Endotoxin and LAL testing for research peptides: what an EU/mg figure on a certificate means, and when it is missing

What bacterial endotoxin is#

Endotoxin is lipopolysaccharide (LPS), the main component of the outer membrane of Gram-negative bacteria such as Escherichia coli. Each LPS molecule has three parts: a hydrophobic anchor called lipid A, a short core oligosaccharide, and an outer polysaccharide chain (the O-antigen). Lipid A is the part that animal immune systems recognise, through the receptor TLR4, and it is responsible for the fever response that gave endotoxin its older name, pyrogen.^[1]

Endotoxin is released when bacteria die, so a solution can carry it after the bacteria are gone, and it is heat-stable, so the autoclaving and filtration that remove or kill organisms do not reliably remove it. A peptide can therefore be sterile and still carry endotoxin, which is why the two are tested separately.

The LAL assay family: gel-clot, turbidimetric and chromogenic#

The standard test is the Bacterial Endotoxins Test, published as USP General Chapter <85> and harmonised with the European and Japanese pharmacopoeias. USP <85> defines it as a test "to detect or quantify endotoxins from Gram-negative bacteria using amoebocyte lysate."^[2] The lysate is made from the blood cells (amoebocytes) of horseshoe crabs, Limulus polyphemus in the Atlantic (hence LAL, Limulus amoebocyte lysate) or Tachypleus tridentatus in Asia (TAL). Horseshoe crab blood clots when it meets endotoxin, through an enzyme cascade that begins with a protein called factor C; the assay watches that cascade in a tube. The chapter describes three ways of reading it.^[2]

Gel-clot is the oldest method, and under the chapter it is the referee: "In the event of doubt or dispute, the final decision is made based upon the gel-clot limit test."^[2] The photometric methods give a quantitative value from a standard curve, which the chapter requires to have a correlation coefficient of at least 0.980 in absolute value.^[2]

Recombinant factor C and USP <86>#

Because factor C is the first enzyme in the cascade, it can be produced recombinantly and used without animal blood. In a recombinant factor C (rFC) assay, endotoxin activates rFC, which cleaves a fluorogenic substrate; newer recombinant cascade reagents (rCR) reproduce more of the pathway and are read in chromogenic or turbidimetric formats.

A 2020 review in the PDA Journal of Pharmaceutical Science and Technology compared rFC with lysate reagents across published studies and reported a high degree of correlation, with one structural difference: rFC does not respond to the factor G pathway that lysate reagents also contain, which makes it more selective for endotoxin.^[5] In July 2024 the USP Microbiology Expert Committee approved a new chapter, <86> Bacterial Endotoxins Test Using Recombinant Reagents, which became official in May 2025; USP stated that manufacturers using LAL "can continue to do so."^[4] An endotoxin line citing rFC under <86> is therefore a compendial method, not an unofficial alternative.

USP <85>, the Endotoxin Unit and what EU/mg means#

Endotoxin is not reported by weight, because LPS from different bacteria differs in potency per microgram, but in Endotoxin Units (EU), a unit of activity defined against a reference standard. USP <85> requires the standard in every run to be calibrated against the USP Endotoxin Reference Standard, itself calibrated to the WHO International Standard, and states that "One USP Endotoxin Unit (EU) is equal to one International Unit (IU) of endotoxin."^[2] EU and IU are the same thing; European reports often write IU.

The assay measures a concentration in the test solution, in EU per millilitre, and the laboratory converts it to the unit that fits the material. For a lyophilised peptide that is EU per milligram: endotoxin found, divided by the mass of peptide dissolved to make the test solution. The chapter lists the common forms as "EU/mL, EU/mg, EU/Unit of biological activity, etc."^[2] A line reading "Endotoxin: < 0.5 EU/mg" means the laboratory dissolved a weighed amount of the lot, tested the solution, and found less than 0.5 Endotoxin Units per milligram of peptide.

Two details decide how much a figure can tell you. The first is the labelled sensitivity (λ) of the reagent, or the lowest point on the standard curve: a gel-clot test with λ = 0.125 EU/mL cannot report anything below that, so a "<" value is a detection floor, not a measurement. The second is dilution. USP <85> defines the maximum valid dilution (MVD) as "the maximum allowable dilution of a specimen at which the endotoxin limit can be determined," and it depends on the limit, the sample concentration and λ.^[2] A sample diluted beyond its MVD can pass a limit it would not pass undiluted, which is why a complete endotoxin line records the dilution used.

How endotoxin limits are expressed, and why they depend on intended use#

There is no universal pass mark for endotoxin. In the pharmacopoeias the limit for a drug is set in its monograph and derived from how the drug is used. USP <85> states that the endotoxin limit for parenteral drugs "equals K/M," where K is a threshold pyrogenic amount of endotoxin per kilogram of body weight and M is the maximum amount of product given per kilogram in the drug's labelled use.^[2] The chapter's footnote fixes K at 5 USP-EU/kg for any route other than intrathecal, and 0.2 USP-EU/kg for intrathecal products, because the central nervous system tolerates far less.^[2] A product used in large volumes therefore gets a tight limit per millilitre, and the same molecule can carry different limits in different presentations.

