Cryo-EM Reveals Why P2X7 Drugs Fail in Rodents but Work in Humans
Structural biology uncovers the exact amino acid motif causing species-specific drug failures, enabling transgenic mouse models for better P2X7 drug testing.
Summary
P2X7 receptor drugs for inflammation, depression, and pain have repeatedly failed clinical trials partly because they work differently in humans versus lab rodents. This study used cryo-electron microscopy to map exactly why: a specific pocket region called PCP1 contains a key amino acid (valine-312 in humans, alanine-312 in mice) that determines whether a drug binds effectively. The researchers designed a new compound, PSFL1191, that potently inhibits human P2X7 but is inactive in mice. They then engineered transgenic mice carrying the human-like V312 variant, which responded to PSFL1191 normally, validating the model for preclinical drug testing. This work provides a structural roadmap for designing better P2X7 drugs and testing them more accurately before human trials.
Detailed Summary
P2X7 receptors are central mediators of inflammation, pain, depression, and immune responses, making them attractive drug targets. Yet despite dozens of clinical candidates, no P2X7 antagonist has successfully completed Phase II trials. A major but poorly understood obstacle is that compounds highly potent against human P2X7 often show dramatically reduced activity against rodent P2X7 — the very species used in preclinical testing. This study set out to define the structural basis for these interspecies differences and propose a practical solution.
The researchers focused on the portal of the central pocket (PCP), the only confirmed allosteric binding site on P2X7. Using cryo-EM, they resolved structures of human and giant panda P2X7 receptors bound to two distinct inhibitors: PSFL1191, a newly synthesized berberine analog, and JNJ-54175446, a clinical candidate currently in Phase II trials for depression. The PCP was found to contain three sub-pockets: PCP1 (a deep, rigid base pocket), PCP2 (a conserved middle cavity), and PCP3 (an upper region). PSFL1191 penetrates deeply into PCP1, while JNJ-54175446 occupies only PCP2.
Species selectivity mapped precisely to a five-residue motif in PCP1: V312-Y295-M105-F103-P96. In rodents, position 312 carries alanine instead of valine. This single substitution creates a subtly different steric environment in PCP1 that prevents PSFL1191 from binding effectively. Functional assays confirmed that PSFL1191 inhibits human P2X7 with an IC50 in the nanomolar range but is essentially inactive against mouse and rat P2X7. By contrast, JNJ-54175446, which does not reach PCP1, showed no species-dependent differences in potency — explaining why it has advanced further in clinical development.
To validate this mechanism in vivo, the team generated P2rx7^A312V/A312V knock-in mice, in which the murine alanine-312 was replaced with valine to mimic the human receptor. These transgenic mice showed normal baseline physiology, immune function, and P2X7 channel properties. However, when treated with PSFL1191, they exhibited significant alterations in macrophage-mediated bacterial clearance and wound healing — biological effects completely absent in wild-type mice receiving the same compound. This demonstrated that the transgenic model faithfully recapitulates human pharmacology.
The study also showed that the PCP architecture can be exploited deliberately: by designing compounds that penetrate to different depths within the pocket, researchers can engineer inhibitors with or without interspecies differences as needed. Two additional compounds, PSFL2780 (species-selective) and PSFL1633 (species-agnostic), were synthesized to demonstrate this principle. The findings establish a clear structural framework for understanding why so many P2X7 drugs fail in translation and provide both a mechanistic explanation and a practical transgenic mouse tool to improve preclinical-to-clinical predictability for this important drug class.
Key Findings
- AZ10606120 shows 500-fold higher potency at human P2X7 vs mouse P2X7 and 1587-fold higher vs rat P2X7, illustrating the scale of interspecies divergence
- PSFL1191 and its stereoisomer PSFL1190 differ only in chirality yet show >50-fold difference in apparent affinity for human P2X7, demonstrating critical steric complementarity in PCP1
- Species selectivity maps to a single amino acid difference: valine-312 in humans vs alanine-312 in rodents within the deep PCP1 sub-pocket
- JNJ-54175446 binds only PCP2 (the conserved middle cavity) and shows no species-dependent potency differences, explaining its superior clinical translatability
- P2rx7^A312V/A312V knock-in mice carrying the human-like valine-312 responded to PSFL1191 with significant changes in macrophage bacterial clearance and wound healing, effects absent in wild-type mice
- Transgenic A312V mice maintained normal baseline physiology and P2X7 channel properties, confirming the single substitution does not disrupt receptor function
- Rational design produced PSFL2780 (species-selective) and PSFL1633 (species-agnostic) as proof-of-concept that PCP depth of binding can be tuned to control interspecies pharmacology
Methodology
The study combined cryo-EM structural determination of human and giant panda P2X7 receptors bound to PSFL1191 and JNJ-54175446, site-directed mutagenesis, conformational sampling simulations, and functional electrophysiology/calcium flux assays across multiple species orthologs. CRISPR-based knock-in mice (P2rx7^A312V/A312V) were generated and characterized for baseline physiology, immune function, macrophage bacterial clearance, and wound healing. Statistical analyses included IC50 determinations from concentration-response curves and comparison of in vivo outcomes between transgenic and wild-type animals.
Study Limitations
The study was conducted primarily in cell-based and mouse models, and the in vivo transgenic experiments, while compelling, do not yet include disease models beyond bacterial clearance and wound healing. The paper does not report full pharmacokinetic or toxicology profiles for PSFL1191 in the transgenic mice. No conflicts of interest were declared by the authors, though the work was funded by the National Natural Science Foundation of China and a Jiangsu Province innovation program.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.