Endotoxin Testing for Research Peptides: Why "99% Pure" Isn't the Whole Story

The 30-second answer: Endotoxin is a molecule shed by certain bacteria. It survives after the bacteria die, so a sterile peptide can still carry it. Tiny amounts switch on immune cells and can skew lab results. Labs measure it with the LAL or rFC test and report it in EU/mg.

Here's something easy to miss. A peptide can test 99% pure and still carry enough bacterial residue to set off immune cells in a dish. Purity tests simply aren't built to see it.

This guide explains laboratory quality testing. It is not guidance on using any product.

In this article you'll learn:

• What endotoxin is, and why "sterile" doesn't mean "endotoxin-free"
• How the horseshoe-crab test and its lab-made rival work
• How to read the endotoxin line on a certificate of analysis (COA)

What is endotoxin?

Endotoxin is a fat-and-sugar molecule from the outer wall of Gram-negative bacteria, such as E. coli. Its chemical name is lipopolysaccharide, or LPS. The toxic part is lipid A, a fatty anchor that sits in the bacterial wall.

Think of endotoxin as the smoke a fire leaves behind. Put out the fire, and the smoke still hangs in the room. Kill the bacteria, or filter them out, and their endotoxin stays in the sample. ^[13]

Your body spots it with a sensor called Toll-like receptor 4, or TLR4. A 1998 Science paper found that mouse strains with broken Tlr4 genes resist endotoxin. ^[6]

Three things make endotoxin a quality problem:

1. It's everywhere. Water, buffers, glassware and human hands all carry these bacteria or their fragments.
2. It outlasts the bacteria. A 2007 review called removing it from biological samples "challenging." ^[13]
3. It's potent. Immune cells react to it at a few trillionths of a gram per millilitre (picograms per mL).

How does endotoxin get into a synthetic peptide?

Making a peptide by chemistry, called solid-phase synthesis, uses no bacteria at all. Endotoxin sneaks in later. It rides in on water, solvents, purification columns, containers and handling.

A 1993 paper in Peptides gave a blunt warning. Endotoxins, it said, "are easily introduced into peptides produced under standard laboratory conditions." It flagged them as a source of false results in physiology studies. ^[7]

Peptides and proteins grown in E. coli start with a much bigger load. There, the factory itself is the source.

Why doesn't a purity test catch it?

HPLC, the standard purity test, is like a race. Every molecule crosses the finish line at its own time, and the lab counts who arrives. It tells you how much of the sample is the right peptide.

But endotoxin is present in trace amounts that HPLC does not detect. Think of a stadium headcount. Counting 10,000 fans won't tell you whether three of them have the flu. Endotoxin needs its own test.

Did you know? A peptide can be ≥99% pure by HPLC and still carry enough endotoxin to activate immune cells. Purity and endotoxin are two separate questions.

How does the horseshoe-crab test work?

Horseshoe crabs have blood cells, called amebocytes, that clot when they touch endotoxin. It's the crab's built-in germ alarm. Lab tests borrow that alarm.

Here's the chain reaction. Endotoxin switches on an enzyme called Factor C. Factor C sets off a cascade of other enzymes. The cascade ends in a gel.

The reagent is called Limulus Amebocyte Lysate, or LAL: crab blood cells broken open. The test is defined in USP chapter <85>, the US standard for bacterial endotoxin testing. ^[1]

FDA accepted LAL in 1977 as a finished-product test. It replaced the older rabbit fever test. ^[4]

LAL comes in three formats. One checks whether a gel forms. One tracks cloudiness. One tracks a color change. If results ever disagree, USP <85> says the gel-clot test settles it. Details are in "The evidence in detail."

What is rFC, and is it as good?

rFC stands for recombinant Factor C. It's a lab-made copy of the crab's first alarm enzyme, built with recombinant DNA instead of harvested blood. When endotoxin switches it on, it cuts a substrate that glows.

It has two practical advantages:

• Fewer false alarms. Crab lysate also contains Factor G, which reacts to β-glucans, sugars from sources like fungi. Glucans don't cause fever, so a Factor G reaction is a false positive. rFC has no Factor G. ^[11]
• No crab harvest. Bleeding horseshoe crabs carries reported mortality of up to 30%. ^[12]

A 2024 study from Korea's National Institute of Food and Drug Safety Evaluation compared four rFC reagents with two LAL reagents. It concluded rFC was "as reliable as the LAL method" for biopharmaceuticals. ^[12]

rFC became an official USP method, in chapter <86>, on May 1, 2025. In March 2026, FDA updated its guidance to make room for recombinant reagents. But FDA says sponsors using recombinant reagents "should verify that the assay method is suitable for its intended purpose." ^[5] Recombinant doesn't mean validation-free.

