How Peptides Are Made: Manufacturing, Synthesis and Production

Updated August 2026 · Patriot Labs Research Team · ~32 min read

The short answer

Here is the whole story in three sentences. Peptides come from living things — every cell builds them from instructions encoded in DNA. Peptides made for laboratory work are built two ways: assembled chemically, one amino acid at a time, on a polymer bead inside a reactor, or grown biologically by putting a synthetic gene into a microorganism and fermenting it. Everything after that is separation and proof — purify the target away from the near-identical junk manufacturing always produces, exchange it into a stable salt, freeze-dry it, and test it hard enough to say what is in the vial.

That is the map, and this guide escalates deliberately: the first sections answer “where do peptides come from” in plain language, and after that it becomes a description of an industrial process, with the numbers, failure modes and economics most explainers leave out. New to the subject? Start with What Are Peptides? and come back.

Where peptides come from in nature

A peptide is a short chain of amino acids joined by amide bonds — the same chemistry that makes proteins, just fewer links. Insulin, oxytocin, glucagon and ghrelin are all peptides, manufactured inside cells on demand. The honest answer to “where do peptides come from” is: from the machinery of life, still the most sophisticated peptide factory that exists.

That machinery follows a consistent recipe. A peptide begins as a gene — a stretch of DNA — which the cell copies into messenger RNA (transcription). The mRNA travels to a ribosome, which reads the code three bases at a time and links the matching amino acids into a chain (translation). Anyone searching “where are peptides synthesized” is looking for that answer: at the ribosome, in the cytoplasm, and for secreted peptides on the rough endoplasmic reticulum. The ribosome always builds from the amino end (N-terminus) toward the acid end (C-terminus); chemists in glassware ended up going the opposite way, C-terminus first. That reversal is not trivia — it is why the chemistry needs protecting groups at all.

Cells rarely stop at the raw chain. Many peptides are made as a longer pre-pro precursor: a signal sequence steers the chain into the endoplasmic reticulum and is clipped off, and a pro-region keeps the molecule inactive until prohormone convertases cut at specific basic residues inside secretory granules. Along the way the chain may be folded and locked with disulfide bonds, amidated at the C-terminus, or glycosylated. Insulin is the worked example: a single-chain precursor, folded and bridged, then cut to release the connecting peptide.

It is also why post-translational modification is a recurring manufacturing headache: chemistry installs some of these features trivially, biology installs others trivially, and the two lists are not the same. That mismatch is why the industry runs two completely different production routes.

From glands to the lab: how sourcing evolved

The way peptides are obtained has moved through three eras, each transition driven by a limit of the last.

The extraction era. The first medically important peptide, insulin, was not synthesized at all — it was harvested from the pancreas glands of cattle and pigs in the early 1920s, and for roughly six decades animal glands were the world's supply. The constraints are brutal: supply is tied to slaughterhouse throughput, the product carries tissue-derived impurities, and the sequence you get is the animal's, not the human one. Those differences are small in a diagram and not small in an immune system.

The synthesis revolution. In 1963 R. B. Merrifield published a method for building a peptide chemically on a solid support — solid-phase peptide synthesis, or SPPS.^[1] The 1984 Nobel Prize in Chemistry followed, “for his development of methodology for chemical synthesis on a solid matrix.”^[2] For the first time a chemist could construct any chosen sequence with no gland involved, and SPPS remains the workhorse of peptide manufacturing today.

The recombinant era. In 1979 a Genentech team reported expression in Escherichia coli of chemically synthesized genes for the human insulin A and B chains, purified and joined afterwards.^[18] By 1982 recombinant human insulin was on the market as the first medicine ever made this way; since 1986 the dominant commercial process has instead expressed a single proinsulin gene and excised the connecting peptide enzymatically.^[19] Instead of extracting a peptide or building it bond by bond, you hand the blueprint to a cell.

Those two approaches account for essentially all peptide made for research and medicine today. Everything that follows is about how they actually run.

The two industrial routes

The choice between the two routes is not ideological. It comes down to chain length, whether the sequence contains anything a cell cannot make, and how many kilograms per year the plant has to deliver.

Chemical synthesis assembles the peptide in a reaction vessel, one protected amino acid at a time. It is fast to start, indifferent to how “unnatural” the sequence is, and it can incorporate D-amino acids, N-methylated residues, C-terminal amides and lipid conjugates that living cells will not build. Its ceiling is length: routine Fmoc-SPPS is reliable below roughly 50 residues, and large-scale manufacture of longer chains remains genuinely difficult.^[24] Recombinant biosynthesis instead puts a synthetic gene into a host such as E. coli or a yeast and grows it in a fermenter. It wins decisively for long chains and very large annual volumes, because a fermenter scales on biology rather than on stoichiometric reagent consumption.

A third category sits between them and is increasingly where large-volume manufacturing lives: hybrid processes, in which a backbone is produced by fermentation or fragment assembly and then finished chemically.

