Self-Assembling Peptide Hydrogels: From a Yeast Protein to a Surgical Gel

Biomaterials

Self-assembling peptide hydrogels began as an accident: a sixteen-residue repeat spotted inside a yeast DNA-binding protein around 1990 turned out, when dropped into salt water, to knit itself into a membrane that shrugged off heat, acid, detergent and proteases. Three decades later a descendant of that sequence is squeezed through endoscopes to stop bleeding in the gut, and the same design rules underpin scaffolds for growing cells in three dimensions. This guide explains how a peptide of 16 amino acids becomes a gel that is more than 99 percent water, what triggers the change, which products have cleared regulators, and where the evidence stops.

What are self-assembling peptide hydrogels?

A self-assembling peptide hydrogel is a water-swollen network of nanofibers built spontaneously from short synthetic peptides, with no polymer backbone, no chemical crosslinker and no animal-derived material. The classic examples are “ionic self-complementary” peptides: sequences in which positively charged, negatively charged and hydrophobic residues alternate in a strict pattern. RADA16, the best-known, is simply arginine-alanine-aspartate-alanine repeated four times, capped with an acetyl group at one end and an amide at the other. FDA documents describe it as acetyl-(arginyl-alanyl-aspartyl-alanyl)4-amide.

The alternation is the whole trick. In an extended beta-strand, consecutive side chains point to opposite sides of the backbone, so an alternating sequence produces one face lined with charged groups and one face lined with alanines. Strands stack into beta-sheets, sheets pair with their hydrophobic faces buried, and the result is a nanofiber with a charged, water-loving surface. The original 1993 paper reported interwoven filaments roughly 10 to 20 nanometres across by electron microscopy; the research-grade product literature gives an average pore size of 50 to 200 nanometres for the finished network. At the standard 1 percent working concentration (10 mg/mL) the gel is 99.0 to 99.5 percent water.

The yeast protein that started it

The first peptide of this family was not designed. Shuguang Zhang, then working on a yeast protein called zuotin that binds left-handed Z-DNA, noticed an unusual repetitive segment in its sequence: AEAEAKAKAEAEAKAK, sixteen residues alternating glutamate, lysine and alanine. Synthesised on its own and named EAK16, it showed a beta-sheet signature in water and, when salt was added, assembled spontaneously into a macroscopic membrane. That membrane, reported in PNAS in 1993, survived heat, acid, alkali, guanidine hydrochloride, SDS, urea and a range of proteases. The peptide aggregation and fibrillation guide covers the same beta-sheet physics as a problem to be avoided; here it is the point.

RADA16 came later as a deliberate variant, swapping glutamate and lysine for aspartate and arginine so that the sequence carried the RAD motif reminiscent of the RGD cell-adhesion sequence in fibronectin, and using alanine rather than glycine to stiffen the chain. Zhang’s group also showed in 2005 that RADA16 fibers broken apart by sonication reassemble on their own, which is the property that makes the gels workable after they have been sheared through a needle.

How self-assembling peptide hydrogels are triggered

The stock material is a clear, acidic, salt-free solution (pH about 2 to 2.5 for the research product). In that state the fibers already exist but repel one another, so the liquid is viscous but flows. Add ions, whether from culture medium, buffer, tissue fluid or blood, and the charges on the fiber surfaces are screened; the fibers entangle and the solution becomes a transparent gel without changing volume. The research-grade product literature warns that gelation starts at salt concentrations around 1 millimolar, which is why the material must be kept away from any salt-containing buffer until gelation is wanted. Sequence, concentration, pH, temperature and ionic composition all shift the outcome.

For the surgical product the trigger is blood itself. The FDA clearance summary describes exposure to physiological fluids causing the peptide solution to “quickly form a transparent gel without expansion in volume” that covers the bleeding site as a physical barrier. The device is delivered through a standard endoscopic catheter, and the agency’s bench testing included injection force and viscosity in simulated body fluid.

The product that made it to the clinic

PuraStat, made by 3-D Matrix, is a 2.5 percent RADA16 solution in a prefilled syringe. It is a CE-marked class III device in Europe for surgical haemostasis and for reducing delayed bleeding after colonic endoscopic submucosal dissection, and it is sold across Europe, Australia, Canada and several other markets. In the United States the path was two 510(k) clearances. PuraStat-GI (K210098) was cleared on 25 June 2021 for mild and moderate bleeding after endoscopic resection, using Hemospray as the predicate. PuraStat (K222481) followed on 28 February 2023 with an expanded indication that added primary non-variceal gastrointestinal bleeding, while stating explicitly that it is not indicated for spurting arterial (Forrest 1a) bleeding.

The research counterpart, PuraMatrix, is the same peptide at 1 percent, sold for laboratory use only. It has been used to grow hepatocyte progenitors, neurons and endothelial tubes in three dimensions, and its main selling point over the traditional mouse-tumour-derived matrices is that it is fully synthetic: no animal material, no batch of unknown growth factors, and degradation products that are simply amino acids.

