Eight nanobodies from an alpaca and a llama neutralised 17 African snake venoms

Researchers immunised an alpaca and a llama with the venoms of 18 different snakes — mambas, cobras and a rinkhals among them — built phage display libraries from the animals, and screened for nanobodies that bound broadly and tightly across venom toxins [s1]. Eight of the resulting nanobodies were combined into a defined oligoclonal mixture [s1].

The result, published in Nature on 29 October, is an experimental polyvalent recombinant antivenom capable of neutralising seven toxin families or subfamilies [s1]. In mice it prevented venom-induced lethality across 17 African elapid snake species and markedly reduced venom-induced dermonecrosis for all cytotoxic venoms tested [s1]. It performed better than a currently used plasma-derived antivenom [s1].

Why the existing product is the problem

Snakebite envenoming claims thousands of lives each year and causes severe injury across sub-Saharan Africa, where many victims depend on antivenoms derived from animal plasma as their only treatment option [s1].

Those plasma-derived antivenoms are made by immunising large animals — typically horses — and purifying antibodies from their blood. The paper lists four problems with the approach: the products are expensive, they can cause adverse immunological reactions, they offer limited efficacy against local tissue damage, and they are often ineffective against all medically relevant snake species [s1].

The last two are the clinically important ones. A treatment that saves a life but does not prevent the venom from destroying the flesh around the bite leaves survivors with amputations and permanent disability. And an antivenom raised against one region's snakes may do little against another's.

Why breadth is hard

Developing a broad-spectrum treatment is difficult because of the sheer diversity of venomous snakes and the complex, variable composition of their venoms [s1]. Venom is not one toxin; it is a mixture whose components differ between species, between populations of the same species, and sometimes with the age of the individual snake.

The strategy in this paper attacks that problem from the toxin side rather than the species side. Instead of matching one antivenom to one snake, it identifies nanobodies that bind conserved features shared across toxin families, so that a fixed mixture of eight covers seven toxin families or subfamilies and, through them, 17 species [s1].

Nanobodies are single-domain antibody fragments of the kind found in the camelid immune repertoire the study sampled [s1]. Because a fixed set of them can be produced recombinantly, the resulting mixture has a defined composition — which plasma, whose antibody content varies with the individual animal, does not.

Where this sits in a fast-moving field

This is the second high-profile attempt in 2025 to replace plasma-derived antivenom with something designed rather than harvested.

In January, a separate group used deep learning to design proteins de novo that bind short-chain and long-chain α-neurotoxins and cytotoxins from the three-finger toxin family [s2]. Those designs neutralised all three 3FTx subfamilies in vitro and protected mice from a lethal neurotoxin challenge [s2]. That paper notes that snakebite envenoming claims over 100,000 lives annually and that existing plasma-derived polyclonal antibodies have high cost and limited efficacy against three-finger toxins [s2].

The two approaches converge on the same diagnosis and diverge on method: one screens an immunised animal's repertoire for broadly neutralising binders, the other designs binders computationally without an animal. Both end at the same place — a defined, recombinant, manufacturable product instead of purified horse plasma.

What has not been shown

No human has received this antivenom. The evidence is neutralisation of lethality and dermonecrosis in mice [s1]. Mouse lethality assays are the standard preclinical model for antivenom, and they are also a poor proxy for the clinical reality of a bite: real envenoming involves an unknown dose delivered at an unknown time before treatment, in a person whose tissue damage may already be underway.

The species coverage is African elapids — cobras, mambas, rinkhals. It does not extend to vipers, whose venoms cause a different set of injuries and which fall outside the elapid family this mixture was built against. The paper's own claim is protection against bites by all medically relevant African elapids [s1], which is a specific and bounded claim.

Cost and manufacturing are also unresolved by this paper. Recombinant production is more controllable than plasma, but "controllable" and "affordable at the volumes rural African clinics need" are different propositions, and nothing here establishes the second.

What to watch

The meaningful next steps are toxicology and a first-in-human safety study, and evidence on whether a fixed eight-nanobody mixture holds up against venom from snake populations not represented in the 18 venoms used for immunisation. Geographic venom variation is the standard way broad-spectrum antivenoms fail in the field.

The structural point stands regardless of what happens to this particular mixture. Two independent groups have now shown, in animals, that a defined recombinant product can match or beat plasma-derived antivenom. The bottleneck for snakebite is shifting from whether such a product can exist to whether anyone will pay to develop and distribute one.

Sources

• Nanobody-based recombinant antivenom for cobra, mamba and rinkhals bites — Nature, 29 October 2025
• De novo designed proteins neutralize lethal snake venom toxins — Nature, 15 January 2025