EPA Report Number: 600/J-99/225
APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Nov. 1993, p. 3534-3544
0099-2240/93/113534-ll$02.00/0
Copyright © 1993, American Society for Microbiology
Vol. 59, No. 11
New Medium for the Simultaneous Detection of Total
Coliforms and Escherichia coli in Water
KRISTEN P. BRENNER,1* CLIFFORD C. RANKIN,1 YVETTE R. ROYBAL,1
GERARD N. STELMA, JR.,1 PASQUALE V. SCARPING,2
AND ALFRED P. DUFOUR1
Environmental Monitoring Systems Laboratory, U. S. Environmental Protection Agency,
Cincinnati, Ohio 45268,1 and Department of Civil and Environmental Engineering,
University of Cincinnati, Cincinnati, Ohio 452212
Received 5 February 1993/Accepted 13 August 1993
A new membrane filter agar medium (MI agar) containing a chromogen, indoxyl-p-n-glucuronide, and a
fluorogen, 4-rnethylumbelliferyl-p-n-galactopyranoside, was developed to simultaneously detect and enumerate
Escherichia coli and total coliforms (TC) in water samples on the basis of their enzyme activities. TC produced
P-galactosidase, which cleaved 4-methylumbelliferyl-p-D-galactopyranoside to form 4-methylumbelliferone, a
compound that fluoresced under longwave UV light (366 nm), while E. coli produced p-glucuronidase, which
cleaved indoxyl-p-n-glucuronide to form a blue color. The new medium TC and E. coli recoveries were
compared with those of mEndo agar and two E. coli media, mTEC agar and nutrient agar supplemented with
4-methylumbelliferyl-p-D-glucuronide, using natural water samples and spiked drinking water samples. On
average, the new medium recovered 1.8 times as many TC as mEndo agar, with greatly reduced background
counts (<7%). These differences were statistically significant (significance level, 0.05). Although the overall
analysis revealed no statistically significant difference between the E. coli recoveries on MI agar and mTEC
agar, the new medium recovered more E. coli in 16 of 23 samples (69.6%). Both MI agar and mTEC agar
recovered significantly more E. coli than nutrient agar supplemented with 4-methyIumbelliferyl-p-D-gIucu-
ronide. Specificities for E. coli, TC, and noncoliforms on MI agar were 95.7% (66 of 69 samples), 93.1% (161
of 173 samples), and 93.8% (61 of 65 samples), respectively. TheE. coli false-positive and false-negative rates
were both 4.3%. This selective and specific medium, which employs familiar membrane filter technology to
analyze several types of water samples, is less expensive than the liquid chromogen and fluorogen media and
may be useful for compliance monitoring of drinking water.
Chromogens and fluorogens, substrates that produce color
and fluorescence, respectively, upon cleavage by a specific
enzyme, have been used for many years to detect and
identify coliform bacteria, including the fecal pollution indi-
cator Escherichia coli (3, 22, 28). Compounds such as
o-nitrophenyl-p-D-galactopyranoside (ONPG), p-nitrophe-
nyl-p-D-galactopyranoside (PNPG), and 4-methylumbel-
liferyl-p-D-galactopyranoside (MUGal) have been included
in a variety of media (3) to demonstrate the presence of
p-galactosidase, an enzyme produced by coliforms, and
4-methylumbelliferyl-p-D-glucuronide (MUG) has been used
to detect E. coli in milk and dairy products (14,19), food and
shellfish (14,19, 20, 25, 31, 32, 40), water and wastewater (8,
9, 13, 19, 21, 27, 29, 31, 40), and urine and other clinical
samples (7, 23, 24, 27, 35) by means of its p-glucuronidase
production. Recently, other chromogens, such as indoxyl-p-
D-glucuronide (IBDG) and 5-bromo-4-chloro-3-indolyl-p-D-
glucuronide (X-Gluc), have also been used to detect or
enumerate E. coli in water (26, 40), urine (11), and food (20,
40). Some of the methods utilize chromogens and fluorogens
in liquid media in a most-probable-number (or multiple-
fermentation-tube) test, presence-absence format, or some
other type of tube test, while others use agar media for direct
plating or membrane filter (MF) technology.
* Corresponding author.
Drinking water regulations under the Final Coliform Rule
(16) state that total coliform (TC)-positive drinking water
samples must be examined for the presence of E. coli or fecal
coliforms. Currently approved MF technology (16) for de-
tecting TC and fecal coliforms uses several different types of
media and two different incubation temperatures. The MF
test for E. coli (17, 29, 39), which uses nutrient agar
supplemented with MUG, is a confirmatory test, not a
primary isolation medium. The combined procedures (TC
test and either fecal coliform or E. coli test) can take 28 to 72
h, and the standard most-probable-number fecal coliform
test (1, 5) can take up to 72 h. The most recently approved E.
coli test, (18), Colilert, is a liquid most-probable-number or
presence-absence medium that can detect both coliforms
and E. coli within 24 to 28 h.
The need for a rapid MF method to detect both types of
bacteria, the increased cost of drinking water testing under
the new regulations, and the potential delay in the detection
of fecally contaminated drinking water with currently used
methods provided the impetus for research to develop an
MF method to simultaneously detect coliforms and E. coli
within 24 h. Other potential uses, such as recreational water,
monitoring source water for potable water, and groundwa-
ter, were also considered during this study. In this paper we
describe the development and evaluation of a new MF
medium, containing the fluorogen MUGal and the chro-
mogen IBDG, that can be used to detect both types of
organisms in a variety of water types.
3534
-------
VOL. 59, 1993
TC AND E. COLI IN WATER
3535
TABLE 1. Parameters tested during medium development
Parameter or type of
compound
Compound used and sources(s)
Concn or levels tested
pH
Chromogen or fluorogen
lac operon inducer
Antibiotic
Buffer
Medium ingredients
Not applicable
ONPG; Boehringer-Mannheim GmbH, Mannheim,
Germany
CPRG, sodium salt; Boehringer-Mannheim GmbH,
Mannheim, Germany
RBDG; Research Organics, Inc., Cleveland, Ohio
MUGal; Boehringer-Mannheim GmbH, Mannheim,
Germany
IBDG, sodium salt; Marcor Development Corp.,
Hackensack, N.J., Arthur Ley, Saul Wolfe
IBDG, cyclohexylammonium salt; Sigma Chemical
Co., St. Louis, Mo.
IPTG; Bethesda Research Laboratories,
Gaithersburg, Md.
Cefsulodin, sodium salt; Marcor Development Corp.,
Hackensack, N.J.
Tris; Fisher Scientific Co., Fair Lawn, N.J.
Sodium lauryl sulfate; BDH Biochemicals, Ltd.,
Poole, United Kingdom
Sodium desoxycholate; Difco Laboratories, Detroit,
Mich.
fi-D-Lactose; Eastman Kodak Co., Rochester, N.Y.
7.0, 7.5, 8.0
100, 200, 400, 800, and 1,600 jig/ml
10, 20, 40, 80, and 160 |i.g/ml
10, 25, 50, 100, 200, and 400 tig/ml
50, 100, and 200 u.g/ml
50, 100, 200, 250, and 320 u.g/ml
50, 100, 200, 250, 320, 400, 600, and 800 tig/ml
0.001 M
2, 3, 4, 5, 7, 10, 15, and 25 u.g/ml
0.005, 0.01, and 0.05 M (at pH 8.0)
50, 100, and 200 tig/ml
25, 50, and 100 fig/ml
0, 0.5, and 1 g/liter
MATERIALS AND METHODS
Medium development, (i) Bacterial cultures. The four co-
liform cultures used in the medium development study
included E. coli ATCC 25922 obtained from Bactrol disks
(Difco Laboratories, Detroit, Mich.) and three U. S. Envi-
ronmental Protection Agency bacterial strains, E. coli EPA
206, Enterobacter aerogenes EPA 202 (also preserved as
ATCC 49701), and Klebsiella pneumoniae EPA 207. Other
cultures used in this study were Flavobacterium multi-
vorum, Flavobacterium thalophilum, and two Aeromonas
hydrophila isolates, all of which were obtained from
Eugene Rice, Risk Reduction Engineering Laboratory, U. S.
