False-negative β-d-glucuronidase reactions in membrane lactose glucuronide agar medium used for the simultaneous detection of coliforms and Escherichia coli from water


Colin R. Fricker, Child’s Acre, Church Lane, Three Mile Cross, Reading, RG7 1HD, UK. E-mail: colinfricker@aol.com


Aims:  Testing for β-d-glucuronidase activity has become the basis of many methods for the detection of Escherichia coli in both food and water. Used in combination with tests for the presence of β-d-glucuronidase, these tests offer a simple method for simultaneously detecting coliforms and E. coli. The purpose of this study was to determine the effectiveness of several different procedures in detecting β-d-glucuronidase activity and hence in detecting E. coli.

Methods and Results:  The ability of membrane lactose glucuronide agar (MLGA), Colilert-18®, MI agar, Colitag® and Chromocult agar to detect β-d-glucuronidase activity was tested with over 1000 isolates of E. coli recovered from naturally contaminated water samples. Four of the media gave very similar results but MLGA failed to detect glucuronidase activity in 15·6% of the cultures tested.

Conclusions:  MLGA had very poor sensitivity for the detection of β-d-glucuronidase activity in strains of E. coli isolated from naturally contaminated water. This is probably because of the fact that β-d-glucuronidase activity is pH-sensitive and that acid is formed by E. coli during fermentation of lactose in MLGA.

Significance and Impact of the Study:  The detection of E. coli in drinking water is the primary test used to establish faecal contamination. The poor sensitivity of MLGA in detecting β-d-glucuronidase activity suggests that this medium and others containing high concentrations of fermentable carbohydrate should not be used for the detection of E. coli.


The use of media containing substrates for the enzyme β-d-glucuronidase for the detection of Escherichia coli in water has become widely adopted in many regions of the world (Cowburn et al. 1994; Eckner 1998; Edberg et al. 1988; Manafi et al. 1991) and such methods have been shown to be more specific than methods based upon fermentation of lactose and production of indole from tryptophan at 44°C (Fricker et al. 2008). However, it is clear that not all such media perform equally (Fricker and Fricker 1996; Niemela et al. 2003; Fricker et al. 2008; Olstadt et al. 2007).

There are obvious benefits in using media that can simultaneously detect both coliforms and E. coli that have led to their widespread application. The enzyme β-d-glucuronidase is widespread in nature and while E. coli is not the only bacterium that possesses the enzyme, media have been formulated to minimize the number of false-positive reactions. Such media usually contain inhibitors to gram-positive bacteria and may rely on the detection of the enzyme β-d-galactosidase to discriminate between E. coli and other organisms such as Shigella spp. and Salmonella spp. False-positive reactions have been noted, particularly in marine water (Pisciotta et al. 2002) but these can be alleviated by incubation at 44°C (Fricker et al., unpublished observations). Furthermore, false-positive β-d-glucuronidase reactions are uncommon when examining water although some other organisms possess the uid A gene, which is responsible for the expression of the enzyme (Fricker and Fricker 1994). While false-positive reactions are undesirable, false-negative reactions are of far more concern particularly when examining drinking water.

As part of a larger study of the performance characteristics of media approved for use of the examination of drinking water in the United States and United Kingdom, this study has examined the performance of a variety of media that utilize the detection of the enzyme β-d-glucuronidase in detecting E. coli isolated from a variety of sources including sewage and water used for the production of drinking water.

Materials and methods

Samples of raw water (52) and sewage effluent (11) that were collected from a variety of sites across the United States (examined for routine purposes) were used in this study. All routine water samples were processed using Colilert-18® together with QuantiTray®. After incubation at 35°C for 18–22 h, wells that were yellow in colour and fluoresced under ultraviolet light were subcultured directly onto membrane lactose glucuronide agar (MLGA; Oxoid CM 1031; Sartory and Howard 1992), MI agar (BD Difco™ 214883), Chromocult agar (EMD™ 1·10426·0500) and ColiScan MF medium (Micrology Laboratories Kit C #CMFKC). All media were incubated according to manufacturer’s instructions as shown in Table 1. A total of 914 wells that were yellow and fluoresced in Colilert (indicating the presence of E. coli) were subcultured in this way.

Table 1.   Incubation conditions used for each of the media tested
MediumIncubation temperatureRecommended Incubation period
  1. MLGA, membrane lactose glucuronide agar.

Chromocult37°C23–25 h
Colilert-1837°C18–22 h
Coliscan35°C18–24 h
MI agar35°C20–24 h
MLGA37°C18–24 h

In addition, a total of 312 blue colonies (so called faecal coliforms) isolated on mFC agar (BD Difco™ 267720) from diluted sewage effluent (seven samples collected from locations across the USA) after incubation at 44°C for 24 h were used to test the media. The colonies were checked for purity on MacConkey agar (BD Difco™ 212123) and then subcultured onto the same four media as were used before and also into Colilert-18®.

After incubation for the appropriate period, all plates were examined for reactions indicating the activity of the β-d-glucuronidase enzyme and the results were recorded.


