- Email: [email protected]
Phenotypic and genotypic characterization of human and nonhuman escherichia coli
PII: S0043-1354(00)00269-4
Wat. Res. Vol. 35, No. 2, pp. 379–386, 2001 # 2000 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/00/$ - see front matter
PHENOTYPIC AND GENOTYPIC CHARACTERIZATION OF HUMAN AND NONHUMAN ESCHERICHIA COLI SALINA PARVEEN1,2*, NANCY C. HODGE3, ROBERT E. STALL3, SAMUEL R. FARRAH1M and MARK L. TAMPLIN4 1
Departments of Microbiology and Cell Science, P.O. Box. 110700, University of Florida, Gainesville, FL 32611-0700, USA; 2 Food Science and Human Nutrition, P.O. Box. 110700, University of Florida, Gainesville, FL 32611-0700, USA; 3 Plant Pathology, P.O. Box. 110700, University of Florida, Gainesville, FL 32611-0700, USA and 4 Water Examination Technologies Inc, Gainesville, FL 32609, USA (First received 1 December 1999; accepted in revised form 1 May 2000)
Abstract}Estuarine waters receive fecal pollution from a variety of sources, including humans and wildlife. Escherichia coli is one of several fecal coliform bacteria that inhabit the intestines of many warmblooded animals that sometime contaminate water. Its presence does not specifically implicate human fecal input, therefore it is necessary to differentiate contamination sources to accurately assess health risks. E. coli were isolated from human sources (HS) and nonhuman sources (NHS) in the Apalachicola National Estuarine Research Reserve and analyzed for fatty acid methyl ester (FAME), O-serogroup, and pulsedfield gel electrophoresis (PFGE) profiles. For FAME and PFGE analyses, there was no relationship between profile and isolate source. Human source PFGE profiles were less diverse than NHS isolates, and conversely for FAME. In contrast, O-serogrouping showed less diversity for HS vs. NHS isolates, and the predominant HS O-serogroups differed significantly (P50:01) from those of NHS isolates. # 2000 Elsevier Science Ltd. All rights reserved Key words}fecal pollution, human and nonhuman sources, Escherichia coli, FAME, O-serogroup, PFGE
INTRODUCTION
Fecal pollution can be a problem for estuaries which are associated with wildlife and human populations. Such entities can introduce fecal pollution that not only degrades water quality, but also restricts its use for harvesting seafoods and recreational activities. Escherichia coli is a fecal coliform that has been extensively used to indicate the presence of human enteric pathogens in water (Geldreich, 1966). In more recent years, it has been established that E. coli is widely distributed in the intestines of numerous warm-blooded animals (Orskov and Orskov, 1981). Therefore, since the presence of E. coli in water does not always indicate human fecal input, methods are needed to identify the source as being human (HS) or nonhuman (NHS). Recently, we reported that multiple antibiotic resistance (MAR) and ribotype (RT) profiles can discriminate HS and NHS E. coli (Parveen et al., 1997, 1999). In an effort to investigate other potential differentiating methods, we have studied fatty acid methyl ester (FAME), O-serogroup, and pulsed-field gel electrophoresis (PFGE) profiles of E. coli.
*Author to whom all correspondence should be addressed. Tel.: +1-352-392-1885; fax: +1-352-392-5922; e-mail: [email protected] 379
Gas-liquid chromatography of bacterial cellular FAME has been used in clinical microbiology as a primary, or adjunctive method, for identifying medically important bacteria (Kotilainen et al., 1991; Mukwaya and Welch, 1989). It has been well established that the fatty acid composition of a microorganism is an important taxonomic character, and it can be quantitatively analyzed to provide useful taxonomic information at the species level and in some cases, sub-species (Kotilainen et al., 1991; Mukwaya and Welch, 1989; Vauterin et al., 1992). O-serogrouping of microorganisms is based on the presence or absence of somatic (O) antigenic determinants and their reaction with specific antisera. Several investigators have used this method for discriminating E. coli from different sources (Crichton and Old, 1979; Gonzalez, 1989). The distribution of different E. coli serotypes appears to be different for humans and animal isolates (Bettelhiem et al., 1976; Hartly et al., 1975; Orskov and Orskov, 1981), although many human serotypes can also be associated with isolates from nonhuman sources (Orskov and Orskov, 1981). Pulsed-field gel electrophoresis (PFGE) is a genotypic method that directly detects variations in nucleotide sequences of chromosomal DNA. It has been shown to resolve genomic restriction fragments ranging from microorganisms responsible for
380
Salina Parveen et al.
