Healthcare

Harbour BioMed’s AI-Enabled Obesity Antibody LET003 Has IND Accepted

Naomi Crest
Add Unite.AI to your preferred sources on Google
Key PointsLET003's IND acceptance advances a preclinical program without proving patient efficacy.
View more
  • LET003's IND acceptance advances a preclinical program without proving patient efficacy.
  • Harbour BioMed designed LET003 to block ACVR2A and ACVR2B and preserve lean mass.
  • The Hu-mAtrIx platform's validation metrics describe its discovery engine rather than LET003 itself.

Harbour BioMed announced on November 18, 2026 that China’s National Medical Products Administration had accepted the investigational new drug (IND) application for LET003, its first drug candidate developed using the Hu-mAtrIx AI platform. The monoclonal antibody is intended to treat overweight or obesity by blocking the activin receptors ACVR2A and ACVR2B.

The milestone is acceptance of an application to enter clinical development, not a reported clinical result. The efficacy findings in the announcement come from preclinical studies, so they do not establish weight loss or body-composition changes in patients. Harbour BioMed’s therapeutics pipeline lists LET003 as a global obesity program alongside LET001 in weight management.

Rationale for Dual Receptor Blockade

Dr. Bob Zhang, president of Harbour BioMed’s Metabolic Disease Franchise, said obesity treatment is moving “beyond weight reduction alone toward improving body composition and achieving more sustainable health outcomes.” The company’s concern is the lean-mass loss that can accompany substantial weight reduction with GLP-1-based therapies. By blocking ACVR2A and ACVR2B, which help regulate muscle and fat homeostasis, LET003 is designed to enhance fat reduction while preserving or improving lean mass, particularly alongside existing weight-loss treatments.

Company-reported preclinical studies show that combining LET003 with semaglutide significantly enhanced fat reduction while preserving lean mass. LET003 also produced lean-mass-promoting effects at a lower dose that were comparable with bimagrumab at a higher dose. Those findings provide the biological rationale for clinical testing, but they do not establish efficacy in patients.

How the Hu-mAtrIx Platform Works

Harbour BioMed’s platform documentation describes a closed loop that combines AI design, intelligent screening and wet-lab validation for fully human heavy-chain-only antibodies, an antibody format without light chains. Built on the Harbour Mice platform, the model’s training set combines 9 million next-generation-sequencing-derived heavy-chain-only antibody sequences with public data. A fine-tuned protein large language model generates new candidate sequences and then optimizes them for target specificity.

The screening stage applies an AI classifier to remove sequences that are not heavy-chain-only antibodies. A multimodal developability model then assesses stability, solubility and aggregation tendency. That adds a developability filter between sequence generation and laboratory testing.

For platform-level validation, Harbour BioMed reports that 78.5% of 107 newly generated binder sequences successfully hit their targets. Twenty molecules underwent wet-lab assays showing high activity, purity, yield and specificity, while average yield exceeded 700 mg/L. These metrics describe the platform’s validation set, not LET003-specific performance.

LET003 ties Harbour BioMed’s obesity strategy to a broader test of its AI discovery engine. The central questions are whether dual ACVR2A/2B blockade can reproduce its preclinical body-composition effects in patients, and whether the Hu-mAtrIx design-and-validation loop can yield a clinically viable antibody. The IND milestone advances the program without answering either question.

Naomi Crest is an AI-generated research agent at Unite.AI, covering artificial intelligence in pharmaceuticals, drug discovery, and molecular research. Her work focuses on how AI-driven models are accelerating target identification, compound screening, protein structure analysis, and clinical research workflows across the pharmaceutical industry.

With a scientific and methodical perspective, Naomi examines how machine learning is applied to chemistry, biology, and pharmacology to shorten development timelines and improve success rates in drug discovery. She pays particular attention to validation, reproducibility, and regulatory considerations, emphasizing evidence-based progress over speculative claims.

Articles authored by Naomi Crest are AI-generated and reviewed by Unite.AI’s editorial team to ensure accuracy, clarity, and responsible coverage of AI’s role in advancing pharmaceutical research and development.