Trehalose Crystallization Undermines Protein Stability in Lyophilized Drug Formulations

Trehalose, one of the most widely used protective agents in freeze-dried protein medicines, can quietly betray the very proteins it is meant to shield. A new open-access study published in AAPS Open on 12 August 2026 shows that when trehalose crystallizes during lyophilization, it can leave a monoclonal antibody and highly monomeric albumin formulations measurably less stable during accelerated storage, even when the sugar makes up as much as 96 percent of the formulation solids. The work, led by Rachana Sapkota and Eric J. Munson and colleagues, combined powder X-ray diffraction, size-exclusion chromatography, and solid-state nuclear magnetic resonance spectroscopy to track exactly what happens inside a vial when the protective glass breaks down.

Lyophilization, the freeze-drying process used for protein drugs that cannot survive in liquid solution, removes water under low pressure and temperature in three stages: freezing, primary drying by sublimation, and secondary drying to desorb remaining bound water. An optional annealing step briefly raises the temperature during freezing to encourage the growth of ice crystals and of crystallizing excipients such as mannitol or glycine. Disaccharides like sucrose and trehalose are added as lyoprotectants, and two mechanisms explain how they work. The water replacement theory holds that the sugars form hydrogen bonds with protein residues, replacing the water interactions lost during drying. The vitrification theory holds that they form a rigid glassy matrix that immobilizes the protein, restricting the molecular mobility that drives denaturation, unfolding, and aggregation. For either mechanism to succeed, the sugar and protein must be intimately mixed and the matrix must stay glassy.

Crystallization is a known failure mode. Earlier work by Sundaramurthi and Suryanarayanan demonstrated that trehalose can crystallize during lyophilization while still yielding a final product that appears amorphous: crystalline peaks of trehalose dihydrate appeared by powder X-ray diffraction during annealing and primary drying, then vanished after secondary drying, likely because water desorbed from the crystal structure. That earlier study, however, did not explore what transient crystallization means for protein stability. The new study set out to answer that question directly, asking whether crystalline trehalose that survives into the dried cake compromises the protein it left behind.

The researchers built three protein/trehalose systems with a total solids content of 50 mg/mL and weight ratios ranging from 30 percent protein with 70 percent trehalose down to 4 percent protein with 96 percent trehalose. The first, called BSA60, contained bovine serum albumin of which roughly 60 percent was monomer and 40 percent aggregates, a consequence of pandemic-era supply-chain issues. The second, BSA95, contained about 95 percent monomer and 5 percent aggregates, prepared once higher-quality material became available, specifically to test whether the pre-existing aggregates were stabilizing the mixture. The third used a monoclonal antibody supplied by Merck with less than 1 percent aggregation, chosen for its relevance to therapeutic proteins. All formulations were frozen at minus 45 degrees Celsius, annealed at minus 15 degrees Celsius for 0, 2, 4, 12, 24, or 72 hours, refrozen, primary dried at minus 27 degrees Celsius, and secondary dried at 30 degrees Celsius. Stability samples were then stored at an accelerated 50 degrees Celsius for up to 12 months.

The diffraction data drew a sharp line. In BSA60 formulations, no crystalline peaks appeared without annealing or with annealing up to 12 hours, but most samples annealed for 24 hours showed peaks characteristic of trehalose dihydrate, with intensity rising as trehalose content increased. The exception was instructive: the 30BSA60_70trehalose formulation, with the least trehalose, stayed predominantly amorphous even after 24 hours of annealing. In contrast, BSA95 and monoclonal antibody formulations showed no crystallinity at 24 hours and required 72 hours of annealing before trehalose dihydrate peaks emerged, again growing with trehalose content. The monoclonal antibody formulations crystallized more extensively than either albumin system. The authors suggest that the high-molecular-weight aggregates in BSA60 may reduce trehalose-protein interactions, letting crystallization proceed more readily, while predominantly monomeric proteins resist it.

