Endotoxin, a novel biomarker for the rapid risk assessment of faecal contamination of coastal and transitional waters

Water-based recreational activities are important for both mental and physical health (Overbury et al. 2023). Additionally, local economies benefit greatly from the tourism generated by such activities, with Europeans estimated to spend over €800 billion annually on recreational visits to open waters (Börger et al. 2021). Unfortunately, microbial pollution from various sources, such as sewage discharge and agricultural runoff, poses a substantial risk to public health. Bathers are 3.3 times more likely to show signs of gastrointestinal illness, respiratory infections, skin infections, ear and eye infections than non-bathers (Leonard et al. 2020). The risk of infection is made more serious by the increased prevalence of antimicrobial resistance associated with wastewater (Fouz et al. 2020). Furthermore, the perception of poor water quality costs an estimated €100 billion in lost annual revenue in Europe (Börger et al. 2021), as water quality has been identified as the primary reason for tourists' choice of destination (Dodds & Holmes 2018). Therefore, monitoring bathing water quality is vital to mitigate the health risks associated with the recreational use of bathing waters and to improve public perception and confidence in water quality.

Coastal water quality is monitored for faecal pollution using faecal indicator bacteria (FIB). The most widely used indicators are Escherichia coli and intestinal Enterococci. These bacteria are normally found in the intestinal tract of warm-blooded animals, including humans. When present in water, they indicate faecal contamination. The majority of FIB are not pathogenic but indicates the presence of faecal contamination that is a potential source of pathogenic organisms such as Salmonella, Shigella, Clostridium, Legionella, Yersinia, Mycobacterium, adenovirus, norovirus, rotavirus and coronavirus (Motlagh & Yang 2019), helminths and fungal species (Shah et al. 2011). Using FIB as a proxy for other potentially more harmful pathogens eliminates the need for multiple tests, saving resources.

In the EU, bathing water quality is currently regulated by monitoring E. coli and intestinal Enterococci (Bathing Water Directive, BWD, 2006/7/EC), and locations are given an annual classification of Excellent, Good, Sufficient or Poor based on the levels of these indicator bacteria present in the waters (Table 1) (Bathing Water Regulations 2013).

Culture-based methods to enumerate FIB are the current gold standard and have been used since the mid-20th century (Dufour 2021). Although these methods are well established, they have several major drawbacks. The most notable issue is that they require growth of the organism that takes at least 18–44 h, meaning that any result obtained is retrospective and delays proactive measures to address concerning polluted waters. In addition, the current methods require sample transportation to specialised laboratories and trained personnel to carry out these methods in a reproducible way. Moreover, by focussing solely on FIB, it is possible that other pathogens that are not necessarily of faecal nature may go undetected (Topić et al. 2021). These drawbacks to the current culture-based approaches highlight the need for not only rapid, but also more complete solutions to better indicate a broader set of pathogens.

Several investigators have suggested that endotoxin may be a useful biomarker for rapidly determining bacterial biomass and water quality (Jorgensen et al. 1973; Watson et al. 1977; Evans et al. 1978; Jorgensen et al. 1979; Haas et al. 1983). Endotoxin is the lipopolysaccharide present in the outer membrane of Gram-negative bacteria and some cyanobacteria. Previous work by our group has shown the applicability of using endotoxin as a marker of faecal contamination of seawater (Sattar et al. 2013) and surface water (Sattar et al. 2022) due to the high concentrations of Gram-negative bacteria, notably coliforms, in the natural gut microbiota and in sewage discharge (Holcomb & Stewart 2020). The team at Molendotech have developed a portable near real-time assay (Bacterisk) to detect endotoxin in water, which can be conducted by non-specialist staff in situ.

This study was undertaken to evaluate the performance of the Bacterisk assay as a rapid risk assessment tool to assess the quality of coastal and inland bathing waters in the southwest of England. This was achieved by testing samples in parallel using established culture-based methods for FIB, E. coli and intestinal Enterococci, as well as with Bacterisk.

A total of 159 coastal and transitional water samples were collected from various locations in the southwest of England according to the Bathing Water Regulations (2013). Briefly, 500 mL of samples were taken using sterile bottles 30 cm below the water's surface in water at least 1 m deep. The samples were then transported in the dark and tested within 4 h or stored in a fridge (2–8 °C) and tested no later than 24 h after collection.

Appropriate volumes of each water sample (1, 10, and 100 mL) were aseptically filtered through a 0.45 μm membrane (Whatman, UK) using a 6-branch vacuum manifold (Sartorius, UK). Membranes were placed on membrane lactose glucuronide agar (MLGA) (Oxoid, UK) and incubated at 30 °C for 4 h, then at 37 °C for 14 h for the detection of presumptive E. coli or on Slanetz and Bartley medium (Oxoid, UK) and incubated at 36 °C for 44 h for the detection of presumptive intestinal Enterococci. The numbers of colony-forming units (CFUs) were then calculated and expressed as CFU/100 mL.

