Peptides that activate the GLP-1 receptor have attracted enormous scientific attention because of the way they influence appetite signalling, blood-sugar regulation, and metabolic function. The catch is that most of these molecules are fragile. Stomach acid and digestive enzymes break them down quickly, which is why clinical formulations are almost always given by injection. A 2026 study published in ACS Applied Materials and Interfaces set out to ask whether a cleverly engineered mouth film could change that picture.
The research team built what they describe as a buccal delivery platform, meaning a system designed to move a peptide through the lining of the inner cheek and into the bloodstream from there, bypassing the digestive tract entirely. Their approach combined three ideas: wrapping the peptide in tiny nano-scale particles, optimising those particles with statistical design methods, and then locking the whole assembly inside a two-layer dissolvable film. The results offer an early but detailed proof of concept that peptide delivery through cheek tissue is physically achievable.
The problem with peptide delivery
Peptides are chains of amino acids, and most are too large and too water-loving to slip easily through biological membranes. The inner cheek, called the buccal mucosa, is a relatively thin tissue compared with the gut wall, which makes it an attractive target for researchers looking for needle-free routes. Even so, the buccal lining has its own defences: a mucus layer, tightly connected cells, and fat-rich membranes that slow large molecules down.
GLP-1 receptor agonist peptides face an extra challenge. They are not only large but also physically unstable outside of carefully controlled conditions. Leave them in solution at room temperature and their three-dimensional shape begins to fall apart, which matters because shape determines function. Any delivery system has to protect the peptide while simultaneously helping it cross tissue it would normally be blocked by. That is a demanding two-part requirement.
Building the nanocomplexes
The researchers chose a natural polymer called chitosan oligosaccharide, or COS, as their carrier material. COS carries a slight positive electrical charge, and the GLP-1 peptide analogue used in the study carries a negative charge at physiological conditions. When the two are mixed under controlled conditions, they attract each other and spontaneously assemble into tiny spherical particles roughly 100 nanometres across. For context, a single human hair is about 70,000 nanometres wide, so these particles are extraordinarily small.
The team used a statistical technique called design of experiments to systematically vary the mixing conditions and find the combination that produced the most uniform, stable particles. The resulting nanocomplexes were described as monodisperse, meaning the particles were all roughly the same size rather than a scattered mixture of large and small, and they carried a surface charge of around plus 20 millivolts. A positive surface charge is generally considered favourable for sticking to and interacting with mucus membranes, which tend to carry a negative charge.
To make the nanocomplexes storable, the team freeze-dried them through a process called lyophilisation. After one month stored at 4 degrees Celsius, the freeze-dried powder could be redispersed in liquid and the particles reformed with their original size, charge, and, crucially, the same secondary structure of the peptide inside. That last point matters because a peptide that has lost its shape is unlikely to do its job.
How the nanocomplexes interact with tissue
Before testing permeation, the researchers examined what happens when the nanocomplexes meet cells. They used a well-established buccal cell line called TR-146 and found minimal cytotoxicity at a concentration of 1 milligram per millilitre over a three-hour window, suggesting the particles did not meaningfully damage cells at the concentrations tested.
More revealing were the mechanistic studies. The researchers found evidence that the nanocomplexes are taken up by cells through a process called dynamin-dependent endocytosis, one of the cell's standard methods for importing material from outside. Alongside this, they detected what they describe as membrane lipid remodelling and a redistribution of tight junction and desmosomal proteins. Tight junctions are the molecular clamps that hold adjacent epithelial cells tightly together; desmosomes serve a similar structural role. When these proteins redistribute, the seal between cells is temporarily loosened, creating a pathway for material to pass through. The researchers frame this not as damage but as a regulated, reversible modulation of barrier properties.
The bilayer film design
With the nanocomplexes characterised, the team moved on to embedding them in a film designed to sit against the inside of the cheek. The film had two distinct layers, each with a specific job. The inner layer, which faces the cheek tissue, was made from pullulan and sodium carboxymethyl cellulose, two polymers chosen for their ability to stick to mucus. The outer layer, which faces away from the tissue, was made from a polymer called Eudragit RLPO. This backing layer is designed to be water-resistant, so that saliva does not dilute or wash away the peptide before it has a chance to permeate.
This two-layer architecture is intentional. A single-layer film would release the peptide in all directions, wasting a large fraction of the dose into the mouth cavity. The backing layer concentrates release toward the tissue, while the mucoadhesive layer holds the film in place long enough for permeation to occur. The researchers describe this as a directional delivery strategy.
Permeation results across porcine tissue
The researchers measured how much peptide crossed porcine buccal mucosa, a standard laboratory model for human cheek tissue, over three hours. The raw GLP-1 peptide dissolved in plain solution showed essentially no detectable flux across the tissue, which is the baseline expectation and confirms how poorly peptides cross this barrier on their own.
Nanocomplexes delivered in suspension, without being embedded in a film, achieved roughly 13 percent permeation at three hours, with a permeability coefficient of approximately 3.8 times ten to the power of negative six centimetres per second. When the nanocomplexes were incorporated into the bilayer film, permeation was lower at around 3.4 percent with a permeability coefficient of approximately 1.2 times ten to the power of negative six centimetres per second, but still substantially above zero, which represents a meaningful improvement over the plain peptide solution.
The authors note that the lower permeation from the film compared to the suspension is likely related to the additional diffusion step the peptide must take through the film matrix before reaching the tissue. They frame this as an engineering challenge for future optimisation rather than a fundamental limitation, and point out that the film format offers practical advantages in handling, stability, and dosing precision that a suspension cannot match.
Significance for peptide research
The study is an in-vitro and ex-vivo proof of concept, meaning the work was done in cells and excised tissue rather than in living animals or humans. That is an important caveat. Results from porcine buccal tissue and cell lines do not automatically translate into the same performance in a human mouth, and there are further engineering hurdles around scale, shelf life, and dose uniformity before any system like this could move toward clinical investigation.
Even so, the study contributes several specific advances to the scientific literature. It demonstrates that a GLP-1 class peptide can be formed into stable nanocomplexes using a natural polymer, that those complexes can survive freeze-drying and reconstitution with intact peptide structure, that they interact with buccal epithelium through identifiable and apparently reversible mechanisms, and that they can be incorporated into a mucoadhesive bilayer film that retains measurable permeation activity. Each of those steps addresses a real obstacle that has slowed progress in needle-free peptide delivery. Early data like this helps the broader research community understand which approaches are worth pursuing further and which formulation variables matter most.