nosocomial infections (Allardet-Servent et al., 1989), Vibrio species colonizing oysters (Buchrieser et al., 1995; Tamplin et al., 1996), Listeria species in vegetables (Del Rosario et al., 1995), to coliforms isolated from water distribution systems and source water (Edberg et al., 1994). From the literature, it is apparent that PFGE has been mainly used in epidemiological studies, especially for E. coli O157:H7 (Barrett et al., 1994; Johnson et al., 1995). Herbein et al. (1996) have found an association between E. coli PFGE profiles and isolate source. Therefore, the purpose of this study was to determine whether FAME, O-serogroup, and PFGE profiles could be effectively used to characterize HS and NHS E. coli.
MATERIALS AND METHODS
Isolates E. coli were selected from among 700 isolates previously described in a publication of multiple antibiotic resistance within the Apalachicola Bay National Estuarine Research Reserve. HS and NHS isolates were collected from sewage treatment plants and a northeast region of Apalachicola Bay, respectively. The source of NHS isolate was influenced by wildlife species and not by any HS pollution. In the present study, 104 (53 HS and 51 NHS) isolates were selected from the original collection, in proportion to the number of isolates associated with specific MAR clusters. E. coli were isolated and identified using standard procedures. (Parveen et al., 1997). FAME extraction Whole-cell fatty acids were extracted and analyzed as methyl ester derivatives by the method of Mukwaya and Welch (1989). In brief, Bacteria were cultured on trypticase soy broth agar (TSBA, BBL, Cockeysville, MD) for 24 2 h at 288C. The collected bacteria were saponified with 1.0 ml of 1.2 N NaOH in 50% methanol at 1008C for 30 min. Saponified fatty acids were methylated by adding 2.0 ml of 6.0 N HCl in methanol and heating to 808C for 10 min, followed by rapid cooling. FAMEs were extracted with 1.25 ml hexane/methyl-t-butyl ether (1 : 1 v/v), discarding the acidic aqueous phase. The extract was washed and neutralized with 3.0 ml of 0.3 N NaOH. Extracts were then transferred to a chromatography vial and capped. With each batch of extractions, a reagent blank was included along with a control strain of Stenotrophomonas maltophilia grown on TSBA for 24 h at 288C. The profile of this well characterized bacterium was included to ensure reproducibility among different batches. Gas liquid chromatography Fatty acid methyl esters were separated using a HewlettPackard 5890A gas liquid chromatograph fitted with a 25 m 0.2 mm phenyl methyl silicone fused silica capillary column, and a flame ionization detector, a model 3392A integrator, a model 7673A automatic sampler, and a model 310 computer. Initial GC oven temperature was 1708C, and increased at a rate of 58C/min to a final temperature 2708C. The carrier gas was H2 at a flow rate of 20 ml/min. The peaks were automatically integrated, and fatty acid identities and percentages calculated by comparing their retention times to those of a bacterial fatty acid standard (bacterial acid methyl ester mix CP, 4-7080; Supelco, Inc., Bellefonte, PA).
