Modification of the H₂S test to screen for the detection of sulphur- and sulphate-reducing bacteria of faecal origin in water

Water-related diseases are one of the major obstacles to the improvement of people's health, especially in low income countries. Worldwide, there are 785 million people without access to an improved drinking water source, and at least 2 billion people use a drinking water source contaminated with faeces (WHO 2019). It has been estimated that 88% of all incidents of diarrhoea worldwide are caused by microbiologically unsafe water, affecting mostly children below the age of five years (WHO 2008). The World Health Organization (WHO) estimated that inadequate drinking-water sources and the lack of adequate sanitation and hygiene cause 842,000 diarrhoeal disease related deaths per year, especially in low income countries (WHO 2015). The largest increase in water deterioration is expected to occur in low income countries, because of a subsequent increase in population and economic growth, which is especially the case in sub-Saharan Africa (WWAP 2016). Therefore, SDG target 6.1 aims to achieve universal and equitable access to safe and affordable drinking water (or safely managed drinking-water water) for all by 2030 (WHO 2017a). UNEP highlights the very low density of water quality monitoring facilities and laboratories in general in low income countries, as well as the significant inconsistency between global assessment regulations and regional knowledge needs (UNEP 2015).

Millions of people have to rely on unimproved drinking water sources. Many low-resource settings lack adequate finances, reliable energy sources, qualified technicians, and laboratory reagents. Consequently there is a need for straightforward and affordable microbial field tests to substitute for more sophisticated laboratory procedures (McMahan et al. 2011). The H₂S test has the potential to be an affordable alternative in comparison to more sophisticated laboratory-based methods (Bain et al. 2012).

The H₂S test was introduced by Manja et al. (1982) to assess the microbial quality of drinking-water in low-resource settings by testing for hydrogen sulphide (H₂S) producing bacteria. It was promoted as a promising alternative to existing technologies, but the original version lacked sensitivity and especially specificity to the faecal indicator bacteria (FIB) used in microbial drinking-water analysis (Sobsey & Pfaender 2002). Any source of H₂S or the presence of sulphur-reducing and sulphate-reducing bacteria (SRB) in the sample can lead to a false positive test result (Huang & Zira 2011; Sobsey & Pfaender 2002). In contrast, the presence of faecal coliforms with limited ability to reduce sulphate and the absence of related sulphate-reducing bacteria may lead to false-negative results, one of the most concerning limitations of the H₂S test (Sobsey & Pfaender 2002; Huang & Zira 2011). For example, water of poor microbiological quality could be falsely classified as being of acceptable quality on the basis of the existing H₂S test. False negative results, especially when indicator densities are relatively high, are a major cause for concern (Nair et al. 2001). Therefore, further research is needed in order to understand the conditions that lead to this misleading result.

The lack of available alternatives resulted in the H₂S test being promoted and used by many governmental, non-governmental and research organisations including the WHO, UNICEF, USAid, WaterAid, and ACF International, operating in rural communities or in low-resource and emergency settings (Oxfam 2006; UNICEF 2008; WHO 2009, 2010; IFRC 2011; ACF International 2013; WaterAid 2014; USAid 2015; Peletz et al. 2016; Matwewe et al. 2018). Although alternative field tests or potable incubators for bacteriological water quality analysis are available (e.g. Colilert, CBT, DelAgua) their cost is much higher when compared to the H₂S test and access to these tests and their consumables in low-resource settings is difficult.

Previous studies investigating alternative modifications (Venkobachar et al. 1994; Grant & Ziel 1996; Manja et al. 2001; Pant et al. 2002; Pathak & Gopal 2005; Tambekar et al. 2007; Luyt et al. 2012; Khush et al. 2013; Shahryari et al. 2014; Kejariwal et al. 2018) or the microbial sensitivity and specificity of the H₂S test (Jacobs et al. 1986; Dutka & El-Shaarawi 1990; Kromoredjo & Fujioka 1991; Desmarchalier et al. 1992; Castillo et al. 1994; Martins et al. 1997; Nair et al. 2001; Gupta et al. 2008; Izadi et al. 2010; McMahan et al. 2012; Wright et al. 2012 [meta-study]; Khush et al. 2013; Yang et al. 2013; Weppelmann et al. 2014; Murcott et al. 2015; Tambi et al. 2016; Matwewe et al. 2018; Malema et al. 2019) missed assessing the test's performance and its level of accuracy by using a range of known concentrations of confirmed and pure cultured species of organisms (confirmed to subspecies level). Further, the level of sensitivity and specificity was not performed according to accepted method validation protocols. Although McMahan et al. (2012) investigated the specificity of the H₂S test by application of PCR testing, the samples were of environmental origin and not pure cultured. Most studies indicate variability in performance of the H₂S test (Wright et al. 2012), which justifies further investigation.