Environmental Protection Agency, Cincinnati, Ohio. All
coliforms were characterized by using API 20E strips (Ana-
lytab Products, Plainview, N. Y.), and working cultures of all
bacteria were maintained on tryptic soy agar (Difco) slants
and stabs.
(ii) Parameters examined. A variety of ingredients, chem-
icals, chromogens, fluorogens, and pH values were exam-
ined during the development of the medium for TC and E.
coli. These materials and parameters are listed in Table 1
along with the concentrations, amounts, and levels tested.
The base medium consisted of modified mTEC agar (12, 38)
containing a reduced amount of lactose and no pH indica-
tors. Type GN-6 47-mm-diameter MF (pore size, 0.45 u,m;
Gelman Sciences, Ann Arbor, Mich.) were used for most
filtrations. However, in a study of some chromogens, the
effect of MF composition on color development and diffu-
sion was examined by using cellulose mixed-ester type GN-6
and Supor polysulfone filters (pore size, 0.45 M-m; Gelman),
0.4-jj.m-pore-size polycarbonate filters (Poretics Corp.,
Livermore, Calif.), and 0.2-(im-pore-size polycarbonate fil-
ters (Nuclepore Corp., Pleasant on, Calif.).
(iii) Final medium formulation. The final medium formula-
tion (MI agar) was prepared by adding the following ingre-
dients to 1 liter of distilled water: proteose peptone no. 3
(Difco), 5.0 g; yeast extract (Difco), 3.0 g; p-D-lactose
(Eastman Kodak Co., Rochester, N.Y.), 1.0 g; MUGal
(Boehringer-Mannheim Corp., Indianapolis, Ind.), 0.1 g (fi-
nal concentration, 100 u.g/ml); NaCl, 7.5 g; K2HPO4, 3.3 g;
KH2PO4, 1.0 g; sodium lauryl sulfate (BDH Biochemicals,
Ltd., Poole, United Kingdom), 0.2 g; sodium desoxycholate
(Difco), 0.1 g; and agar (Difco), 15.0 g. After the ingredients
were dissolved, the medium was autoclaved for 15 min at
121°C, and the agar was tempered in a 50°C water bath, the
following were added to 1 liter of medium: 20 ml of a freshly
prepared 16-mg/ml filter-sterilized solution of IBDG (final
concentration, 320 ng/ml) and 5 ml of a freshly prepared
1-mg/ml filter-sterilized solution of cefsulodin (Marcor De-
velopment Corp., Hackensack, N.J.) (final concentration, 5
(Ag/ml). Some of the IBDG was obtained from Arthur Ley,
Queens University, Kingston, Ontario, Canada, and from
Saul Wolfe, Simon Fraser University, Burnaby, British
Columbia, Canada, and some was purchased from Marcor
Development Corp. The medium was then pipetted into petri
dishes (9 by 50 mm; 5 ml per plate), allowed to harden,
stored in a refrigerator, and used within 2 weeks. The media
used in the recovery studies were freshly prepared on the
day before use. Some plates from each batch were put aside
and stored at 4°C for later use in a medium storage study.
The slight precipitate that formed in the agar upon autoclav-
ing and the crystals that formed upon refrigeration did not
affect the performance of the medium.
The substrate MUGal was included in MI agar to detect
TC. The p-galactosidase produced by these organisms
cleaved the MUGal, producing 4-methylumbelliferone, a
compound that fluoresced when it was exposed to longwave
UV light (X., 366 nm). Noncoliforms usually do not produce
this enzyme and, hence, did not fluoresce on the medium.
The IBDG was used to detect E. coli. The (3-glucuronidase
produced by this organism cleaved the substrate to form a
blue color (indigo) in the colonies. Since E. coli is also a TC
and produces p-galactosidase as well as p-glucuronidase, the
blue colonies fluoresced a blue-green color under longwave
UV light. Organisms other than E. coli rarely produce the
blue color. The antibiotic cefsulodin was added to inhibit
gram-positive bacteria and some noncoliform gram-negative
organisms that can cause false-positive reactions.
-------
3536
BRENNER ET AL.
APPL. ENVIRON. MICROBIOL.
Filtration procedure, (i) Pure cultures. The four bacterial
cultures used for the medium development study were each
inoculated into 10-ml tubes containing tryptic soy broth
(TSB) (Difco), which were subsequently incubated for 24 to
48 h at 35°C. Each culture was transferred into fresh TSB the
day before the experiment and incubated for 24 h at 35°C;
10-fold dilutions of each culture were made in phosphate-
buffered dilution water (1, 5), and 1-ml aliquots of the 10~7
and 10~8 dilutions were filtered through 0.45-nm-pore-size
cellulose ester MF. The MF funnels and bases were auto-
claved before each experiment and were exposed to germi-
cidal UV light (X, 254 nm) between nitrations. The niters
were placed on 5-ml plates containing MI agar, the standard
TC medium, mEndo agar (1, 5), and an E. colt medium,
mTEC agar (12, 38). The previously described media were
incubated at their recommended temperatures for the rec-
ommended times (1, 5,12, 38), and MI agar was incubated at
35°C for 16 to 24 h. Colonies that grew on MI agar were
inspected for blue color, fluorescence under longwave UV
light (366 nm), both, or neither. For optimal differentiation of
fluorescent colonies with a binocular dissecting microscope
(magnification, xlO to x!5), a 4-Watt Blak Ray model
UVL-21 longwave UV lamp (UVP, Inc., San Gabriel, Calif.)
was placed about 6 in. (15.3 cm) from the plates. Counts
were made of each of the following four different types of
colonies (if they were present): fluorescent, blue (E. coli);
nonfluorescent, blue (probable E. coli); fluorescent, nonblue
(TC other than E. coli); and nonfluorescent, nonblue (back-
ground or noncoliforms). The number of TC was equal to the
total number of fluorescent colonies (i.e., the sum of the
number of fluorescent, blue colonies and the number of
fluorescent, nonblue colonies). Nonfluorescent, blue colo-
nies, if present, were included in the E. coli count because
their lack of fluorescence was an artifact due to overcrowd-
ing of colonies. Background recoveries were obtained by
subtracting the number of fluorescent, nonblue colonies
from the total number of nonblue colonies on the same plate,
counted under ordinary light, because the background or-
ganisms (nonfluorescent, nonblue) could not be counted
under longwave UV light.
(ii) Natural water samples. After the final medium formu-
lation was selected on the basis of the pure-culture results, a
series of natural water samples and wastewater-spiked tap
water samples were used to evaluate the ability of MI agar to
recover a range of concentrations of both target organism
groups. Several volumes or dilutions of each natural sample
(determined in a preliminary analysis the day before by using
mEndo agar and mTEC agar) were filtered in triplicate, and
the filters were placed on MI agar. Similarly, replicate filters
were placed on mEndo agar and mTEC agar. The media
were incubated for the appropriate times and at the appro-
priate temperatures (1, 5, 12, 38), and the target and nontar-
get (background) colonies on the comparison media were
counted and compared. After the mEndo agar target and
nontarget colonies were counted, the locations of the target
organisms were marked, and the MF from some samples
were transferred to plates containing nutrient agar supple-
mented with MUG (Difco). The plates were incubated for 4
h at 35°C (17, 29, 39), and the target and nontarget colonies
that fluoresced under longwave UV light were counted. The
E. coli recoveries on nutrient agar supplemented with MUG
were compared with those obtained on MI agar and mTEC
agar.