Solid-phase peptide synthesis, step by step

SPPS is the most important method in peptide manufacturing and its core idea is elegant. Building a peptide free in solution means separating product from leftover reagent after every coupling — dozens of isolations, each losing material. Merrifield's insight was to anchor the growing chain to an insoluble polymer bead.^[1] Because the chain is tethered you can flood the vessel with excess reagent to drive the reaction to completion, then drain and rinse: the product stays, the mess leaves through the frit.

The chain is built C-terminus first. Every incoming amino acid arrives with its alpha-amino group capped by a temporary protecting group and its side chain masked by a semi-permanent one, so the only bond that can form is the intended one. The cycle, repeated once per residue:

1. Load. Attach the first protected amino acid to the resin through a chemical linker, whose choice determines the finished C-terminus — free acid, amide, or protected fragment.
2. Deprotect. Remove the temporary alpha-amino cap, exposing one reactive amine on every chain. In Fmoc chemistry this is a base treatment, classically piperidine in DMF.
3. Wash. Rinse out the base and the deprotection by-product. Not a formality: residual base carried forward causes side reactions, and washing is where most of the process solvent goes.
4. Couple. Add the next protected amino acid together with a coupling reagent that activates its carboxyl group, usually in several-fold excess to force the reaction toward completion.
5. Wash again. Flush out excess activated amino acid and coupling by-products before the next deprotection.
6. Cap (optional). Acetylate chains that failed to couple so they stop growing, turning a full-length chain missing one residue into a short, easily separated impurity.
7. Repeat, then cleave. Run the cycle for every remaining residue, then treat the resin with a strong acid cocktail that severs the linker and strips the side-chain protecting groups, releasing crude peptide into solution.

Step six deserves emphasis. Capping converts a recoverable yield loss into a certain one, in exchange for an impurity profile you can actually separate — a trade that recurs at every stage of peptide manufacturing. Modern laboratories run the cycle on automated synthesizers that meter reagents into a fritted vessel on a fixed timing program, often with heating to accelerate difficult couplings. What comes off the resin is crude peptide: correct in intent, mixed in fact.

The chemistry underneath: resins, protecting groups, coupling reagents

Three choices define an SPPS process.

Protecting-group strategy. Two schemes dominate. Boc chemistry puts an acid-labile tert-butoxycarbonyl group on the alpha-amine, removed with TFA every cycle, with benzyl-type side-chain protection that survives it. The problem is the endgame: final cleavage and global deprotection require anhydrous hydrogen fluoride, which cannot be handled in glass and needs a dedicated fluoropolymer apparatus, trained operators and defined first-aid procedures.^[7] Fmoc chemistry replaces that alpha cap with the base-labile 9-fluorenylmethoxycarbonyl group introduced by Carpino and Han in 1970 and characterised in 1972.^[3][4] Fmoc comes off with base and side-chain groups with acid, so the two are truly orthogonal and the endgame is a TFA cocktail rather than HF. That single change is why Fmoc took over the field.

One wrinkle rarely appears in vendor copy: piperidine, the standard Fmoc deprotection base, is a controlled chemical, regulated for its use in illicit synthesis, so institutions often need permits to buy it. Interest in 4-methylpiperidine and piperazine is driven partly by that friction and partly by differing rates of aspartimide formation.^[6]

Resin and linker. The bead is not inert scaffolding; it is a reaction environment. Cross-linked polystyrene is the classic support, swelling well in DMF and dichloromethane, while polyethylene-glycol-based resins swell across a wider solvent range including water. Loading matters too: a heavily loaded bead puts growing chains close together and encourages them to associate, so difficult sequences are often run on deliberately low-loading resin. The linker decides the product's C-terminus — hydroxymethylphenoxy linkers give a free acid, amide-forming linkers give a C-terminal amide, and highly acid-sensitive trityl linkers release a still-protected fragment under mild conditions, which is the foundation of fragment condensation at scale.

Coupling reagent. The carboxyl group must be activated before it will form an amide, and how you activate it sets both speed and stereochemical damage. Carbodiimides such as DIC are cheap and normally paired with a racemization-suppressing additive; uronium and phosphonium reagents such as HBTU and HATU are faster and more powerful. HATU owes its performance to the 1-hydroxy-7-azabenzotriazole additive introduced by Carpino in 1993, which accelerates coupling and reduces epimerisation.^[5] Cost, safety and impurity profile all differ: benzotriazole reagents carry energetic-hazard considerations at scale, uronium reagents can cap the chain if left without the amine component, and each leaves a different by-product for the wash.

These choices interact. A resin that swells poorly in a chosen green solvent starves the coupling; a powerful coupling reagent papers over aggregation for a few residues and then fails all at once. Process development is largely the work of finding a combination that survives the sequence in front of you.

Why it gets hard: the arithmetic of yield

This is the part that explains almost everything else about peptide manufacturing. Every cycle has an efficiency: if a coupling is 99% complete, 1% of chains on the bead failed to extend. Unless capped, those failures couple on the next round and produce a chain missing exactly one residue — a deletion sequence, the defining impurity of SPPS, with the same amino acids, nearly the same mass and nearly the same retention time as the product. Now compound it: overall yield is per-step efficiency raised to the number of steps.