O-serogrouping The E. coli isolates were serogrouped with 181 specific Oantisera using standard methods at the E. coli Reference Center, Pennsylvania State University (Wilson and Francis, 1986). In brief, isolates were grown on Veal Infusion Yeast Extract agar (veal infusion, 25 g; yeast extract, 2 g; dextrose, 1 g; agar, 15 g; dH2O, 1 l) and incubated at 378C for 18 to 24 h, suspended in 0.6% formal saline (NaCl, 6 g; 37% formaldehyde, 6 ml; dH2O 1 l) and then heated at 1008C for 2 h. Slide agglutinations were performed by using O-group antisera; the reactions were confirmed by tube titration. When an ‘‘O’’ reaction was negative, the antigen was reheated at 1208C for 60 min and the slide procedure repeated. PFGE PFGE analysis was performed as described by Buchrieser et al. (1995). A single colony of each isolate was inoculated into 10 ml TSB (Difco), and incubated overnight at 378C. Cells were harvested, washed and suspended in agarose plugs. To lyse cells and degrade cellular proteins, agarose plugs were incubated at 508C overnight in proteinase K solution (Fisher Biotech, 2 mg/ml in 0.5 M EDTA and 0.5% N-Lauryl sarcosine; Sigma). Plugs were then treated with phenylmethylsulfonyl fluoride (Sigma) and washed three times in TE buffer. The genomic DNA embedded within agarose plugs was digested with SfiI restriction endonuclease as recommended by the manufacturer (New England Biolabs, Beverly, MA; Stratagene, La Jolla, CA; or Promega Corp.). High-molecular-weight restriction fragments were resolved with a CHEF-DR II pulsed-field system (Bio-Rad Laboratories, Richmond, CA) using 1.0% pulsed-field certified agarose gel (Bio-Rad). An electrophoresis regimen of 200 V for 24 h at a temperature of 188C, and a switching time from 1 to 40 s was used to separate fragments. Low-range PFGE and lamda ladder PFGE (New England Biolabs) markers were used as standard. Gels were stained in ethidium bromide solution (Sigma, 5 mg/ml) and photographed under short-wave UV light. The size of DNA fragments was compared visually. Statistical analysis The quantitative data obtained from the FAME profiles were used as the basis for numerical analysis. Peak-area values for each fatty acid were calculated as percentages of the total-peak area to eliminate the effect of sample-size variation. Differences in the amount of the fatty acids between HS and NHS isolates were evaluated statistically using the two-sided test of binomial proportion (P50:05). Similarities were calculated with the generalized similarity coefficient of Gower (1971) and the coefficient based on the Euclidian distance between the pairs of bacteria. Clustering of strains was achieved by the unweighted pair group method for arithmetic averages with a program provided in the MIDI Library Generation Software (MIDI, Newark, DE), resulting in a dendrogram. Principal-component analysis of the quantitative fatty acid data was performed with the above mentioned statistical program, and the results were plotted graphically in two dimensions. The significant differences between the predominant Oserogroups of HS and NHS isolates were also determined by a two-sided test of binomial proportion (p50:01).
RESULTS
FAME A total of 104 isolates were analyzed for FAME profile. All isolates contained significant amounts of hexadecanoate (30%), cis-7-hexadecenoate (13%),
Phenotypic and genotypic characterization of human and nonhuman Escherichia coli
381
Table 1. Major fatty acids of human source (HS) and nonhuman source (NHS) isolates Name
Dodecanoate Tetradecanoate Hexadecanoate Cis-7-hexadecenoate Cycloheptadecanoate Cyclononadecanoate Summed feature 3 Summed feature 7
Fatty acida
12 : 0 14 : 0 16 : 0 16 : 1 cis 7 17 : 0 cyclo 19 : 0 cyclo
% of fatty acidsb
No. of isolates HS
NHS
HS
NHS
53 53 53 53 53 52 53 53
51 51 51 51 50 47 51 51
4 7 30 13 11 3 8 21
4 8 30 13 13 4 9 18
a
The number before the colon refers to the number of carbon atoms, the number following the colon refers to the number of double bonds. Cyclo indicates a cyclopropane ring, summed feature: a group of isomers which elute so closely in time that they are not differentiated. Summed feature 3 is comprised of 12:0 aldehyde, 16:1 ISO I/14:0 3 OH, 14:0 3 OH/16:1 ISO I, summed feature 7 is comprised of 18:1 w7cis/w9trans/w12trans, 18:1 w9cis/w12trans/w7cis, 18:1 w12trans/w9trans/w7cis. b The P-value for HS and NHS isolates was determined by two-sided binomial test. For all fatty acids P-value was >0.05.
