Rapid Determination of Bacteria in Drinking Water

Using an ATP Assay

Rolf A. Deininger and JiYoung Lee

Department of Environmental Health Sciences, School of Public Health, The University of Michigan,
Ann Arbor, Michigan 48109
Received 20 April 2001; revised 12 May 2001; accepted 22 May 2001

Abstract: The presently used heterotrophic plate count

(HPC) for the evaluation of the total number of bacteria
in a sample of drinking water takes 7 days of incubation.
When the results are known, the water has been consumed and is ineffective for the protection of the health
of the consumers. Operators of water treatment systems
need to know the bacterial water quality in near real time.
Contamination of the system, whether it is intentional,
accidental, or due to an inadequate disinfectant residual
needs to be discovered much sooner because intervention can then take place in the form of flushing low quality
water and/or raising the disinfectant residual. The purpose of this study was therefore to determine if a rapid
ATP assay can estimate the HPC in minutes. Two additional methods were used for some samples. The first
method was the acridine orange direct count (AODC) that
enumerates both viable and nonviable bacteria. The second method was the direct viable count (DVC) that enumerates only viable bacteria. Water samples were obtained from local, national, and international locations.
The sample selection criteria were based on proximity to
the laboratory, cooperating water utilities, and the travel
of the authors. The results of the study show that the
rapid ATP assay is highly correlated with the conventional plate count method and the DVC method. The significance of the ATP assay is that it can determine the
bacterial quality of the drinking water in less than 5 min.

C 2001 John Wiley & Sons, Inc. Field Analyt Chem Technol 5: 185189, 2001
Keywords: drinking water; bacterial quality; rapid method; on-site analysis; ATP bioluminescence

Introduction
Federal regulations regarding the microbiological quality of drinking water focus on a single group of indicator organisms, the coliforms, for the bacteriological safety
of the water.1 Although these standards of water quality
Correspondence to: Rolf A. Deininger; e-mail: rad@umich.edu.
c 2001 John Wiley & Sons, Inc.

have reduced massive outbreaks of waterborne diseases, they

have been questioned whether or not they are adequate to
prevent waterborne illnesses.2 Cases of gastrointestinal illnesses have been reported among individuals drinking tap
water that had met microbiological, physical, and chemical
water-quality criteria. The determination of the total number
of heterotrophic bacteria (HPC) has been known to be a better indicator of water quality than the coliform test because
many opportunistic pathogens are not in the coliform group.3
Furthermore, a high HPC interferes with the determination
of the coliforms.4
The present HPC method using R2A agar is the most sensitive test for enumerating the total number of bacteria in
treated water.5 The disadvantage of the test is that it takes 7
days to complete and when the results are known, the water
has long been consumed. A test is needed that determines
the total bacterial populations in a very short time so that
corrective actions can be taken in a timely manner.
The ATP bioluminescence assay allows an estimation
of bacterial populations within minutes and it can be done
on-site. The estimation of the bacterial count based on a
ATP assay of the water is not new. Standard Methods (1995;
9211C.1) indicate that the test requires 1 h and 1 l of water.6
The test has a sensitivity of 100,000 cells. What is new is that
the method presented here is over 100 times more sensitive,
requires one-hundredth of the sample volume, and is over 10
times faster.
Thus, the purpose of this study was to determine if a rapid
ATP assay could estimate the bacterial populations in a practical and timely manner. To check the accuracy of both the
ATP and the HPC test, a direct enumeration of the bacteria
in a water sample was done using two epifluorescence methods. One was the Acridine Orange Direct Count (AODC)
method to enumerate the total number of bacteria, which
include both the viable cells and the nonviable cells. The
other was the Direct Viable Count (DVC) method, which
selectively enumerates only viable bacteria. These two tests

