Chapter 35 — Organic Chemistry of Drug Design

"Drugs work because of organic chemistry. Every property a drug has — potency, selectivity, half-life, oral absorption, side effects — traces back to its molecular structure. Master the chemistry, and you can rationally design medicines." — paraphrase from a medicinal chemistry text

"There has never been a better time to be a medicinal chemist. Tools we now take for granted — AlphaFold for target structure, AI for retrosynthesis, automation for synthesis — would have been science fiction 20 years ago. Yet the underlying chemistry is still Chapter 24 through 34 of this textbook."

This chapter closes Part VII. It is the culmination of all the anchor-example threads of this textbook: aspirin, ibuprofen, acetaminophen, thalidomide, and the progressive synthesis project. By now you have the chemistry to understand why drugs work at the molecular level, and to begin designing your own.

The chapter: 1. Reviews the molecular mechanisms of three landmark drugs: aspirin, ibuprofen, acetaminophen. 2. Introduces drug-design principles: pharmacophore, SAR, bioisosteres, prodrugs. 3. Discusses Lipinski's rule of 5 and ADME considerations. 4. Closes the thalidomide arc with PROTACs. 5. Looks forward to AI-driven drug discovery and the future of medicinal chemistry.

By the end of this chapter you should be able to: - Explain how aspirin, ibuprofen, and acetaminophen work mechanistically. - Apply Lipinski's rule of 5 to evaluate drug candidates. - Understand SAR (structure-activity relationships) as an optimization process. - Recognize bioisosteres, prodrugs, and covalent inhibitors. - Appreciate how the same molecule (thalidomide) can be both a tragic teratogen and the foundation of a new drug class (PROTACs). - See drug discovery as the integration of chemistry, biology, and (increasingly) AI.

35.1 Drug discovery overview

Modern drug discovery typically follows this pipeline:

1. Target identification: choose a biological target (a protein, often) whose modulation would treat a disease.
2. Hit discovery: find chemical compounds (typically from libraries of millions) that bind the target. Methods: high-throughput screening, fragment-based discovery, virtual screening, etc.
3. Lead optimization: take the best hits and improve their properties — potency, selectivity, ADME (absorption, distribution, metabolism, excretion), toxicity. Iteratively make analogs and test them. Typically takes 1-3 years and $100K-$10M.
4. Pre-clinical development: extensive testing in cell culture and animals; safety studies.
5. Clinical trials: - Phase I (~50 healthy volunteers; safety + dosing). - Phase II (~100-300 patients; efficacy in target disease). - Phase III (~1000-10000 patients; large-scale efficacy + safety).
6. Regulatory approval (FDA, EMA): if successful, the drug is approved.
7. Post-market surveillance: monitor for rare side effects.

Total time from target identification to approval: typically 10-15 years and ~$1-3 billion.

Most drug candidates fail. Of every 10,000 compounds screened, ~1 reaches market. The reasons for failure: lack of efficacy, toxicity, side effects, poor ADME.

This chapter focuses on the chemistry that determines whether a drug candidate succeeds.

35.2 Aspirin's mechanism (revisited)

Aspirin (acetylsalicylic acid) is the archetypal small-molecule drug. We've encountered its synthesis (Ch 26 case study 1) and now deepen the mechanism.

Target: cyclooxygenase

Cyclooxygenase (COX) is an enzyme that converts arachidonic acid (Ch 34, an ω-6 fatty acid) into prostaglandin G2 (PGG2), the precursor of all prostaglandins. The conversion involves: - Cyclization of arachidonic acid's chain. - Insertion of two oxygens.

Prostaglandins are local signaling molecules that mediate: - Pain (inflammatory pain, esp. PGE2). - Fever (PGE2 in the hypothalamus). - Inflammation (PGE2, PGD2, PGI2). - Platelet activation (TXA2, thromboxane). - Gastric protection (PGE2 in stomach).

There are two main COX isoforms: - COX-1: constitutively expressed; in stomach (protective), platelets (TXA2 for clotting), kidneys (homeostatic). - COX-2: induced by inflammation; in inflamed tissues; produces PGE2 for pain/inflammation.

Aspirin's covalent mechanism

Aspirin irreversibly acetylates COX's Ser530 (or Ser529 in some numbering) — the active site serine. The chemistry: nucleophilic acyl substitution (Ch 26).