The FDA's guidance on pyrogen and endotoxin testing, reissued as a second edition in March 2026, takes the same approach: it publishes no limit table of its own and directs firms to the USP or AAMI calculation methods, giving fixed figures only for devices (for example 0.5 EU/mL or 20 EU per device for products contacting the cardiovascular and lymphatic systems, and 0.06 EU/mL or 2.15 EU per device for cerebrospinal fluid contact).^[3] The same edition withdrew the FDA's 1987 LAL validation guideline because it "no longer reflects the Agency's current thinking," and notes that the 1987 limit table is out of date; a report that still cites the 1987 guideline as its basis is citing a withdrawn document.^[3]

Interference: why peptides are awkward samples#

The LAL cascade is an enzyme system, and anything in the sample that changes enzyme behaviour changes the reading. Inhibition makes the assay under-report endotoxin that is present; enhancement makes it over-report. USP <85> therefore requires every product test to include a positive product control (PPC), a portion of the sample spiked with a known amount of endotoxin; for the photometric methods the spike must be recovered within 50% to 200% of what was added, and if it is not, the laboratory must retest at a greater dilution (not beyond the MVD) or remove the interference by a validated treatment.^[2] A report that shows the spike recovery is showing you that the assay worked on that sample. Peptides bring several known interference sources:

• pH and salts. Lyophilised peptides are usually acetate or trifluoroacetate salts and can pull a test solution outside the roughly 6.0 to 8.0 range most reagents need. Divalent cations such as calcium and magnesium alter endotoxin aggregation and the reaction itself. Dilution and buffering are the first remedies.^[9]
• Cationic sequences. Many peptides carry net positive charge from lysine and arginine. Positively charged peptides bind negatively charged lipid A and can stop it activating factor C, so the assay sees less endotoxin than is there. A 2024 study of blood samples found that this masking by cationic antimicrobial peptides survives the dilution and heat steps that remove other interferences.^[8]
• Low endotoxin recovery (LER). In formulations containing chelators (citrate, EDTA) with surfactants (polysorbates), spiked endotoxin can become undetectable over hours to days while still present. A 2021 review in Biopolymers examines the proposed causes (the endotoxin standard used, the protein ingredient, the excipients) and the mitigation strategies proposed by different groups;^[6] sample treatments that restore recovery were reviewed in the PDA Journal in 2019.^[7]
• Beta-glucans. USP <85> notes that "Amoebocyte Lysate reacts to some β-glucans in addition to endotoxins."^[2] Glucans from cellulose filters or fungal contamination activate the factor G pathway and read as false endotoxin; glucan-blocked lysates and rFC reagents, which lack factor G, avoid this.^[2][5]
• Colour and turbidity. Coloured or cloudy solutions interfere with photometric readings; a copper peptide, for example, is blue, and may need a gel-clot method or a sample blank.^[9]

An endotoxin figure for a peptide is therefore only as good as the interference work behind it. A bare "< 1 EU/mg" with no dilution, recovery or method may be correct, but the report gives you no way to know.

What the endotoxin line on a COA looks like, and when it is absent#

On a full Certificate of Analysis, endotoxin is one row among several tests, after the identity and purity rows described in our guides to reading a Certificate of Analysis and to HPLC and mass spectrometry. A complete row carries the method (for example "Bacterial endotoxins, USP <85>, kinetic chromogenic"), any specification applied, the measured value with its unit, the dilution or sensitivity, and the conclusion.

Absence is the case to understand clearly. Most research-peptide certificates report identity (mass spectrometry) and purity (HPLC) and stop there, because those describe the molecule. Endotoxin is a test of the process, the water and the handling, and many vendors do not commission it. A report with no endotoxin row has made no claim about endotoxin, good or bad. If your work is endotoxin-sensitive (cell culture is the usual case), ask whether the lot was tested, commission a test on the lot you hold, or choose a lot whose report already carries the row.

Sterility is a different test#

Sterility and endotoxin testing are routinely confused, and a report can carry one, both or neither. Sterility, under USP <71>, asks whether living microorganisms can be recovered from the material: the sample is filtered through a membrane that is placed in culture medium, or added directly to medium, and the media are incubated for 14 days and inspected for visible growth. It is qualitative; the material passes if nothing grows.^[10] Endotoxin testing asks whether a molecule left behind by Gram-negative bacteria, alive or dead, is present and how much. Because endotoxin survives the heat and filtration that kill or remove organisms, a sterile preparation can carry a high endotoxin load, and a low-endotoxin preparation can still be non-sterile, for instance with Gram-positive organisms or fungi, which do not make LPS. The FDA's guidance treats pyrogen and endotoxin testing and the sterility test as separate requirements.^[3]

A short checklist for an endotoxin line#

1. The method and chapter are named: gel-clot, turbidimetric, chromogenic or recombinant factor C; USP <85> or <86>, or Ph. Eur. 2.6.14.
2. A "<" value is read as a detection floor set by λ and dilution, not as a measurement.
3. The dilution tested is stated, and the positive product control recovery falls within 50% to 200%.
4. Any specification is understood as the vendor's or laboratory's choice, not a pharmacopeial limit for the compound.
5. If there is no endotoxin row, the report is silent on endotoxin, and a sterility row is not read as a substitute.

Renuvia lots are tested by Freedom Diagnostics Testing for HPLC-UV purity, LC-MS identity and measured content; each report appears in the COA library with the tests it contains, and unissued reports show as pending.