How a unique class of molecules is helping us drug the undruggable.
When you think of medicine, you likely picture a traditional pill—a small molecule designed to fit neatly into a specific target within the body. For decades, drug discovery has been guided by the "Rule of Five," a set of principles suggesting that effective drugs should be relatively small and simple. But this rule has a major limitation: it excludes vast biological territories, leaving many disease-causing proteins considered "undruggable".
The term "undruggable" refers to proteins that lack well-defined binding pockets, making them difficult to target with conventional small-molecule drugs.
Enter macrocycles. These unique molecules, defined by a large ring structure of 12 or more atoms, are emerging as powerful therapeutic agents that bridge the gap between traditional small-molecule drugs and larger biologic treatments like antibodies 3 . They combine the precise targeting of biologics with the stability and ease of administration of a pill, opening new frontiers in medicine 8 .
Typically follow the "Rule of Five" with molecular weight < 500 Daltons and simple structures.
Large ring structures that defy traditional rules while maintaining drug-like properties.
Macrocycles occupy a sweet spot in the world of therapeutics. They are substantially larger than typical small-molecule drugs, which allows them to cover a larger surface area and interact more extensively with their targets 3 . This is crucial for targeting large, flat protein-protein interaction interfaces, which are often fundamental to disease processes but notoriously difficult for smaller drugs to disrupt 3 8 .
Their secret weapon is structural pre-organization. Unlike flexible linear molecules, macrocycles are partially constrained into the 3D shape they need to bind to their target. This means they don't have to pay a high energy penalty to adopt the correct conformation upon binding, leading to higher affinity and selectivity 1 4 .
Although many macrocycles violate the traditional Rule of Five, nearly 40% of macrocyclic drugs are orally bioavailable 3 4 . They achieve this through a clever property known as "chameleonicity"—the ability to change their conformation based on their environment 4 . They can shield their polar groups to pass through fatty cell membranes and then expose them to interact with water or their target 1 4 .
| Feature | Traditional Small Molecules | Macrocycles | Biologics (e.g., Antibodies) |
|---|---|---|---|
| Molecular Size | Small (< 500 Daltons) | Intermediate | Large |
| Target Type | Deep, enclosed pockets | Flat surfaces, protein-protein interfaces | Extracellular targets |
| Administration | Oral (typically) | Oral & Parenteral | Injection/Infusion |
| Structural Flexibility | Low | High & Chameleonic | High |
| Example Therapeutics | Aspirin, Metformin | Cyclosporine, Erythromycin | Humira, Keytruda |
Nature has been designing macrocycles long before modern science. A remarkable 88% of FDA-approved macrocyclic drugs are natural products or their derivatives 4 . For example, the immunosuppressant cyclosporine and the antibiotic erythromycin are both naturally occurring macrocycles.
88% of FDA-approved macrocyclic drugs originate from nature.
44.4% of macrocyclic drugs target infections.
20.8% of macrocyclic drugs are used in oncology.
The true power of macrocycles lies in their ability to drug challenging targets. An analysis of macrocycle-target complexes found that the majority bind to targets with difficult-to-drug binding sites—flat, groove-shaped, or tunnel-shaped sites that are often involved in protein-protein interactions 4 .
| Therapeutic Area | Percentage of Macrocyclic Drugs | Example Conditions |
|---|---|---|
| Infectious Disease | 44.4% | Bacterial, viral, and fungal infections |
| Oncology | 20.8% | Various cancers |
| Immunosuppression | 5.6% | Organ transplant rejection, autoimmune diseases |
| Other Indications | 23.6% | Chronic pain, heart failure, genetic obesity |
The journey of a new macrocyclic drug from concept to candidate is being revolutionized by artificial intelligence. A recent pioneering study demonstrated this with the development of a potent JAK2 inhibitor, a target for blood cancers like polycythemia, using a deep learning model named CycleGPT 9 .
The researchers first pre-trained CycleGPT on a massive dataset of 365,063 biologically active compounds from the ChEMBL database. This taught the model the basic "grammar" and "syntax" of chemical structures represented in SMILES notation (a string-based way to describe molecules) 9 .
To overcome the scarcity of macrocyclic data, the model underwent transfer learning with 19,920 known macrocyclic molecules. This step specialized its knowledge from general bioactive compounds to the unique and complex world of macrocycles 9 .