1. Aspirin enters COX's substrate channel.
2. Ser530-OH (positioned near the substrate binding site) attacks aspirin's ester C=O.
3. Tetrahedral intermediate forms.
4. Salicylate leaves; acetyl-Ser is left behind.
5. The acetyl group on Ser530 blocks arachidonic acid from binding.
6. COX is permanently inactivated.

The COX cannot recover until new protein is synthesized. For platelet COX-1, this takes 7-10 days (the platelet's lifetime). For COX-2 in inflamed tissues, ~6-12 hours.

Clinical consequences

• Anti-inflammatory effect: from COX-2 inhibition.
• Pain relief: from COX inhibition (in inflamed tissue and CNS).
• Fever reduction: from CNS COX inhibition.
• Anti-platelet (anticlotting) effect: from irreversible COX-1 inhibition in platelets. Low-dose aspirin (81 mg/day) is used long-term for cardiovascular disease prevention.
• Side effects: from COX-1 inhibition in stomach (irritation, bleeding) and kidneys.

This is why aspirin is uniquely useful for cardiovascular prevention — its irreversible mechanism gives long-lasting anti-platelet effects with low daily doses. Other NSAIDs (ibuprofen, naproxen) are reversible and so don't give the same effect.

35.3 Ibuprofen's mechanism

Ibuprofen is a reversible competitive inhibitor of COX. It binds to the active site (mimicking arachidonic acid's binding pose), blocking the substrate.

Structure

Ibuprofen is 2-(4-isobutylphenyl)propanoic acid. Key features: - Carboxylic acid: mimics arachidonic acid's carboxylate. - Aryl group: occupies the substrate channel. - Isobutyl group: extends into the hydrophobic pocket. - α-methyl: gives one chiral center (the racemic mix is what's sold; only the (S)-enantiomer is active in COX inhibition).

Mechanism of inhibition

1. Ibuprofen enters COX's substrate channel.
2. The COOH binds to the same site that arachidonic acid's COOH would bind (an arginine residue).
3. The aryl + isobutyl groups occupy the substrate channel.
4. While ibuprofen is bound, arachidonic acid cannot bind.
5. When ibuprofen is metabolized and cleared, COX activity returns.

This is reversible competitive inhibition: the drug binds non-covalently and dissociates over time. No covalent modification.

Clinical consequences

• Anti-inflammatory: from COX-2 inhibition.
• Pain relief: comparable to aspirin.
• No cardiovascular benefit: because reversible inhibition doesn't permanently inactivate platelet COX.
• Side effects: gastric (COX-1 in stomach), kidney (COX-1 in kidneys).

Ibuprofen is the prototypical NSAID. Naproxen, ketoprofen, and others have similar mechanisms.

COX-2-selective inhibitors (rofecoxib, celecoxib)

The 1990s/2000s saw development of COX-2-selective inhibitors (rofecoxib/Vioxx, celecoxib/Celebrex, valdecoxib). Goal: block inflammation (COX-2) without blocking gastric protection (COX-1).

Initial success: less GI bleeding. But Vioxx was withdrawn in 2004 due to cardiovascular side effects (increased heart attack risk). The reason: COX-2 also produces prostacyclin (an anti-thrombotic prostaglandin), so blocking COX-2 alone disrupts the COX-1/COX-2 balance.

Celecoxib remains in use; it has fewer of these issues but the cardiovascular caution is real.

35.4 Acetaminophen's (more mysterious) mechanism

Acetaminophen (paracetamol; in the US: Tylenol; elsewhere: Panadol) is one of the most-used analgesics. Its mechanism is more debated than aspirin's or ibuprofen's.

Structure

Acetaminophen is N-(4-hydroxyphenyl)acetamide. It has: - A phenol (-OH). - An amide (-NHCOCH₃).

Putative mechanisms

1. CNS COX inhibition: acetaminophen weakly inhibits COX in the central nervous system, particularly under reduced peroxide tone (which is present in the CNS). It has minimal effect on peripheral COX-1 (so no anti-inflammatory effect; no anti-platelet effect).
2. TRPV1 receptor: some evidence that acetaminophen modulates TRPV1 (a pain-sensing receptor) via its metabolite AM404.
3. Endocannabinoid system: AM404 also affects cannabinoid signaling.