(iii) Preparation of spiked drinking water samples. Spiked
drinking water samples were prepared by adding various
quantities of natural water samples containing both TC and
E. coli to tap water and were used because naturally con-
taminated (i.e., coliform- and£. co//-positive) drinking water
samples were not available. These samples simulated plumb-
ing cross-connections, effluent-contaminated source waters,
and intrusion of fecally contaminated water from the soil into
broken water mains. Preliminary analyses, described below,
were performed the day before the experiment to determine
the amount of natural sample spike and the length of contact
time necessary to achieve an E. coli level of <100 or <10
CFU per 100 ml.
After the tap water was allowed to run for 5 to 10 min, the
preliminary analysis tap water sample was collected asepti-
cally, and free and total chlorine residual levels were deter-
mined by using diethyl-/7-phenylene diamine and a model
CN-66 chlorine test kit (Hach, Loveland, Colo.). Various
amounts or percentages of several natural water samples
were added to measured volumes of the tap water. Following
a suitable contact time, 1 ml of a sterile 10% sodium
thiosulfate solution (1, 5) was added per liter of each sample,
and replicate portions of each sample were filtered. The
filters were placed on mEndo and mTEC agar plates and
incubated at the recommended temperatures (1, 5,12, 38) for
24 h. In general, smaller percentages of sample and longer
contact times were used with effluent samples, while river
water spikes required larger percentages of sample and
shorter contact times because of the lower resistance of the
microorganisms to chlorine.
On the day of the experiment, the tap water line was
flushed, the chlorine residuals were determined, and a
10-liter sample of drinking water was collected and spiked
with the volume or amount of natural sample determined in
the preliminary analysis. After exposure for the predeter-
mined contact time, the chlorine was neutralized, and the
sample was analyzed by the procedure described above for
the natural water samples.
Colony verification and identification. For each of the first
13 natural or spiked water samples analyzed, five isolates (if
present) of each colony type (blue, fluorescent; nonfluores-
cent, blue; fluorescent, nonblue; nonfluorescent, nonblue)
were picked from MI agar for verification and identification.
The isolates were triple-picked to tubes containing TSB,
lauryl tryptose broth (LTB) (Difco), and 2% brilliant green-
lactose-bile broth (BOB) (Difco) if there was sufficient
colonial growth. Whenever there was insufficient growth to
do this, which frequently occurred with the nonfluorescent,
nonblue background colonies, all of the growth from each
colony was transferred to a tube containing TSB. After the
TSB tubes were incubated for 24 to 72 h at 35°C, the cultures
were transferred to tubes containing the verification media.
All of the LTB and BGB tubes were incubated for up to 48
h at 35°C and inspected for growth and gas production (1, 5),
and the LTB tubes with growth (with or without gas produc-
tion) were transferred to EC medium (Difco). The EC
medium tubes were incubated for 24 h at 44.5°C and in-
spected for growth and gas production (1, 5). Each TSB
culture was streaked onto eosin-methylene blue agar (Difco)
plates, which were then incubated for 24 to 72 h at 35°C.
Isolated colonies that were >3 mm in diameter were picked
for identification with API 20E strips. Additional biochemi-
cal tests recommended by the manufacturer were performed
if necessary to identify the organisms. In addition, all of the
colonies on one filter from the tap water sample spiked with
5% Sycamore Wastewater Treatment Plant nonchlorinated
secondary effluent were similarly picked, verified, and iden-
tified. The specificity, selectivity, and false-positive and
-------
VOL. 59, 1993
TC AND E. COLI IN WATER
3537
false-negative rates were calculated by using the American
Society for Testing and Materials procedure (2).
Relative accuracy of MI agar. The three Environmental
Protection Agency stock coliform bacteria used in the me-
dium development study were grown overnight in TSB, and
10-fold dilutions in phosphate-buffered dilution water were
prepared (1, 5). Five 1-ml aliquots of the 10~7 and 10~8
dilutions of each organism were filtered, and the filters were
placed on MI agar plates. A heterotrophic plate count (1, 5)
was performed by using five 1-ml volumes of each of the
same two dilutions of each of the cultures (1, 5), and the
recovery data were used to determine the relative accuracy
of MI agar compared with the reference test.
Medium storage study. Dilutions of EPA 206 E. coll and a
natural sample (Ohio River at the public landing) were
filtered in replicate, and the filters were placed on 5-ml MI
agar plates that had been stored in a refrigerator at 4°C for
various lengths of time (1 day and 2, 4, 7, 9, 13, and 14
weeks). After incubation, counts were made, and the recov-
eries of E. coli, TC, and background organisms were com-
pared to determine the changes in concentration over time
and the maximum length of time that the medium could be
stored and still be effective.
Comparison of media containing autoclaved and filtered
IBDG. To determine whether IBDG could be autoclaved
along with the base medium, thereby simplifying prepara-
tion, base medium was split into two equal portions, one of
which received a final concentration of 320 fig of IBDG per
ml prior to autoclaving and the other received IBDG after
sterilization. Cefsulodin was added to both media after
autoclaving. A series of 10 natural and/or simulated contam-
inated tap water samples were analyzed with these two
media, and the recoveries and colony specificities were
compared.
Comparison of the sodium and cyclohexylammonium salts
of IBDG. To determine whether the cyclohexylammonium
salt of IBDG (Sigma Chemical Co., St. Louis, Mo.) per-
formed as well as the sodium salt of IBDG, base medium was
autoclaved and split into two equal volumes, one of which
received 320 u,g of the sodium salt of IBDG per ml and the
other received 320 u.g of the cyclohexylammonium salt of
IBDG per ml. Cefsulodin was added to both preparations
after autoclaving. A series of 10 natural and/or simulated
contaminated drinking water samples were analyzed with
these two formulations, and the recoveries and colony
specificities were compared. In addition, media containing
autoclaved and filtered cyclohexylammonium salt of IBDG
were prepared and tested as described above for the sodium
salt of IBDG.
Statistical analysis of the data. The recovery data in CPU
per 100 ml for all media were log transformed (base e) and
tested for normality by using the Shapiro-Wilk test (41). The
recoveries of TC and E. coli on MI agar were compared with
those obtained on mEndo agar and mTEC agar, respec-
tively, by using analysis of variance for a randomized block
design (34). In addition, MI agar E. coli recoveries from
12 samples were compared with E. coli recoveries on nutri-
ent agar supplemented with MUG. An analysis of variance
was performed on the means of the data, which were log
transformed (base e). Recoveres of E. coli on media con-
taining filtered and autoclaved IBDG and on media contain-
ing the sodium and cyclohexylammonium salts of IBDG
were also compared by analysis of variance. The differences
were considered to be statistically significant at a signifi-
cance level of 0.05. To examine the medium performance
over a range of sample types and concentrations, the sam-
ples were also grouped into categories by water type (pota-
ble water, which included the spiked tap water and ground-
water samples; surface water; and wastewater effluents) and
by E. coli counts (0 to 10,11 to 100,101 to 1000, and > 1,000
E. coli cells per 100 ml) and compared by analysis of
variance. Scheffe's test for homogeneity of variance (33) was
used to compare the relative precision values for the log-
transformed (base e) data, while the relative accuracy of MI
agar for recovering the three coliform cultures compared
with the reference test was determined by using a variance
ratio test (34) with Bonferroni-adjusted comparisons (30).
RESULTS
Medium development study. When pure cultures of E. coli,
Enterobacter aerogenes, and K. pneumoniae were used, an
examination of media containing various combinations of
chromogens and other ingredients listed in Table 1 showed
that the medium containing MUGal and IBDG (i.e., MI agar)
was the best medium for the simultaneous detection of TC
and E. coli. On this medium, E. coli colonies were blue in
ambient light and exhibited blue-green fluorescence when
the plates were exposed to longwave UV light, while the
colonies of other TC were cream colored in ambient light and
exhibited blue-white fluorescence under longwave UV light.