Given that the use of this method is unlikely to change in the near future and the test's simplicity and low cost makes it popular with many NGOs, efforts are required to improve the H₂S test's microbial sensitivity and specificity. Therefore, the objectives of this research were to (i) evaluate modifications to the culture media of the original H₂S test with the aim of increasing its microbial sensitivity and specificity to E. coli, (ii) assess the sensitivity and specificity of a range of the modifications, and (iii) identify the specific microorganisms that react in the H₂S test and its modifications using confirmed pure culture organisms.

In order to evaluate the performance of the H₂S test and its modifications in terms of sensitivity and specificity to faecal indicator bacteria and faecal and environmental sulphate- and sulphur-reducing bacteria (SRB), the H₂S test's operational range and limit of detection (LOD) was investigated. The H₂S test and each of the 20 modifications were challenged with culture collection strains of known enteric and non-enteric sulphide-, sulphate-, sulphur-reducing, and H₂S-producing bacteria. The incubation period (time in h) to produce a positive reaction (black precipitate caused by the production and reaction of H₂S with ferric iron Fe₂O₃) for each formulation was recorded. In addition, some of the H₂S test modifications were validated against the accepted standard methods for drinking-water analysis as outlined in APHA (2012).

The H₂S test culture medium was prepared according to the original description by Manja et al. (1982). It contains: bacteriological peptone 20 g, dipotassium hydrogen phosphate 1.5 g, ferric ammonium citrate 0.75 g, sodium thiosulphate 1.0 g, liquid detergent (Teepol) 1.0 ml, and distilled and deionised water 50 ml. All heat-resistant components were sterilized by autoclaving before use. 0.5 ml of the medium was placed on a pre-cut sterile filter paper and placed into an oven at 55 °C for 30 minutes to dry. Each strip was then placed in a sterile 10 ml plastic culture tube.

The selective reagents considered to support or inhibit the growth of certain bacteria are shown in Table 1. Sulphur sources of different reduction stages were used, given that the process of sulphur assimilation and dissimilation by most enteric bacteria has rarely been studied nor is it properly understood (La Faou et al. 1990; Cai et al. 2019). All reagents were procured from Fisher Scientific UK Ltd and Sigma-Aldrich^®. Filter papers used for each of the various H₂S tests were produced by Fioroni S.A. in France. Bile salts (LP0055) and bacteriological peptone (LP0037) were manufactured by Oxoid™.

Each of the modified H₂S tests culture media formulae investigated in this study and the rationale for each culture media's modification are outlined in Table 1.

Most bacterial strains used in the testing and analysis were supplied by the National Collection of Type Cultures (NCTC) and the American Type Culture Collection (ATCC^®) as frozen and freeze-dried cultures. Some have been isolated from environmental samples and the species was confirmed by bioMérieux's API^® identification. This study aimed to analyse a range of commonly found and known sulphate-reducing organisms of aquatic/environmental origin to assess the potential for false-positive and false-negative test results in comparison to test results from bacteria of enteric and faecal origin. The author realises that it is impossible to test for all relevant bacterial strains. All of the chosen bacterial strains are of relevance with regards to health-related water quality monitoring and are opportunistic or pathogenic (Percival et al. 2013).

