Whitepaper/App Note

Whitepaper: Detection of naturally occurring bacterial endotoxins in water samples

Posted: 23 December 2014 |

A comparison study using two different Kinetic Chromogenic Limulus Amebocyte Lysate (LAL) Assays and three different Endpoint Fluorescent Recombinant Horseshoe Crab Factor C (rFC) Assays…

Endotoxins (chemically lipopolysaccharides) are pyrogenic, heat-stable outer cell membrane-components of Gram-negative bacteria1, such as Escherichia, Klebsiella, Enterobacter and Citrobacter, which occur in the environment, e.g. in water2. A method comparison study was conducted in order to determine the suitability of different commercially available endotoxin tests to quantify naturally occurring endotoxins in 20 different water samples such as deionized water, tap water and samples from environmental sources like rivers, lakes and springs. The determined endotoxin concentrations ranged from 0.05 EU/ml (tap water) to 258.87 EU/ml (lake water). The results from all endotoxin methods correlated well with each other (>94.4%), demonstrating that recombinant Factor C (rFC) assays are suitable alternatives to Limulus Amebocyte Lysate (LAL) assays for testing naturally occurring endotoxins in water.

Materials & Methods

Water Samples: Water samples were collected in sterile and pyrogen-free 50 ml tubes from different locations and sources such as rivers, lakes, springs, tap water, mineral water and deionized water. All samples were passed through 0.45 µm membrane filters. Subsequently, aliquots were prepared and stored frozen at -20°C until testing. 

Kinetic Chromogenic LAL Assays: The two Kinetic Chromogenic LAL assays used in the study were Kinetic-QCL (Lonza Ltd.) and Pyrochrome (Associates of Cape Cod Inc.), both with a labelled sensitivity of 0.005 EU/ml. The horseshoe crab blood coagulation cascade of LAL assays relies on several enzymes. Factor C, the first component in the cascade, is a protease zymogen which is activated by endotoxin. Factor C in turn activates Factor B. An alternative pathway, via Factor G, can be activated by β-glucan binding. Factor C and Factor G pathways individually process a proclotting enzyme into a clotting enzyme. Chromogenic LAL assays use substrates that are cleaved by the clotting enzyme, resulting in a detectable colour reaction3.

Endpoint Fluorescent rFC Assays: The three Endpoint Fluorescent rFC assays used in the study were EndoZyme, with a labelled sensitivity of 0.005 EU/ml (Hyglos GmbH), PyroGene with a labelled sensitivity of 0.01 EU/ml (Lonza Ltd.) and EndoLISA, with a labelled sensitivity of 0.05 EU/ml (Hyglos GmbH). These methods use recombinant forms of horseshoe crab Factor C (rFC)4, 5, converting a substrate into a detectable fluorescence signal. By using rFC, false-positive β-glucan reactions are eliminated and no horseshoe crabs are exploited. The latter rFC assay EndoLISA differs from the other assays due to its additional sample preparation step consisting of a solid phase pre-coated with an endotoxin-specific phage protein which enables the removal of the sample matrix prior to the detection6, 7.

Quantification of endotoxin: The samples were thawed, freshly diluted (1:10) in depyrogenated water and tested in parallel with the five different endotoxin detection assays according to the manufacturer’s instructions. The possibility of inhibition or enhancement of the tests due to interfering substances in the samples was determined. This was accomplished by spiking (adding) known amounts of the respective Control Standard Endotoxin (CSE) of each kit to control samples, so called Positive Product Controls (PPCs). The spiked amount was 5 EU/ml in all cases except for the PyroGene assay where 1 EU/ml spikes were added. Endotoxin concentrations in the samples and PPC recoveries in the control samples were calculated according to the manufacturer’s instructions. A test results was considered valid, when the corresponding spike was recovered in a range of 50% to 200%.

Results & Discussion

The measured endotoxin concentrations in the 1:10 diluted water samples using the five different endotoxin tests in parallel (Table 1) ranged from 0.005 EU/ml (tap water) to 25.887 EU/ml (lake water). As expected, the unpurified lake and river samples contained higher endotoxin levels than purified water. However, the water sampled from a natural spring also contained low levels of endotoxin contaminants. No significant deviation in results was observed between the rFC tests and the LAL tests. The results from all tests were in the same order of magnitude. Moderate deviations with regards to the mean values were noticed, which can have various reasons. One reason could be the presence of β-glucans from other microorganisms (e.g. fungi), which can lead to higher endotoxin concentrations in LAL methods, but not in rFC methods as previously noted, since β-glucan does not activate the rFC test due to the absence of this side reaction. Another cause may be variances in the formulations of the lysates and recombinant enzymes of the different tests. Furthermore, organic (e.g. humins) and/or inorganic (e.g. metal ions) concomitant components in the samples could have direct impact on enzymatic detection systems and result in over- or under-quantification.