cycloheptadecanoate (>11%) and summed features 7 (>18%). No significant differences (p > 0:05) were observed among type of FAME profile, quantity of a specific FAME, and whether the isolate was HS and NHS (Table 1). The interrelationships of HS and NHS E. coli FAME profiles are depicted by dendrogram in Fig. 1, and are based on the coefficient of similarity generated by the Euclidian distance between pairs of bacteria. On the basis of the MIDI dendrogram software, at a Euclidian distance of approximately 13.0, all isolates were grouped as a single cluster. At a Euclidian distance of approximately 4.0 (the lowest distance that included all isolates), 18 clusters were well separated. Clusters 1, 2, 7, 8, 11, 13, and 17 comprised only HS isolates, whereas clusters 3, 5, 6, 9, 10 and 12 comprised only NHS isolates. Clusters 4, 14, 15, 16, and 18 contained isolates from both sources. The number of isolates in each cluster was too low to indicate significance differences. Based on less than 10% variance for 85–95% of the FAME profiles, 87 of the 104 isolates were divided into 10 subgroups. It is interesting that subgroups 7, 8, and 10 contained only HS isolates and subgroup 9 only NHS isolates. The remaining subgroups comprised isolates from both sources. The relationship of FAME profiles is also illustrated by principal-component analysis of the data in two dimensions (Fig. 2). HS isolates were more widely distributed throughout the plot compared to NHS isolates. No well-separated HS and NHS groups were observed.
O-Serogrouping A total of 100 HS and NHS isolates were tested for O-serogroup; 77% could be typed. The predominant O-serogroups are shown in Table 2. Human source isolates exhibited 19 serogroups, with 48% belonging to seven serogroups (O2, O20, O28, O79, O148, O153, O159) and the remaining isolates to 12
Fig. 1. Dendrogram of FAME profiles of human source (HS) and nonhuman source (NHS) E. coli isolates. Bottom line indicates number of clusters.
382
Salina Parveen et al.
Fig. 2. Two-dimensional plot of FAME profiles of E. coli isolates generated by principal-component analysis.
Table 2. O-Serogroups of human source (HS) and nonhuman source (NHS) isolates O-Serogroups
2 8 11 13 19 20 28 79 148 150 152 15/143 153 159 Othersa MR NT a
Number (%) of E. coli isolates HS (n ¼ 50)
NHS (n ¼ 50)
3(6) 1(2) 1(2) 0 0 2(4) 2(4) 7(14) 3(6) 0 0 0 2(4) 5(10) 10(20) 4(8) 10(20)
1(2) 3(6) 2(4) 2(4) 5(10) 0 0 1(2) 0 2(4) 2(4) 2(4) 0 0 17(34) 0 12(24)
10 and 17 serogroups of HS and NHS isolates, respectively, that contained only one isolate are not shown. MR: multiple reaction. NT: nontypeable. : not accepted by World Health Organization. One NHS isolate was autoagglutinable.
