Session XII — Nobel Laureate Panel Discussion II
When the Molecules Don’t Behave
Three laureates — Thomas Südhof (2013), Brian Kobilka (2012), and David Baker (2024) — on the A-beta paradox, allosteric GPCR modulators, and de novo protein design.
Südhof — The A-beta Paradox
- Lecanemab made brains plaque-free; cognitive decline only slowed by 27%
- iPSC neurons with APP Swedish mutation: increased synapse density (not decreased)
- A-beta 40: consistently increased synapse formation; A-beta 42 toxic only at aggregation concentrations
- Pre-synaptic compartment selectively shrinks before neuron death
“We don’t actually know whether the little beneficial effect was due to the decrease in plaques — or the increase in free A-beta — or both.”
Kobilka — Allosteric Modulators for the Mu-Opioid Receptor
GPCRs: ~800 genes, ~35-40% of all approved drugs act on them.
Negative Allosteric Modulator
- From 4-billion-compound DEL screen; dramatically extends naloxone’s duration in mice
- Addresses clinical problem: fentanyl outlasts emergency naloxone
Positive Allosteric Modulator (AP58)
- From 40-billion-compound screen; produces dose-dependent pain relief without exogenous opioid
- Amplifies endogenous met-enkephalin; no addiction signal in conditioned place preference
- Preserves temporal regulation of analgesic signaling (vs. chronic elevation with morphine/fentanyl)
Baker — Proteins by Design
De Novo Binders
- Diffusion-based generative models: TNF receptor binder at 6-picomolar affinity
- Survives gut proteases; promising in vivo anti-inflammatory activity
Intrinsically Disordered Proteins
- Co-diffuse disordered target + binder simultaneously → tau binders that suppress/disassemble fibrils
- Fused to FBOX → bioPROTAC that achieves complete tau clearance from cells
Designed Proteases
- Entirely new folds; enhance amide bond hydrolysis by >10 orders of magnitude
- TDP-43-targeting protease: could degrade target at far lower doses than binding alone
Cancer Applications
- Designed mesothelin binders → CAR-T; bioPROTACs targeting DNMT3A reduce T-cell exhaustion
- Conditionally active interferon beta (masked until encountering integrin in tumor microenvironment)
- Mini-proteins discriminating individual peptide-MHC neoantigens
Immunogenicity
- De novo proteins: counterintuitive safety advantage — even if antibodies form, cannot cross-react with any human protein (no autoimmunity risk)
Shared Insight: On-Target Binding Is Becoming Solved; Biology Is Not
- Structural data: 300,000 atomic-resolution structures. Clinical data: “tiny, tiny, tiny”
- “If we had full clinical trial results on everything being tested all over the world, maybe we would start generating enough data” — Baker
- Predicting adverse effects in an integrated organism remains fundamentally harder than target binding
Unlimited Budget Thought Experiment
- Baker: Scale protein design across medicine, sustainability (PFAS, plastics), industrial chemistry, sensing
- Südhof: Integrative cell biology bridging atomic and whole-organism scales; reconcile Alzheimer’s camps
- Kobilka: Fund expensive next steps in GPCR drug development — compound libraries, in vivo pathway-selectivity studies
Comments