The confirmed and pure cultured bacteria used in this research were:

• Aeromonas hydrophila (unknown strain)

• Campylobacter jejuni NC11168

• Citrobacter freundii ATCC^® 8090™

• Clostridium difficile ATCC^® 9689™

• Clostridium perfringens (unknown strain)

• E. coli NCEMB 10240 (ATCC^® 23744™)

• E. coli O157:H7 NCTC 12900 (shigatoxin negative)

• E. coli NCTC 10418

• E. coli NCTC 5933

• E. coli (unknown strain)

• Edwardsiella tarda NCIMB 2056

• Enterococcus faecalis ATCC^® 29212™

• Klebsiella pneumoniae (unknown strain)

• Proteus mirabilis ATCC^® 43071

• Salmonella enterica serovar enteritidis ATCC^® 13076™

• Salmonella typhimurium NC12023

• Serratia marcescens (unknown strain)

• Staphylococcus aureus NCTC10788

• Vibrio cholerae NCTC 10256 (biotype El Tor, non-toxic)

• Yersinia enterocolitica ATCC^® 9610™

All bacterial strains used for testing the H₂S tests and its modifications were re-cultured in brain-heart infusion broth (Oxoid CM1135), and stored in 10–15% glycerol at −80 °C until needed. Before each round of testing, the pure cultures were defrosted, and 0.1 ml inoculated into 100 ml autoclaved brain-heart infusion broth, followed by 24 hour incubation at the strain's required temperature. 0.1 ml of each pure culture was used to produce dilutions of 10⁻², 10⁻⁴, and 10⁻⁶ in sterile water, which when in use in the test produced final concentrations of 10⁻⁴/10⁻⁶/10⁻⁸. The Miles and Misra method (or surface viable count method) was subsequently used to determine the concentration of bacterial cells in the diluted inoculate (Miles et al. 1938). Clostridium spp. cultures and H₂S tests containing Clostridium spp. have been cultivated in an anaerobic jar.

To perform the H₂S test, a piece of filter paper (2 × 8 cm) inoculated with 0.5 ml of the culture medium (see Figure 1) was placed in a disposable, sterile 10 ml plastic culture tube (see Figure 2) to which 9.9 ml of deionised, autoclaved water containing the pure cultured bacteria at the appropriate dilution was added. Tubes were incubated in the dark at ambient temperature (20 °C) for up to 48 hours. For each version of the H₂S test, one tube contained 10 ml of deionised, autoclaved water without bacteria as a blank control. If the filter paper changed colour from a light yellow to black (see Figure 3), this indicated a positive reaction for bacteria able to produce hydrogen sulphide. A change of colour was observed between 12 and 48 hours, depending on the concentration of bacteria. The different test formulations were examined for colour change continuously (once every hour). Tests in which the content remained light yellow in colour without any production of black precipitate after 48 hours were recorded as a negative result.

It should be highlighted that Manja et al. (1982) used a 20 ml volume glass bottle for the original H₂S test. However, research from Yang et al. (2013) indicates that a smaller test volume lowers the sensitivity, but raises the specificity. A smaller sample volume also reduces the price per test, and disposable culture tubes do not require sterilisation and are very cheap to procure, and are generally cost effective.

During the second part of this study, the H₂S test and its modifications were tested against unknown samples as suggested by the method validation protocol of EPA (2009). It was considered important to validate the original H₂S test and any modifications against accepted standard methods such as membrane-filtration with semi-selective culture medium to detect indicator bacteria. This step was also necessary to assess any differences in the performance of the original H₂S test and the modifications in different environmental and field conditions, and especially with different types of water. Unimproved water samples were collected from 10 sites along the River Ouse in East Sussex (UK), and from one rain harvesting point also located in East Sussex.

Validation based on a sensitivity/specificity analysis was conducted on the original H₂S test and the modified versions. Membrane filtration using the semi-selective media m-FC and m-Enterococcus, as described by Standard Methods for the Examination of Water and Wastewater (APHA 2012) was used as the test method. Each different H₂S test was assessed at incubation temperatures of 20 °C and at 37 °C, to assess any relationship between the incubation temperature and the reaction time.