The mechanism is "mostly central CNS COX inhibition + secondary effects."

Acetaminophen toxicity (NAPQI)

At overdose (typically >4g/day for adults), acetaminophen forms a toxic metabolite via cytochrome P450 enzymes (CYP2E1):

$$\text{acetaminophen} \xrightarrow{\text{CYP2E1, oxidation}} \text{NAPQI (N-acetyl-p-benzoquinoneimine)}$$

NAPQI is an α,β-unsaturated quinone imine — a powerful Michael acceptor (Ch 29). It rapidly reacts with glutathione (a tripeptide thiol present in liver cells) via thia-Michael addition, depleting glutathione stores.

Once glutathione is depleted, NAPQI then reacts with cellular proteins (cysteines, lysines) → covalent modification → liver cell death.

Acetaminophen overdose is the leading cause of acute liver failure in the U.S.

Treatment: N-acetylcysteine (NAC), which replenishes glutathione. NAC must be given within ~8-10 hours of overdose to be effective. Late presentation requires liver transplant.

Biological Connection 35.1: Acetaminophen + alcohol = liver damage.

Chronic alcohol use induces CYP2E1, the enzyme that produces NAPQI. So alcoholics on therapeutic doses of acetaminophen can experience NAPQI-induced liver damage.

Combined: chronic alcohol + acetaminophen (even at recommended doses) = potential liver damage.

The chemistry: alcohol-induced CYP2E1 generates more NAPQI; depleted glutathione (often present in alcoholics) cannot detoxify it; NAPQI Michael-attacks liver proteins; cells die.

35.5 Lipinski's rule of 5

In 1997, Christopher Lipinski (Pfizer) analyzed the FDA-approved oral drugs and proposed empirical criteria for "drug-like" molecules:

1. Molecular weight ≤ 500 Da.
2. LogP ≤ 5 (octanol-water partition coefficient; measures lipophilicity).
3. Hydrogen bond donors ≤ 5 (count: -OH and -NH groups).
4. Hydrogen bond acceptors ≤ 10 (count: -O and -N atoms).
5. (Sometimes added: rotatable bonds ≤ 10.)

Compounds violating two or more rules tend to have poor oral bioavailability. The reasons: - Too high MW or logP: poor absorption. - Too many H-bond donors/acceptors: strong solvation; doesn't cross membranes. - Too few H-bonds: too lipophilic; partitions into fats.

Most successful oral drugs satisfy 4 or 5 of the rules. Exceptions exist (some peptide drugs, antibiotics) but tend to require special delivery methods.

Beyond Lipinski

Newer rules: - Rule of 3 (for fragment-based drug discovery): smaller fragments, MW ≤ 300, logP ≤ 3. - Veber's rules: rotatable bonds ≤ 10 + polar surface area (PSA) ≤ 140 Ų. - CNS rules (for blood-brain barrier crossing): MW ≤ 400, logP 2-4, H-bond donors ≤ 3.

These rules are heuristics, not strict requirements. They guide the chemist's optimization but don't replace experimental testing.

35.6 SAR (Structure-Activity Relationships)

When optimizing a lead compound, medicinal chemists make systematic structural variations and measure how each variation affects activity. This is the structure-activity relationship (SAR) approach.

Common SAR tweaks

• Add/remove methyl groups: tune lipophilicity and steric profile.
• Replace H with F: F is small and similar in size to H, but increases metabolic stability (CYP enzymes don't easily oxidize C-F bonds).
• Change ring size: 5- vs 6-membered rings have different geometry and pharmacophore positioning.
• Add/remove OH: H-bond donor; affects solubility and metabolism.
• Swap amine for amide or vice versa: amine is basic and protonatable; amide is non-basic and stable.
• Change ester to amide: amide is more metabolically stable (esters are easily hydrolyzed by esterases).
• Add chirality: one enantiomer is usually more active than the other.

Reading a SAR table

Medicinal chemistry papers typically include SAR tables: a list of analogs and their measured activities (IC₅₀, EC₅₀, Ki). The chemist reads: - Which positions can be modified without losing activity? - Which modifications improve activity? - Which modifications improve ADME without losing activity?

These insights guide the next round of synthesis.

35.7 Bioisosteres

A bioisostere is a structural replacement that retains the biological activity of the original group but changes some other property (often metabolic stability).

