Ep209: Mathai Mammen on Developing Drugs Against Undruggable Targets

What It Covers

Mathai Mammen, CEO of Parabolas Medicines, details how the company uses stabilized alpha helical peptides incorporating over 3,000 synthetic amino acids to drug previously inaccessible intracellular proteins. The conversation covers the science behind targeting beta catenin and TCF, early clinical results in desmoid tumors showing 74% objective response rate, and plans following a record $770M biotech IPO in 2026.

Key Questions Answered

• •Drugging flat protein surfaces: Roughly 80% of intracellular proteins lack the pockets or clefts required for conventional small molecule drugs, making them inaccessible to existing modalities. Parabolas addresses this by using alpha helical peptides that can traverse cell membranes and bind flat, smooth protein surfaces. The alpha helix structure works because its self-satisfied hydrogen bonds do not require water removal during membrane crossing, a property borrowed directly from naturally occurring transmembrane proteins.
• •Platform differentiation from stapled peptides: Earlier stapled peptide approaches like Aileron's used a single central chemical lock on the helix, leaving the peptide ends unstructured and limiting amino acid diversity to a small set of helix-compatible natural residues. Parabolas employs hundreds of distinct stabilization chemistries distributed across the full peptide, enabling use of over 3,000 synthetic amino acids. This produces a chemical search space of 3,000 to the power of 20 possible compounds, creating a structurally distinct and proprietary drug class.
• •Indication selection strategy for novel mechanisms: When a drug targets a previously undrugged mechanism, choosing a late-line, heavily mutated cancer population as the first indication risks obscuring efficacy signals. Parabolas learned this in colorectal cancer, where fifth- and sixth-line patients had accumulated dozens of additional driver mutations beyond beta catenin. The corrective strategy was to pivot to genetically purer, earlier-stage cancers like desmoid tumors, where beta catenin is the dominant driver, yielding a 74% objective response rate across 38 patients.
• •Portfolio filtering by replaceability: Mammen applies a "missing man" test to drug development decisions: if a program disappeared overnight, would any patient population go unserved? Programs that are nth-in-class on a validated mechanism fail this test because competitors will cover that patient need. This filter, applied at both Merck and J&J, led to terminating incremental programs and redirecting resources toward first-in-class mechanisms. At J&J, 21 deals were executed under this philosophy, with 18 still advancing.
• •Safety profile of beta catenin blockade: Prior Wnt pathway inhibitors targeting porcupine produced significant bone density loss and taste disturbance, creating concern about the entire pathway's therapeutic window. Zolucatotide shows only mild, manageable bone effects addressable with standard osteoporosis medications, and no taste disturbance, suggesting the earlier side effects were off-target rather than mechanism-driven. Aldosterone reduction appears only at doses well above those needed for efficacy, further widening the usable therapeutic index.
• •AI moat through proprietary unnatural amino acid data: General-purpose AI protein design tools cannot model peptides containing thousands of synthetic amino acids because no training data exists for those structures. Parabolas has accumulated over a decade of experimental data on its helicon peptides, which it uses to fine-tune generative models through reinforcement learning. This creates a defensible data moat: the predictive models improve with each new compound synthesized, reducing the need to physically make and test every candidate before identifying leads.
• •Hiring for curiosity, hunger, and mission alignment: Across Theravance, Merck, and J&J, Mammen identifies three non-negotiable hiring criteria: genuine intellectual curiosity, sustained hunger to achieve outcomes, and authentic belief in the mission of making medicines. Curiosity correlates with humility because learners continuously revise prior assumptions. Mission alignment prevents the short-term distractions that derail programs. At Theravance, this culture produced 16 eventual biotech CEOs from the first 200 employees, and 9 approved products from 30 clinical nominations.