BGPT: Paper Review: Human liver glycogen phosphorylase inhibitors bind at a new allosteric site

Explore by Goal

Quick Explanation Copied

This Pfizer structure-based design study identified a novel allosteric site at the dimer interface of human liver glycogen phosphorylase a (HLGPa), and exploited it rationally: linking two chloroindole carboxamide ligands boosted potency ~2000-fold to 6 nM (CP-526,423) . The work is a landmark of ligand-linking drug design, though its central limitation—absence of direct in vivo validation in this paper—was later partly addressed by independent canine efficacy data .

Long Explanation

The Central Discovery: A Novel Allosteric Site at the Dimer Interface

Rath et al. (Pfizer) solved two X-ray crystal structures of human liver glycogen phosphorylase a (HLGPa) complexed with indole-2-carboxamide inhibitors: HLGPa/GlcNAc/CP-403,700 at 2.4 Å resolution (PDB 1EXV) and HLGPa/GlcNAc/CP-526,423 at 2.2 Å (PDB 1EMG) . The site is genuinely novel: two inhibitor molecules bind near the molecular 2-fold axis within a solvent-filled cavity at the dimer interface, forming one continuous composite binding pocket spanning both subunits. This was contrary to the authors' initial hypothesis that these glucose-sensitizing inhibitors bound the caffeine site—synergy with caffeine correctly indicated a distinct locus.

Structurally, the site has two distinct environments: a hydrophobic chloroindole pocket (Arg60, Lys191 aliphatic chains plus six hydrophobic residues, burying 266 Å² of hydrophobic surface) and a second, polar/solvent-exposed region built from both subunits. Critically, in the active conformation, the side chains of Arg60, Val64, and Lys191 sterically occlude this site and must move to accommodate binding—providing a plausible mechanistic basis for how binding stabilizes the inactive conformation.

The Masterstroke: Structure-Guided Ligand Linking

Recognizing that the two bound CP-403,700 molecules sit within 6 Å of each other, the team applied Jencks' ligand-additivity principle and covalently joined two chloroindole carboxamide units. The resulting bis-indole CP-526,423 inhibits HLGPa with an IC50 of 6 nM—a ~2000-fold improvement over its unlinked analog CP-305,494—and the crystal structure confirms a single molecule spans both pockets (chloroindole r.m.s.d. of 0.34 Å versus CP-403,700) . This is a textbook demonstration of entropic advantage in fragment linking—rarely executed so cleanly then.

The graph shows the glucose-synergy data: within the physiologic glucose range, potency strengthens nearly 8-fold, a mechanism the authors emphasize as hypothecemia-protective—inhibition deepens precisely when glucose is abundant.

Allosteric Mechanism and Independent Validation

Functionally, CP-403,700 behaves as a classic Monod-Wyman-Changeux allosteric effector: it doubles the rate of PP1-mediated dephosphorylation of Ser14 and synergizes with active-site inhibitors . Later thermodynamic work on related indole-site inhibitors (AVE5688, AVE2865, AVE9423) confirmed enthalpy-driven binding preferential to the inactive b-form, supporting the site's physiological relevance . In vivo validation came from canine studies using CP-316819, where GP inhibition completely suppressed basal hepatic glucose output and blocked glucagon-stimulated output by diverting flux into glycogen synthesis .

Critical Limitations and Blindspots

Species and isoform gaps: The concurrent rabbit-muscle GPb structure of CP-320,626 (Oikonomakos et al.) confirmed site conservation, but muscle isoform relevance matters differently—skeletal muscle concerns were later tested in rats, where CP-316819 reduced GP activation without impairing force production .
Crystal artifacts: Both structures required GlcNAc co-crystallization (CP-403,700 alone failed to yield crystals), so the reported conformation may partially depend on the active-site ligand; the authors acknowledge minimal effort to crystallize without it.
In vitro IC50 vs pharmacology: All potency data derive from malachite-green phosphate-release assays in purified enzyme; hepatocyte-level and human PK data were absent from this paper—arguably its largest gap at publication.
Industry context: As a Pfizer-discovery paper with an accompanying EPO patent, positive-result framing toward therapeutic potential is expected; claims of "most potent inhibitors to date" were accurate then but did not survive into approved drugs—no GP inhibitor for diabetes ever reached the clinic.

Bottom line: The structural work is rigorous, the ligand-linking result is reproducible from reported data, and the mechanistic claims were independently corroborated. The paper overpromised only in therapeutic translation—a caveat of the era, not a flaw of the science.

Feedback:

Updated: November 01, 2026