• At 99% per step, a 30-residue peptide finishes at 0.99³⁰ ≈ 74%.
• At 98%, the same 30-mer drops to ≈ 55%.
• At 95%, it collapses to ≈ 21%.

The Nobel committee's illustration made the point at 100 steps: 90% per step gives an overall yield of roughly 0.003%, while the 99.5%-or-better per-step efficiency solid-phase methods made possible gives about 61%.^[2] The entire value of SPPS sits in the third decimal place of one number.

Two consequences follow. First, a single bad coupling in the middle of a long sequence is unrecoverable — you cannot fix it later, only separate around it. Second, the purification burden grows with length faster than the synthesis burden does, because you are separating a 30-mer from a 29-mer that differs by one residue in the middle.

Difficult sequences make it worse. Hydrophobic runs of leucine, valine, isoleucine and phenylalanine, and glycine-containing motifs, cause resin-bound chains to associate into beta-sheet-like aggregates that bury their reactive N-termini, so deprotection and coupling become incomplete no matter how much reagent is added.^[8] The countermeasures are structural rather than brute-force: pseudoproline dipeptides, which install a temporary ring at a serine or threonine that kinks the backbone and prevents sheet formation; backbone amide protection; lower resin loading; solvent mixtures such as NMP/DMSO that improve swelling; and elevated temperature.^[8]

None of this is edge-case chemistry. A significant fraction of commercially interesting sequences are “difficult” by this definition, and whether a manufacturer has solved a given one is a large part of what you are paying for.

Side reactions that ruin batches

Beyond incomplete coupling, three side reactions do most of the damage.

Aspartimide formation. Under the repeated base treatments of Fmoc chemistry, an aspartic acid residue can cyclise onto the backbone nitrogen of the next residue to form a five-membered succinimide ring. That ring reopens — sometimes to the correct alpha-linked peptide, sometimes to a beta-linked isomer, sometimes with loss of stereochemistry, sometimes trapped by the deprotection base itself. Asp-Gly is the notorious motif; Asp-Asn, Asp-Ser and Asp-Thr are also prone. The result is a family of impurities with the same molecular mass as the product, so mass spectrometry alone will not catch them — one practical reason piperazine and 4-methylpiperidine are studied as substitutes.^[6]

Racemization and epimerisation. Activating a carboxyl group makes the alpha-carbon acidic and opens a route to an oxazolone intermediate that can lose and regain a proton on either face. Reported levels span the full range: carbodiimide coupling with an azabenzotriazole additive around 4% epimerisation in a test system where a different activation gave roughly 25%, and glycosylated serine building blocks anywhere from under 1% to over 70% depending on structure.^[9] Cysteine and histidine are chronically at risk, as are N-methylated residues and any residue whose side chain stabilises the anion; hindered tertiary amines suppress alpha-proton abstraction.^[9] An epimerised peptide is a diastereomer — again, identical mass, different molecule.

Cleavage-stage damage. The final TFA treatment generates highly reactive carbocations as side-chain protecting groups come off, and left alone those cations alkylate tryptophan, tyrosine, methionine and cysteine. Cleavage cocktails therefore contain scavengers — thiols, silanes, water, phenol — chosen for the residues present. Get the cocktail wrong and the batch is modified peptide.

Together these explain a fact worth internalising: the hardest impurities in a peptide are the ones that most resemble the peptide. That is why a certificate of analysis must report purity by chromatographic separation and identity by mass, and why neither alone is sufficient — our guide to peptide COA testing covers how to read those numbers.

Scaling up: from milligrams to metric tons

A benchtop synthesizer and a commercial peptide plant run the same chemistry and share almost nothing else.

At manufacturing scale the vessel becomes a jacketed, agitated reactor with a bottom filter plate holding tens to thousands of litres of resin slurry. Mixing must be gentle enough not to fracture the beads — broken resin blinds the filter and the batch stops draining — so filtration time, not reaction time, often sets the cycle. Temperature control matters more because deprotection is exothermic and a large bed does not shed heat like a 10 mL vessel. And reagent excesses that are trivial in a research lab become the dominant cost line: a three-fold excess of protected amino acid at 5 kg scale is a purchasing decision.

The capacity being installed gives a sense of the industry's shape: single-company programmes are adding tens of thousands of litres of solid-phase synthesis capacity.^[33] These are chemical plants, not laboratories.

Where liquid phase comes back. Stepwise SPPS is structurally inefficient at scale: you pay a full reagent excess and a full set of washes per residue regardless of chain length. Large-scale manufacture has therefore leaned on fragment condensation — assemble several short protected fragments, purify them while they are still small, then join them.^[11] A defect in one fragment is removed before it contaminates the whole molecule.