additional serogroups. In contrast, NHS isolates displayed 26 serogroups, with 36% in seven serogroups (O8, O11, Ox13, O19, O150, O152, O15/143), and the remaining isolates in 19 additional serogroups. Only 2 of 50 NHS isolates shared the seven most frequently detected serogroups for HS isolates (P50:01). This indicated that there was a association between O-serogroups and the isolate source. PFGE A total of 32 E. coli isolates were analyzed for PFGE profile. Preliminary experiments examined several restriction enzymes (SfiI, NotI, and XbaI) with a panel of E. coli isolates to determine the appropriate enzyme for this study. Results showed that SfiI yielded 12–18 discernable bands ranging from 9 to 528 kb, and thus was used for subsequent studies. Patterns were considered to be unique when differentiated by one or more bands. HS and NHS isolates formed 9 and 18 PFGE profiles, respectively. There was no association between PFGE profile and
Phenotypic and genotypic characterization of human and nonhuman Escherichia coli
383
Fig. 3. Representative PFGE profiles of human source (HS) and nonhuman source (NHS) E. coli isolates digested with SfiI restriction endonuclease. Lanes 1 and 20: Low-range PFG marker in kb; 2: SPS 35, 3: SPS 80; 4: SPS 42, 5: SPS 45; 6: SPS 118, 7: SPS 205; 8: SPS 466, 9: SPS 493;10: SPS 601, 11: SPS 602;12: SPS 561,13: SPS 573;14:SP 5; 15: SP 50, 16: SP 93;17: SP 6, 18: SP 112.
source of isolate. In addition, we used PFGE to determine whether isolates which shared RT and MAR profiles were from the same clonal origin. Out of 32 isolates only four pairs shared the same MAR, RT and PFGE profiles (Fig. 3), indicating the existence of multiple E. coli strains rather a few clones in the ANERR. DISCUSSION
When attempting to identify specific health risks associated with polluted water, it is necessary to understand input sources. This approach implies that methods exist that can discriminate the pollution source. Because human feces can carry various human enteric pathogens, such as Salmonella spp., Shigella spp., E. coli, hepatitis A and Norwalk group viruses. In contrast, most of these pathogens do not colonize nonhuman species (Guzewich and Morse, 1986; Orskov and Orskov, 1981). For microbiological hazards, early attempts to differentiate fecal pollution included the ratio of fecal coliform to fecal streptococci, and fecal streptococci species associated with humans, livestock and wildlife (Geldreich and Kenner, 1969). After much research, these methods have proven unreliable (Pourcher et al., 1991). More recently, research has investigated multiple antibiotic resistance (MAR) profiles of E. coli, and fecal streptococci (Parveen et al., 1997; Wiggins, 1996). This approach has shown some promise, and
is certainly a simple and cost-effective method for most laboratories. Concerns about the method include the stability of resistance (i.e. plasmids), geographical variation in MAR, and lack of specificity due to the use of similar antibiotics among humans and other animals. Our laboratory has investigated the ability of ribotype profiling to discriminate E. coli sources, and found that, like MAR, specific ribotype patterns are associated with HS and NHS E. coli isolates (Parveen et al., 1999). However, research has yet to establish the broad geographical application of ribotyping and other possible limitations. More than likely, the combined use of phenotypic and genotypic methods will produce the highest level of discrimination, compared to basing conclusions on a single microbial trait. For this reason, we examined the applicability of FAME, O-serogroup and PFGE profiles as potential discriminatory tools. A FAME profile is a stable phenotypic expression of a bacterial genotype when cells are grown under controlled conditions (Welch, 1991). Fatty acid metabolism is constitutive and directed by the chromosome, not plasmids. The presence of certain bacterial fatty acids has been shown to correlate with certain taxonomic conventions. Hence, fatty acids profiles can be used to produce bacterial ‘‘fingerprints.’’ This observation has led to the use of FAME profiling for chemotaxonomic classifications (Drucker, 1976; Lechevalier, 1977).
384
Salina Parveen et al.