What does EU/mg mean on a COA?

An endotoxin unit (EU) measures what endotoxin does, not how much it weighs. It's like rating chili peppers by heat instead of grams. Labs calibrate it against a USP reference standard, set at 10,000 EU per vial.

Labs first measure EU per millilitre. Then they convert to EU per milligram of peptide. EU/mg is the useful number because it doesn't change with how much liquid the peptide is dissolved in.

Why does the result say "<" instead of a number?

Things in a sample can block the test. A 2024 study lists pH, salts, detergents, chelating agents and proteins as known culprits. ^[12]

So labs water the sample down until the blocking stops. That dilution sets the lowest value they can report. USP <85> caps how far they can dilute, with a limit called the maximum valid dilution.

That's why a COA says "<0.5 EU/mg" rather than "0." It's like a bathroom scale that can't read below a certain weight. "<0.5" means "below what this test could measure," not zero.

How do you know the test wasn't fooled?

The lab adds a known amount of endotoxin to the sample and checks the test finds it. Think of burying a coin to test a metal detector. This step is the positive product control (PPC), or spike recovery.

USP <85> requires the test to find 50% to 200% of the spike. Outside that window, the result is invalid, whatever the headline number says.

Who sets the limit?

For approved injectable drugs, USP sets endotoxin limits with a formula based on the drug's labeled dose. Research peptides have no approved dose and no USP monograph. So there's no official limit for them.

Any limit on a research-peptide COA is a supplier specification. The COA should state it.

What this means for you: A good endotoxin result names the test, reports EU/mg, shows the spec beside it and prints the spike recovery. A bare "pass" tells you almost nothing.

Why does endotoxin matter in cell studies?

Many immune cells, including macrophages, monocytes and dendritic cells, react to endotoxin through TLR4. Endotoxin in a peptide stock adds a second, uncontrolled signal to every well.

The classic cautionary tale comes from 2003. Researchers saw a lab-made protein, Hsp70, trigger an inflammatory signal (TNF-α) in mouse immune cells. The signal came entirely from endotoxin in the sample. A highly purified batch did nothing. ^[8]

Did you know? In a 2003 study, a protein's "immune effect" turned out to come entirely from bacterial endotoxin in the sample. A cleaner batch of the same protein had no effect. ^[8]

Supplier specs aren't always low enough, either. In a 2014 study, most suppliers of E. coli-made proteins quoted contamination below 1 EU. Yet endotoxin at levels found in some commercial proteins still activated human dendritic cells. The cells responded from 0.02 ng/mL. ^[9]

Tests of inflammatory signals are the most exposed. These include TNF-α and IL-6 assays and NF-κB reporter cells.

What this means for you: If your readout is an immune or inflammatory signal, endotoxin data on the peptide lot isn't optional. It's part of your controls.

Why does endotoxin matter in animal studies?

Endotoxin causes fever and inflammation. In mice it drives cytokine release and a drop in circulating lymphocytes, a type of white blood cell. Many peptide studies measure those same things.

Species differ sharply. In a 2005 study, mice needed about 250 times more endotoxin per kilogram than humans to reach the same IL-6 level. Humans developed fever and a faster heart rate; mice did not. ^[10]

That leads to two design rules:

1. "The animals looked fine" is not a control. Mice can show cytokine and white-cell changes with no visible fever.
2. Strain matters. C3H/HeJ mice carry a Tlr4 mutation that blunts endotoxin signaling. If endotoxin is present, results in that strain will differ from strains with normal TLR4. ^[6]

Did you know? Mice needed about 250 times more endotoxin per kilogram than humans to show the same IL-6 response. A calm-looking mouse can still be reacting. ^[10]

Myth vs fact

What should I look for on a COA?

Use this as your checklist for any research peptide headed for cell or animal work.

And three questions to ask any supplier:

• [ ] Which endotoxin method did you use, and which lab ran it?
• [ ] What's your EU/mg spec, and where is it printed?
• [ ] What was the spike recovery on this lot?

ProGrade COA library

Read next:

• How to read a certificate of analysis
• HPLC vs. mass spectrometry
• What ≥99% purity means
• Peptides explained: how research peptides are made and tested
• Peptide storage and handling

The evidence in detail

For readers who want the technical specifics.