The calculation of positive and negative predictive values state the probabilities that an individual (or a test) is truly positive given that it tested positive, or truly negative given that it tested negative. In this study, sensitivity marks the proportion of those H₂S tests showing a positive reaction and which correctly identified a water sample containing faecal indicator bacteria. Specificity is the proportion of those H₂S tests which show no reaction (negative reaction) to a water sample that does not contain any faecal indicator bacteria (zero CFU) in a 100 ml sample (Peacock & Peacock 2013). The diagnostic sensitivity and specificity, the PPV and NPV respectively, were calculated using MedCalc^® (MedCalc Software Ltd). The PPV and NPV were defined as follows (Peacock & Peacock 2013):

Sensitivity:

True positive/(true positive + false negative)

Specificity:

True negative/(false positive + true negative)

Positive predictive value:

True positive/(true positive + false positive)

Negative predictive value:

True negative/(false negative + true negative)

Prevalence of disease:

(True positive + false negative)/grand total

Single-operator characteristics’ and ‘method detection level determination’ are suggested tests in the development of new diagnostic methods by APHA (2012). The performance of each H₂S test (original and modifications) was investigated using pure cultures of selected confirmed strains of enteric and non-enteric sulphide-, sulphate-, sulphur-reducing, and H₂S-producing bacteria. Performance indicators assessed were (i) the strength of the reaction at different dilutions and (ii) the time taken to produce a positive reaction.

20 pure cultures of bacterial strains were tested against the original H₂S test and modifications at concentrations of 10⁻⁴/10⁻⁶/10⁻⁸ plus one blank (no bacteria) test. This experiment was repeated twice to produce data for analysis of 3,400 H₂S tests. As shown in Tables 2 and 3, the original H₂S test's positive reaction is triggered only when the sample water contains Citrobacter freundii, Proteus mirabilis, Salmonella typhimurium, and to a lesser extent Salmonella enterica. With regards to the positive reaction caused by the Citrobacter freundii ATCC^® 8090™, Proteus mirabilis ATCC^® 43071™, and Salmonella typhimurium ATCC^® 14028™ strains, analysis suggests that the original H₂S test is not able to detect thermotolerant faecal coliforms or any other faecal indicator bacteria apart from Citrobacter freundii. This is an important finding C. freundii is regarded as thermotolerant, but is also ubiquitous in nature (Holt et al. 1994).

All of the tested E. coli culture collection strains plus one unclassified strain isolated from the River Ouse demonstrated the production of H₂S from sulphur sources other than thiosulphate, as used in the original H₂S test. The preferred sources of (organic) sulphur are the amino acids L-cysteine and its oxidised disulphide-bond L-cystine. When L-cysteine was provided in combination with 2-mercaptopyridine and tested in the presence of E. coli NCTC 5933, positive results were observed at ambient temperature after only 18 hours. Also, L-cystine produced faster (12 hours on average) results compared to tests prepared with L-cysteine. The H₂S test with added L-cystine delivered fast and strongly readable results when tested with bacteria from the family of Enterobacteriaceae in particular.

The H₂S test modification containing L-glutathione showed no reaction to any organism used in this study. Also, it was not observed that a modified H₂S test containing L-cysteine or L-cystine is able to detect for the presence of V. cholerae.

The modified H₂S tests containing the organic sulphur compounds D-cysteine, L-methionine, L-glutathione, taurine, and pyruvate plus the tests containing the inorganic sulphur sources sodium sulphate and tetrathionate showed, mostly, either no positive result, or only reacted very weakly. The H₂S test containing L-glutathione reacted weakly in the presence of Citrobacter freundii ATCC^® 8090™ only. However, no other organism tested was observed to be capable of reducing this amino acid to H₂S. Also, using ferrous chloride instead of ferrous citrate, as suggested by Barrett & Clark (1987), yielded no convincing results. Only Citrobacter freundii ATCC^® 8090™, E. coli O157:H7 NCTC 12900, Proteus mirabilis ATCC^® 43071™, and Salmonella Typhimurium NCTC 12023/ATCC^® 14028™ triggered some very weak reaction to this compound.

The modified H₂S tests containing penicillin G (benzylpenicillin) showed promising results, as it allowed for the growth of all tested E. coli strains and other organisms from the family of Enterobacteriaceae. Since penicillin G prevents the growth of Gram-positive and many Gram-negative bacteria (such as, e.g. Klebsiella pneumoniae) but allows the growth of most other Enterobacteriaceae, it makes the H₂S test very specific to detect for faecal contamination (Sigma Aldrich/Merck 2017a). The modified H₂S tests containing gentamicin in contrast prevents the growth of Gram-negative bacteria and Gram-positive Staphylococcus spp. and would theoretically allow the growth of Enterococcus faecalis (Sigma Aldrich/Merck 2017b). However, Enterococcus faecalis ATCC^® 29212™ showed no reaction to any of the H₂S tests.