The variants proliferate from there. Hybrid SPPS/LPPS assembles fragments on solid support and condenses them in solution. Soluble-support methods anchor the chain to a hydrophobic tag that keeps it in solution during reaction but lets it be precipitated and filtered between steps — liquid-phase kinetics with solid-phase workup. Manufacturers have reported liquid-phase approaches cutting solvent use on the order of 60% versus the corresponding SPPS route, with process details withheld.^[15]

Chemical ligation for long chains. When a target exceeds what stepwise synthesis can deliver, the classic solution is native chemical ligation: react an unprotected peptide thioester with a second unprotected fragment bearing an N-terminal cysteine, and the transient thioester-linked intermediate rearranges spontaneously into a genuine peptide bond. Dawson and colleagues demonstrated it in 1994 by assembling a disulfide-containing cytokine from two unprotected pieces and folding it.^[10] A newer industrial variant is chemo-enzymatic synthesis, in which an engineered peptide ligase joins fragments in water with no protecting groups at the coupling site — enzymatic amide formation does not racemise and needs no coupling reagent.^[11][15]

Flow chemistry and the speed frontier

Batch SPPS spends most of its wall-clock time waiting — for reagent to diffuse into a bead, for a vessel to drain, for a wash to finish. Continuous-flow synthesis attacks that by packing the resin into a heated column and pumping reagents through it, so fresh reagent is continuously delivered and by-products continuously swept out.

The numbers are startling. A fully automated flow platform reported in 2017 achieved amide bond formation in about 7 seconds and a complete cycle of roughly 40 seconds per residue, with crude purities and isolated yields comparable to standard batch SPPS — a 30-mer in about twenty minutes.^[12] A 2020 extension produced fully synthetic single-domain proteins, including proinsulin, barnase and HIV-1 protease variants with non-canonical amino acids, that folded to activity comparable to recombinant material.^[13]

Flow has not displaced batch in commercial manufacturing, and the honest reason is that plants are built around batch. But the direction of travel matters: the constraint that used to define what could be made chemically is being relaxed.

The solvent problem

If you take one number from this guide, take this one. Peptide manufacturing's process mass intensity — total mass consumed per unit mass of product — has been reported at roughly 3,000 to 15,000 kg per kg for peptides in the 1,000–5,000 Da range, with solvent making up 80 to 90 percent of the waste.^[15] Analysis published through the ACS Green Chemistry Institute Pharmaceutical Roundtable cites commercial peptides generating on the order of 34 tonnes of waste and 118 tonnes of CO₂ equivalent per kilogram of active ingredient.^[16] Small molecules typically run one to two orders of magnitude lower. Peptides are, by mass balance, among the least efficient things the pharmaceutical industry makes.

Where does it all go? Overwhelmingly into washing. Standard protocols use roughly five washes between steps, and post-deprotection washing alone has been estimated at about 90% of the waste generated, with research-scale conditions consuming on the order of 4.25 mL of solvent per amino acid per cycle.^[14]

The specific solvents are the second problem. DMF and NMP are the workhorses because they dissolve everything and swell polystyrene resin well. Both are classified as substances of very high concern under REACH on the basis of reproductive toxicity and are described in the field as requiring urgent replacement; global DMF consumption across all uses has been put above 400,000 tonnes annually.^[15] The ACS roundtable lists DMF, DMAc and NMP among solvents to avoid where possible.^[16]

The response has three strands. Replace: a substantial literature evaluates greener solvents and binary mixtures tuned for polarity and viscosity so resin swelling and coupling kinetics survive the substitution.^[15][17] Reduce: methods that cut or eliminate washing have reported up to 95% waste reduction versus traditional SPPS, and about 80% at 25 mmol scale — 28.4 L of waste against 139.7 L.^[14] Recover: distillation-based recycling has been reported to cut process mass intensity by more than 60%, with 85–95% solvent recovery for some systems.^[15] This is not an environmental footnote — solvent purchase, handling and disposal are a large fraction of what a kilogram of peptide costs, and a plant's recovery capability constrains its output.

Recombinant production, in depth

“Recombinant peptide” is one of the most searched and least well-explained phrases in this field. Instead of building the amide bonds yourself, you install the instructions in a cell and let its ribosomes do it: write a DNA sequence coding for the target, codon-optimised for the host, clone it into an expression plasmid behind a controllable promoter, transform that into a host strain, and grow the strain in a bioreactor under defined feeding, aeration, pH and temperature control until expression is induced.

The small-peptide problem. Here is the thing almost nobody mentions: you usually cannot express a short peptide on its own. Small peptides are highly susceptible to intracellular proteolysis because of their size and, for cationic sequences, their charge, and any peptide with antimicrobial character is toxic to the bacterium being asked to make it.^[21] Both problems have one solution: express the peptide as a fusion with a carrier protein that hides it from proteases, neutralises its activity, and often improves folding and solubility.

Fusion carriers and where the product ends up. Common solubility-enhancing partners include thioredoxin, glutathione S-transferase, SUMO and green fluorescent protein; silk-derived domains have been reported to give several-fold more soluble product than conventional tags.^[21] The opposite strategy is to drive the fusion deliberately into inclusion bodies, the dense insoluble aggregates E. coli forms on over-expression. These look like failure and are frequently an advantage: they shield the product from proteolysis, isolate easily by centrifugation, and start you at high purity. The price is that you must solubilise and refold with a chaotrope and redox buffer, and refolding yield is where a lot of material is lost.

