How Are Peptides Made?

Most peptides sold for research don't come from a living cell at all. They're built by hand, one amino acid at a time, on a bead of plastic resin — then cut free, washed of the chemicals that built them, tested, and dried into powder. Only the longest chains, closer to full proteins, are usually grown inside an engineered cell instead.

Building a chain backward, one link at a time

The method behind most synthetic peptides is solid-phase peptide synthesis (SPPS), and it traces back to one chemist's idea. Robert Bruce Merrifield first published the concept in 1962 and followed with the paper that defined the field in 1963 — work the American Peptide Society's history of his career credits as the fifth most-cited paper in the 125-year history of the Journal of the American Chemical Society and a 2006 ACS Chemical Breakthrough Publication.

SPPS anchors the first amino acid — what becomes the peptide's tail end, its C-terminus — to a bead of insoluble resin. Each cycle after that couples one more amino acid onto the free end, washes away everything that didn't react, strips a temporary cap off the newest link so the next one can attach, and repeats. The chain grows toward what becomes its head end, the N-terminus, which is the reverse direction from the one a living cell's ribosome uses when it builds a protein strand.

Merrifield showed the method could scale almost immediately. With Keith Stewart and Clark Jernberg, he built an automated version of the synthesizer, and in 1969, working with Bernd Gutte, he used it to make ribonuclease A, a 124-amino-acid enzyme — the first enzyme ever built this way. The Royal Swedish Academy of Sciences awarded him the 1984 Nobel Prize in Chemistry "for his development of methodology for chemical synthesis on a solid matrix."

Two chemistries, and why one largely took over

Every amino acid added to a growing chain needs its reactive end capped until its turn to react, or the chain would branch instead of growing in one line. According to a technical overview from instrument maker Biotage, SPPS runs on two competing ways to do that capping. The original Boc method, which Merrifield used, protects the chain's growing end with a group removed by a moderately strong acid, trifluoroacetic acid (TFA), and originally cut the finished chain from the resin using hydrogen fluoride. The newer Fmoc method, developed by Atherton and Sheppard, protects the same spot with a group removed instead by a mild base, saving the acid step for the very end.

Fmoc chemistry is now the more common approach, Biotage notes, largely because it avoids repeated handling of concentrated acid and hazardous hydrogen fluoride during the build, and because its protecting-group scheme fits more easily with the on-resin modifications many research peptides now carry.

The step that leaves an unexpected label on the vial

Whichever scheme was used, the finished chain still has to come off the resin, and that final cut needs a strong acid: TFA. A 2025 peer-reviewed paper in Pharmaceuticals lays out why that one reagent shows up twice in a peptide's life — once to cleave the finished chain from its resin and strip the last protecting groups, and again as the standard ion-pairing additive in the reversed-phase HPLC step used to purify it. A peptide that goes straight from that purification into the freeze-dryer comes out as a peptide–TFA salt, not as the free peptide alone.

That matters because residual TFA adds to the dried powder's weight without being peptide, which is one reason net content is measured directly rather than assumed and can read differently from the label strength. The same paper's authors note evidence that residual TFA can affect a peptide's shape and behavior in cell-based assays, which is part of why some manufacturers exchange it for a different counter-ion, usually chloride or acetate, by washing the peptide with dilute hydrochloric acid before the final freeze-dry. There's no settled standard for how that exchange should be done: the authors' own review of the literature found protocols using HCl concentrations anywhere from 2 to 100 millimolar and one to five wash cycles, with no field-wide agreement on which is correct. In their own test, a single wash at 10 millimolar HCl brought TFA below their instrument's detection limit while the peptide's purity stayed above 96%.

Checking what actually came off the resin

Synthesis on a bead is reliable but not perfect — a small share of chains fail to couple correctly at any given step, and those small losses compound over dozens of cycles. So before a batch ships, two separate tests ask two separate questions. Reversed-phase HPLC asks how clean the sample is: peptide manufacturer GenScript's quality control page describes running the sample through a C18 column past a UV detector tuned to around 220 nanometers, where the peptide bond itself absorbs light, and the resulting purity figure is the target peptide's peak as a percentage of everything the detector saw. Mass spectrometry asks a different question: whether the molecule's mass actually matches what its sequence should weigh. GenScript gives a worked example — an ion carrying two charges that lands at 865.05 on the instrument's mass-to-charge scale corresponds to a molecule weighing roughly 1,728.10 daltons, read off that charge state. Neither test alone is conclusive, and GenScript states plainly that even a well-purified peptide can still carry small amounts of impurity after multiple purification passes.

Once identity and purity are confirmed, the purified peptide is almost always freeze-dried before shipping, the same step used across injectable drug manufacturing generally, because the dry powder resists the water-driven reactions that break a peptide down in solution.

When a cell does the work instead

Not every peptide-length molecule is built on a bead. Once a chain gets long enough, chemical synthesis becomes slow and costly — each coupling step's small failure rate compounds with every amino acid added — and biological production, the same ribosomal process a normal cell already runs, becomes the more practical route. Insulin, a 51-amino-acid, two-chain molecule, is the standard example. A review of recombinant insulin manufacturing in Bioresources and Bioprocessing describes the human insulin gene being cloned into a plasmid as part of a larger fusion protein and expressed inside engineered E. coli, which pack the protein into dense, insoluble clumps called inclusion bodies. Those have to be washed out of the cells, dissolved in a strong denaturant such as urea or guanidine hydrochloride, chemically cleaved away from their fusion partner, and refolded in a controlled buffer before enzymes trim it down to the finished two-chain hormone, which is then purified by chromatography. A second route uses yeast — Saccharomyces cerevisiae or Pichia pastoris — engineered to secrete a soluble insulin precursor directly into the growth broth, skipping the inclusion-body and refolding steps entirely. E. coli remains the industry's default, the review notes, mainly for its fast growth, inexpensive growth medium and high yield.

What's settled, and what still isn't

The chemistry of each step above is well characterized: SPPS's couple-and-deprotect cycle, the Boc/Fmoc split, HPLC's separation by stickiness, mass spectrometry's mass-to-charge measurement, and E. coli's inclusion-body pathway for insulin are all documented in the primary literature cited here. What's still unsettled is some of the finishing work. There's no field-wide standard for whether, or exactly how, a TFA salt should be exchanged for chloride or acetate before a peptide ships, which is one reason a lab report's stated net content is worth reading rather than assumed.

What this means for researchers

A peptide's synthesis route shapes what's actually in the vial in ways the label alone won't show. A chemically synthesized peptide can carry a TFA or acetate counter-ion that adds to its dried weight without adding peptide — exactly what a net-content result is measured against rather than guessed at. A recombinantly produced one went through an entirely different set of steps, from cell lysis through refolding to enzymatic trimming, each with its own failure modes and its own reasons one batch can differ from the next.

None of that is a reason for concern on its own; it's the reason the test results on a certificate of analysis exist at all. Purity by HPLC, identity by mass spectrometry and net content against label strength each answer a question synthesis alone can't, and reading them together is still the only way to know what a given lot actually is.