Skip to content
PeptideHormone

When the mix goes solid

A peptide blend that clouds, strings, or sets into a gel has not “thickened.” It has left solution. Opposite net charge builds an insoluble complex in seconds; a sequence parked near its isoelectric point, or one that wants a β-sheet, gels over hours; the molecules that stay dissolved keep reacting — deamidation, oxidation, disulfide scrambling — and each of those edits changes what a receptor can still read. Why the brick is a missing dose, why a designed precipitate (insulin glargine) is not the same object as an accidental one, and why the clear supernatant is not proof the hormone survived.

12 min read · reviewed November 2026

The solid is a different object

People who combine peptides in one syringe have a vocabulary for the failure: the vial bricked, crashed, gelled, went stringy, threw a snowstorm. The words sound like texture. They are phase changes. A hormone that was a population of dissolved monomers has become an insoluble aggregate, and the receptor never sees an aggregate the way it sees a monomer. Whatever is left in the liquid is a depleted, still-reacting remainder.

This is not an exotic research-chemical quirk. It is ordinary peptide physical chemistry, documented for the molecules pharmacies already handle. Glucagon gels. Amylin fibrillates. Insulin glargine is deliberately insoluble at tissue pH, and is forbidden from sharing a syringe for that reason. A review of therapeutic-peptide stability walks through exactly these liabilities — pH, net charge, concentration, interfaces, and sequence — for glucagon, amylin, insulin, calcitonin, and GLP-1 (Zapadka et al., 2017 · PMID 29147559). Blending two of them, or blending either with a basic repair peptide, is how you ask that chemistry to happen on a kitchen counter.

The governing fact

Solubility and chemical identity are set by pH, partners, and time. A shared diluent does not give two peptides a shared formulation. The moment they occupy one solution they share one pH, one ionic strength, and one clock.

Charge is the first incompatibility

A peptide is a chain of acids and bases. Aspartate and glutamate are negative once deprotonated. Lysine, arginine, and the N-terminus are positive once protonated. Histidine flips across the physiologic window. Add those up and you get a net charge that is a function of pH, and a pH — the isoelectric point, pI — where the sum is zero.

Two consequences follow, and they fail on different clocks.

  • Opposite signs attract. At any pH between their isoelectric points, an acidic peptide and a basic peptide carry opposite net charge. Polycation plus polyanion is a polyelectrolyte complex. The complex can phase-separate as a cloudy suspension, a stringy coacervate, or a plug — the brick — in seconds to minutes, long before a covalent bond has broken. Each peptide may have been perfectly soluble in its own vial, because each vial was holding it on the safe side of its own charge.
  • Zero net charge stops the repulsion. Like-charged monomers repel, which is a large part of what keeps them in solution. Near its pI that repulsion collapses and solubility usually bottoms out. A diluent chosen because it dissolved peptide A can sit directly on the pI of peptide B. Self-association then does not need a partner.
  • Hydrophobic patches finish the job. Charge gets the chains close. Sticky side chains — and any sequence that can donate a strand to a β-sheet — lock them there. That second step is why some crashes redissolve if you change pH immediately and others never do: the electrostatic complex is reversible in principle; the hydrogen-bonded fibril often is not.
net charge, qualitative0+−2467.4911pH of the shared solutionacidic peptidebasic peptidepIpIopposite signs — the crash window
Schematic, not a measured titration. Between the two isoelectric points the peptides attract. At each pI, that peptide’s own repulsion is gone and its solubility is usually worst.

The window in that figure is why “both of them dissolve in bacteriostatic water” is not a compatibility test. Water is not a pH. The acidic peptide’s formulation and the basic peptide’s formulation were each a compromise between solubility, chemical stability, and the preservative. Averaging them is a third condition nobody studied.

Three medicines already ran the experiment

Glucagon is the textbook gel. In aqueous solution near neutral pH it assembles into amyloid-like fibrils and the liquid sets. That is why the older emergency kits are lyophilized and the label says to use the mixed solution immediately and discard the rest (DailyMed glucagon labeling). The newer products exist because the native sequence could not be left in water. Putting glucagon into a neutral blend and waiting is how you repeat the problem the kits were built to avoid.

Amylin is worse. It is one of the most amyloidogenic peptides the body makes; its fibrils are the islet amyloid of type 2 diabetes (Westermark et al., 2011 · PMID 21742788). The drug version, pramlintide, is formulated acidic so it stays monomeric, and the label is unambiguous about partners: pramlintide and insulin are always separate injections and must never be mixed. Premixing them changed pramlintide’s peak concentration by as much as 40 percent and shifted exposure for both drugs (SYMLIN prescribing information). The incompatibility is not a style preference. Raise the pH of an amylin-family peptide toward the insulin vial and you hand it the condition it fibrillates in.

Insulin glargine is the case that separates a designed precipitate from a mistake. Two extra arginines shift its isoelectric point so the analog is soluble in its acidic vial and insoluble at tissue pH. The precipitate is the long-acting depot. The label therefore forbids diluting it or mixing it with any other insulin or solution: another liquid changes the pH trajectory and the time-action profile the depot was qualified at (LANTUS prescribing information). An accidental brick in a blended vial has already precipitated, at an uncharacterized particle size, before tissue ever buffers it. There is no pharmacokinetic study of that solid. There is only the word “cloudy.”