Getting the peptide off the carrier. Three approaches:

• Proteolytic cleavage with a site-specific protease such as enterokinase or a SUMO protease — clean and precise, but the enzyme is expensive at scale and must be removed afterwards.^[21]
• Chemical cleavage, classically cyanogen bromide at methionine — cheap and effective, but acutely toxic and incompatible with internal methionine residues.^[21]
• Self-cleaving tags, principally inteins — protein elements that excise themselves under a pH or thiol trigger, combining purification and cleavage in one operation. The recurring risks are premature cleavage during expression and incomplete cleavage at harvest.^[21]

Yields, realistically. Reported production levels for host-defence peptides span roughly 2–600 mg per litre in inducible E. coli, under 0.1 mg/L to several hundred mg/L in Pichia pastoris, and single-digit to low tens of mg/L in insect cells.^[21] That spread is the honest picture: recombinant expression is not a fixed-yield process, it is a research project per target.

Yeast, and why secretion changes everything. The methylotrophic yeast Pichia pastoris, reclassified as Komagataella phaffii, is the other major workhorse, and its appeal is that it secretes: the product leaves the cell into the culture medium, far cleaner than a bacterial lysate. It also performs eukaryotic folding, disulfide formation, proteolytic processing and glycosylation.^[22] Expression is usually driven by the strong AOX1 promoter, induced with methanol and repressed by glycerol, ethanol and glucose. Gram-per-litre titres are achievable, fermentations typically run 72–96 hours, and commercial products made this way include a recombinant insulin.^[22] Saccharomyces cerevisiae is also used industrially for insulin precursors secreted with an alpha-factor signal sequence.^[19]

The insulin case study. Insulin is the best-documented recombinant peptide process in existence. The dominant E. coli route expresses proinsulin as inclusion bodies, which are recovered, solubilised, subjected to oxidative sulfitolysis, refolded to the correct disulfide pattern, purified chromatographically, and converted to insulin by enzymatic excision of the connecting peptide with trypsin and carboxypeptidase B, with pH precipitation and zinc crystallisation in the sequence.^[19][20][34] Ten or more unit operations stand between fermenter and vial, and that downstream train is where most of the cost lives.

Beyond microbes. Plant-based systems occupy a real if niche position. Molecular farming uses transient expression in Nicotiana benthamiana via viral or agrobacterial vectors, or stable expression in crops such as rice, barley and maize. The arguments in favour are low cost, easy scalability and reduced risk of human or animal pathogen contamination; reported yields are modest — hundreds of micrograms per kilogram of barley seed for one antimicrobial peptide — and toxicity, misfolding and proteolysis remain limiting.^[23]

Choosing a route: synthesis vs fermentation

Manufacturers do not pick a route by preference; the decision tree is fairly mechanical.

Note how many rows turn on a single structural feature rather than on economics. That is the practical meaning of the earlier point about post-translational modification: the route is chosen by the molecule, not by the accountant.

Purification and counterion exchange

Whichever route builds the peptide, what comes out is a mixture: synthesis leaves deletion sequences, truncations, epimers, aspartimide-derived isomers and alkylated by-products, while fermentation leaves host cell protein, DNA, endotoxin, misfolded species and clipped variants. Purification is not a polish step — it is where a large share of manufacturing cost sits.^[25]

Preparative reversed-phase HPLC is the dominant tool for synthetic peptides. Crude material is loaded onto a hydrophobic stationary phase, typically C18-modified silica, and eluted with an increasing gradient of an organic modifier — almost always acetonitrile — in water containing an acidic ion-pairing additive such as trifluoroacetic acid. Retention of a molecule that size falls off very steeply with organic content, so gradient elution is essential.^[25] The ion-pairing acid pairs with basic side chains and sharpens peaks, which is precisely what you need when the impurity differs from the product by one residue.

Scale-up is its own discipline. Loading is pushed until resolution begins to fail, then fractions are cut conservatively — the peak edges, where product overlaps impurity, are discarded or recycled. Purity and yield trade directly against each other, and every percentage point of specified purity costs recovered mass. The mobile phase is itself a major consumable.^[25]

Counterion exchange. Most explainers omit this step entirely. A peptide purified with TFA in the mobile phase comes out as its trifluoroacetate salt: basic side chains such as lysine and arginine pair with trifluoroacetate anions.^[26] That is a consequence of the purification chemistry, not a choice, and TFA is often undesirable in the final material — it can interfere with physicochemical characterisation and has been reported to affect cell-based experiments, including inhibition of cell proliferation.^[26] Manufacturers therefore exchange the counterion, typically to acetate, using ion-exchange resin, a chromatographic step run in the target salt system, or repeated lyophilization from the corresponding acid.^[26]

Counterion identity is not cosmetic. It has been reported to influence secondary structure and helicity, solubility and aggregation behaviour, and stability in formulated systems, so the salt form belongs on the certificate of analysis alongside purity.^[26] Two vials of the same peptide at the same stated purity but different salt forms contain different masses of actual peptide, which is why peptide content is reported separately. A desalting or buffer-exchange step usually follows before the solution goes to the freeze-dryer.

