ASCEND Hidden-Druggability Atlas · Case Report · Multidrug-Resistant Gram-Negative Infections
A new structural route against multidrug-resistant (MDR) bacterial infections — targeting LpxH in Gram-negative pathogens.
From structure and motion alone — with no prior knowledge of where any drug binds — ASCEND mapped the druggable landscape of an essential Gram-negative antibiotic target, designed candidates de novo, and docked them into the discovered pockets.
Summary. Given only an experimental LpxH structure — and no inhibitor, binding, or co-crystal information — ASCEND modeled the enzyme’s intrinsic motion, mapped its druggable landscape, generated de-novo candidate molecules, and docked them into the discovered pockets. The blind analysis independently recovered the true catalytic site; beyond it, a non-active-site pocket accepts a designed candidate with binding comparable to the active site — the differentiated opportunity current LpxH chemistry lacks.
Figure 1 — interactive. Drag to rotate, scroll to zoom. The LpxH surface is colored by ASCEND’s blind, motion-derived druggability evidence (warmer = stronger); the pale forms are de-novo candidate molecules docked into the discovered pockets. Molecular envelopes only — exact chemistry, residues and coordinates are withheld.
With no binding information provided, the strongest druggability signal fell on LpxH’s true catalytic center — a rigorous internal control that the map reflects real structure, not fitted answers.
De-novo candidates were designed for every region and docked back into their pockets. A distal, non-active-site pocket accepts a drug-like candidate with binding comparable to the active site — the foothold needed for chemistry with better properties and reduced exposure to existing resistance, where known LpxH inhibitors cannot go.
At least one identified region is supported by a separate experimental structure in which the site genuinely opens, indicating the opportunity is a physical conformational feature, not an artifact.
De-novo candidates were generated for every discovered region and docked into the corresponding pockets.
Docking (AutoDock Vina) places the designed candidates within the discovered pockets. Notably, a non-active-site pocket binds a designed candidate almost as strongly as the catalytic site — the key signal that a differentiated, non-orthosteric inhibitor is feasible on this target.
LpxH (UDP-2,3-diacylglucosamine hydrolase) catalyzes a committed step in building lipid A, the anchor of the lipopolysaccharide outer membrane that defines Gram-negative bacteria. Inhibiting LpxH prevents correct assembly of this membrane, which is lethal to the cell.
LpxH is essential across most Gram-negative pathogens yet has no human homolog — the profile of a clean, low-host-toxicity antibiotic target.
Organisms dependent on this pathway include the most difficult hospital infections: Klebsiella pneumoniae, Pseudomonas aeruginosa, Acinetobacter baumannii and E. coli — carbapenem-resistant, ESKAPE-class bacteria the WHO designates critical priority for new antibiotics.
The outer membrane that makes these bacteria so resistant is the structure LpxH helps build, which is what makes the target so compelling.
Antimicrobial resistance is projected to be among the leading causes of death worldwide, with Gram-negative bacteria driving its most urgent and least-treatable infections. On LpxH, the field has largely converged on one difficult active-site pocket.
The significance of this analysis is where it points: reading motion rather than a frozen structure reveals druggable opportunity a static view misses — and ASCEND has already advanced it to de-novo candidates that dock into a non-active-site pocket with active-site-comparable binding. That is a credible, differentiated starting point for next-generation antibiotics with improved properties and a mechanism less exposed to today’s resistance.