Antibody Engineering and Developability: Building a Molecule That Survives Manufacturing

An antibody can bind its target well and still be a poor medicine. It may clump together, turn thick at high concentration, clear too fast, or degrade on the shelf. Antibody and protein engineering is about building a molecule that works in the body and in the plant. Developability assessment is how teams find those problems while they are still cheap to fix.

The Antibody Landscape Today

The Antibody Society counts more than 200 marketed antibody therapeutics. In 2024, 21 antibodies received a first approval in at least one country or region, and 178 were in the late-stage pipeline (The Antibody Society, 2025). The 2024 class included several bispecifics, such as tarlatamab and zanidatamab, and antibody–drug conjugates such as sacituzumab tirumotecan.

That growth brings competition. Sponsors now need molecules that stand out on dosing, route and manufacturability, not only on target.

Formats and Engineering Choices

Most modern therapeutic antibodies are humanized or fully human IgGs, which lowers the immunogenicity seen with early murine antibodies. Beyond the standard IgG, the main formats are:

• Bispecifics. Two binding arms engage two targets, or bridge immune cells to tumour cells. They add chain-pairing and assembly challenges.
• Antibody–drug conjugates (ADCs). A cytotoxic payload is attached through a linker. Conjugation adds hydrophobicity and heterogeneity that have to be controlled.
• Fc-engineered antibodies. Changes to the Fc region tune half-life or effector functions such as ADCC and CDC, either boosting them for oncology or silencing them where they add risk.
• Fragments and fusions. Fab, scFv, single-domain and Fc-fusion proteins trade half-life for size, tissue penetration or new functions.

Half-Life Extension

Fc changes that raise binding to the neonatal Fc receptor (FcRn) keep antibodies in circulation longer. Nirsevimab (Beyfortus) carries the M252Y/S254T/T256E (YTE) substitution. Sanofi states that YTE extends the anti-RSV antibody half-life from 21–28 days to 87–117 days, supporting a single dose that protects infants for about five months (Sanofi, 2024).

Ravulizumab (Ultomiris) carries the M428L/N434S (LS) substitutions (Antibody Society database). It is dosed every 8 weeks, against every 2 weeks for its predecessor eculizumab (Center for Biosimilars, 2019). That is a real benefit to patients and a strong lifecycle argument.

“An antibody that cannot be made at scale, or stored for two years, is a research reagent, not a product.”

Discovery Platforms and Optimization

Three platform families dominate antibody discovery:

• Hybridoma. The classic route: immunize mice, fuse B cells with myeloma cells, and screen clones. Murine antibodies then need humanization.
• Display technologies. Phage and yeast display screen very large libraries in vitro, with tight control over selection conditions and the option to add developability filters.
• Transgenic animals. Mice engineered with human antibody genes produce fully human antibodies through a natural immune response, including in vivo affinity maturation.

Lead antibodies are then refined. Affinity maturation uses targeted or random mutagenesis of the complementarity-determining regions (CDRs) to improve binding. De-immunization uses in silico and in vitro T cell epitope tools to remove sequences likely to trigger anti-drug antibodies. Each change should be rechecked for developability, because a mutation that lifts affinity can also raise hydrophobicity or create a new chemical liability.

What Developability Means

Developability is the likelihood that an antibody can be manufactured, formulated, stored and dosed without major problems. The main risk areas are:

Risk Area	What Can Go Wrong	Common Assays
Aggregation	Immunogenicity, lost potency	SEC, DLS
Thermal stability	Unfolding, short shelf life	DSF, DSC
Self-interaction	High viscosity at dose	AC-SINS
Hydrophobicity	Aggregation, fast clearance	HIC
Polyreactivity	Fast clearance, off-target	BVP ELISA, PSR
Expression	Costly, low-yield supply	Titre measurement

Chemical liabilities deserve specific mention. Certain sequence motifs are prone to degradation: asparagine-glycine (NG) sites can deamidate, aspartate-glycine (DG) sites can isomerize, and exposed methionine and tryptophan residues can oxidize. When these sit in or near the CDRs, they can erode binding over a product’s shelf life. Engineers screen for them and, where possible, remove them early.

Viscosity is the other quiet killer. Many antibodies are dosed subcutaneously at high concentration, and a thick solution is hard to inject. One industry rule of thumb limits viscosity to about 20 cP for a 1-mL prefilled syringe, lower still for patients with limited dexterity (Pharmaceutical Technology, 2018). Self-interaction, measured by AC-SINS, is an early warning sign.

Assays and In Silico Screening

A modern developability panel combines wet-lab and computational methods. Jain and colleagues ran 137 clinical-stage antibodies through 12 biophysical assays covering self-interaction, polyspecificity, hydrophobicity, thermal stability, expression and aggregation. Approved antibodies had the fewest red flags, and a clean, low-risk cluster held 34 of the 48 approved therapeutics (Jain et al., PNAS, 2017).

• SEC measures aggregate and monomer content.
• DSF and DSC report thermal unfolding, a proxy for stability.
• AC-SINS flags self-association that drives viscosity (Zhou et al., Pharmaceutical Research, 2021).
• HIC ranks surface hydrophobicity.
• BVP ELISA and polyspecificity reagents detect nonspecific binding linked to fast clearance.

In silico tools screen before any protein exists. The Therapeutic Antibody Profiler (TAP) flags five computational metrics derived from a large set of clinical-stage antibodies: total CDR length, surface hydrophobicity, positive charge in the CDRs, negative charge in the CDRs, and asymmetry between heavy- and light-chain surface charge (Raybould et al., PNAS, 2019). Machine learning models increasingly predict chemical liabilities and hydrophobicity from sequence alone, letting teams rank candidates before the bench work begins.

Why Early Developability Saves Money Later

The logic is simple. A liability found at candidate selection costs a sequence change. The same liability found during clinical manufacturing can mean a failed batch, a reformulation, a delayed filing, or a product that cannot be delivered the way the commercial plan assumed. Viscosity that forces an intravenous infusion instead of a subcutaneous injection can reshape the entire value proposition.

What This Means for Antibody Developers

For discovery teams, the lesson is to screen for developability alongside potency from the first lead panel, not after a single favourite has been chosen. In silico filters are cheap, so there is little reason to wait. For platform companies, a documented developability package is a licensing asset that de-risks the molecule for a partner.

For leaders and investors, the quality of this assessment is a strong signal. A molecule with clean biophysical data, no obvious sequence liabilities, a manufacturable titre and a formulation path is far more valuable than one with a slightly higher affinity and unknown developability.

Terraforme Biosciences helps antibody developers plan developability strategy, select discovery and testing partners, interpret biophysical data, and prepare technical packages for investors and licensing partners. Learn more about our strategy and business development services.

Conclusion

Engineering a therapeutic antibody is a balancing act between binding, biology and the realities of manufacturing and dosing. The formats and Fc tools now available give developers real room to differentiate a molecule. Developability assessment is what keeps that creativity grounded, so the antibody that works at the bench still works in a vial and in a patient. The time to find the problems is now, not in Phase III.