FIELD ANALYTICAL CHEMISTRY AND TECHNOLOGY 5(4):185--189, 2001

are very time consuming and require high technical skills

and are not normally done by operators of water-treatment
systems.
Materials and Methods
Water Samples
Several water utilities in various states (Michigan, Ohio,
California, Colorado, Georgia, and Florida) cooperated in
the study and provided samples of their drinking water.
Additional water samples were taken from airport locations (Illinois, California, Oregon, Washington, Kentucky,
Tennessee, Maryland, Texas, and New York). Other samples were taken from drinking water fountains in Michigan, Ohio, Illinois, Washington, DC, California, Colorado,
Florida, Georgia, Oregon, Washington, Kentucky, Tennessee,
Maryland, Texas, and New York. No attempt was made to
have a sampling plan to cover the entire United States and
thus all samples may be considered convenience samples. To
see if the test would also work with drinking water samples
from abroad, a number of worldwide samples were obtained
over a period of 2 years from Hungary, Germany, Switzerland,
The Netherlands, Austria, United Kingdom, France, Ukraine,
Lithuania, Japan, Korea, Egypt, Saudi Arabia, Argentina,
Peru, Brazil, Panama, and Australia. Again this set of samples
does not cover the entire world, and must also be considered
as a convenience-sample set.
Filtration of the Water Samples
Based on the expected bacterial count of the water sample
an amount between 0.1 and 10 ml was drawn into a sterile syringe. The usual volume of water filtered was typically 10 ml.
After drawing the water into the syringe, a filter holder with a
FiltravetteTM was attached to the syringe and the water sample was pressed through the filter. The Filtravette is a combination of a filter with pore size of 0.45 mm and a cuvette.
The filter holder used in the study was a Swinex filter holder
(13 mm; Millipore Corporation, Bedford, MA). After filtration the Filtravette was removed from the filter holder and
placed onto a sterile blotting paper. The water remaining
inside the Filtravette was removed with a converted 3-ml
syringe by applying gentle air pressure.
ATP Bioluminescence
A two-step procedure is necessary to separate the
somatic-cell ATP from the bacterial-cell ATP. A somaticcell-releasing agent (New Horizons Diagnostic Corporation
[NHD], Columbia, MD) was used to lyse all non-bacterial
cells and release their ATP. Air pressure was used to remove
the nonbacterial ATP through the filter. At this stage, the
Filtravette retained the bacteria on its surface, and the bacterial ATP remained within the bacterial cell membranes
through this step of the procedure. The Filtravette was then
inserted into the microluminometer (Model 3550, NHD,
Columbia, MD) and the bacterial-cell-releasing agent was

186

FIELD ANALYTICAL CHEMISTRY AND TECHNOLOGY----2001

added to lyse the bacterial cells retained on the surface of

the filter. The released bacterial ATP was mixed with 50
ml of luciferin-luciferase (NHD, Columbia, MD) and the
drawer of the microluminometer was closed. The light emission was recorded over a 10-s time interval and the light
impulses were integrated. The unit of measurement of the
luminometer is relative light units (RLU). The result was
expressed as RLU/ml by dividing the RLU values by the filtered water volume. The detection limit and sensitivity of
the luminometer was tested with serially diluted ATP solutions (NHD, Columbia, MD). Distilled deionized water
was used for the dilution at a pH of 7.8. The activity of the
luciferin-luciferase was checked by using an ATP standard
(NHD, Columbia, MD). The RLUs are proportional to the
amount of ATP, and the amount of ATP is proportional to
the number of viable bacteria. The luminometer and the required accessories fit nicely onto a clipboard and are shown in
Fig. 1.
Bacterial Enumeration: AODC, DVC, HPC
Three different methods were used to enumerate the
bacteria. The first two methods count the number of individual cells. The latter method counts the number of
colonies growing on an agar dish after an incubation time of
7 days.
AODC. The AODC is a method that determines the total number of viable and nonviable bacteria. The bacteria
are fixed with formaldehyde (2%, v/v, final concentration).7
The bacterial cells are then stained with acridine orange
(0.01%, w/v, Fluka, Switzerland) after filtration onto a 0.2 mm
pore-size black polycarbonate membrane filters (Poretics,
Livermore, CA). The cells were then counted under a magnification of 1000 with an Olympus Provis epifluorescence
microscope (Olympus Optical Co., Japan) equipped with a
mercury arc lamp and a 460490 nm excitation filter. The
number of bacteria was counted in 10 microscopic fields by
using three sub samples and was averaged. The number of
bacteria per milliliter of sample was calculated using the formula in Standard Methods.6 The method requires high technical skills and also requires judgment in identifying which
of the stained images represents a cell. The method cannot
distinguish living cells from dead cells.
DVC. The DVC is a method that identifies only living cells.8 The samples were incubated with yeast extract (0.005%, w/v, Difco, Detroit, MI) and nalidixic acid
(10 mg/L, Sigma, St. Louis, MO) without dilution for 24 h at
20 C. The modifications from the original method consisted
of using a lower concentration of yeast extract and no dilution. After incubation, fixation, counting, and calculation of
elongated bacteria, followed the AODC method. Under the
microscope, only living cells will show their typical elongated form. Thus this method is the most accurate, but also
most time-consuming method.
HPC. The HPC was determined for each water sample in triplicate using R2A medium (Difco, Detroit,
Michigan). The bacterial colonies on the agar plates

FIG. 1. Portable microluminometer with all accessories on a clipboard.