The team introduced an innovative sampling algorithm, HyperTemp, which encouraged the model to explore new structural variations. Instead of always choosing the most probable next step in molecule generation, HyperTemp increased the probability of "suboptimal" but innovative choices, ensuring the generated macrocycles were both valid and novel 9 .
The fine-tuned AI was then used to generate new macrocyclic candidates targeting JAK2. The most promising compounds were synthesized and tested for their ability to inhibit the JAK2 kinase in biochemical and cellular assays, with the best candidate evaluated in an animal model of disease 9 .
The study was a resounding success. CycleGPT-HyperTemp outperformed other state-of-the-art molecular generation models, with 55.8% of its generated compounds being valid, unique macrocycles not present in the training data 9 .
Novel & Unique Macrocycles
Far superior to other models (e.g., CharRNN: 11.76%)Top JAK2 Inhibitor Potency (IC50)
Exceptional binding strength to the targetMore importantly, the AI-designed macrocycles were highly effective. Three potent JAK2 inhibitors were identified with IC50 values in the low nanomolar range (1.65 nM, 1.17 nM, and 5.41 nM), indicating exceptional potency. One candidate, in particular, showed a better kinase selectivity profile than two marketed drugs, Fedratinib and Pacritinib, meaning it was less likely to cause off-target side effects 9 .
In a mouse model of polycythemia, the discovered macrocycle effectively treated the disease at a lower dose than the established drugs, highlighting its promising therapeutic potential 9 .
| Metric | CycleGPT-HyperTemp Performance | Significance |
|---|---|---|
| Novel & Unique Macrocycles | 55.80% | Far superior to other models (e.g., CharRNN: 11.76%) |
| Top JAK2 Inhibitor Potency (IC50) | 1.17 nM | Exceptional binding strength to the target |
| Kinase Selectivity | Inhibited only 17 of 17 wild-type kinases | More selective than marketed drugs, suggesting fewer side effects |
| In Vivo Efficacy | Effective at lower doses than Fedratinib/Pacritinib | Promising therapeutic window and efficacy in a living organism |
Advancing macrocyclic drug discovery relies on a sophisticated toolkit that combines cutting-edge synthetic chemistry, computational power, and analytical techniques.
Ring-Closing Metathesis, Solid-Phase Synthesis, DNA-Programmed Chemistry
AI/Deep Learning Models, Molecular Dynamics Simulations
mRNA Display, High-Throughput Screening (HTS)
NMR, X-ray Crystallography, Cryo-Electron Microscopy
| Tool Category | Specific Technologies | Function |
|---|---|---|
| Synthetic Chemistry | Ring-Closing Metathesis 1 , Solid-Phase Synthesis , DNA-Programmed Chemistry (DPC) , Modular Biomimetic Assembly 1 | Enables efficient construction of complex macrocyclic rings and large libraries for screening. |
| Computational Design | AI/Deep Learning Models (e.g., CycleGPT 9 ), Molecular Dynamics Simulations 1 , Structure-Based Design | Predicts properties, generates novel scaffolds, and models interactions with biological targets. |
| Screening Platforms | mRNA Display , High-Throughput Screening (HTS) 5 | Rapidly identifies potential lead compounds from vast libraries. |
| Structural Analysis | Nuclear Magnetic Resonance (NMR) 3 , X-ray Crystallography, Cryo-Electron Microscopy (cryo-EM) 3 | Determines 3D structure and conformational dynamics in solution and when bound to targets. |
Macrocycles are more than just a scientific curiosity; they represent a paradigm shift in how we approach drug discovery. By successfully targeting proteins once deemed undruggable, they offer new hope for treating a wide range of complex diseases. The future of the field lies in the continued integration of innovative synthetic methods with powerful computational tools like AI, which will dramatically accelerate the design and optimization of these promising molecules 1 9 .
"As platforms for synthesizing and screening macrocycles become more advanced, and as computational models provide ever-deeper insights into their chameleonic behavior, we can expect a new wave of macrocyclic therapeutics to emerge."
These tiny rings are poised to make a monumental impact on human health.
Expanding the druggable proteome
Accelerating design and optimization
Treating previously untreatable conditions