The three LAL formats

USP <85> defines the reagent as Limulus Amebocyte Lysate, extracted from Limulus polyphemus or Tachypleus tridentatus. One USP endotoxin unit equals one international unit (IU) of endotoxin. ^[1]

Where the standards stand (checked October 29, 2026)

Sources: ^[2] · ^[3] · ^[5]

Maximum valid dilution (MVD)

USP <85> caps the dilution with this formula:

MVD = (endotoxin limit × sample concentration) ÷ λ

Here λ is the labeled sensitivity of the reagent. Here are illustrative numbers, not a ProGrade spec. A 5 EU/mg limit, a 1 mg/mL sample and a reagent sensitivity of 0.05 EU/mL give an MVD of 100.

For approved injectable drugs, the USP limit formula is K/M. It is based on a threshold pyrogenic dose of endotoxin (K) and the drug's labeled dose (M).

The key cell studies

Gao and Tsan, 2003. Two preparations of recombinant human Hsp70 were compared. One carried 0.2 ng LPS per µg of protein and drove TNF-α release from mouse macrophages. A highly purified one, at 1.4 pg/µg, did not, even at 5 µg/mL. Polymyxin B, which binds LPS, removed the activity. Their controls make a useful template. They used polymyxin B both as a removal column and added directly to the medium. They also spiked the same LPS into the clean preparation. ^[8]

Schwarz et al., 2014. Human CD1c+ dendritic cells responded to LPS starting at 0.02 ng/mL, at levels found in some commercial recombinant proteins. ^[9]

---

For research use only. Not for human consumption. These statements have not been evaluated by the Food and Drug Administration. These products are not intended to diagnose, treat, cure, or prevent any disease. Products are sold for research purposes only. This article explains laboratory quality testing and is not medical advice.

---

References

1. United States Pharmacopeia. <85> Bacterial Endotoxins Test (USP 29 text, archived; method descriptions, MVD, 50–200% recovery, gel-clot referee, USP-EU = IU). www.uspbpep.com/usp29/v29240/usp29nf24s0_c85.html
2. USP. Bacterial Endotoxins: PDG harmonization page (official date 1 Dec 2012; gel-clot limit test as referee). www.usp.org/harmonization-standards/pdg/general-methods/bacterial-endo…
3. USP. Expert Committee approves endotoxin testing using non-animal derived reagents. Jul 26, 2024. www.usp.org/news/expert-committee-approves-endotoxin-testing-using-non…
4. FDA. Guidance for Industry: Pyrogen and Endotoxins Testing: Questions and Answers. June 2012. www.fda.gov/files/drugs/published/Pyrogen-and-Endotoxins-Testing--Ques…
5. FDA. FDA Clarifies Current Thinking on Pyrogen and Endotoxins Testing (Edition 2). Mar 18, 2026. www.fda.gov/science-research/advancing-alternative-methods-fda/fda-cla…
6. Poltorak A, et al. Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene. Science. 1998;282(5396):2085–2088. PMID 9851930. doi.org/10.1126/science.282.5396.2085
7. Majde JA. Microbial cell-wall contaminants in peptides: a potential source of physiological artifacts. Peptides. 1993;14(3):629–632. PMID 8332558. doi.org/10.1016/0196-9781(93)90155-a
8. Gao B, Tsan MF. Endotoxin contamination in recombinant human heat shock protein 70 (Hsp70) preparation is responsible for the induction of tumor necrosis factor alpha release by murine macrophages. J Biol Chem. 2003;278(1):174–179. PMID 12403778. doi.org/10.1074/jbc.M208742200
9. Schwarz H, Schmittner M, Duschl A, Horejs-Hoeck J. Residual endotoxin contaminations in recombinant proteins are sufficient to activate human CD1c+ dendritic cells. PLoS One. 2014;9(12):e113840. pmc.ncbi.nlm.nih.gov/articles/PMC4257590/
10. Copeland S, et al. Acute inflammatory response to endotoxin in mice and humans. Clin Diagn Lab Immunol. 2005;12(1):60–67. pmc.ncbi.nlm.nih.gov/articles/PMC540200/
11. Bolden J, et al. Currently available recombinant alternatives to horseshoe crab blood lysates: are they comparable for the detection of environmental bacterial endotoxins? A review. PDA J Pharm Sci Technol. 2020;74(5):602–611. PMID 32817324. doi.org/10.5731/pdajpst.2020.012187
12. Kang DH, et al. A study on the application of recombinant factor C (rFC) assay using biopharmaceuticals. Microorganisms. 2024;12(3):516. pmc.ncbi.nlm.nih.gov/articles/PMC10972375/
13. Magalhães PO, et al. Methods of endotoxin removal from biological preparations: a review. J Pharm Pharm Sci. 2007;10(3):388–404. PMID 17727802. pubmed.ncbi.nlm.nih.gov/17727802/