For this study, we used FAME profiles to characterize HS and NHS E. coli. We found that there were no significant differences between FAME profiles of HS and NHS isolates. At a Euclidian distance of 10.0, two major groups were formed, indicating that all isolates were likely members of the same species (Sasser, 1990). It was determined that HS isolates were more widely distributed throughout the plot than NHS isolates. This suggested a high level of phenotypic diversity within the E. coli population. Havelaar et al. (1992), studying Aeromonas strains from patients with diarrhea and from drinking water, observed that FAME profiles could be used to compare strains from different sources. However, they could not identify FAME profiles that discriminated human and drinking water isolates. Similarly, other investigators have found that FAME analysis did not provide a high level of discrimination. For example, FAME profiles could not differentiate pathovars of Xanthomonas campestris, Streptomyces scabies and members of Enterobacteriaceae at the intraspecies level. The treatment of results by principal-component analysis or by other methods of classification did not improve the separation of Xanthomonas campestris and Saccharomyces scabies strains (Chase et al., 1992; Steele et al., 1997). Kotilainen et al. (1991) used this method to identify and type coagulase-negative Staphylococcus, and concluded that FAME could identify isolates at the species level. In addition, FAME was able to differentiate among multiple coagulase-negative staphylococcal blood isolates. We conclude that FAME analysis was unable to differentiate HS and NHS E. coli. However, it may depend on the species examined. For example, some species may exhibit specific fatty acids that are associated with certain microbial characteristics. Similarly, PFGE profiles of E. coli were not able to differentiate HS and NHS isolates. This was not surprising since this method detects small sequence differences that may not associate with a specific bacterial characteristic, such as host source. This technique has also been used by Edberg et al. (1994) to differentiate Enterobacter cloacae in a county drinking water distribution system, hospital source waters, and associated clinical samples. A single PFGE pattern was found in the water distribution system, which differed from that of the isolates found in hospital source water and the clinical E. cloacae isolates. Herbein et al. (1996) reported PFGE profiles for three racoons, one otter, one goose, three humans, and one muskrat. Within this limited sample size, they found that PFGE profiles of racoons differed significantly from those of the otter and goose. There was no significant difference between PFGE profiles of racoons, humans and a muskrat.
From this study, it was determined that the Oserogroups of HS isolates were rarely found in NHS isolates, similar to previously published studies (Orskov and Orskov, 1981). The majority of Oserogroups that we observed for NHS isolates were reported for cattle, chicken, and swine (Orskov and Orskov, 1981). These results also agree with studies by Hartley et al. (1975) who compared the prevalent O-serogroups of diseased animals and humans, and found that most strains associated with diarrhea in pigs and humans were different, while infantile and piglet diarrhea have many common traits. Bettelheim et al. (1976) reported that although the distribution of different E. coli strains identified by serotype appears to differ in human and animals, many serotypes are shared among humans and other animals. This situation was illustrated by analyzing the O : H serotypes of 13,139 strains of E. coli isolated from humans, and 1076 strains isolated from animals, 689 of the latter obtained from cow-pats sampled at 22 sites in England and Wales. Typing with a virtually complete set of 154 O and 53 H antisera led to the identification of 708 distinct O : H serotypes, among which 520 were only found in humans, 130 only in animals, and 58 in both. The great diversity of O : H serotypes is illustrated by the finding that, of 78 serotypes identified in cow-pat specimens, 65 were found only in one site and 12 were found only at both sites. Moreover, approximately half of the isolates from animals could not be typed in spite of the large number of antisera used. In the present study, we found that 22% of the isolates were nontypeable, which is consistent with Blanco et al. (1994), who showed that 19% isolates could not be typed. Moreover, the percentage of untypeable isolates from both HS and NHS was very similar (20 and 24%). This result is supported by other investigators who noted that the frequency of untypeable animal isolates was approximately the same as the frequency of untypeable human isolates obtained in a study of the remote Yanomama Indians of South America (Eveland et al., 1971). Serogrouping may be a useful tool for differentiating HS and NHS of E. coli. Drawbacks include (1) the need for a large bank of antisera, many of which are not commercially available (2) many serotypes are shared among HS and NHS isolates, and that some isolates cannot be typed.