What the human evidence shows

The clinical record is real but narrow. Across the studies tabulated by FDA reviewers, 223 patients were treated with no adverse events attributed to the device. The largest single series, Branchi and colleagues’ prospective multicentre study across 15 German endoscopy departments, treated 111 patients between 2017 and 2018 and reported initial haemostatic success in 94 percent when the gel was used as primary therapy (74 of 79) and 75 percent when used after other methods had been tried (24 of 32), with an overall 30-day rebleeding rate of 16 percent. Subramaniam’s 2019 series of 100 endoscopic resections reported 75 percent haemostasis and 3 percent delayed bleeding.

What those numbers cannot tell you is how the material compares with clips, coagulation or powder haemostats, because none of the tabulated studies were randomised against them. The indication is deliberately limited to oozing and mild-to-moderate bleeding, and the label excludes arterial spurting. Read the evidence as “safe and usable as a barrier in this setting”, not as “better than the alternatives”.

Beyond RADA16: the other design families

Ionic self-complementary peptides are one of several ways to make a peptide gel. Peptide amphiphiles from Samuel Stupp’s laboratory attach a hydrophobic alkyl tail to a short peptide so that the molecules pack into cylindrical nanofibers with the peptide on the outside; in a 2004 Science paper, fibers presenting the laminin-derived IKVAV sequence at high density pushed neural progenitor cells toward neurons and away from astrocytes. The MAX1 family from Schneider and Pochan uses a designed beta-hairpin, VKVKVKVKV-DPP-TKVKVKVKV, whose folding can be switched by pH, salt or temperature and whose gels shear-thin through a syringe and recover immediately, which is the property that makes them attractive for cell delivery. Fmoc-diphenylalanine shows that even a protected dipeptide can gel, through a stacked arrangement in which the aromatic Fmoc groups lock adjacent beta-sheets together. D-amino-acid versions of EAK16 resist the proteases that would digest the natural form; the D-amino acids in peptides guide explains why.

Research uses: scaffolds, depots and a hamster’s eyesight

Three applications recur in the literature. The first is three-dimensional cell culture, where a transparent, tunable, animal-free matrix with a stiffness overlapping that of nervous tissue is a useful alternative to tumour-derived gels. The second is drug delivery: Koutsopoulos and Zhang showed in 2009 that functional proteins can be released in a controlled way from a RADA16 gel, and in 2012 that layered gels can sustain antibody release. The third, and the most striking, is tissue repair. In 2006 Ellis-Behnke and colleagues cut the optic tract of hamsters at the superior colliculus, injected the peptide solution into the lesion, and reported axons regrowing through the gap with functional return of vision. Kisiday’s 2002 work had earlier shown chondrocytes producing cartilage matrix inside a peptide scaffold. These are animal and in-vitro results; none has become an approved regenerative therapy.

Advantages, limits and cost

The advantages are the ones the regulators cite: the material is fully synthetic, contains no animal or plant components and no preservatives, and breaks down to amino acids that the body simply resorbs. The limitations are equally concrete. The gels are soft and easily disrupted, which is fine for a haemostatic barrier and for culturing cells but rules out load-bearing uses without reinforcement. The acidic stock has to be equilibrated quickly once cells are added. And the peptide has to be made by chemical synthesis at pharmaceutical purity; Zhang’s own retrospective notes that in the early 2000s peptide production was still too expensive and complicated even for small-scale industrial production, and that affordable GMP scale-up remains one of the key issues. The solid-phase peptide synthesis guide explains where that cost comes from.

Frequently asked questions

Is a self-assembling peptide hydrogel the same as a collagen or gelatin gel?

No. Collagen and gelatin are animal-derived proteins of hundreds of residues; these are synthetic peptides of about sixteen, and their gels form by nanofiber entanglement rather than by protein melting and setting.

Why does the gel not swell?

Gelation happens by charge screening and fiber entanglement, not by absorbing extra water, so the volume placed is the volume that gels. The FDA summary describes the transition as occurring “without expansion in volume”.

Can the same peptide be used for research and in patients?

The chemistry is the same, the products are not. PuraMatrix is sold at 1 percent for research use only; PuraStat is a 2.5 percent sterile device manufactured and cleared for a specific clinical indication. Grade, sterility, endotoxin control and regulatory status differ.

Do the fibers really reform after being broken?

For RADA16, yes: a 2005 PNAS study showed that fibers fragmented by sonication reassemble over time. That reassembly is why the material can be sheared through a catheter and still form a coherent gel at the target.

Informational only — not medical advice · 21+. Nothing here is treatment, wound-care or dosing guidance. Decisions about haemostatic devices or any medical procedure belong with a qualified healthcare professional.