Journal Publication

AI-designed antibody GB-0669 neutralizes SARS-CoV-2 in first human trial

A first-in-human study of an AI-optimized antibody targeting a conserved spike region showed dose-dependent virus neutralization and no dose-limiting toxicity in 51 healthy adults.

By AppliedXLJuly 14, 2026

Executive Summary

A first-in-human trial of an AI-designed antibody targeting a conserved, mutation-resistant region of the SARS-CoV-2 spike protein showed dose-dependent virus neutralization without dose-limiting toxicity.
Earlier antibody therapies against the virus's receptor-binding domain lost effectiveness as resistant variants emerged, and this antibody was designed against a region under less selective pressure to avoid that failure mode.
The tolerability and pharmacokinetic profile support advancing a specific middle dose into further testing in a population that cannot clear infection as readily as healthy adults.
Laboratory testing showed the antibody's neutralization improved when paired with existing antiviral drugs, pointing toward combination use rather than a standalone replacement for current therapy.

The stake

Monoclonal antibodies directed at the SARS-CoV-2 spike receptor-binding domain worked as COVID-19 treatments until resistant variants emerged and eroded their activity. GB-0669 was engineered against the spike S2 stem helix instead, a region the abstract describes as subject to limited selective pressure from antibody responses generated by natural infection or vaccination. The premise being tested is whether targeting a more conserved viral region can produce an antibody less vulnerable to the escape mutations that undercut earlier RBD-directed therapies. Randomized

How it was done

The trial was a randomized, double-blind, placebo-controlled, single-ascending-dose study in healthy adults aged 18 to 55. Participants received a single intravenous dose of GB-0669 or placebo across five ascending cohorts of 100, 300, 600, 1200, and 2400 mg, and were followed for 43 weeks to assess safety, pharmacokinetics, and pharmacodynamics measured by serum live virus neutralization. Preclinical safety work was conducted in cynomolgus monkeys before the human dosing began. Separate in vitro experiments tested GB-0669's neutralization activity when combined with the antivirals remdesivir, nirmatrelvir, and molnupiravir. Randomized

The results

Of 51 participants, 36 received GB-0669 and 15 received placebo. GB-0669 was well tolerated with no dose-limiting toxicities, and all adverse reactions were Grade 1 or 2. Pharmacokinetics were dose-proportional up to 2400 mg, with a half-life of 54 days. Serum live virus neutralization increased in a dose-dependent manner at the 600 mg and 1200 mg doses, separating from placebo. A neutralizing index that adjusts serum concentration for in vitro potency supported therapeutic activity for two weeks after a single dose. Randomized

The combination signal

The in vitro experiments found that combining GB-0669 with remdesivir, nirmatrelvir, or molnupiravir improved the antibody's neutralization profile relative to GB-0669 alone. Paired with a two-week window of single-dose activity and a tolerability profile clean enough to reach 2400 mg without dose-limiting toxicity, that combination result is what points the program toward pairing with existing antivirals rather than replacing them. Randomized

Next step in development

The authors conclude the data support testing GB-0669 at the 1200 mg dose in a Phase 2 trial for treating COVID-19 in immunocompromised individuals, a population less likely to clear the virus without added antiviral support. That population choice follows directly from the mechanism: a durable, half-life-extended antibody against a conserved target is positioned as most useful where a patient's own immune response is least able to neutralize the virus on its own. Randomized

This analysis was produced using AI-assisted reporting systems, AppliedXL data, and official public records. These systems undergo editorial review, quality checks, and regular audits by human experts. Errors may still occur, as with any automated system. Always consult the linked primary sources. Read our AI Editorial Policy.