CONCLUSIONS
From this study we conclude that (1) there were no significant differences among FAME profiles, quantity of a specific FAME, and whether the isolate was HS and NHS, (2) PFGE profiles did not show any correlation with source of fecal pollution, (3) HS PFGE profiles were less diverse than NHS isolates, and conversely for FAME, (4) O-serogrouping showed less diversity for HS vs. NHS isolates, and
Phenotypic and genotypic characterization of human and nonhuman Escherichia coli
the predominant O-serogroups differed significantly from those of NHS isolates, and (5) O-serogrouping may be useful tool for differentiating sources of fecal pollution in conjunction with MAR and RT profiling. Acknowledgements}This publication was supported by the U.S. Department of Commerce, National Oceanographic and Atmospheric Administration, Sanctuaries and Reserves Division, grant #NA370R0166 and Engineering Research Center (ERC) for Particle Science and Technology at the University of Florida. We thank Lee Edmiston and Chip Bailey for assistance with sample collection, Dr. J. K. Jackson for technical assistance and K. Robinson for statistical analysis of data.
REFERENCES
Allardet-Servent A., Bouziges N., Carles-Nurit M. J., Bourg G., Gouby A. and Ramuz M. (1989) Use of lowfrequency-cleavage restriction endonucleases for DNA analysis in epidemiological investigations of nosocomial bacterial infections. J. Clin. Microbiol. 27, 2057–2061. Barrett T. J., Lior H., Green J. H., Khakhria R., Wells J. G., Bell B. P., Greene K. D., Lewis J. and Griffin P. M. (1994) Laboratory investigation of a multistate food borne outbreak of Escherichia coli 0157 : H7 by using pulsed-field gel electrophoresis and phage typing. J. Clin. Microbiol. 32, 3013–3017. Bettelheim K. A., Ismail N., Shinebaum R., Shooter R. A., Moorhouse E. and O’Farrell S. (1976) The distribution of serotypes of Escherichia coli in cow-pats and other animal materials compared with serotypes of E. coli isolated from human sources. J. Hyg. 76, 403–406. Blanco J. E., Blanco M., Blanco J., Rioja L. and Ducha J. (1994) Serotypes, toxins, and antibiotic resistance of Escherichia coli strains isolated from diarrhoeic and healthy rabbits in Spain. Vet. Microbiol. 38, 193–201. Buchrieser C., Ganger V. V., Murphree R. L., Tamplin M. L. and Kaspar C. W. (1995) Multiple Vibrio vulnificus strains in oyster as demonstrated by clamped homogenous electric field gel electrophoresis. Appl. Environ. Microbiol. 61, 1163–1168. Chase A. R., Stall R. E., Hodge N. C. and Jones J. B. (1992) Characterization of Xanthomonas campestris strains from Aroids, using physiological, pathological, and fatty acid analyses. Phytopathology 82, 754–759. Crichton P. B. and Old D. C. (1979) Biotyping of Escherichia coli. J. Med. Microbiol. 12, 473–486. Del Rosario L., Barth S. S., Rutter M. and Ray B. (1995) Molecular Analysis of Listeria monocytogenes Isolates Associated with the Consumption of Listeria Contaminated Mixed Vegetables. Abstract from 95th General Meeting of ASM. Washington, DC. Drucker D. B. (1976) Gas liquid chromatography. Methods Microbiol. 9, 51–125. Edberg S. C., Patterson J. E. and Smith D. B. (1994) Differentiation of distribution systems, source water, and clinical coliform by DNA analysis. J. Clin Microbiol. 32, 139–142. Eveland W. C., Oliver W. J. and Neel J. V. (1971) Characterization of Escherichia coli serotypes in the Yanomama, a primitive Indian tribe of South America. Infect. Immun. 4, 753–756. Geldreich E. E. (1966) Sanitary Significance of Fecal Coliforms in the Environment. Water Pollution Control Research Series, Publication WP-20-3. Federal Water Pollution Control Administration, U.S. Department of the Interior, Cincinnati, OH.