Lyophilization and the finished vial

The final operation converts a purified aqueous solution into a dry solid, because a peptide in water is a peptide slowly hydrolysing. That operation is lyophilization — freeze-drying — and it is a designed thermal cycle, not simply drying.

The cycle has three stages.^[27] In freezing, the solution is cooled until the water crystallises, sometimes with an annealing hold that briefly warms the frozen product to grow larger ice crystals and leave a more open, faster-drying pore structure. In primary drying, the chamber is evacuated and heat supplied so ice sublimes directly to vapour onto a cold condenser, with shelf temperature set to keep the product several degrees below its collapse temperature and chamber pressure at a modest fraction of the ice vapour pressure. In secondary drying, shelf temperature is raised to desorb the water that never froze, commonly targeting residual moisture below about 1%.^[27]

Three thermal parameters govern the design: the glass transition temperature of the maximally freeze-concentrated solute, the collapse temperature, and the eutectic temperature of any crystalline component.^[27] Get them wrong and you get collapse, meltback, long reconstitution times, or a cake that looks fine and carries too much residual moisture to be stable. What emerges is the familiar dry cake or thin film at the bottom of a vial, stoppered and sealed in a controlled fill-finish environment. Peptide quantities are small enough that the visible cake may be almost nothing — which is why peptide content on the COA matters more than what the vial looks like.

From there the material is stored cold and dark; shipping and long-term holding are covered in cold-chain handling and how peptides degrade. Before laboratory work it must be reconstituted into a sterile diluent such as bacteriostatic water, and our reconstitution calculator handles the arithmetic. Where several peptides share one vial, see why peptides are blended.

Release testing: proving what is in the vial

A purified peptide has to earn its label, and the two load-bearing tests are complementary for reasons that follow directly from the side reactions described earlier.

Mass spectrometry answers identity. The measured molecular mass is compared against the mass calculated from the sequence; because every sequence has a characteristic mass, agreement is strong evidence the intended molecule was built. What it cannot distinguish is an isomer — an epimerised residue or an aspartimide-derived beta-linked variant has exactly the same mass. Chromatography answers purity. Analytical reversed-phase HPLC separates species by hydrophobicity and reports the target as a percentage of total peak area, catching the isomers mass spectrometry misses and quantifying deletion sequences.

Serious release testing adds more, and each test answers a specific manufacturing question: amino acid analysis confirms composition; peptide content states how much of the vial's mass is peptide rather than salt and water; water content by Karl Fischer checks the lyophilization cycle; counterion assay identifies the salt form; residual solvents catch carryover from synthesis and purification; and bacterial endotoxin testing is essential for anything that came out of a fermenter. Our guide on understanding peptide COA testing covers how to read each of these, and what a real batch-specific certificate looks like versus a generic one.

Research grade vs GMP grade

“Research grade” is a commercial description, not a regulatory standard.

GMP is a defined framework. For active pharmaceutical ingredients, ICH Q7 — adopted by FDA as guidance — requires an independent quality unit with release authority, qualified equipment, controlled raw materials, validated processes, complete batch records, defined specifications, deviation and change control, and retention samples, all subject to inspection.^[29] The point of GMP is not that the chemistry is better; it is that the evidence is complete and auditable.

Expectations for synthetic peptides specifically are equally developed. FDA's guidance on abbreviated applications for certain highly purified synthetic peptides referencing recombinant listed drugs — covering glucagon, liraglutide, nesiritide, teriparatide and teduglutide — requires that peptide-related impurities shared with the reference product be present at the same level or lower; that any new specified peptide-related impurity above 0.5% of the drug substance takes the product outside that pathway; that impurities at or above 0.10% be identified; and that applicants assess whether new impurities introduce sequences with increased MHC affinity or raise aggregation propensity or innate immune stimulation relative to the reference.^[28] That is the level of specificity a regulated peptide process is held to.

The practical consequence: since “research grade” guarantees nothing on its own, the only meaningful signal is the batch-specific certificate of analysis, ideally with independent third-party testing tying HPLC purity and mass-spec identity to the exact lot in hand. How to evaluate a supplier on that basis is covered in How to Buy Research Peptides and what to look for in a vendor.

What actually drives the cost

Peptide pricing varies by orders of magnitude for material that all looks like white powder. Four factors explain nearly all of it.

Length. Cost is superlinear in residue count. Each residue adds a coupling with a multi-fold reagent excess, a deprotection, several washes and a new chance to generate a deletion impurity — while overall yield decays geometrically. A longer chain consumes more input and delivers less output.

Sequence difficulty. An aggregating sequence may need pseudoproline dipeptides, backbone protection, low-loading resin or specialised solvents, and those building blocks cost far more than standard protected amino acids.^[8] Multiple cysteines need a defined oxidative folding step; an Asp-Gly motif needs deliberate aspartimide suppression.^[6] None of that is visible in the finished formula.