Solid-state NMR told the mechanistic story. Measuring the proton spin-lattice relaxation time in the laboratory frame, 1H T1, the team probed molecular mobility: shorter times indicate a more mobile matrix, and in a homogeneous mixture the protein and sugar converge on similar values because spin diffusion equalizes their magnetization. In non-annealed BSA60 samples, protein and trehalose relaxation times were nearly identical, confirming molecular-level mixing. In the 24-hour annealed samples the values diverged, and in the crystallizing monoclonal antibody formulations the divergence was striking. In the 16mAb_84trehalose formulation, trehalose 1H T1 rose from 4.61 seconds at 24 hours of annealing to 5.26 seconds at 72 hours, while the antibody’s relaxation time fell from 4.28 seconds to 3.30 seconds. The most likely explanation, the authors write, is that as trehalose crystallizes, less amorphous sugar remains to stabilize the protein, effectively raising the protein-to-trehalose ratio in the glass that is left.

The stability results split cleanly along the same line. BSA60 formulations, despite containing crystalline trehalose after 24-hour annealing, showed no additional aggregation on storage; monomer and aggregate content remained largely unchanged over 12 months at 50 degrees Celsius, with monomer loss in the 30BSA60_70trehalose formulation of only a few percentage points, from 61.60 percent to 56.60 percent in the worst case. The authors propose that enough amorphous trehalose remained to protect the monomeric BSA, or that the abundant aggregates themselves helped stabilize the system. The predominantly monomeric systems behaved very differently. All BSA95 formulations annealed for 72 hours lost monomer over three months at 50 degrees Celsius, with the greatest initial loss, though still below 10 percent, in the 4BSA95_96trehalose sample. The monoclonal antibody formulations annealed for 72 hours showed much more significant aggregation after just one month, while the 24-hour annealed, amorphous formulations stayed stable, with aggregate increases below 1 percent for most compositions over six months.

That destabilization appeared even at a trehalose-to-protein weight ratio of 24:1 is the study’s most consequential finding. A high excipient ratio alone does not guarantee stability if inhomogeneity develops. The authors weigh two mechanisms: crystal surfaces may directly interact with the antibody, creating interfaces for unfolding and aggregation, or crystallizing trehalose may abandon part of the protein, leaving it unprotected. Because the amorphous trehalose peaks in the 72-hour spectra decreased only slightly compared with the 24-hour samples, most of the sugar remained amorphous and, in principle, sufficient, which points toward the crystal surfaces as the likelier culprit, though the authors note both possibilities remain open.

The team also highlights the predictive value of the relaxation measurements. Samples whose protein and trehalose shared similar 1H T1 values remained stable over six months, while those with larger differences showed more aggregation and monomer loss. Because solid-state NMR detects subtle changes in mobility and homogeneity before instability becomes detectable on long-term storage, the authors argue it could serve as an early screening tool for formulation design and lyophilization process optimization. They caution, appropriately, that the NMR spectra were not fully quantitative for crystalline trehalose, since a short repetition delay was chosen to maximize the protein signal, and that PXRD of the final cake cannot reveal trehalose that crystallized transiently and later reverted to the amorphous form.

The overall message for formulation scientists is twofold. First, trehalose crystallization is genuinely hard to provoke: it generally requires prolonged annealing, 24 hours or more, and a high trehalose-to-protein ratio, so it should be rare under typical lyophilization conditions. Second, when it does occur and persists into the dried product, the resulting trehalose-rich and protein-rich domains measurably erode protein stability at accelerated storage temperatures. For a biopharmaceutical industry that relies on freeze-dried monoclonal antibodies, the practical lesson is that annealing conditions deserve the same scrutiny as the formulation itself, and that a relaxation-time measurement early in development may flag trouble long before an stability study does.

Subject of Research: The impact of trehalose crystallization on protein aggregation and stability in lyophilized protein formulations

Article Title: Investigating the impact of trehalose crystallization on protein aggregation in lyophilized formulations

Article References: Sapkota, R., Perez Mejia, I., Tower, C. W., Huls, N. J., Renawala, H. K., & Munson, E. J. (2026). Investigating the impact of trehalose crystallization on protein aggregation in lyophilized formulations. AAPS Open, 12(1), Article 51. https://doi.org/10.1186/s41120-026-00183-1

Image Credits: AI Generated

DOI: 10.1186/s41120-026-00183-1

Keywords: trehalose, lyophilization, protein aggregation, solid-state NMR, monoclonal antibody, BSA, crystallization, lyoprotectant, PXRD, formulation stability, annealing, freeze-drying