Two clocks, and only one of them you can see

Physical collapse and chemical rewrite run together, and people notice only the first. Cloudiness is light scattering off objects hundreds of nanometers across. A deamidated side chain is a dalton-scale edit. The vial can look perfect while the hormone is already a different molecule.

What you can seeWhat the receptor still readssecondsOpposite charge locksCloud, strings, or a plug.Monomer concentration collapses.The receptor sees less hormone.minutes–hoursβ-sheet takes overGlucagon- and amylin-type sequences gel.A nucleus recruits the rest of the vial.The solid is not a ligand.hours–daysThe quiet rewriteisoAsp, scrambled disulfides, oxidation.The liquid stays clear.The message does not.
Physical failure scatters light. Chemical failure does not. Both change the dose the receptor can respond to. Times are order-of-magnitude, and they shorten as concentration, temperature, and a bad pH go up.

Read the figure as two failure modes with different biologic meanings.

On the fast clock, opposite charge or a near-pI condition removes monomer from the liquid. What you draw is some unknown fraction of the intended concentration, plus particles. The receptor does not bind a fibril productively. Dose becomes a guess, biased low, and biased differently every time the aggregation has progressed a little further. For glucagon-type and amylin-type sequences the fibril, once nucleated, templates more fibril — a little gel makes the rest of the vial catch up. That is why “it was only slightly cloudy yesterday” is not a stable state.

On the slow clock, the chains that remain dissolved keep reacting. Three edits matter most for what a receptor can still do, and all three accelerate when a blend forces the wrong pH.

Deamidation rewrites the backbone

Asparagine, especially in Asn–Gly, deamidates through a succinimide ring. The ring reopens two ways: ordinary aspartate, or isoaspartate, in which the backbone itself runs through the side-chain carbon. The peptide gains one extra methylene in the main chain and a new negative charge, at a site that used to be neutral. Half-times for unstructured Asn–Gly can be on the order of a day under warm, neutral-to-alkaline conditions and much longer in the cold or in a folded protein; other sequences take months or years. Sequence and pH are the whole clock (Robinson & Robinson, 2001 · PMID 11172027).

A receptor does not bind “a peptide of about the right weight.” It reads a surface: a precise spacing of charges and backbone hydrogen bonds. Isoaspartate changes both. Depending on whether that asparagine sat in a contact, the edited chain can lose affinity, become a weaker partial agonist, or — if the contact was not important — keep most of its activity while presenting a new epitope. You cannot see which of those you have. Insulin’s own pharmaceutical history is this reaction: deamidation at specific A- and B-chain asparagines is a primary solution-state pathway, pH-dependent, and one reason insulin formulations are buffered where they are rather than where a second peptide might prefer (Brange & Langkjær, 1993 · PMID 8257962).

Oxidation and scrambled disulfides change the message

Methionine, tryptophan, histidine, and cysteine oxidize. The classic product, methionine sulfoxide, is bulkier and more polar than methionine. If that methionine was a receptor contact or a structural staple, potency drops; if it was not, the molecule can keep working and still fail a purity specification. Trace metals, light, and dissolved oxygen drive it. A second peptide’s vial is a fine way to import all three — another stopper, another excipient, another history of air.

Disulfides fail differently, and pH is the switch. Thiol–disulfide exchange is carried by the thiolate anion, so the rate climbs sharply as pH approaches the cysteine pKa, near 8, and is effectively frozen under the acidic conditions many peptides are deliberately stored at (Nagy, 2013 · PMID 23121320). Mix two cysteine-containing peptides, or a free-thiol peptide with a disulfide peptide, and push the blend toward neutral or alkaline pH, and the bridges reshuffle. Insulin is two chains and three disulfides; the wrong pairing is not insulin. Oxytocin and vasopressin are single disulfides whose ring is the pharmacophore. A scrambled ring is a different instruction, or no instruction, delivered at the same nominal milligram.

Hydrolysis is the slowest of the common cuts and the one that most resembles the design trick this site keeps meeting: a hormone broken into clauses. Asp–Pro bonds and a few other sequences cleave, and the piece that remains can still occupy part of a receptor. That fragment is not “a weaker version of the drug.” It is an untested molecule that may activate, partially activate, or sit in the site and block the intact peptide — an accidental antagonist made out of the agonist. The keep-one-clause move is hard when a chemist does it on purpose and checks the clause in a trial. Done by a bad pH in a vial, nobody checks.

Intact monomer

The surface the receptor was built for. Full agonist: the second messenger fires.

isoAsp or sulfoxide

Same nominal mass, wrong spacing or a bulkier side chain. Weaker binding, partial agonism, or a new epitope.

Scramble or fragment

Wrong disulfide, or a clipped clause. Can be silent — or can occupy the site and block the intact hormone.

Aggregate

Not a ligand. Missing dose, undesigned depot, and a repetitive particle the immune system is better at seeing.

