Letters in Applied Microbiology 1994, 18, 346-348

Use of a commercial gene probe assay kit for rapid MPN enumeration of Escherichia coli in drinking water

P. Gale and P.J. Broberg Water Research Centre, WRc Medmenham, Marlow, Buckinghamshire, UK

DMG/093: accepted 19 January 1994

P . GALE AND P.J. BROBERG. 1994. A commercial gene probe assay kit for presence/absence determination of Escherichia cola in food samples has been used in the standard UK six tube format most probable number (MPN) method for enumerating E . coli in drinking water samples. Presence/absence analysis with the gene probe kit (requiring 3 h) of all MPN tubes after a 21-24 h incubation (minerals modified glutamate; 37°C) enumerated confirmed E. coli in 24-27 h which offered an improvement of up to 48 h over the standard UK MPN method. MPNs determined by the gene probe method and the standard UK method agreed in nine of the 16 water samples which were analysed and for which E . coli concentrations were within the detection limits of the six tube MPN format. This was consistent with the gene probe method detecting one E. coli in a tube. For the other seven water samples, the gene probe method registered positive only 20 of the 30 tubes which the standard UK method determined to be positive. The sensitivity of the gene probe method for drinking water samples, although encouraging, needs improvement perhaps through kit quality control procedures.

iNTRODUCTlON

In the UK, two methods are approved for the detection and enumeration of Escherichia coli in drinking waters (DOE et al. 1983). These are the membrane filtration method (MF) and the most probable number (MPN) method and require up to 48 h or 72 h, respectively, to complete includ- ing confirmation. Gene probes offer the potential for more rapid detection of micro-organisms in water samples because the specificity of the probe eliminates the require- ment for additional confirmation.

A commercial gene probe assay kit, marketed by GENE- TRAK Systems, is available for the presence/absence determination of confirmed E. coli in food samples and, after sufficient culture steps to amplify the numbers of E. coli, requires 3 h to complete. The development and mechanism of operation of the assay are described by Parsons (1988). The objective of the study here is to assess the potential of using the gene probe assay kit in an MPN method for drinking water samples such that confirmed E. coli could be enumerated more rapidly than by standard UK methods. The sensitivity requirement of any method for the analysis of E. coli in drinking water samples is one organism in a 100 ml volume (DOE et al. 1983). The ability

Correspondence t o : Dr Pauf Gale, Water Research Centre, WRc Medmenham, Henley Road, Marlow, Buckinghamshire SL7 ZHD, UK.

of the gene probe MPN method (method A) to enumerate E. coli is compared with the standard UK MPN method (method B).

MATERIALS AND METHODS

Enumeration of E. CON by the standard UK and gene probe MPN methods

Volumes (0.10, 0.25, 0-50 or 1.00 ml) of river water, likely to contain at least 100 E. coli organisms per 100 ml, were made up to 100 ml with Ringers solution (Oxoid BR52) and cultured in double strength minerals-modified gluta- mate (MMG) medium base (Oxoid CM607) with sodium glutamate (Oxoid L124) in the six tube MPN format (1 x 50 ml plus 5 x 10 ml) of the standard UK enumer- ation method for E. coli in drinking waters (DOE et al. 1983). Each tube was tested for E. coli by both method A and method B. For method A, a volume of 0.5 ml was removed from each MPN tube after 21-24 h at 37°C for presence/absence determination by the GENE-TRAK E. coli assay kit as described in the manufacturer’s instructions (Digen Ltd, Wheatley, Oxford, UK). The assay is colori- metric and a portable photometer is available to measure the optical density (O.D.). Negative and positive controls

RAPID GENE PROBE ENUMERATION OF E . COLl 347

are supplied with the kits to monitor performance during each experiment. GENE-TRAK stipulates that the nega- tive control O.D. reading must be less than 0.15 unit and the positive control must be greater than 1-00 unit for results from an experiment to be valid. The assay takes about 2.5 h to perform and GENE-TRAK states that samples with O.D. greater than 0.10 unit are positive for E. coli. Samples with O.D. less than 0.10 unit are negative for E. coli. Confirmed E. coli MPNs were calculated from the number of positive tubes (DOE et al. 1983). For method B (DOE et al. 1983), MMG tubes producing acid and gas after 24 h or 48 h at 37°C were confirmed for E. coli by the production of gas on culture in lauryl tryptose broth (Oxoid (34451; 24 h, W C ) and by the production of indole on culture in tryptone water (Oxoid L42; 24 h, 44°C).