The background or noncoliform colonies, which were also
cream colored in ambient light, did not fluoresce when they
were exposed to UV light.
The chromogen ONPG was not useful in the agar medium
because of the natural cream to pale yellow color of the
bacterial colonies. TC were difficult to distinguish even when
concentrations up to 1,600 u,g/ml were used. In addition,
extensive lateral diffusion of color further hindered target
colony discrimination.
The combination of chlorophenol red-|3-D-galactopyrano-
side (CPRG) and IBDG did not provide enough color con-
trast between the orchid TC colonies and the blue E. coli
colonies, especially when enzyme production was reduced
or slow. Raising the pH to 7.5 or 8.0 and adding the
gratuitous lac operon inducer isopropyl-(3-D-thiogalactopyr-
anoside (IPTG) (10) did not improve CPRG color develop-
ment, nor did changes in the concentrations of other ingre-
dients result in any further improvements. However,
addition of cefsulodin greatly reduced the background
counts obtained with all of the formulations and combina-
tions tested.
Although a medium containing a combination of resorufin-
(J-D-galactoside (RBDG) and IBDG showed promise ini-
tially, the large amount of lateral diffusion of pink color away
from the TC colonies made the differentiation of target
colonies against the pink MF difficult. Reduction of the
concentration of RBDG to the point at which there was little
or no color diffusion resulted in pale pink colonies that were
hard to distinguish from the nontarget organisms. Attempts
to decrease the diffusion by using reduced amounts of
sodium lauryl sulfate and/or sodium desoxycholate, MF
pretreatments, and filters having different compositions were
also unsuccessful (6).
Medium storage study. The results of the medium storage
study showed that the recoveries of E. coli from the natural
sample remained constant on MI agar stored at 4°C for up to
14 weeks. Similarly, pure culture recoveries of E. coli did
not change as the medium aged, but the sizes of the colonies
decreased. TC recoveries on MI agar that was refrigerated
for 2 weeks or more increased to twice the original value, but
-------
3538
BRENNER ET AL.
APPL. ENVIRON. MICROBIOL.
then remained constant on medium kept up to 14 weeks. The
increase was primarily due to the growth of green fluorescent
(false-positive) organisms. Media stored for more than 1
month produced an increase of 1 Iog10 in background
growth, but no further increase occurred on media stored up
to 14 weeks. However, the elevated levels of nontarget
organisms were still less than those found on freshly pre-
pared mEndo agar.
Evaluation of the final medium formulation with natural
samples and simulated contaminated drinking water samples.
(i) TC. Recoveries of TC, E. coli, and background or
noncoliform colonies on MI agar from natural and spiked
drinking water samples were compared with those obtained
with the comparison media. All 26 samples analyzed were
TC positive, and 23 contained E. coli (Tables 2 and 3). On
the average, the new medium recovered 1.8 times as many
TC as mEndo agar, and this difference was statistically
significant (P = 0.0001). When the data were analyzed by
water type (surface water, wastewater effluent, or potable
water, which included groundwater and spiked tapwater),
MI agar recovered significantly more TC than mEndo agar
with potable water samples (P = 0.0002) and surface water
samples (P = 0.0001), but not with wastewater effluent
samples (P = 0.5399).
(ii) E. coli. Although MI agar recovered more E. coli from
16 of 23 water samples (69.6%), the overall analysis revealed
no significant difference between the E. coli recoveries on
MI agar and mTEC agar, a result that might have been
expected since MI agar is a modification of mTEC agar.
Most of the samples with greater recovery on MI agar
contained less than 1,000 E. coli per 100 ml, while three of
the four samples with counts greater than 1,000 had de-
creased recovery compared with mTEC agar. When the high
counts were omitted from the analysis, the new medium
recovered 21% more E. coli than mTEC agar on the average.
This difference was statistically significant (P = 0.0307).
More samples will be needed to determine whether the
difference observed with high counts with effluents is real or
an artifact produced by too few samples.
Results of an analysis using the four E. coli count catego-
ries (0 to 10, 11 to 100, 101 to 1,000, and > 1,000 E. coli per
100 ml) showed that MI agar recovered significantly more E.
coli than mTEC agar only with the 11 to 100 catetory (P =
0.0012). When the data were analyzed by water type, no
differences were found among the three types of water
samples and between the two subgroups of potable water
samples (i.e., spiked tap water and groundwater).
Using 12 samples, the E. coli recoveries on nutrient agar
supplemented with MUG, a medium utilized with mEndo
agar in a newly promulgated MF E. coli method (17, 29, 39),
were compared with those on MI agar and mTEC agar.
Recovery on nutrient agar supplemented with MUG was
significantly lower than on either of the other two media (P
< 0.0004). In addition, the precision of this medium, deter-
mined by Scheffe's test for homogeneity of variance, was
significantly lower than the precision of MI agar (P =
0.0019), but it was not different from the precision of mTEC
agar.
(iii) Background. The background counts on the new
medium averaged <7% of the background counts on mEndo
agar, and this difference was highly significant (P < 0.0001).
The background counts on MI agar were significantly lower
than those on mEndo agar with all three water types and all
four count categories tested.
Precision of MI agar. The coefficients of variation for TC
enumeration on MI agar and mEndo agar were 17.6 and
19.5%, respectively, while the coefficients of variation for E.
coli enumeration on MI agar and mTEC agar were 25.1 and
33.7%, respectively. These differences were not statistically
significant.
Medium specificity. Of the 265 isolates picked for identifi-
cation with API 20E strips, 238 were successfully identified,
23 failed to grow after subculturing from MI agar, and 4 had
repeatedly unacceptable profile numbers. The identities of
the four types of colonies are shown in Table 4. A total of 57
of 59 fluorescent, blue colonies (96.6%) and 9 of 10 nonflu-
orescent, blue colonies (90%) were E. coli. The nonfluores-
cent, blue colonies were found only on overcrowded plates,
and 8 of 9 of these organisms demonstrated fluorescence
when they were retested on MI agar. Since the nonfluores-
cence appeared to be an artifact due to overcrowding, the
two groups were combined. The specificity for the combined
blue groups was 95.7% (66 of 69 colonies). Only 74.2% of the
E. coli isolates (49 of 66 isolates) were positive for growth
and gas production in EC broth. The remaining 25.8% grew,
but were anaerogenic.
Three blue colonies on MI agar were not E. coli, giving a
false-positive rate of 4.3% (3 of 69 colonies), and three
organisms that demonstrated fluorescence but were not blue
(i.e., TC) were actually E. coli, resulting in a false-negative
rate of 4.3% (3 of 69 colonies). The false-negative isolates
were further tested by using an E. coli 0157 latex test (Oxoid,
Ltd., London, United Kingdom) and H7 antiserum (Difco)
to determine whether enterohemorrhagic strains were
present. These strains do not produce p-glucuronidase or
grow at 44.5°C (22). However, they would be detected as TC
on MI agar because it is incubated at 35°C. The false-
negative isolates tested in this study were not E. coli
0157:H7.
The specificity of the medium for TC, which included all
blue colonies and all fluorescent, nonblue colonies, varied
depending on how TC were defined. When only the four
typical genera (Escherichia, Enterobacter, Citrobacter, and
KLebsiella) (5) were considered, the new medium TC speci-
ficity was 87.9% (152 of 173 colonies). However, if the
organisms that verified in LTB and BOB (i.e., LTB+ BGB+)
with genus names other than the four listed above were
added, the specificity was 93.1% (161 of 173 colonies). The
average coliform verification rate for typical coliforms was
77.6% (118 of 152 colonies), and the overall verification rate
was 74.6% (129 of 173 colonies).