Of the 914 cultures that were plated from Colilert-18® to four media used for routine water testing, four (0·4%) subcultures gave negative reactions on MI agar and Chromocult agar and six (0·7%) subcultures were negative on Coliscan medium. In contrast, 114 (12·5%) subcultures were negative for β-d-glucuronidase activity on the MLGA medium. The results are summarized in Table 2. All of the subcultures that failed to produce green colonies on MLGA gave reactions typical of coliform organisms, i.e. a positive lactose reaction but a negative β-d-glucuronidase reaction.

Table 2.   Number of cultures taken from 914 Colilert-18® positive wells that failed to give a positive glucuronidase reaction on different media
MediumNumber of negative culturesPercentage of negative cultures
  1. MLGA, membrane lactose glucuronide agar.

Chromocult4 0·44
Coliscan6 0·66
MI agar4 0·44

Of the 312 isolates purified from mFC agar, 51 (16·3%) were negative for β-d-glucuronidase activity on all of the media tested, indicating that they were non-E. coli coliforms (all gave positive β-d-galactosidase reactions on all media). Of the remaining 251 isolates, 68 (27·1%) were negative for β-d-glucuronidase activity on MLGA, while the other media had a much lower frequency of false-negative results as shown in Table 3.

Table 3.   Number of cultures taken from 251 Escherichia coli isolates grown on mFC medium that failed to give a positive glucuronidase reaction on different media
MediumNumber of negative culturesPercentage of negative cultures
  1. MLGA, membrane lactose glucuronide agar.

Chromocult0 0
Colilert-18®1 0·40
Coliscan3 1·20
MI agar8 3·19

A random selection of organisms that failed to produce green colonies on MLGA medium but were positive by the other methods were identified using standard biochemical tests using the Vitek microbial identification system (Biomerieux, France). All were shown to be E. coli.


Escherichia coli is widely used as a faecal indicator when examining many types of water and the sensitivity of methods is of particular concern when examining drinking water. The vast majority of drinking water samples that are analysed are free from E. coli but the presence of a single cell is indicative of some form of contamination that requires immediate investigation. Ideally, methods used for the examination of drinking water are both highly specific and highly sensitive and give a ‘confirmed’ result within a 24-h period. However, recent work (Fricker et al., unpublished data) has suggested that there may be significant differences in the performance of some of the methods used to test drinking water. This study was performed to determine if any of the differences in sensitivity seen between different methods was because of an inability of some strains of E. coli to produce a demonstrable β-d-glucuronidase reaction.

In this study, the medium MLGA failed to detect β-d-glucuronidase activity in a significant proportion of strains of E. coli that gave positive reactions with other media. The false-negative rate was highest in isolates recovered from mFC agar (27·1%) and was also high in cultures taken from Colilert-18® (12·5%). The difference in the magnitude of the false-positive reactions may be because of the fact that cultures grown initially in Colilert-18® had been exposed to a glucuronide substrate. This previous exposure may have resulted in an increased production of β-d-glucuronidase in some strains such that they were more easily detected on MLGA. The difference in false-negative rates between MLGA and the other media tested was remarkable. The other media tested do not have a significant source of fermentable carbohydrate while MLGA contains 30 g l−1 of lactose.

The enzyme β-d-glucuronidase is widespread in nature and enzymes from different organisms have different pH optima (Ho and Ho 1985; Fang et al. 1995) and that of E. coli is most active at neutral or slightly alkaline pH (Fang et al. 1995; Togo et al. 2006). The presence of large amounts of lactose in MLGA would result in the production of significant amounts of organic acids during fermentation. Typically, coliforms growing in MLSB medium (the basis of MLGA) lower the pH to c. 6·5. The pH within a colony may be even lower. This is the probable reason for the failure of MLGA to detect β-d-glucuronidase activity in a large proportion of strains tested. While there can be no doubt that β-d-glucuronidase offers a simple and reliable method of detecting the presence of E. coli not all methods work effectively. Growth of E. coli in the presence of large amounts of lactose appears to significantly reduce the number of strains that produce a demonstrable β-d-glucuronidase reaction. This study has shown that a large proportion of E. coli strains fail to produce characteristic green colonies on MLGA and instead appear as yellow colonies that would be recognized as coliforms. While further testing would demonstrate that these organisms were indeed E. coli, the real benefit of using β-d-glucuronidase-containing media is the ability to recognize the presence of E. coli without resorting to further confirmatory tests. Similarly, glucuronide-containing media are used for routine testing of sewage sludge and recreational waters. The use of MLGA in these situations would require all lactose-fermenting colonies to be tested to see if they were in fact E. coli. Such a large amount of confirmatory testing negates the benefit of using a glucuronide-containing medium. The results presented here suggest that MLGA should be used with caution when examining drinking water (and other matrices) for the presence of E. coli because a significant proportion of E. coli strains may appear to be β-d-glucuronidase-negative on MLGA.


This manuscript has been made possible through funding from the Awwa Research Foundation and the UK Drinking Water Inspectorate. The information contained herein is based upon Intellectual Property that is jointly owned by ASI, DWI and AwwaRF. AwwaRF retains its right to publish or produce the Jointly Owned Intellectual Property in part or in its entirety.