The Bacterisk assay (Molendotech Ltd, Brixham, UK; www.bacterisk.com) was performed following the manufacturer's instructions. Briefly, the samples were diluted 1 in 100 in dilution buffer and then combined with the detection reagent in equal parts in endotoxin-free micro-cuvettes at a final volume of 200 μL. The samples were then incubated at 37 °C for 25 min using the integrated Bacterisk incubator and the reader. An endotoxin risk (ER) score was then calculated by the device based on the absorbance of the sample at 405 nm. Some samples were incubated and read using a Biotek ELx808 plate reader (Agilent) for increased throughput. Both a negative control (endotoxin-free water) and a positive control (endotoxin from E. coli 055:B5) were run with every assay.

Endotoxin activity (EU/mL) was determined using a Kinetic Chromogenic Limulus Amoebocyte Lysate (LAL) assay according to the manufacturer's instructions.

Statistical analyses on the distribution of samples and the correlation between Bacterisk data and culture-derived water quality were performed with Fisher's exact test and the Chi-square test using GraphPad prism version 9 for Windows, GraphPad Software, Boston, Massachusetts, USA (www.graphpad.com).

A total of 159 coastal and transitional water samples were analysed in parallel by the Bacterisk assay to calculate ER and by membrane filtration to enumerate the levels of E. coli and intestinal Enterococci (CFU/100 mL). The samples were further classified into two water quality groups, ‘sufficient or better’ or ‘poor’, based on the levels of E. coli (CFU/100 mL). The threshold chosen for ‘poor’ water quality was >500 E. coli CFU/100 mL based on the Bathing Water Directive 2006/7/EC. Using this threshold, there were a total of 107 ‘sufficient or better’ samples with a median ER of 3,696, and 25th and 75th percentiles of 2,219 and 6,108, respectively, and 52 ‘poor’ samples with a median ER of 12,475, and 25th and 75th percentiles of 8,827 and 20,316, respectively. Though intestinal Enterococci are also used as FIBs, they are Gram-positive bacteria. Therefore, they were not used to classify water samples but instead to understand if there was a correlation between intestinal Enterococci and ER.

Of the 159 water samples analysed, there was strong concordance between the Bacterisk ER results and both E. coli and Enterococci culture results (TP E. coli = 42; TP Enterococci = 46; TN E. coli = 94; TN Enterococci = 91; OP E. coli = 21; OP Enterococci = 17; FN E. coli = 2, FN Enterococci = 5).

Using the ER threshold value, a contingency table was produced and analysed using Fisher's exact test to examine the distribution of samples and the correlation between the chosen threshold and the two categories of water quality (Table 2), showing that ER is dependent on water quality (p < 0.0001).

To further improve the characterisation of water samples into each water quality group, three risk groups were created with different ER thresholds. These three risk groups are comprised of low, medium and high based on the likelihood of the sample being characterised as ‘poor’. The low-risk group used the original threshold determined by the ROC analysis. These risk groups were put into a contingency table (Table 3) and analysed using a Chi-square test for trend (X² = 88.93, degrees of freedom = 1, p-value = <0.0001).

Faecal pollution of coastal waters impairs water quality and poses a serious health threat by promoting the spread of infectious diseases among humans and marine ecosystems. With the prevalence of illness in coastal bathers being 3.3 times higher than that of non-bathers (Leonard et al. 2020), it is unsurprising that the public's perception of poor water quality has a detrimental effect on the economy (Börger et al. 2021). Effective monitoring of faecal pollution is thus essential to protect human health and prevent or mitigate the spread of faecal pollution to the ocean and restore public confidence.

Current routine methods for testing water for faecal contamination detect the indicator bacteria E. coli and intestinal Enterococci. This requires the transportation of water samples to qualified laboratories, sample filtration, incubation, and biochemical tests and identification by specialist staff. This is a time-consuming process often taking 18–44 h to obtain results. The results are therefore retrospective, which inhibit proactive measures to be taken concerning polluted waters. While this approach can be used to develop a risk profile for a designated bathing location over time, it cannot give information on water quality at the present time, and this is a major limitation considering that the levels of FIB are known to vary throughout the day (Kim et al. 2009).

Previous work by our group has shown the applicability of using endotoxin as a marker of faecal contamination of seawater (Sattar et al. 2013) and surface water (Sattar et al. 2022). The Bacterisk assay, which detects the bacterial biomarker endotoxin, was developed to address the limitations of current culture-based testing. It is portable and produces results in under 30 min, allowing for near real-time monitoring of water quality. This makes it possible to gather information for daily beach management decisions. Additionally, it can notify the public on whether the water is suitable for bathing prior to a major social event (WHO 2018).