Statistical analysis

In this study a presencelabsence result was obtained by both method A and method B for each MPN tube. The null hypothesis that the two methods were comparable in detection ability was tested by McNemar’s test (Sprent 1989). This is a non-parametric, matched-pair test which compares the numbers of tubes registering presence/ absence discrepancies, i.e. tubes which were determined to be positive for E. coli by one method, but negative by the other method. If the two methods are comparable in detec- tion ability then the number of tubes in which method A does not register an E. coli (A-) detected by method B (B’) would be similar to the number of tubes in which method B does not detect an E. coli (B-) registered by method A (A’), i.e. the number of A-, B+ pairs should equal the number of A+, B- pairs. McNemar’s test deter- mines the probability of obtaining a particular combination of A-, B+ and A+, B- pairs if the null hypothesis that the two methods are comparable is true.

Speclflclty of the GENE-TRAK E. CON assay kit

The specificity of the GENE-TRAK assay kit used in method A was tested by comparison with API 20E on 50 presumptive coliform colonies recovered from river water by membrane filtration (membrane lauryl sulphate broth Oxoid MM615; 20 h, 37°C). Each colony was cultured in nutrient broth (Oxoid CMl; 48 h, 37°C) to amplify bac- teria such that sensitivity did not limit detection (Parsons 1988) and a 0.5 ml volume tested by the GENE-TRAK E. coli kit. Each nutrient broth culture was identified by API 20E (Bio-Mirieux, Basingstoke, UK).

RESULTS AND DISCUSSION

PresenceJabsence testing of MPN tubes with the GENE- TRAK assay kit after a 21-24 h culture in MMG (method

A) provided confirmed E. coli concentrations in 2427 h which is up to 48 h more rapid than the standard UK MPN method (method B). Confirmed E. coli MPNs deter- mined from the analyses of 16 water samples by both methods are compared in Table 1. These samples regis- tered E. coli MPNs between 1 and 16 per 100 ml which was within the detection limits of the six tube MPN format (i.e. for each sample there were both positive and negative tubes).

Results from samples which registered < 1 or > 18 E. coli per 100 ml by both methods and which were beyond the detection limits of the six tube MPN format are not presented. The MPN method of enumeration is dependent on the analytical method being capable of detecting a single target organism in a tube. For nine of the 16 samples, the MPNs determined by methods A and B were the same, with presencelabsence results agreeing for all MPN tubes. This was consistent with method A being sensitive enough to detect one E. coli in a tube. Indeed in one water sample, method A detected one E. coli per 100 ml (although the 95% confidence intervals are 0.54.0 E. coli per 100 mi; DOE et al. 1983). In six of the seven water samples for which the MPN values determined by the two methods did not agree, method B registered a higher value than method A (Table 1). For these seven samples, method A registered as positive only 20 of the 30 tubes which method B deter- mined to be positive.

The analysis of 16 water samples by the six tube MPN format requires 108 tubes. These tubes are categorized according to presencelabsence determinations by each of the two methods in Table 2. Agreement in presence/

Table 1 Escherichia coli MPNs determined for 16 water samples by the gene probe method (A) and by the standard UK method (B)

Method A Method B Difference (B - A)

1 3 3 3 4 6 6 9

16 2 2 3 6

16 9 9

1 3 3 3 4 6 6 9

16 9 9 9 9 9

16 16

0 0 0 0 0 0 0 0 0

+7 + 7 + 6 + 3 - 7 + 7 +7

348 P. GALE AND P.J . BROBERG

Table 2 Statistical analysis of Escherichia coli presence/absence determinations for individual MPN tubes by the gene probe method (A) and by the standard UK method (B)