Four species one blend can hand the body. Only the first is the molecule the label, the protocol, or the receptor pharmacology was about.

What the body does with the wreckage

Once the liquid is injected, tissue buffer pulls the pH to about 7.4 within moments. Protonation state in the syringe is not the binding event. The binding event is whatever species already exist: remaining monomer, chemical edits, and particles. Four outcomes follow, and they are not mutually exclusive.

  • A missing dose. Precipitated peptide is not in the volume you measured. The pharmacologic effect, if the monomer was the active species, is smaller than the label arithmetic by an unknown amount. Chasing a weak effect with a larger draw the next day compounds the error, because the vial has also had another day of both clocks.
  • An undesigned depot. Some of the solid may redissolve slowly in tissue, the way glargine is meant to. Unlike glargine, the particle size, the local pH path, and the release rate were never measured. The time course stops being the one in any protocol and becomes a property of that particular crash.
  • A better immune stimulus. A repetitive aggregate is easier for the immune system to see than a soluble monomer. Regulators treat aggregates and subvisible particles as immunogenicity risk factors for therapeutic proteins for that reason (FDA immunogenicity guidance). Neutralizing antibodies do not only cancel the injected molecule. For a hormone, they can cross-react with the endogenous copy — the failure mode that made protein aggregates a clinical, not just a pharmaceutical, problem. Most research peptides have no such surveillance at all. Absence of a case report is not a clean bill.
  • A local irritant. A gel or a crystalline crash injected into subcutaneous tissue is a foreign particle load. Inflammation at the site is a biologic response to the physical form, independent of whether any monomer still binds its receptor.

None of this is fixed by the solution looking acceptable an hour later. Fibrils and chemical edits do not reliably reverse because a vial sat on the counter and “cleared a bit.” Warming, vigorous shaking, forcing the plug through a needle, or dripping in acid or extra water changes shear, pH, and concentration again. It does not assay identity. The reconstitution rule still holds: if the product was supposed to be clear and it is not, you do not use it (the vial is not a recipe).

What the brick is telling you

The crash is the formulation’s assay, and it already reported. The molecules that left solution are not the dose. The molecules that stayed behind have been held at a pH and next to a partner their chemistry was not qualified for. Discard it. Do not titrate a rescue.

Why “it worked for them” survives the gel

A blend can brick halfway, leave a fraction of intact monomer, and still do something. That something gets reported as evidence that the combination is fine. It is evidence that some monomer survived long enough to bind, which is the least surprising result in pharmacology. It says nothing about the particles, the isoaspartate, the scrambled disulfides, or the dose the next vial will actually deliver after a different dwell time.

Pharmaceutical combinations that do share a container — a few premixed insulins, a co-formulated incretin — earned that container with compatibility studies at a defined pH, with defined excipients, for a defined hold time. The studies are the product. Two powders and a bottle of bacteriostatic water are not a study. They are the conditions under which charge, β-sheet, and the slow chemical clock are free to write a new molecule and then hide most of it in a solid the syringe cannot measure.

Common questions

What is peptide “bricking” or gelling?
Both words describe the same physical failure: peptide that was supposed to stay dissolved has become an insoluble aggregate. A brick is a plug or a cloudy crash, often from opposite charges locking two peptides into a complex, or from a peptide driven to its isoelectric point. A gel is usually a network of β-sheet fibrils, the failure glucagon and amylin are famous for. Either way the solid is no longer a measured liquid dose.
Why do two peptides that each dissolve on their own crash when combined?
Each peptide’s net charge depends on the pH it is sitting in. A basic sequence and an acidic sequence at the same pH carry opposite signs and attract. That polyelectrolyte complex can phase-separate in seconds. Separately, a shared pH that is fine for one sequence can sit on the isoelectric point of the other, where repulsion disappears and the peptide self-associates.
If I can still see liquid, did the molecule survive?
Clarity only says nothing large enough to scatter light has formed. Deamidation, oxidation, and disulfide scrambling rewrite a peptide while it stays invisible. The receptor reads the edited chain, not the fact that the vial looked clear.
Can a bricked vial be rescued by warming, shaking, filtering, or adding acid?
No. Those maneuvers do not restore the original monomers, the original concentration, or the original chemical identity, and they add shear, new pH, and new surfaces. A preparation that has left solution, or whose diluent and partners were never on its label, is a failed formulation. The labeled instruction is to discard it, not to repair it.
Is an accidental precipitate the same thing as a long-acting insulin depot?
No. Insulin glargine is engineered so that a pH shift after injection precipitates a depot with a studied release profile, and its label forbids mixing precisely because another liquid would change that profile. An accidental brick precipitates before injection, at an unknown particle size, with no measured release and a higher chance of being seen by the immune system as a repetitive particle.

Educational reference on mechanism, summarized from public scientific literature and product labeling and simplified in places. Not individualized medical advice or a prescription. Nothing here is a mixing procedure, a rescue procedure, or a dosing instruction. Follow the approved product label and the dispensing pharmacist. Specific products are named because their labels already document the incompatibility; verify any claim against the linked sources.