From destabilizing to stabilizing Ebola glycoprotein: how MWAC-3634 redefines entry inhibition

The prior frontier: why destabilizing GP was the dominant entry-inhibition strategy

Before this paper, the small-molecule landscape against Ebola GP was defined by modest potency and a shared mechanism. Compounds identified through cell-based screening, virtual docking, or drug repurposing generally bound GP with millimolar-range Kd values and blocked entry with micromolar IC50 values, with only a few reaching approximately 100 nM [1]. Structurally characterized compounds—including the repurposed cancer drug toremifene—all occupied the same GP1/GP2 interface cavity and neutralized entry by destabilizing the glycoprotein [1]. Earlier work had established that endosomal proteolysis of GP is a required entry step and that small molecules could interfere with this process, laying the conceptual groundwork for entry inhibition [3]. A separate line of evidence showed that natural polymorphisms can decrease GP stability in a species-dependent manner, reinforcing the idea that stability is functionally consequential [4]. The field thus converged on a model in which pushing GP toward premature conformational change was the way to block fusion.

How DNA-encoded library screening changed the chemical starting point

The anchor study departed from prior discovery methods by screening purified EBOV GP ectodomain directly against 4.73 billion DNA-encoded compounds, a scale inaccessible to conventional high-throughput screening [1]. The top hit, MWAC-2600, contained three chiral centers; two rounds of stereochemical optimization separated trans isomers and flipped one configuration from S to R, yielding MWAC-3634 [1]. This optimization improved binding affinity from a Kd of 69.6 nM for the isomer mixture MWAC-3006 to 40.7 nM for MWAC-3634, and increased thermal stabilization of GP from 6.2 °C to 7.2 °C at 200 µM [1]. The structural basis for this improvement was later confirmed by cryo-EM, which showed that the optimized chiral configurations make precise contacts within the binding pocket [1]. DNA-encoded library screening has similarly enabled rapid discovery of covalent inhibitors against SARS-CoV-2 nonstructural proteins, validating the approach as a generalizable strategy for challenging antiviral targets [8][9]. The same technology has been applied to identify HR1-targeting fusion inhibitors for SARS-CoV-2, further demonstrating its versatility for membrane-fusion targets [5].

Structural evidence for stabilization rather than destabilization

The 2.54 Å cryo-EM structure of the ternary complex comprising EBOV GP, MWAC-3634, and a GP2-binding nanobody revealed that the compound occupies a hydrophobic cavity at the GP1/GP2 interface, interacting with 18 GP residues across both subunits [1]. Unlike toremifene, which engages primarily the western portion of the pocket (13 of 17 contacts), MWAC-3634 forms extensive contacts across both western and eastern regions, with 8 residues from the western portion and 10 from the eastern portion [1]. Binding displaces the DFF lid—a structural element that normally seals the pocket and stabilizes the prefusion state—and pushes the β1-β2 loop outward [1]. Despite these local rearrangements, the overall effect is increased thermostability of the prefusion trimer, raising the energy barrier for transition to the post-fusion conformation [1]. This mechanism contrasts sharply with prior GP-destabilizing inhibitors and aligns with single-molecule FRET studies showing that viral envelope glycoproteins sample an ensemble of prefusion conformations whose stability is thermodynamically tuned by host environmental cues [10]. The finding that GP requires an optimal level of stability—neither too rigid nor too labile—reframes the target as a conformational landscape rather than a simple lock-and-key binding site.

Potency gains and the gap between mouse efficacy and human translation

MWAC-3634 inhibits wild-type EBOV infection with an IC50 of 0.65 nM, at least one to two orders of magnitude more potent than previously reported GP-targeting small molecules [1]. Against a mouse-adapted strain bearing the S246P GP substitution and eight additional genomic changes, potency was reduced to an IC50 of 2.87 nM, a difference the authors attribute to strain-specific GP variation [1]. In the lethal mouse challenge, intraperitoneal dosing achieved 100% survival (7/7) with a 4019-fold reduction in serum viral load by day 4, while oral dosing achieved 85.7% survival (6/7) with a 176-fold reduction that did not reach statistical significance at that early time point [1]. The oral bioavailability of 69% and sustained plasma concentrations above 2 µM for eight hours support further development, though the authors note that the mouse-adapted strain differs from wild-type virus and that murine disease kinetics differ from human infection [1]. A separate study using recombinant vesicular stomatitis viruses pseudotyped with filoviral glycoproteins in Ifnar1-knockout mice provides a BSL-2-compatible surrogate model that could facilitate broader testing of GP-targeting countermeasures, but it does not substitute for non-human primate validation [7]. Natural product-based entry inhibitors identified through computational screening have shown micromolar potency (IC50 of 1.72 µM for the best hit), highlighting the potency advantage of the DNA-encoded library approach [2]. Configurationally locked ridaifen-B analogs have achieved nanomolar potency against EBOV GP-mediated entry, but their binding site at the GP1/GP2 fusion loop and their destabilizing or stabilizing mechanism remain to be fully characterized [11].

What remains uncertain: mechanism boundaries and the path to clinical relevance

The claim that GP stabilization is an effective anti-filovirus strategy rests on a single compound in a single animal model. The authors explicitly state that efficacy evidence comes from a lethal mouse challenge and that translation of potency will require evaluation in additional preclinical models, particularly non-human primates [1]. The mouse-adapted EBOV strain used for in vivo testing differs from wild-type virus at GP residues 65 and 544, which are located at or close to the MWAC-3634-binding region, and the compound showed reduced potency against this strain (IC50 of 2.87 nM versus 0.65 nM for wild-type) [1]. Whether the stabilization mechanism generalizes to other filoviruses remains untested, though the GP1/GP2 interface pocket is conserved in Sudan virus and Marburg virus [1]. The DFF lid displacement and β1-β2 loop rearrangement observed upon binding raise the question of whether long-term exposure could select for escape mutations that restore lid function or alter pocket geometry. Additionally, the compound has lower aqueous solubility than toremifene, and the authors identify solubility and metabolic stability as targets for future analog development [1]. The broader concept that envelope glycoproteins require an optimal stability window is supported by single-molecule imaging studies across multiple viruses, but whether this principle can be exploited pharmacologically across different fusion protein classes remains an open question [10]. Griffithsin, a lectin that aggregates virions by binding high-mannose glycans, represents an entirely different entry-inhibition mechanism and shows stronger activity against Lassa, Lujo, and Crimean-Congo hemorrhagic fever viruses than against Ebola and Marburg, underscoring that GP-targeting strategies must account for structural differences among hemorrhagic fever viruses [6].

About These Sources

This research page is built on 11 peer-reviewed studies — published from 2011 to 2026, 8 from 2024 or later, collectively cited 155 times — selected as the most relevant from 13 studies that passed quality screening, drawn from 58 papers retrieved from a database of over 500 million.

Sources used in this answer