Endotoxin Testing Explained: USP <85> and Why It Matters for Reconstituted Peptides

A peptide certificate of analysis with a 99.5% HPLC purity number on it tells you what peptide-related impurities the lot does and does not contain. It tells you nothing about endotoxin. Endotoxin is bioactive at picogram concentrations — roughly four orders of magnitude below the lowest impurity an HPLC chromatogram can resolve at normal sample loadings — and it survives the standard autoclave cycle that sterilizes most labware^[5][14]. The only way to know whether a finished peptide lot carries an acceptable endotoxin burden is a separate bacterial endotoxins test (BET), measured against a compendial chapter, on the finished material.

For research peptides that will be reconstituted and injected — even at the animal-model scale — the relevant compendial chapter is USP <85> Bacterial Endotoxins Test, formally interchangeable with Ph.Eur. 2.6.14 and JP 4.01 under ICH Q4B Annex 14^[6]. The newer USP <86>(official May 2025) elevates recombinant reagents from an “alternative method” to a compendial standard^[2]. This article is a reading guide to both, the calculation that turns a per-kilogram pyrogenic threshold into a per-mL drug specification, and the reconstitution-water choices that determine the second half of the contamination story.

Why endotoxin is the contamination HPLC cannot see

Three properties separate endotoxin from peptide-related impurities and explain why it has its own compendial chapter rather than living inside the purity assay.

• Picogram-scale bioactivity. One endotoxin unit (1 EU) is approximately 100 pg of LPS. HPLC at typical UV/ELSD/MS sample loadings cannot resolve impurities at that concentration; endotoxin contamination is invisible to the purity number.
• Heat stability.Endotoxin is heat-stable through standard autoclaving (121 °C, 15 min). Depyrogenation requires dry heat (typically ≥250 °C) or validated ultrafiltration / charge-based removal^[5].
• Filtration does not remove it. A 0.22 µm sterile filter retains bacteria but does not retain LPS, which is far smaller. Sterile filtration sterilizes a solution; it does not depyrogenate it.

What endotoxin actually is

Endotoxins are lipopolysaccharides (LPS) shed from the outer membrane of Gram-negative bacteria. The bioactive moiety is lipid A, which is recognized by TLR4 and triggers the innate-immune cascade responsible for the fever, hypotension, and disseminated intravascular coagulation seen in Gram-negative sepsis. From the compendial-testing standpoint, two practical features dominate:

• The biological response is dose-dependent and triggers at low picogram concentrations, which is why the per-kilogram pyrogenic-dose threshold (K = 5 EU/kg/h for non-intrathecal parenteral) drives the limit calculation^[1].
• The aggregation state of LPS in solution influences detection. The compendial methods deliberately use reference-standard LPS (RSE, CSE) calibrated to defined potency to make plate-to-plate, lab-to-lab comparison meaningful.

Bang, Levin, and the horseshoe-crab story

The story behind LAL is one of the cleanest accidents in clinical chemistry. Frederick Bang, working at the Marine Biological Laboratory at Woods Hole in the 1950s, observed that Limulus polyphemus— the Atlantic horseshoe crab — suffered fatal intravascular coagulation when challenged with Gram-negative infection^[8]. Together with Jack Levin, he subsequently identified endotoxin as the trigger and proposed that horseshoe-crab amebocyte lysate could be used as a sensitive diagnostic for endotoxin contamination^[9]. The amebocyte serine-protease cascade became the basis of the Limulus Amebocyte Lysate (LAL) test that has dominated parenteral-drug release testing since.

Two consequences of this origin are worth keeping in mind. First, the assay is a biological coagulation cascade, not a chemical reaction; method validation must account for matrix interference (the “inhibition / enhancement” check in USP <85>). Second, every LAL test is, in the strict sense, a horseshoe-crab reagent test, and that fact has driven the pharmacopeial movement toward recombinant Factor C reagents on both sustainability and assay-determinism grounds^[11].