Common bioisosteres

Example: ester-to-amide bioisostere

Many drug candidates have ester groups that get cleaved by esterases (rapid metabolism). Replacing the ester with an amide retains the geometry and most properties while extending the half-life. Many drug classes (e.g., some β-blockers) use this bioisostere.

Example: -COOH to tetrazole

Carboxylic acids (-COOH, pKa 4-5) are common in drug structures (they often bind to arginines or lysines via salt bridges). But COOH groups can be glucuronidated and excreted rapidly. Tetrazoles (5-membered ring with 4 N + 1 C) have a similar pKa (~4) and shape, but are metabolically stable. The angiotensin receptor blocker losartan uses a tetrazole instead of COOH for this reason.

35.8 Prodrugs

A prodrug is an inactive compound that is metabolized in the body to the active drug. Prodrugs are used to:

1. Improve oral bioavailability: mask polar groups for membrane crossing.
2. Increase plasma half-life: slow release by metabolism.
3. Target specific tissues: activated only in certain organs.
4. Reduce side effects: prodrug is non-toxic; only the active form has effect at the target site.

Examples

• Enalapril → enalaprilat: an ACE inhibitor for hypertension. Enalapril is the ester (orally absorbable); in the liver, esterases cleave to give enalaprilat (the active diacid form). This is an ester prodrug.
• Valacyclovir → acyclovir: anti-herpes drug. Valine ester of acyclovir; rapidly absorbed; cleaved in vivo to acyclovir (the active form). 5x better bioavailability than acyclovir alone.
• L-DOPA → dopamine: Parkinson's drug. L-DOPA crosses the blood-brain barrier; in the brain, it is decarboxylated to dopamine (the active neurotransmitter). Co-administered with carbidopa (an inhibitor of peripheral L-DOPA decarboxylase) to prevent peripheral activation.
• Heroin → morphine: heroin (diacetyl morphine) crosses the BBB faster than morphine because of the acetyl groups. In the brain, esterases cleave to give morphine. Heroin is a "self-prodrug" — though it is also an active drug at opioid receptors.

Prodrug design principles

• Choose a metabolically labile group (ester, amide, glycoside, etc.) that the body cleaves.
• Ensure the metabolic activation is fast enough.
• Minimize off-target activity of the prodrug itself.
• Ensure the active metabolite is the only species reaching the target.

Prodrug design is an art; many compounds are too rapidly metabolized or not metabolized enough.

35.9 Pharmacokinetics (PK) and Pharmacodynamics (PD)

Pharmacokinetics (PK): what the body does to the drug. Measures: - Absorption (A): how the drug enters the bloodstream from the dosing site. - Distribution (D): how the drug spreads to tissues. - Metabolism (M): chemical transformation by liver enzymes (mostly CYP enzymes). - Excretion (E): how the drug leaves the body (via kidneys, bile).

Together, ADME determines: dose, frequency, duration of action, drug interactions.

Pharmacodynamics (PD): what the drug does to the body. Measures: - Receptor binding (Kd, IC₅₀): how tightly the drug binds the target. - Functional response: what biological effect is produced. - Selectivity: does it bind only the intended target, or also off-targets? - Dose-response curve: how much drug for what effect.

Drug discovery balances PK (the molecule must reach the target at sufficient concentration for sufficient time) with PD (the molecule must engage the target with appropriate effect).

35.10 PROTACs: the thalidomide comeback

The thalidomide story (introduced in Ch 1, deepened in Ch 27 case study 1) closes here.

What is a PROTAC?

A PROTAC (proteolysis-targeting chimera) is a heterobifunctional molecule that has: - One end: binds a target protein (a disease-causing one, often a drug-resistant cancer driver). - The other end: binds an E3 ubiquitin ligase (specifically, cereblon is the most-used E3). - A linker connecting the two ends.

When a PROTAC engages both targets simultaneously, the E3 ligase tags the target protein with ubiquitin. The ubiquitinated target is then degraded by the proteasome.

The PROTAC catalytically destroys the target — one PROTAC molecule can recruit cereblon to ubiquitinate hundreds of target molecules, since the PROTAC-target complex dissociates after the ubiquitination.

Why thalidomide?