Some typical coliforms did not verify in both LTB and
BOB, although they were usually able to utilize one of
the lactose-containing media. This may have been due to
injuries sustained from prolonged exposure to the longwave
UV light used during the counting and colony-picking pro-
cedures. A longwave UV light exposure study (6) showed
that verification of fluorescent, blue E. coli colonies was not
affected by exposure times up to 60 min, but the verification
rate for the fluorescent, nonblue TC colonies decreased
rapidly after 15 min of exposure. The background colony
verification rate was also affected, but to a lesser degree.
Only 49.2% of the background isolates (32 of 65 isolates)
were members of typical noncoliform genera (i.e., genera
other than Escherichia, Enterobacter, Citrobacter, and
Klebsielld). However, when the background organisms that
did not verify (i.e., LTB+ BOB", LTB~ BGB+, and LTB~
BGB~ organisms) were added, the specificity was 93.8% (61
of 65 isolates).
More than half of the background isolates (18 of 33
isolates) were identified as Enterobacter agglomerans, an
organism previously classified in the noncoliform genus
-------
VOL. 59, 1993
TC AND E. COLI IN WATER 3539
TABLE 2. Comparison of the numbers of TC detected and selectivity characteristics of MI agar and mEndo agar
Sample used"
Sycamore WTP
nonchlorinated
secondary effluent
Ludlow Springs
Tylersville Artesian Well
Ohio River (Anderson Ferry)
Clough Creek
East Fork Artesian Well
Great Miami River
(Dayton, Ohio)
Clough Springs
Ohio River (public landing)
Winton Woods Lake
Sharon Woods Lake
Ohio River (public landing)
Tap water + 5%
Muddy Creek WTP
nonchlorinated
secondary effluent
Tap water + 5%
Sycamore WTP
nonchlorinated
secondary effluent
Tap Water + 10%
Mill Creek WTP
nonchlorinated
secondary effluent
Tap water + 25%
Great Miami River water
Tap water + 40%
Ohio River water (Anderson Ferry)
Muddy Creek WTP
nonchlorinated
secondary effluent
Muddy Creek WTP chlorinated
secondary effluent
Ohio River (Anderson Ferry)
Tap water + 10% Muddy Creek
WTP primary effluent
Muddy Creek WTP
primary effluent
Clough Springs
Stonelick Lake (beach)
Stonelick Lake (campground)
Great Miami River
(Dayton, Ohio)
No. of TC per 100 ml on6:
MI agar
3.56 x 106
1.06 x 104
0.33
1.62 x 104
9.19 x 102
1.64 x 102
7.85 x 103
1.67 x 102
2.27 x 103
9.22 x 102
6.95 x 103
1.64 x 10"
46.4
9.53
6.59 x 103
4.76 x 102
2.81 x 102
5.78 x 10"
6.35 x 103
3.96 x 105
1.31 x 10"
2.72 x 106
2.25 x 102
1.42 x 103
1.75 x 104
2.32 x 104
mEndo agar
1.77 X 106
1.05 x 103
0.11
7.43 x 103
9.67 x 102
1.51 x 102
3.32 x 103
57
2.39 x 103
2.11 x 102
5.53 x 102
6.20 x 103
31.0
14.2
3.60 x 103
2.00 x 102
25
7.30 x 104
5.70 x 103
1.43 x 105
2.23 x 103
1.74 x 106
60.3
1.17 x 102
1.29 x 103
1.61 x 103
No. of noncoliforms per
100 ml on*:
MI agar
3.00 x 105
4.17 x 103
0.0
9.40 x 103
2.50 x 102
1.30 x 102
2.03 x 103
2.67 x 102
4.87 x 102
2.53 x 102
4.67 x 103
9.63 x 103
6.33
1.00
1.97 x 103
3.37 x 102
1.33
1.07 x 104
2.87 x 103
7.00 x 104
1.19 x 104
8.67 x 105
1.37 x 102
8.23 x 103
2.80 x 103
1.37 x 103
mEndo agar
1.18 x 107
6.57 x Iff
15.9
1.82 x 10*
1.67 x 103
2.89 x 102
1.02 x 10*
3.93 x 102
5.20 x 103
1.89 x 103
9.67 x 103
2.68 x 104
85.3
11.8
1.79 x 104
7.40 x 102
84.7
1.04 x 105
1.88 x 10*
3.27 x 105
1.81 x 104
6.13 x 106
2.87 x 102
2.28 x 10"
2.33 x 104
2.06 x 104
MITC/
2.01
10.09
3.00
2.18
0.95
1.09
2.36
2.93
0.95
4.37
12.57
2.65
1.50
0.67
1.83
2.38
11.24
0.79
1.11
2.77
5.87
1.56
3.65
12.14
13.57
14.41
Recovery ratios^
mEX
0.03
0.63
0.00
0.52
0.15
0.45
0.20
0.68
0.09
0.13
0.48
0.36
0.07
0.08
0.11
0.46
0.02
0.10
0.15
0.21
0.66
0.14
0.48
0.36
0.12
0.07
MI^
MINT
11.87
2.54
NDd
1.72
3.68
1.26
3.87
0.63
4.66
3.64
1.49
1.70
7.33
9.53
3.35
1.41
211.28
5.40
2.21
5.66
1.10
3.14
1.61
0.17
6.25
16.93
mEndoNT
0.15
0.16
0.01
0.41
0.58
0.52
0.33
0.15
0.46
0.11
0.06
0.23
0.36
1.20
0.20
0.27
0.30
0.70
0.30
0.44
0.12
0.28
0.21
0.01
0.06
0.08
* WTP, Wastewater Treatment Plant.
6 Values are the means of three replicates.
c Ratios of the recoveries of TC and/or nontarget organisms on two different media. The values indicate that there were significant differences in the recovery
and selectivity on the different media. Abbreviations: MITC> number of TC recovered on MI agar; mEndOrc, number of TC recovered on mEndo agar; MIfn-,
number of background or nontarget organisms recovered on MI agar; mEndOfn-, number of background or nontarget organisms recovered on mEndo agar. The
mean recovery ratios were as follows: MI-rc/mEndOrc, 4.56; MlN-p/mEndo,^, 0.26; MI-rc/MIfj-r, 12.50; mEndOrc/mEndOOT, 0.30.
d ND, not determined.
Erwinia (15). Currently, these bacteria are considered
members of the genus Enterobacter on the basis of a
DNA homology of 70% (15) and, hence, are typical
coliforms. However, the reactions of the isolates with con-
ventional media (6) were those of noncoliforms; these iso-
lates failed to ferment lactose in LTB and/or BOB and were
unable to produce the typical golden green metallic sheen
when growth was streaked onto mEndo agar. The lack of
fluorescence of the colonies observed on MI agar may have
been caused by insufficient p-galactosidase production due
to injury to the organisms or by blocking or quenching of the
fluorescence by the natural carotenoid pigment found in the
organisms (37). Alternately, these bacteria may be Erwinia
species that are misclassified as coliforms.
Use of various forms of IBDG in MI agar. Studies (6) of the
salt form of the chromogen and the method of sterilization
revealed no significant differences in chromogen color de-
velopment and E. coli recovery. A study of the TC, E. coli,
-------
3540
BRENNER ET AL.
APPL. ENVIRON. MICROBIOL.