USP <85>: the harmonized compendial chapter

USP <85> is the U.S. compendial chapter for the bacterial endotoxins test. Its text is harmonized with Ph.Eur. 2.6.14 and Japanese Pharmacopoeia 4.01 under ICH Q4B Annex 14, meaning that a BET measurement performed under any of the three chapters is formally interchangeable across the U.S., EU, and Japanese regulatory regions^[1][6]. The chapter recognizes three LAL-based methods, defines a reference-standard endotoxin (RSE) traceable potency unit, and defines the maximum valid dilution (MVD) framework that products with intrinsic matrix interference must operate within.

The companion informational chapter USP <1085>(and its 2020 proposal <1085.1>) provides guidance on test selection, reagent choice (LAL vs rFC), and method validation under USP General Notices 6.30^[3]. USP <86>, official May 1, 2025, elevates recombinant reagents to compendial status^[2]. The combined effect is that, as of 2026, a parenteral product can be released against the endotoxin specification using either LAL methods (USP <85>) or recombinant methods (USP <86>) without an alternative-method justification.

The three LAL methods (and rFC)

Four BET methods are currently in routine use across compendial laboratories. The differences are not academic: detection range, susceptibility to optical interference, and per-test cost shape which method ends up on a contract laboratory’s scope of accreditation^[14].

Two practical consequences:

• Gel-clot is the referee method. When a kinetic LAL or rFC result is contested, the gel-clot endpoint reading is the compendial tiebreaker. Gel-clot is also the lowest-cost method and the least susceptible to matrix interference.
• Kinetic chromogenic LAL is the biologics workhorse.Its broad dynamic range (typically 0.005–50 EU/mL) makes it the standard release-test format for parenteral biologics, including peptides reconstituted at typical research-dose concentrations.

USP <86> and recombinant reagents

Recombinant Factor C was cloned by Ding, Ho, and colleagues from the Singapore horseshoe crab Carcinoscorpius rotundicauda in 1995^[10]. The practical advantage over native LAL is twofold: it removes the dependence on annual horseshoe-crab bleeding (estimated at >500,000 animals annually in the U.S.^[11]), and it eliminates Factor G — the side branch of the amebocyte cascade that is responsible for β-glucan false positives in LAL. The FDA accepted rFC on a case-by-case alternative-method basis from 2018; Ph.Eur. 2.6.14 added Method G (rFC fluorimetric) effective 2021; and USP <86>, official May 1, 2025, makes the recombinant approach compendial^[2][4][11].

The endotoxin-limit calculation (K/M)

USP <85> defines the endotoxin limit as a function of dose, body weight, and a per-kilogram pyrogenic-dose constant K. The formula is:

Endotoxin limit = K / M
K = 5 EU/kg/h  (non-intrathecal parenteral)
K = 0.2 EU/kg/h  (intrathecal)
M = maximum dose per kg per hour

The maximum valid dilution (MVD) caps how heavily a sample can be diluted before the method’s sensitivity λ becomes the limiting factor. Sensitive lysate (low λ) is essential when product matrix self-interferes and must be diluted heavily — a common situation with concentrated peptide solutions.

A worked example

Step 1 — M (dose per kg per hour):

M = (1 mg/mL × 0.5 mL) / 70 kg = 0.5 mg / 70 kg = 0.00714 mg/kg

Step 2 — Endotoxin limit:

Limit = K / M = 5 EU/kg ÷ 0.00714 mg/kg ≈ 700 EU/mg of peptide

Equivalently, the assembled 1 mg/mL solution at ≤700 EU/mL would meet the compendial threshold, and the 0.5 mL injected dose would carry up to 350 EU at the threshold (= 5 EU/kg × 70 kg).

In practice, finished sterile injectables are released far below this value — typical biologic specifications cluster at 0.5–10 EU/mg — because (a) repeat or multi-component dosing compounds the load, (b) endotoxin assays carry roughly 2× method variability, and (c) MVD considerations require headroom. A research peptide CoA reporting <10 EU/mg by kinetic chromogenic LAL or rFCsits comfortably inside the USP <85> envelope for the worked example above and is the practical benchmark to look for.