Detection of naturally occurring bacterial endotoxins in water samples

Table 1: Measured endotoxin values in 1:10 diluted water samples. Valid test results fulfilled the validity criterion of a Positive Product Control (PPC) recovery of 50 to 200%.  Highlighted in orange are values which did not fulfill the validity criterion (see Table 2). Endotoxin values below the detection limit of each test are listed as <0.05 (EndoLISA), <0.01 (PyroGene) and <0.005 (EndoZyme, Kinetic-QCL and Pyrochrome).

* not calculated

In Table 2, the calculated spike control values for each sample and method are shown. All methods except EndoLISA showed 10-25% invalid spike recoveries due to test inhibition or enhancement. The EndoLISA assay was the only method giving valid spike recoveries for all 20 samples, which can be explained by its heterogeneous test format, enabling higher robustness against sample matrix interferences. Invalid spike recoveries were observed using the other four tests, but the interference was not sample specific. Thus, such effects may also depend on different formulations of the individual detection systems. Interestingly, when comparing the invalid spike recoveries between rFC and LAL methods, contrary interfering effects were observed. The LAL methods tended to enhanced spike recoveries, whereas the rFC methods tended to inhibited spikes in the affected samples. Since native Factor C (LAL) and recombinant Factor C are assumed to be equivalent, the different effects might have been caused by differences in the reaction pathways, e.g. substrate conversion, in the assays.

Detection of naturally occurring bacterial endotoxins in water samples - Table 2

Table 2: Positive Product Controls (PPC, %): Sample spiking was performed by addition of defined amounts of endotoxin standard from the kits to the samples. Subsequently, the EU/ml values were measured and the PPC recovery in % was calculated. Highlighted in orange are values which did not fulfill the validity criterion of 50 to 200% recovery of the added standard endotoxin.

In summary, the correlation of all results obtained with each test method, compared to the mean of each sample, was higher than 94.4% in all cases (Figure 1). As a comparison, previous studies comparing Kinetic Chromogenic LAL with Gel Clot LAL for endotoxin testing in air samples resulted in correlations of as low as 50.3% and 87.2%8. The good correlation of all assays in this study implies that they are suitable for the detection of endotoxin in water samples.

Detection of naturally occurring bacterial endotoxins in water samples - Figure 1

Figure 1: Correlation between the results from the different endotoxin assays in the 20 water samples. The mean EU/ml values for each sample of the five different tests were calculated (invalid results are not included in the mean value). The logarithms of the determined EU/ml values for each detection system were plotted against the logarithm of the mean EU/ml value. The red line represents a theoretical 100% correlation line. Below the graph is a summary of parameters for each method: displayed: the coefficient of correlation (%), the slope and the y-intercept of the linear regression as well as the number of valid samples according to the spiking control experiment.


This study showed that the three rFC tests produced results comparable to the two LAL tests in detecting naturally occurring endotoxins of unknown origin in water samples. As only method, the EndoLISA assay gave valid Positive Product Control results for all 20 samples, which confirmed the benefit of its heterogeneous test format. It can finally be concluded that rFC methods are suitable and sustainable alternatives to the LAL methods for measuring endotoxin in water samples.


  1. Wang, X. et al. Endotoxins: lipopolysaccharides of gram-negative bacteria. Subcell Biochem. 53:3-25. (2010).
  2. Guentzel, MN. Escherichia, Klebsiella, Enterobacter, Serratia, Citrobacter, and Proteus, in Baron’s Medical Microbiology (Baron S et al., eds.), 4th ed., Univ of Texas Medical Branch, USA (1996).
  3. Pfeiffer, M. et al. (eds) Detection of Bacterial Endotoxins with the Limulus Amoebocyte Lysate test. Liss, New York, USA (1987).
  4. Ding, J.L. et al. Endotoxin detection – from limulus amebocyte lysate to recombinant factor C. Subcell Biochem. 53:187-208. (2010).
  5. Nakamura, T. et al. Interaction between lipopolysaccharide and intracellular serine protease zymogen, factor C, from horseshoe crab (Tachypleus tridentatus) hemocytes. J. Biochem. 103(2):370-4 (1988).
  6. Rieth, M. Chapter 7, p. 166 in Pharmazeutische Mikrobiologie. Wiley-VCH, Weinheim, Germany (2012).
  7. Grallert, H. et al. EndoLISA: A novel and reliable method for endotoxin detection. Nature Methods, October (2011)
  8. Górny, R. et al. Application of the Classic Limulus Test and the Quantitative Kinetic Chromogenic LAL Method for Evaluation of Endotoxin Concentration in Indoor Air. Ann Agric Environ Med, 6, 45-51 (1999).  

Related organisations

Send this to a friend