Clostridium perfringens (River Ouse) reacted to the H₂S test that contained L-cysteine only. Positive reactions at ambient temperature were observed after between 22 and 40 hours. However, growth was very slow, and the full reaction was not completed before 72 hours of incubation time. Klebsiella pneumoniae (River Ouse isolate) reacted weakly after between 40 and 46 hours to the H₂S test that contained L-cysteine only. Both Clostridium perfringens and Klebsiella pneumoniae showed no reaction to any other H₂S test modification within 48 h.

A. hydrophila (River Ouse isolate), C. perfringens (River Ouse isolate), C. jejuni NCTC 11168 /ATCC^® 700819™, E. faecalis ATCC^® 29212™, S. aureus NCTC 10788, and S. marcescens did not show a positive reaction to any of the tests, and have not been included in Tables 2 and 3.

The reaction time (the time until the production of a black precipitate) of pure cultures was considerably longer (about 12 h) compared to mixed cultures (undiluted raw water from the River Ouse). Apart from E. coli O157, Proteus mirabilis ATCC^® 43071™ and Salmonella Typhimurium ATCC^® 14028™ mostly showed a positive reaction after +/ − 20 hours at a low dilution (high concentration) of 1 × 10⁻⁴ (approximately 42,025–28,900 CFU, see Tables 4 and 5). The other bacteria demonstrated positive reaction times of between 24 and 48 hours at ambient temperature when tested with a pure culture diluted at 1 × 10⁻⁴.

At ambient temperature (20 °C), the H₂S tests producing the fastest reactions are those containing L-cysteine and 2-mercaptopyridine rather than thiosulphate (28.6 hours). The original H₂S test (29 hours), that containing L-cysteine and thiosulphate (29.2 hours), and that containing L-cystine rather than thiosulphate (31.2 hours).

The H₂S test with bile salts presented stronger results than the same test prepared with liquid detergent instead, as done by Manja et al. (1982). Results came in even slightly faster (about one hour) when the concentration of bile salts was (in a different test modification) increased from 2% to 6%. Both H₂S tests prepared with bile salts instead of detergent yielded a positive reaction to Citrobacter freundii ATCC^® 8090™, Proteus mirabilis ATCC^® 43071™, and Salmonella Typhimurium NCTC 12023/ ATCC^® 14028™, very similar to that demonstrated by the original H₂S test, but reacted additionally to most of the tested E. coli strains apart from E. coli NCIMB 10240/ ATCC 23744 when tested with 6% bile. Overall and among all H₂S tests, Salmonella Typhimurium NCTC 12023/ ATCC^® 14028™, Proteus mirabilis ATCC^® 43071™ and Citrobacter freundii ATCC^® 8090™ presented the strongest and fastest positive reactions.

This step involved evaluating the performance of the original H₂S test and its variants against water samples of unknown microbial and chemical composition from the environment. This is in compliance with ‘standard methods’ for the examination of water and wastewater (EPA 2009; APHA 2012). This step was necessary to assess how the original H₂S test and its new modifications function under different environmental and field conditions, and especially with different types of waters: unimproved vs. improved.

With regards to the results collected above on sensitivity, the analysis of the various H₂S test modifications has been reduced to eight H₂S tests including the original H₂S test:

• Original H₂S test

• H₂S test with L-cysteine/ no thiosulphate

• H₂S test with L-cysteine and thiosulphate

• H₂S test with 2% bile salts instead of detergent

• H₂S test with 6% bile salts instead of detergent

• H₂S test with L-cystine instead of thiosulphate

• H₂S test with penicillin G

• H₂S test with 2-mercaptopyridine

The results from the seven H₂S test modifications and the original test against the accepted standard methods (membrane-filtration with semi-selective media for presumptive thermotolerant faecal coliforms and Enterococcus) and by calculating and analysing the positive-predictive (PPV) and the negative-predictive value (NPV) are presented in Tables 6 and 7.