TABLE 3. Comparison of E. coli recoveries on three media
No. of E. coli per 100 ml onfc:
Sample used"
Sycamore WTP
nonchlorinated
secondary effluent
Ludlow Springs
Ohio River
(Anderson Ferry)
Clough Creek
Clough Springs
Ohio River
(public landing)
Winton Woods Lake
Sharon Woods Lake
Ohio River
(public landing)
Tap water + 5%
Muddy Creek WTP
nonchlorinated
secondary effluent
Tap water + 5%
Sycamore WTP
nonchlorinated
secondary effluent
Tap water + 10%
Mill Creek WTP
nonchlorinated
secondary effluent
Tap water + 25%
Great Miami River water
(Dayton, Ohio)
Tap water + 40%
Ohio River water
(Anderson Ferry)
Muddy Creek WTP
nonchlorinated
secondary effluent
Muddy Creek WTP
chlorinated
secondary effluent
Ohio River
(Anderson Ferry)
Tap water + 10%
Muddy Creek WTP
primary effluent
Muddy Creek WTP
primary effluent
Clough Springs
Stonelick Lake
(beach)
Stonelick Lake
(campground near
sewage lagoon)
Great Miami River
(Dayton, Ohio)
MI agar
5.67 x 104
22.3
817
35.7
0.11
72.7
9.07
85.3
477
0.87
0.40
125
7.25
0.78
3.77 x 103
180
9.00 x 103
48.3
1.53 x 10s
1.58
144
257
16.3
mTEC agar
4.67 x 104
5.67
630
11.3
0.11
49.3
8.00
80.0
410
0.53
0.60
167
3.42
0.44
4.37 x 103
150
9.33 x 103
7.22
1.67 x 105
2.92
134
225
15.7
Nutrient agar
supplemented with MUG
ND-*
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
167
6.67
0.67
4.33 x 103
(1.00 x lO4)'
100
(1.30 x 103)
9.67 x 103
(4.37 x 104)
33.3
(66.7)
1.30 x 105
(6.33 x 104)
0.67
(4.33)
167
(TNTCy
53.7
(TNTC)
28.0
Recovery ratios'
MI agar/mTEC agar "^t
1.21
3.93
1.30
3.16
1.00
1.47
1.13
1.07
1.16
1.64
0.67
0.75
2.12
1.77
0.86
1.20
0.96
6.69
0.92
0.54
1.07
1.14
1.04
ir/nutrient agar
aining MUG
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
0.75
1.09
1.16
0.87
1.80
0.93
1.45
1.18
2.36
0.86
4.79
0.58
" WTP, Wastewater Treatment Plant.
b Values are the means of three replicates.
c Ratios of the recoveries of E. coli on two different media. The mean recovery ratios were as follows: MI agar/mTEC agar, 1.60; MI agar/nutrient agar
supplemented with MUG, 1.49.
d ND, not done.
" The values in parentheses are the counts for nontarget or background organisms on mEndo agar that gave positive reactions on nutrient agar supplemented
with MUG.
f TNTC, too numerous to count.
-------
VOL. 59, 1993
TC AND E. COLI IN WATER
3541
TABLE 4. Identification of colonies picked from MI agar
Species
Escherichia colt
Escherichia vulneris
Enterobacter species
Enterobacter amnigenus 2
Enterobacter cloacae
Enterobacter intermedium
Enterobacter agglomerans
Enterobacter aerogenes
Klebsiella species
Klebsiella pneumoniae
Klebsiella oxytoca
Klebsiella ozaenae
Citrobacter freundii
Citrobacter amalonaticus
Serratia species
Serratia liquefaciens
Serratia fonticola
Serratia pfymutheca
Hafnia alvei
Kluyvera species
Cedacea lapagei
Yersinia pestis (presumptive)
Alcaligenes species
Acinetobacter calcoaceticus subsp. Iwoffi
Salmonella species
Fluorescent Pseudomonas group
Centers for Disease Control group 5-E1
(Pseudomonas luteola)
Centers for Disease Control group 5-E2
(Flavimonas oryzihabitans)
Pseudomonas putrefaciens
Pseudomonas fluorescens
Pseudomonas maltophilia
Aeromonas hydrophila
Achromobacter species
No. of isolates identified in each category
(% of total no. of isolates in the category)
Fluorescent, „ „ . Fluorescent, nonblue
. , , . Nonnuorescent, , . ,™ ,
blue colonies . . . . ' colonies (TC other
,,. ... blue colonies .. .i ...
(E. colt) than E. coll)
57 (96.6) 9 (90.0) 3 (2.9)
2 (1.9)
6 (5.8)
1 (1.7) 2 (1.9)
14 (13.5)
2 (1.9)
7 (6.7)
3 (2.9)
1 (1.0)
19 (18.3)
21 (20.2)
1 (1.0)
4 (3.8)
1 (1.0)
4 (3.8)
6 (5.8)
3 (2.9)
1 (1.0)
1 (10.0) 2 (1.9)
1 (1.7) 2 (1.9)
Nonfluorescent,
nonblue colonies
(background)
7 (10.8)
4 (6.2)
1 (1.5)
18 (27.7)
1 (1.5)
1 (1.5)
1 (1.5)
2 (3.1)
2 (3.1)
3 (4.6)
1 (1.5)
1 (1.5)
1 (1.5)
10 (15.4)
1 (1.5)
1 (1.5)
2 (3.1)
1 (1.5)
2 (3.1)
1 (1.5)
2 (3.1)
1 (1.5)
1 (1.5)
Total no. of
isolates
(% of total)
69 (29.0)
2 (0.8)
13 (5.5)
3 (1.3)
18 (7.6)
3 (1.3)
25 (10.5)
3 (1.3)
1 (0.4)
20 (8.4)
22 (9.2)
2 (0.8)
4 (1.7)
1 (0.4)
4 (1.7)
8 (3.4)
5 (2.1)
1 (0.4)
6 (2.5)
3 (1.3)
1 (0.4)
1 (0.4)
1 (0.4)
10 (4.2)
1 (0.4)
1 (0.4)
2 (0.8)
1 (0.4)
2 (0.8)
1 (0.4)
2 (0.8)
1 (0.4)
1 (0.4)
Total"
59
10
104
65
238*
" The percentages of the total number of isolates in the four categories were as follows: fluorescent, blue colonies, 24.8%; nonfluorescent, blue colonies, 4.2%;
fluorescent, nonblue colonies, 43.7%; and nonfluorescent, nonblue colonies, 27.3%.
6 This value does not include organisms that failed to grow and organisms that repeatedly produced unacceptable profiles.
and background specificities (6) obtained for the four differ-
ent media revealed few differences in the types and numbers
of species found in each colony category.
Relative accuracy of the new medium. The new medium
recovered 97.9% of the E. coli recovery of the heterotrophic
plate count. The recoveries of the TC K. pneumoniae and
Enterobacter aerogenes were 85.7 and 87.5%, respectively,
of the reference test values. None of these differences was
statistically significant.
DISCUSSION
The new medium, MI agar, is sensitive, selective, and
specific and has low false-positive and false-negative rates.
In addition, it is precise and accurate in recovering the two
target organisms. The combined E. coli specificity rate of
95.7% is similar to that reported for other media (19, 22, 24),
but differs from the results of Chang et al. (7). However,
after subculturing and storage on tryptic soy agar slants and
stabs, 29% of the MI agar E. coli isolates (20 of 69 isolates)
restreaked on the new medium were p-glucuronidase nega-
tive (i.e., they were not able to produce the same fluores-
cent, blue color formed during primary isolation); 19 of these
isolates were also MUG negative in the Colilert test. Simi-
larly, subculturing and/or storage may explain the large
number of (i-glucuronidase-negative isolates found by Chang
and coworkers (7), or their use of a lactose-containing me-
dium for isolation (MacConkey agar) may have inhibited the
MUG reaction (36).
In addition, familiar MF technology is used to simulta-
neously detect both TC and E. coli from a variety of water
samples. Use of this method with potable water should
simplify drinking water laboratory compliance with the Final
Coliform Rule (16), eliminate the additional time, labor, and
expense of repeat or serial analyses that can delay the
detection of contaminated drinking water, and obviate the
need for a second incubator set at the elevated temperature
-------
3542
BRENNER ET AL.
APPL. ENVIRON. MICROBIOL.
needed with some media. The plates can be observed for the
presence or absence of fluorescence indicative of TC and
blue color indicating the presence of E. coli, or actual counts
can be made to monitor distribution lines or treatment plant
effectiveness. Furthermore, MI agar is capable of recovering
E. coli from water samples containing high particulate con-
centrations and is less expensive than the liquid media
containing chromogens and/or fluorogens.