were counted after an incubation period of 7 days at

28 C.

Results and Discussion

Correlation Between Parameters
The detection limit of ATP assay was determined with
multiple serial dilutions. It showed that the microluminometer was able to determine ATP concentrations as low as
0.2 pg.9 It is known that the average ATP content of one
bacterial cell is about 1015 g (1 fg).10 Thus the detection
limit of 0.2 pg corresponds to about 200 bacterial cells.
About 120 water samples from the United States alone
were analyzed with ATP bioluminescence, HPC, DVC, and
AODC methods (each in triplicate). A statistical analysis of
the data showed that there is a high correlation between ATP,
HPC, DVC, and AODC. The correlation coefficient between
ATP and HPC was 0.84, and the correlation coefficient between ATP and DVC was 0.8. Both were statistically highly
significant ( p < 0.05). In Fig. 2 the relation between HPC
and ATP is shown for samples from the United States, and in
Fig. 3 the same relation from worldwide samples is shown.
The relation between DVC and ATP is shown in Fig. 4. The
fact that the regression line does not go through the origin
may be due to the detection limit of the ATP. That is, over
200 cells are necessary to get a response from the luminometer. And finally, in Fig. 5 the relationship between AODC
and ATP is shown. Again the intercept at the y-axis may be
due to the fact that the AODC method counts both living

and dead cells, while the ATP assay determines only living
cells.
Intentional Contamination of a Water Supply System
Contamination of water supply systems has its root in antiquity. In biological warfare the preferred route is airborne
through aerosols, but the water route cannot be excluded.
Targets in the latter route are the sources of the water supply
system, lakes and rivers, the treatment plant, and the water

FIG. 2. HPC vs. ATP for U.S. water supplies.

FIELD ANALYTICAL CHEMISTRY AND TECHNOLOGY----2001

187

FIG. 3. HPC vs. ATP for international water samples.

FIG. 5. ATP vs. AODC.

distribution system. The most likely point of attack would be

after the treatment plant. Prime targets are the reservoirs and
the water distribution pipes. Consider a biological agent such
as bacteria that has an infectious dose of a million cells. To be
effective, that dose would have to be contained in a glass of
water (200 ml). If one takes a 10-ml sample of the contaminated water, one would collect about 50,000 organisms. Since
the detection limit is well below 1000 organisms, detection
poses no problem. If the infectious dose is 1000 organisms,
then a 60-ml sample (a 60-cc syringe) of the glass of water
would contain about 300 organisms and one would be at the
detection limit of the method. Operating at this level would
require a concentration step using larger sample volumes and
an additional filtration step. Thus, an intentional contamination of a water supply system can be detected. The question
is if it can be done in time. A near continuous monitoring

of the water is necessary. An instrument exists which can do

this, but the instrument and chemicals are expensive and not
practical for the everyday monitoring of a supply, but quite
possible for a high profile event.
Conclusions
The estimation of the heterotrophic plate count can be
done in minutes. A strong correlation exists between the HPC
and an ATP assay using bioluminescence allowing the prediction of the HPC from the ATP test. The samples from the
United States and a large number of worldwide samples show
the same relationship. For the U.S. water samples, the additional tests of a bacterial count using both the AODC and the
DVC methods indicate that the ATP method is a valid test. The
method is rapid and sensitive and can be done on-site with a
portable power supply. The volume of water required for the
test is small. Typical volumes are less than 10 ml. For very
clean waters the size of the sample is dictated by the commercially available syringes. Thus, a traveler wishing to determine the bacterial water quality of a drinking water supply in
the United States or a foreign country can do this in less than
5 min.
References

FIG. 4. ATP vs. DVC.

188

FIELD ANALYTICAL CHEMISTRY AND TECHNOLOGY----2001

1. Federal Register. 1996. Vol 61, No 94, p. 24372.

2. Payment P, Richardson L, Siemiatycki J, Dewar R, Edwardes M, Franco
E. A randomized trial to evaluate the risk of gastrointestinal disease
due to consumption of drinking water meeting current microbiological
standards. Am J Public Health 1991;81:703708.
3. Geldreich EE, Nash HD, Reasoner DJ, Taylor RH. The necessity of
controlling bacterial populations in potable waters; community water
supply. J Am Water Works Assoc 1972;64:596602.
4. LeChevallier MW, McFeters GA. Interactions between heterotrophic
plate count bacteria and coliform organisms. Appl Environ Microbiol
1985;49:13381341.

5. Reasoner DJ, Geldreich EE. A new medium for the enumeration and
subculture of bacteria from potable water. Appl Environ Microbiol 1985;
49:17.
6. APHA-AWWA-WEF. Standard Methods for the Examination of Water
and Wastewater 19th edition. Washington, DC: American Public Health
Association, Inc.; 1995.
7. Hobbie JE, Daley RJ, Jasper S. Use of Nuclepore filters for counting bacteria by fluorescence microscopy. Appl Environ Microbiol
1977;33:12251228.

8. Coallier J, Prevost M, Rompre A. The optimization and application

of two direct viable count methods for bacteria in distributed drinking
water. Can J Microbiol 1994;40:830836.
9. Lee JY, Deininger RA. A rapid method for detecting bacteria in drinking
water. J Rapid Methods Automation Microbiol 1999;7:135145.
10. Crombrugge J, Waes G. ATP method. In: Heeschen W, editor. Methods for Assessing the Bacteriological Quality of Raw Milk from the
Farm. Brussels, Belgium: International Dairy Federation; 1991. pp 53
60.

FIELD ANALYTICAL CHEMISTRY AND TECHNOLOGY----2001

189