Method A

Present Absent

Method B Present 57 10 Absent 1 40

Data from 16 water samples with E. coli concentrations within the detection limits of the six tube MPN format. Null hypothesis: the two methods are comparable in ability to detect E. coli. Rejected by McNemar’s test; two-sided P = 0.012.

absence result by the two methods was observed for 97 of the 108 tubes, with 57 registering A+, B+ and 40 A-, B-. Discrepancies in presencelabsence results determined by the two methods were observed in 11 MPN tubes; 10 regis- tering A-, B+ and one A’, B-. The null hypothesis that the two methods are comparable in detection ability is rejected by McNemar’s test (Table 2). The gene probe method (A) is more likely to give a negative result than the standard UK MPN method (B).

The specificity of GENE-TRAK kit for E. coli agreed with API 20E suggesting that the larger number of negative results by method A was not due to the specificity of the gene probe. In total, 12 of the 50 presumptive coliform colonies recovered from the river water used for spiking the samples were determined to be E. coli by the GENE- TRAK kit. Of these 12 colonies, 10 were identified as E. coli by API 20E; the other two being identified as E. coli/ Shigella and E. coli/Citrobacter. The GENE-TRAK kit did not detect Aeromonas hydrophila, Citrobacter freundii, Ci t . diversus, Vibrio fluvialis, Enterobacter cloacae, Ent. inter- medium or Klebsiella oxytoca colonies which were recovered from the river water and identified by API 20E.

Differences in sensitivity between the two methods would produce presencelabsence discrepancies in tubes with very low E. coli numbers where the less sensitive method is at the limit of its detection ability and fails to detect an E. coli registered by the more sensitive method. Method A requires the single 21-24 h enrichment step to produce at least lo5 E. coli in the 0.5 ml volume used for analysis with the GENE-TRAK assay kit (Parsons 1988). The GENE-TRAK protocol for food samples, which has

two consecutive 24 h enrichment steps, was not adopted because the total detection time (51 h) would offer no time advantage over the standard U K M F method (48 h) for drinking water samples. For food samples after two 24 h enrichments, a sensitivity of three E. coli is reported (Digen Ltd, instructions for GENE-TRAK E. cob assay kit). As discussed above, method A (single enrichment) may detect one E. coli in a tube for some of the water samples. This suggests that a second enrichment step may not be neces- sary for water samples.

All GENE-TRAK negative controls registered optical densities of <Om15 unit, ruling out false positives from method A by kit malfunction or improper operation. The positive controls, however, supplied with the GENE- TRAK assay kits used for analysis of the water samples in Table 1 registered optical densities of < 1.00 unit, suggest- ing that the kits were faulty and/or were operated improp- erly. (Positive controls with the GENE-TRAK Salmonella kits operated correctly.) This could account for the larger number of false negatives by method A. Despite this, method A registered as positive 57 of the 67 tubes which were determined as positive by method B (Table 2). It is concluded that the GENE-TRAK Systems E. coli assay kits offer potential for more rapid analyses of drinking water samples than standard UK methods, but that the sensitivity, although encouraging, needs improvement perhaps through kit quality control procedures.

ACKNOWLEDGEMENT

We thank the Foundation for Water Research for funding this work.

REFERENCES

Department of the Environment, Department of Health and Social Security and Public Health Laboratory Service (1983) Technical Methods. The Bacteriological Examination of Drink- ing Water Supplies 1982. Reports on Public Health and Medical Subjects No. 71, pp. 32-57. London: HMSO.

Parsons, G. (1988) Development of DNA probe-based com- mercial assays. Journal of Clinical Immunoassay 4, 152-160.

Sprent, P. (1989) Methods for paired samples. In Applied Non- parametric Statistical Metbods. pp. 71-85. London : Chapman and Hall.