Purity specification. Purity is bought with yield. Moving from 95% to 98% to 99% means cutting fractions ever more conservatively, and the marginal cost of the last percentage point is much higher than the first.^[25]

Scale. Fixed costs — process and analytical development, cleaning validation, documentation — are amortised over the batch, so the same molecule can differ enormously in per-gram cost between a milligram research order and a multi-kilogram campaign. This is also where the solvent mass balance stops being an abstraction: at PMI figures in the thousands of kilograms per kilogram of product, solvent purchase, handling and disposal are a first-order cost line.^[15][16]

The recombinant route reshuffles all four: its costs sit in strain construction, media, and above all in the ten-plus downstream unit operations between fermenter and vial.^[20] Per-gram cost falls steeply with volume in a way stepwise synthesis cannot match — which is why the world's insulin is fermented and most short modified peptides are not.

The GLP-1 capacity story

The most consequential thing happening in peptide manufacturing right now is a capacity build-out driven by demand for GLP-1 receptor agonist medicines, and the scale of the spending is unusual even by pharmaceutical standards. Eli Lilly announced in May 2024 that investment at a single Lebanon, Indiana site would rise to $9 billion — $5.3 billion on top of $3.7 billion already committed — specifically to produce active pharmaceutical ingredient for tirzepatide and pipeline medicines, with production beginning late 2026 and scaling through 2028.^[31] In February 2025 it said it would more than double US manufacturing investment since 2020 to more than $50 billion across four new sites, three of them API plants.^[32] Novo Nordisk announced a $4.1 billion expansion in Clayton, North Carolina in June 2024, adding 1.4 million square feet with completion between 2027 and 2029, and said 2024 production investment would rise to about $6.8 billion from $3.9 billion.^[30]

The contract manufacturing sector has followed. Industry reporting from May 2025 describes a peptide and oligonucleotide CDMO market projected to reach roughly $5.67 billion by 2030 at about 12.5% annual growth, with peptides the larger share; CordenPharma committing over €1 billion across multiple sites, including adding 30,000 L of solid-phase capacity and a Colorado investment above $500 million intended to roughly double SPPS capacity past 42,000 L by 2028; PolyPeptide investing around €100 million to double capacity in Malmö; plus expansions from Bachem, SK Pharmteco and others.^[33]

Two things follow. First, this is chemical capacity — reactors, solvent handling, chromatography and lyophilization at industrial scale — which is why the mass balance described above became a strategic problem rather than an academic one. Second, capacity of this kind takes years to build, which is the structural reason peptide supply cannot respond quickly to demand; when a peptide is in shortage the constraint is usually in purification or fill-finish rather than synthesis.

Can you make peptides at home?

No — and the reason is worth understanding, because it is not the one people usually assume. This guide will not give a procedure; it will explain honestly where the wall is. The obstacle is not that the recipe is secret — Fmoc-SPPS is published in exhaustive detail. The obstacle is that peptide synthesis is a chain of operations that each require controlled conditions and controlled materials, and failing any one of them produces something that looks exactly like success.

• The building blocks are not amino acids. You need protected derivatives — an alpha-amino protecting group plus orthogonal side-chain protection matched to the cleavage strategy — and difficult sequences additionally need pseudoproline or backbone-protected dipeptides.^[8]
• The chemistry is anhydrous. Couplings run in dry aprotic solvents; water hydrolyses activated esters and quietly lowers coupling efficiency, and as the arithmetic above shows, small per-step losses compound into heavy deletion-sequence contamination.
• Deprotection uses a controlled chemical. Piperidine, the standard Fmoc deprotection base, is regulated as a controlled substance precursor and typically requires permits to purchase.^[6]
• Cleavage is the dangerous step. Fmoc chemistry ends with concentrated trifluoroacetic acid plus scavengers, requiring engineered ventilation and proper waste handling. Boc chemistry ends with anhydrous hydrogen fluoride, which cannot be handled in glass at all and requires a dedicated fluoropolymer apparatus, trained operators and specific first-aid provision.^[7]
• Purification is not optional. Crude peptide is a mixture of the target and structurally similar failures, and separating them needs preparative reversed-phase HPLC with gradient control and fraction collection.^[25] No filtration or recrystallisation shortcut resolves a 29-mer from a 30-mer.
• Verification is not optional either. Without mass spectrometry you cannot confirm identity; without analytical HPLC you cannot state purity. And because epimers and aspartimide-derived isomers share the product's exact mass, a mass spectrum alone is not sufficient.

Put together, the barrier is not one exotic piece of knowledge. It is that the last two steps — separation and proof — decide whether you have a defined substance or an unknown mixture, and they are precisely the steps that cannot be improvised. Anyone who assembles a peptide without them has produced a powder of unknown composition. That is also why third-party analytical data matters so much when acquiring material rather than making it: the chemistry is only as good as the evidence attached to the specific lot.

What this guide does not establish

A description of a manufacturing process is not a claim about a molecule's properties.