Bacterisk gives a measure of endotoxin as ‘ER’ units that correlate precisely with endotoxin quantity. Results presented here show that Bacterisk can effectively monitor coastal water quality and the results have significant correlation with levels of the FIB E. coli and Enterococci. The significant correlation with Enterococci, a Gram-positive bacterium that does not contain endotoxin, was surprising but most probably reflects the detection of Gram-negative bacteria that are also present in faecal contamination. ROC analysis determined the optimal ER threshold for coastal and transitional water samples of ≥7,300, which corresponds to a sensitivity of 95.45% and a specificity of 83.48%. Using this threshold, we found a significant correlation between Bacterisk ER values and the presence of E. coli or Enterococci. When looking at E. coli, 60.38% of the samples tested were classified as ‘sufficient or better’, with 26.42% being classified as ‘poor’. Of the samples, 11.95% were detected as positive by Bacterisk but negative by culture (‘OP’). This is due to Bacterisk detecting high levels of endotoxin produced by Gram-negative bacteria or Gram-negative bacteria in a VBNC state. The issue of VBNC bacteria can result in significant underestimation of viable cell counts when assessing bathing water quality. Enterotoxigenic E. coli has been shown to enter a VBNC state during stressful conditions yet remain infectious after long-term incubation in both sea and fresh water (Lothigius et al. 2010). This VBNC state would therefore negatively impact the accuracy of any assay that uses the enumeration of E. coli as an indicator of water quality. Bacterisk can therefore not only alert to the presence of VBNC bacteria but also other potential pathogens such as Salmonella spp. that have been found in bathing waters classified as ‘good’ by current indicator detection methods (Mansilha et al. 2010). Likewise, pathogens such a Aeromonas ssp. that are not necessarily of faecal nature and would not be detected by current methods either directly or as a proxy (Janda & Abbott 2010) but can be detected by Bacterisk due to their expression of endotoxin (Magaña et al. 2013).

Endotoxin may also be derived from non-pathogenic bacteria; however, our data show that for the ‘sufficient or better’ category, we see a median ER value of 3,696 compared to 12,475 for the ‘poor’ category, with 25th and 75th percentile values of 2,219 and 6,108, respectively, for the ‘sufficient or better’ category and 8,827 and 20,316, respectively, for the ‘poor’ category. This indicates that a background level of endotoxin from Gram-negative bacteria is detected in coastal and transitional waters, though it is during a pollution event that we observe a significant increase in endotoxin and not merely from fluctuations in the normal microbial community.

Of the 159 samples tested, two were detected as negative by Bacterisk but positive by culture giving an FN rate of 4.5%. These FN results could be due to a phenomenon known as low endotoxin recovery that can be due to the presence of surfactants, chelating agents or heavy metal ions in high concentrations leading to the disruption of endotoxin aggregates (Gorman & Golovanov 2022). The Bacterisk method incorporates a 100-fold sample dilution that is specifically optimised to circumvent this problem, except extreme cases (Milton et al. 1992). It is worth noting that the current standard method, Colilert-18^®, was shown to have an FN rate of 7.3% when testing bathing waters in Finland (Tiwari et al. 2016), making the Bacterisk FN rate superior.

It is worth mentioning that many of the current reference methods for the selective isolation and enumeration of E. coli rely on the detection of the enzyme β-D-glucuronidase such as chromogenic agar (MLGA, MI, and Chromocult Coliform^®) and different formulations of Colilert^® (notably Colilert-18^®). This reliance on detecting β-D-glucuronidase can result in high FN rates due to the absence or low levels of the enzyme in up to 34% of E. coli strains in human faeces (Chang et al. 1989) and 10–20% of E. coli strains isolated from environmental sources (Shadix & Rice 1991). Furthermore, as MLGA was used to determine the number of presumptive E. coli in this study, it may be that Bacterisk results that were classified as ‘OP’ may, in fact, be TP. Therefore, as all strains of E. coli, regardless of β-D-glucuronidase activity, express endotoxin, Bacterisk will be able to detect them.

In conclusion, we have shown that the Bacterisk assay provides a rapid and easy to use in situ method to assess bathing water quality. The strong statistical corelation between Bacterisk results and traditional culture results for FIB will give confidence for using Bacterisk as a rapid screening tool for coastal bathing water quality. With an estimated annual loss of €100 billion in European tourism resulting from the perception of poor water quality, the need for improved infrastructure and testing is vital to bolster public confidence in bathing water quality. The economic and health value of clean recreational waters combined with record fines for polluting companies has stimulated a desire for new, rapid testing solutions to monitor and risk assess bathing waters. By providing a near real-time assessment of water quality, Bacterisk will allow regulators and companies responsible for bathing water quality the ability to take immediate action to protect public health and the environment.

Financial support for J. Brandão was provided by CESAM (UID/AMB/50017-POCI-01-0145-FEDER-007638) and CITAB (UID/AGR/04033/2019) via FCT/MCTES using national funds (PIDDAC). This support was also cofounded by FEDER (PT2020 Partnership Agreement and Compete 2020).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.