Endotoxin ≠ pyrogen

A common reading error is to treat “endotoxin” and “pyrogen” as synonyms. They are not. Endotoxin (Gram-negative LPS) is a subset of pyrogens. Other pyrogenic substances — Gram-positive cell-wall components (lipoteichoic acid, peptidoglycan), viral particles, fungal cell-wall β-glucans — trigger the same febrile response in vivo but are not detected by LAL or rFC^[5].

For a finished injectable where the intended population is human and where non-endotoxin pyrogenic contamination is a real possibility, the relevant tests are:

• Rabbit Pyrogen Test (RPT): the historical reference; in vivo; retained in some compendial contexts.
• Monocyte Activation Test (MAT): Ph.Eur. 2.6.30; an in-vitro alternative that uses human blood monocytes to detect both endotoxin and non-endotoxin pyrogens^[13].

For research peptides, a CoA that reports BET (USP <85> or USP <86>) but not RPT/MAT is normal and adequate: the BET result captures the dominant contamination risk in a synthetic peptide reconstituted in compendial water. RPT and MAT come into play for biologically derived materials, vaccine intermediates, and finished injectable products where non-endotoxin pyrogen risk is structural.

Bacteriostatic vs sterile water — what they are and are not

For reconstituted peptides, the diluent is the second contamination vector and the one researchers most often misread.

• Sterile Water for Injection USP— preservative-free, endotoxin specification < 0.25 EU/mL. Appropriate for single-use reconstitution where the assembled solution will be administered immediately or stored for a short window^[12].
• Bacteriostatic Water for Injection USP— contains 0.9% (9 mg/mL) benzyl alcohol as preservative. Inhibits microbial regrowth after the first needle stick and is therefore the correct choice for multi-dose vials reconstituted over several days. Contraindicated in neonates and at large cumulative volumes due to benzyl-alcohol toxicity.

Reading endotoxin numbers on a peptide CoA

Translating the compendial framework to what should appear on a research peptide certificate of analysis:

• Method.Either “Kinetic chromogenic LAL per USP <85>” or “Recombinant Factor C per USP <86>” is appropriate. Gel-clot alone is acceptable but qualitative. The method should be named, not implied.
• Result. Reported in EU per mg of peptide (preferred) or EU per mL at the tested concentration. A single qualitative pass/fail without a quantitative number is less informative than a measured EU/mg.
• Specification.<10 EU/mg is a reasonable benchmark for research peptides intended for animal-model parenteral use, sitting comfortably inside the K/M envelope for typical reconstitution scenarios. Tighter specifications (<1 EU/mg, <0.5 EU/mg) are normal for clinical-grade biologics and are achievable for synthetic peptides; they cost more.
• Laboratory accreditation. A contract laboratory performing BET should hold ISO/IEC 17025:2017 accreditation with BET on its scope of accreditation^[7]. The accreditor (ANAB, A2LA, UKAS, IAS) maintains a public registry that lists every accredited lab and its accredited methods — a five-minute verification step.
• Validation.USP <85> requires demonstration of inhibition/enhancement on each new product matrix. A CoA that reports a BET result without indicating that the matrix has been validated for non-interference is incomplete — the sample concentration tested must be within the maximum valid dilution.

Further reading

For broader context on what every line on a peptide certificate of analysis means — HPLC chromatograms, mass-spectrometric identity, water content, and the fields most stores omit — see our companion methods articles on how to read a peptide certificate of analysis and how we verify peptide purity. For the analytical conventions behind the headline percentage, what ≥99% purity actually means addresses the framing question that BET sits alongside.

The primary documents referenced above — USP <85>^[1], USP <86>^[2], USP <1085>^[3], the FDA 2012 pyrogen-and-endotoxins Q&A guidance^[5], ICH Q4B Annex 14^[6], and ISO/IEC 17025:2017^[7]— are the operational sources. Williams (2007) is the standard textbook on LAL and depyrogenation methodology^[14].