This medium is not the first medium to detect the two
groups of organisms simultaneously. The agar medium de-
veloped by Petzel and Hartman (31) combined a selective
medium for TC identification with detection of E. coli with
MUG. Problems encountered with this medium included the
inability to use standard diluents, a high false-positive rate
when high levels of flavobacteria or other oxidase-positive
organisms (e.g., Aeromonas spp.) were present in the water
samples, and difficulty in distinguishing the natural fluores-
cence of pseudomonads from the fluorescence produced
during substrate breakdown. Furthermore, this medium
could not be used for precise enumeration of the target
organisms because of the large number of other gram-
negative bacteria that grew on it. Some similar problems
were experienced by other workers (21). Unlike the media of
these investigators (21, 31), MI agar inhibited the growth of
Flavobacterium and Aeromonas species, background organ-
isms that can cause false-positive results on coliform media
(21). A study (6) using two cultures of each genus showed
that the recoveries of aeromonads that were able to grow on
the base medium without cefsulodin were reduced by more
than 4 and 5 Iog10 when the antibiotic was included in the
formulation. The Flavobacterium species were unable to
grow on the base agar alone.
Berg and Fiksdal (4) incorporated the fluorogen MUGal
into liquid and agar media to detect TC in water within
15 min and fecal coliforms within 6 h, respectively. How-
ever, the TC method required expensive and complex
equipment, and the level of detection was unsuitable for use
with drinking water. The lack of specificity of MUGal for
fecal coliforms and the lower, nonstandard incubation tem-
perature (41.5°C) may result in an increased false-positive
rate.
Several commercially available liquid presence-absence or
most-probable-number media (Colilert [Environetics, Inc.,
Branford, Conn.] and Colisure [Millipore Corp., Bedford,
Mass.]) have been developed to detect TC and E. coli in
water samples within 24 to 28 h, but only the Colilert test has
been approved for drinking water analysis (18). p-Galacto-
sidase and p-glucuronidase reactions of the TC, E. coli, and
background MI agar isolates identified in this study showed
92.2% (118/128), 97.1% (67/69), and 91.5% (43/47) agree-
ment, respectively, with those of Colilert.
An agar medium (27) using two different enzyme sub-
strates and a different base medium was developed in
Germany to simultaneously detect both TC and E. coli. TC
colonies were identified by the production of a blue color
from p-galactosidase cleavage of the substrate 5-bromo-4-
chloro-3-indolyl-p-D-galactopyranoside (X-Gal), while E.
coli colonies were detected by the fluorescence of 4-methyl-
umbelliferone, produced by the cleavage of MUG by p-glu-
curonidase. Little is known about the performance of this
medium, as it has not been used or tested in the United
States, and the base medium is not available in this country.
However, these two enzyme substrates, as well as several
other combinations of chromogens and/or fluorogens, also
work in MI base agar medium (6).
The new medium is not completely without problems. For
example, the cefsulodin must be added after the base agar
medium is autoclaved, as this antibiotic is destroyed by
excessive heat. The presence of some lateral diffusion of
blue color away from the target E. coli colonies can make
enumeration and colony picking more difficult on over-
crowded plates (i.e., plates containing >200 colonies of all
types). However, this problem should not affect filters
containing low counts, such as those obtained with drinking
water, and filtration of multiple volumes or dilutions of other
water types should result in filters containing colonies that
can be easily enumerated.
With a few samples, tiny flat or peaked pinpoint blue
colonies (<0.5 mm in diameter on plates containing <200
colonies) were found along with the usual large fluorescent,
blue E. coli colonies (1 to 3 mm in diameter). Ten colonies
were picked and identified with API 20E strips as Hafnia
alvei (three colonies), Enterobacter amnigenus 2 (one colo-
ny), Escherichia vulneris (four colonies), and Citrobacter
freundii (two colonies). The reason for the production of
(J-glucuronidase by these organisms is not known, but other
researchers (22) have also reported p-glucuronidase activity
in some of these species. Alternately, the reaction may be
plasmid mediated. When only the tiny pinpoint blue colonies
are present on filters, water samples should not be consid-
ered E. coft-positive until the identity of at least one colony
has been verified by another method (e.g., EC medium
supplemented with MUG [17, 32] or API 20E strips). The
large, very pale blue colonies occasionally observed did not
appear to be a problem, as the colonies picked for identifi-
cation were shown to be E. coli.
With some samples, a few bright green, fluorescent,
nonblue colonies were observed along with the typical
blue-white fluorescent TC colonies. Several of the green
colonies were picked and identified as Flavobacterium men-
ingosepticum (1 of 7 colonies), Achromobacterxylosoxidans
(4 of 7 colonies), and fluorescent Pseudomonas species (2 of
7 colonies). Unlike the fluorescent green organisms found by
Petzel and Hartman (31), those found on MI agar could be
distinguished from the blue-white fluorescent TC and, when
present, should be eliminated from the TC count. These
organisms generally represented <5% of the population and
were never observed in the absence of the typical blue-white
fluorescent coliforms. An increase in the number of green
colonies may indicate an unusual sample population or a
breakdown of the cefsulodin in the medium.
Furthermore, with some samples, the ideal volume for
E. coli enumeration may not be optimum for TC enumera-
tion and vice versa. However, since blue E. coli colonies
were clearly visible on a TC background that was too
numerous to count, this should be of minor importance for
drinking water compliance purposes, and multiple volumes
or dilutions of other water types should provide accurate
enumeration.
Last and most important, because of the increased recov-
ery of the two target organism groups and the greater
accuracy in detecting E. coli with the new medium, drinking
water facilities may be out of compliance more often if this
medium is used. However, use of other less sensitive or less
specific methods could result in the potential use of contam-
inated water by the public or the rejection of acceptable
water by water treatment facilities.
In conclusion, the new medium, MI agar, shows promise
for use in monitoring several different types of water,
including drinking water. The results of a study initiated to
confirm the ability of MI agar to recover chlorine-stressed or
-------
VOL. 59, 1993
TC AND E. COLI IN WATER
3543
damaged organisms in drinking water will be reported else-
where.
ACKNOWLEDGMENTS
We thank Larry Wymer and Manohari Sivaganesan of Computer
Services Corp. for the statistical analysis of the data and Doris
Morris for typing the manuscript. We also thank Gabriele Reischl
for literature translations.
REFERENCES
1. American Public Health Association. 1989. Standard methods for
the examination of water and wastewater, 17th ed. American
Public Health Association, Washington, D.C.
2. American Society for Testing and Materials. 1988. Standard
practice for establishing performance characteristics for colony
counting methods in bacteriology, designation D 3870-79, p.
465-467. In 1988 Annual book of ASTM standards, section 11:
water and environmental technology, vol. 11.02. American
Society for Testing and Materials, Philadelphia.
3. Bascomb, S. 1987. Enzyme tests in bacterial identification.
Methods Microbiol. 19:105-160.
4. Berg, J. D., and L. Kiksdal. 1988. Rapid detection of total and
fecal coliforms in water by enzymatic hydrolysis of 4-methyl-
umbelliferone-p-D-galactoside. Appl. Environ. Microbiol. 54:
2118-2122.
5. Bordner, R., J. Winter, and P. Scarpino (ed.). 1978. Microbio-
logical methods for monitoring the environment: water and
wastes. Publication EPA-600/8-78-017. Environmental Monitor-
ing and Support Laboratory, U. S. Environmental Protection
Agency, Cincinnati.
6. Brenner, K. P., and C. C. Rankin. Unpublished data.
7. Chang, G. W., J. Brill, and R. Lum. 1989. Proportion of
p-D-glucuronidase-negative Escherichia coli in human fecal
samples. Appl. Environ. Microbiol. 55:335-339.