Thalidomide and its analogs (lenalidomide, pomalidomide) bind cereblon. The "thalidomide comeback" came from the discovery (around 2010) that cereblon was the target of thalidomide-induced birth defects: thalidomide bound cereblon, recruiting it to ubiquitinate and degrade specific transcription factors needed for limb development.

Researchers realized: if thalidomide could recruit cereblon to degrade unwanted proteins, why not use it deliberately to degrade cancer-driving proteins?

This insight launched the PROTAC field around 2015. By 2024, dozens of PROTACs were in clinical trials for various cancers.

Mechanistic details

1. PROTAC enters the cell.
2. Binds to the target protein (e.g., an aberrant kinase) at one end.
3. Binds to cereblon at the other end.
4. The target-PROTAC-cereblon ternary complex forms.
5. Cereblon's activity tags the target protein with ubiquitin.
6. The ubiquitinated target is degraded by the proteasome.
7. The PROTAC dissociates and is recycled.

The chemistry: the thalidomide ligand binds cereblon; the linker projects out to a separate domain; the second ligand binds the disease-causing protein.

The thalidomide arc

• 1957: Thalidomide introduced as anti-nausea drug.
• 1957-1961: ~10,000 cases of phocomelia in babies whose mothers took thalidomide.
• 1961: Withdrawn from market.
• 1962: Kefauver-Harris Amendment (FDA pre-marketing testing required).
• 1990s: Thalidomide derivatives (lenalidomide) used for multiple myeloma and other cancers.
• 2010: Cereblon identified as the thalidomide target.
• 2015+: PROTACs designed using thalidomide-like ligands.
• 2024: Multiple PROTACs in late-stage clinical trials.

The molecule that caused birth defects in 1960 now forms the foundation of a new drug class. This is the closing arc of the thalidomide story — a tragic chemistry made productive through deeper understanding.

35.11 The future: AI-driven drug discovery

Drug discovery is being transformed by AI: - Target prediction: AI models predict which proteins are druggable. - Hit identification: virtual screening of billions of compounds. - Lead optimization: ML models predict ADME and potency. - Synthesis planning: AI retrosynthesis (Ch 31 case study 2). - Clinical trial design: ML for patient stratification.

Companies like Recursion (lab automation + AI), Insitro (computational biology + AI), Atomwise (virtual screening), and the big pharma R&D groups are aggressively integrating AI.

The chemistry is still Chapter 24-34 chemistry. But the speed of design-build-test cycles is increasing dramatically.

35.12 Summary

1. Drug discovery pipeline: target ID → hit discovery → lead optimization → preclinical → clinical → approval. 10-15 years; $1-3 billion.
2. Aspirin irreversibly acetylates COX's Ser530 (Ch 26 mechanism). Used for inflammation, pain, fever, and cardiovascular prevention (low-dose).
3. Ibuprofen reversibly competes with arachidonic acid for COX's active site. Anti-inflammatory + analgesic; no cardiovascular benefit.
4. Acetaminophen: weakly inhibits CNS COX; analgesic + antipyretic but not anti-inflammatory. Toxic at overdose via NAPQI Michael acceptor + glutathione depletion.
5. Lipinski's rule of 5: MW ≤ 500, logP ≤ 5, H-bond donors ≤ 5, H-bond acceptors ≤ 10. Heuristic for oral drug-like properties.
6. SAR (structure-activity relationship): systematic structural variation to optimize lead compounds.
7. Bioisosteres: structural swaps preserving biology but changing chemistry (e.g., -COOH ↔ tetrazole).
8. Prodrugs: inactive compounds metabolized to active drugs. Examples: enalapril, valacyclovir, L-DOPA.
9. PROTACs: heterobifunctional degraders. Use cereblon-binding ligands (thalidomide-derived) to recruit E3 ligase to a target protein. Degrade the target.
10. Thalidomide arc: from teratogen (1957-1961) → multiple myeloma drug (1990s) → PROTAC ligand foundation (2015+).
11. AI is transforming drug discovery: target prediction, virtual screening, retrosynthesis, ADME prediction. Speed of design-build-test cycles accelerating.

This concludes Part VII. Part VIII covers advanced topics: oxidation/reduction (Ch 36), organometallic chemistry (Ch 37), the art of synthesis (Ch 38), pericyclic reactions (Ch 39), and green chemistry / future directions (Ch 40).