• Manufacturing quality is not biological activity. A high HPLC purity figure and a matching mass tell you what the substance is; they say nothing about what it does in any biological system.
• Process descriptions are general. Specific manufacturers use specific resins, reagents, solvents and purification conditions, and those details are usually proprietary. Nothing here describes any one supplier's validated process.
• Yield and PMI figures are literature ranges. The process mass intensity, waste and solvent numbers cited are reported for particular classes of peptide and particular processes;^[14][15][16] they indicate scale, not a value for every product.
• Regulatory thresholds apply to their stated context. The impurity thresholds described come from a specific FDA guidance covering five named synthetic peptides;^[28] they are not universal specifications.
• Nothing here describes use in a person. The material discussed on this site is supplied for in-vitro research and laboratory use only.

Frequently asked questions

How are peptides manufactured?

Industrially, by one of two routes. Chemical manufacturing uses solid-phase peptide synthesis: protected amino acids are coupled one at a time onto a polymer resin in a reactor, then cleaved, purified by preparative reversed-phase HPLC, converted to a stable salt and freeze-dried. Biological manufacturing uses recombinant expression, growing an engineered microorganism carrying a synthetic gene in a fermenter. Chain length, sequence difficulty, non-natural residues and annual volume decide which route is used.

What is the peptide manufacturing process, step by step?

A typical synthetic process runs: load the first protected amino acid onto resin; repeat a deprotect–wash–couple–wash cycle for every residue; cleave the chain from the resin and strip side-chain protecting groups with an acid cocktail; precipitate the crude peptide; purify by preparative reversed-phase HPLC; exchange the counterion; lyophilize the purified fractions; then test identity, purity, peptide content, water content, residual solvents and endotoxin before release.

What is recombinant peptide synthesis?

Recombinant peptide synthesis means letting a living cell build the peptide from a gene rather than assembling it chemically. A synthetic DNA sequence is cloned into a plasmid and transformed into a host such as Escherichia coli or the yeast Komagataella phaffii, which is then grown in a fermenter. Short peptides are expressed fused to a carrier protein, because alone they are degraded by host proteases or toxic to the cell; the fusion is later cleaved by a protease, a self-splicing intein or a chemical reagent.

Where do peptides come from?

Peptides come from living systems. Every cell transcribes genes into messenger RNA and translates that RNA on the ribosome into chains of amino acids, many of which are then trimmed, folded and chemically modified. Peptides sold as research material are laboratory-made copies of those same molecules, produced to a defined sequence by chemical synthesis or by fermentation, then purified and characterised so that identity and purity are known.

Where are peptides synthesized in the body?

At the ribosome, in the cytoplasm and on the rough endoplasmic reticulum. The ribosome reads messenger RNA three bases at a time and links the matching amino acids into a chain running from the amino terminus toward the carboxyl terminus. Many peptides are first made as longer pre-pro precursors, then processed by signal peptidases and prohormone convertases in the endoplasmic reticulum, Golgi and secretory granules before release.

What are peptides formed from?

Peptides are formed from amino acids joined by amide bonds, called peptide bonds, between the carboxyl group of one residue and the amino group of the next. In biology the twenty canonical amino acids are used and the bond is made by the ribosome. In chemical manufacturing the same bond is made by activating a carboxyl group with a coupling reagent, and the building-block menu is much larger.

Can you make peptides at home?

Realistically, no. Peptide synthesis is not a difficult recipe with hard-to-find ingredients; it is a chain of operations that each require controlled equipment and materials. Assembly needs anhydrous aprotic solvents and protected amino acid derivatives, deprotection typically uses piperidine, a controlled chemical,^[6] and cleavage uses concentrated trifluoroacetic acid or, in Boc chemistry, anhydrous hydrogen fluoride in a fluoropolymer apparatus.^[7] Even a perfect assembly gives a mixture that is useless without preparative HPLC to purify it and mass spectrometry to confirm what was made.

How are peptides made in a lab?

In a research laboratory, peptides are usually made on an automated synthesizer running Fmoc solid-phase peptide synthesis at millimole scale. The instrument delivers deprotection base, activated amino acid and wash solvent to a fritted vessel packed with resin, repeating that cycle once per residue, so a short peptide can be assembled overnight. The crude product is then cleaved, precipitated, purified on a preparative reversed-phase column, and characterised by HPLC and mass spectrometry.

Why is peptide production so expensive?

Because cost compounds with length. Every added residue means another coupling with a several-fold excess of a protected amino acid, another deprotection, several washes, and another chance to create an impurity that purification must later remove. Yield falls geometrically with chain length and purification discards material. Solvent dominates the mass balance: reported process mass intensity runs from roughly 3,000 to 15,000 kilograms per kilogram of product, with solvent 80 to 90 percent of the waste.^[15]

What is the difference between research-grade and GMP-grade peptides?

The chemistry can be identical; the control system is not. GMP manufacturing under ICH Q7 requires an independent quality unit, validated processes, qualified equipment, controlled raw materials, full batch records, defined impurity specifications and change control, all subject to inspection.^[29] Research-grade material carries no such obligation. It may be made well and documented honestly with a batch-specific certificate of analysis, but the label alone guarantees nothing.

For in-vitro research and laboratory use only. Not for human consumption.