8. Clark, D. L., B. B. Milner, M. H. Stewart, R. L. Wolfe, and
B. H. Olson. 1991. Comparative study of commercial 4-methyl-
umbelliferyl-p-D-glucuronide preparations with the Standard
Methods membrane nitration fecal coliforrn test for the detec-
tion of Escherichia coli in water samples. Appl. Environ.
Microbiol. 57:1528-1534.
9. Covert, T. C., L. C. Shadix, E. W. Rice, J. R. Haines, and R. W.
Freyberg. 1989. Evaluation of the Autoanalysis Colilert test for
detection and enumeration of total coliforms. Appl. Environ.
Microbiol. 55:2443-2447.
10. Dalbow, D. G., and H. Bremer. 1975. Metabolic regulation of
fi-galactosidase synthesis in Escherichia coli: a test for consti-
tutive ribosome synthesis. Biochem. J. 150:1-8.
11. Delisle, G. J., and A. Ley. 1989. Rapid detection of Escherichia
coli in urine samples by a new chromogenic p-glucuronidase
assay. J. Clin. Microbiol. 27:778-779.
12. Dufour, A. P., E. R. Strickland, and V. J. Cabelli. 1981.
Membrane filter method for enumerating Escherichia coli. Appl.
Environ. Microbiol. 41:1152-1158.
13. Edberg, S. C., M. J. Allen, D. B. Smith, and the National
Collaborative Study. 1988. National field evaluation of a denned
substrate method for the simultaneous enumeration of total
coliforms and Escherichia coli from drinking water: comparison
with the standard multiple tube fermentation method. Appl.
Environ. Microbiol. 54:1595-1601.
14. Entis, P. 1989. Hydrophobic grid membrane filter/MUG meth-
od for total coliform and Escherichia coli enumeration in
foods: collaborative study. J. Assoc. Off. Anal. Chem. 72:936-
950.
15. Ewing, W. H. 1986. Edwards and Ewing's identification of
Enterobacteriaceae, 4th ed., p. 391^15. Elsevier Science Pub-
lishing Co., Inc., New York.
16. Federal Register. 1989. Drinking water; national primary drink-
ing water regulations; total coliforms (including fecal coliforms
and E. coli); final rule. Fed. Regist. 54:27544-27568.
17. Federal Register. 1991. National primary drinking water regula-
tions; analytical techniques; coliform bacteria; final rule. Fed.
Regist. 56:636-643.
18. Federal Register. 1992. National primary drinking water regula-
tions: analytical techniques; coliform bacteria; final rule. Fed.
Regist. 57:24744-24747.
19. Feng, P. C. S., and P. A. Hartman. 1982. Fluorogenic assays for
immediate confirmation of Escherichia coli. Appl. Environ.
Microbiol. 43:1320-1329.
20. Frampton, E. W., L. Restaino, and N. Blaszko. 1988. Evaluation
of p-glucuronidase substrate 5-bromo-4-chloro-3-indolyl-p-D-
glucuronide (X-GLUC) in a 24-hour direct plating method for
Escherichia coli. J. Food Prot. 51:402-404.
21. Freier, T. A., and P. A. Hartman. 1987. Improved membrane
filtration media for enumeration of total coliforms and Esche-
richia coli from sewage and surface waters. Appl. Environ.
Microbiol. 53:1246-1250.
22. Hartman, P. A. 1989. The MUG (glucuronidase) test for Esch-
erichia coli in food and water, p. 290-308. In A. Balows, R. C.
Tilton, and A. Turano (ed.), Rapid methods and automation in
microbiology and immunology. Brixia Academic Press, Brescia,
Italy.
23. Iritani, B., and T. J. Inzana. 1988. Evaluation of a rapid tube
assay for presumptive identification of Escherichia coli from
veterinary specimens. J. Clin. Microbiol. 26:564-566.
24. Kilian, M., and P. Billow. 1979. Rapid identification of Enter-
obacteriaceae: II. Use of a p-glucuronidase detecting agar
medium (PGUA agar) for the identification of E. coli in primary
cultures of urine samples. Acta Pathol. Microbiol. Scand. Sect.
B 87:271-276.
25. Koburger, J. A., and M. L. Miller. 1985. Evaluation of a
fluorogenic MPN procedure for determining Escherichia coli in
oysters. J. Food Prot. 48:244-245.
26. Ley, A. N., R. J. Bowers, and S. Wolfe. 1988. Indoxyl-p-D-
glucuronide, a novel chromogenic reagent for the specific de-
tection and enumeration of Escherichia coli in environmental
samples. Can. J. Microbiol. 34:690-693.
27. Manafi, M., and W. Kneifel. 1989. Ein kombiniertes Chro-
mogen-Fluorogen-Medium zum simultanen Nachweis der
Coliformengruppe und von E, coli in Wasser. Zentralbl. Hyg.
189:225-234.
28. Manati, M., W. Kneifel, and S. Bascomb. 1991. Fluorogenic and
chromogenic substrates used in bacterial diagnostics. Micro-
biol. Rev. 55:335-348.
29. Mates, A., and M. Shaffer. 1989. Membrane filtration differen-
tiation of E. coli from coliforms in the examination of water. J.
Appl. Bacteriol. 67:343-346.
30. Neter, J., and W. Wasserman. 1974. Applied linear statistical
models: regression, analysis of variance, and experimental
designs, p. 480-482. Richard D. Irwin, Inc., Homewood, 111.
31. Petzel, J. P., and P. A. Hartman. 1985. Monensin-based medium
for determination of total gram-negative bacteria and Esche-
richia coli. Appl. Environ. Microbiol. 49:925-933.
32. Rippey, S. R., L. A. Chandler, and W. D. Watkins. 1987.
Fluorometric method for enumeration of Escherichia coli in
molluscan shellfish. J. Food Prot. 50:685-690.
33. Scheffe, H. 1959. The analyses of variance, p. 83-87. John Wiley
and Sons, Inc., New York.
34. Snedecor, G. W., and W. G. Cochran. 1967. Statistical methods,
6th ed., p. 116-118 and 299-308. Iowa State University Press,
Ames.
35. Thaller, M. C., F. Berlutti, B. Dainelli, and R. Pezzi. 1988. New
plate medium for screening and presumptive identification of
gram-negative urinary tract pathogens. J. Clin. Microbiol. 26:
791-793.
36. Thompson, J. S., D. S. Hodge, and A. A. Borczyk. 1990. Rapid
biochemical test to identify verocytotoxin-positive strains of
Escherichia coli serotype 0157. J. Clin. Microbiol. 28:2165-
2168.
37. Tuveson, R. W., R. A. Larson, and J. Kagan. 1988. Role of
cloned carotenoid genes expressed in Escherichia coli in pro-
tecting against inactivation by near-UV light and specific pho-
totoxic molecules. J. Bacteriol. 170:4675-4680.
38. U. S. Environmental Protection Agency. 1985. Test methods for
Escherichia coli and enterococci in water by the membrane filter
-------
3544
BRENNER ET AL.
APPL. ENVIRON. MICROBIOL.
procedure. Publication EPA-600/4-85-076, Environmental Mon-
itoring and Support Laboratory, U.S. Environmental Protection
Agency, Cincinnati.
39. U. S. Environmental Protection Agency. 1991. Test methods for
Escherichia coli in drinking water. Publication EPA/600/4-91/
016. Environmental Monitoring Systems Laboratory, U.S. En-
vironmental Protection Agency, Cincinnati.
40. Watkins, W. D., S. R. Rippey, C. R. Clavet, D. J. Kelley-Reitz,
and W. Burkhardt III. 1988. Novel compound for identifying
Escherichia coli. Appl. Environ. Microbiol. 54:1874-1875.
41. Wilk, M. B., and S. S. Shapiro. 1968. The joint assessment of
normality of several independent samples. Technometrics 10:825-839.
NOTE: The final pH of the medium should be 6.95.
------- |