminates drug binding but also breaks the enzyme is worthless to the bacterium. The most successful resistance mutations selectively disrupt drug binding while leaving enzyme activity intact.
Why resistance is constrained: The binding site for the inhibitor overlaps with the active site. Mutations that affect one tend to affect the other. But evolution is patient — over millions of generations, bacteria find the rare mutations that thread this needle.
Strategies to overcome resistance
Strategy A — Recover the lost binding energy
Design a new drug that compensates for the lost interaction by forming stronger interactions elsewhere — using hot spots NOT affected by the resistance mutation.
Strategy B — Design around the mutation
Identify which hot spots are mutation-prone and which are stable. Design a new drug that does not rely on vulnerable positions. Use alanine scan data (Project 5) to find alternative hot spots to exploit.
Strategy C — Combination therapy
Use two drugs targeting different binding sites simultaneously. Resistance to both requires two independent mutations at once — probability ≈ (10⁻⁷)² = 10⁻¹⁴. In 10¹⁰ bacteria, one mutation is inevitable; two simultaneous mutations are essentially impossible.
Six projects tell one connected story
Project
What you did
Scientific question
1
Scored and mutated Lysozyme
Is this protein stable? Which residues matter?
2
Ranked five crystal structures
Which structure is high enough quality to use?
3
Docked three drugs into COX-2
Which drug candidate binds most tightly?
4
Mapped the COX-2 binding site
Which residues line the pocket and what chemistry?
5
Alanine-scanned the binding site
Which residues are hot spots for drug binding?
6
Modelled resistance mutations
How do clinical mutations weaken inhibitor binding?
You started by learning to measure protein stability. You applied that to drug binding. You mapped where the drug sits. You found which residues it depends on. Now you will see what happens when bacteria evolve to disrupt exactly those residues.
Computation vs experiment
Question
What computation tells you
What it cannot tell you
Does the drug bind?
Predicted binding strength in a static structure
Actual binding in a living cell
Which residues matter?
Predicted energetic contribution (ΔΔG)
Whether the mutant folds correctly in vivo
Will resistance emerge?
Predicted weakening of binding
Which mutations bacteria will actually evolve
Is the drug safe?
Nothing
Side effects, toxicity, metabolism
The key message: Computation is fast, cheap, and screens thousands of possibilities. Experiment is slow, expensive, and measures reality. The power comes from combining them — using computation to decide which experiments are worth running.
The drug design cycle never ends
Identify target protein
↓
Choose highest-quality structure (Project 2)
↓
Dock candidates → rank by binding energy (Project 3)
↓
Map binding site chemistry (Project 4)
↓
Identify hot spots via alanine scanning (Project 5)
↓
Model resistance mutations (Project 6)
↓
Redesign drug → loop back to docking ↑
Summary — five lectures in five sentences
Lecture 1: A drug works by binding to a specific protein — choosing the right target and the right crystal structure is the foundation of everything.
Lecture 2: Rosetta measures protein stability as a sum of non-covalent forces — lower REU always means more stable.
Lecture 3: Drug binding depends on shape and chemical complementarity — binding energy is the complex score minus the sum of its parts.
Lecture 4: Alanine scanning identifies hot spot residues by measuring ΔΔG — the energetic cost of erasing each side chain's contribution.
Lecture 5: Resistance arises when bacteria evolve mutations at hot spots — overcoming it requires designing drugs that engage stable, mutation-resistant positions.
Closing thought:
The proteins, mutations, and drugs in this module are all real. The structures come from real crystallography. The resistance mutations come from real patients. The computational methods are used in real pharmaceutical research. You have not been doing exercises — you have been doing science.