385
Geldreich E. E. and Kenner B. A. (1969) Concepts of fecal streptococci in strem pollution. J. Water Pollut. Control Fed. 41, R336–R352. Gonzalez E. A. and Blanco J. (1989) Serotypes and antibiotic resistance of verotoxigenic (VTEC) and necrotizing (NTEC) Escherichia coli strains isolated from calves with diarrhoea. FEMS Microbiol. Lett. 60, 31–36. Gower J. C. (1971) A general coefficient of similarity and some of its properties. Biometrics 27, 857–874. Guzewich J. J. and Morse D. L. (1986) Sources of shellfish in outbreaks of probable viral gastroenteritis: implications for control. J. Food Prot. 49, 389–394. Hartly C. L., Howne K., Linton A. H., Linton K. B. and Richmond M. H. (1975) Distribution of R plasmids among O-antigen types of Escherichia coli isolated from human and animal sources. Antimicrob. Agents Chemother. 8, 122–131. Havelaar A. H., Schets F. H., Silfhout A. V., Jansen W. H., Wieten G. and Kooij D. V. D. (1992) Typing of Aeromonas strains from patients with diarrhea and from drinking water. J. Appl. Microbiol. 72, 435–444. Herbein S. A., Simmons Jr. G. M. and Myers S. L. (1996). Use of Pulsed-Field Gel Electrophoresis (PFGE) to investigate nonpoint fecal coliform sources to tidal inlets on the Eastern shore of the Chesapeake Bay. 96th General Meeting of the American Society for Microbiology, p. 467, May 19–23. New Orleans, LA. Johnson J. M., Weagant S. D., Jinneman K. C. and Bryant J. L. (1995) Use of pulsed-field gel electrophoresis for epidemiological study of Escherichia coli O157 : H7 during a food-borne outbreak. Appl. Environ. Microbiol. 61, 2806–2808. Kotilainen P., Huovinen P. and Eerola E. (1991) Application of gas liquid chromatographic analysis of cellular fatty acids for species identification and typing of coagulase-negative staphylococci. J. Clin. Microbiol. 29, 315–322. Lechevalier M. P. (1977) Lipids in bacterial taxonomy. A taxonomist’s view. Critical Rev. Microbiol. 5, 109–210. Mukwaya G. M. and Welch D. F. (1989) Subgrouping of Pseudomonas cepacia by cellular fatty acid composition. J. Clin. Microbiol. 27, 2640–2646. Orskov F. and Orskov I. (1981) Enterobacteriaceae. In Medical Microbiology and Infectious Diseases, ed. A. I. Broude, pp. 340–352. W.B. Saunders Co., Philadelphia. Parveen S., Murphree R. L., Edmiston L., Kaspar C. W., Portier K. M. and Tamplin M. L. (1997) Association of multiple-antibiotic resistance profiles with point and nonpoint sources of Escherichia coli in Apalachicola Bay. Appl. Environ. Microbiol. 63, 2607–2612. Parveen S., Portier K. M., Robinson K., Edmiston L. and Tamplin M. L. (1999) Discriminant analysis of ribotype profiles of E. coli for differentiating human and nonhuman Sources of fecal pollution. Appl. Environ. Microbiol. 65, 3142–3147. Pourcher A. M., Devriese L. A., Hernandez J. F. and Delattre J. M. (1991) Enumeration by a miniatured method of Escherichia coli, Streptococcus bovis and enterococci as indicators of the origin of fecal pollution of waters. J. Appl. Bacteriol. 70, 525–530. Sasser M. (1990) ‘‘Tracking’’ a Strain Using the Microbial Identification System. Technical Note No. 102 MIDI, Newark, Delaware. Steele M., McNab W. B., Read S., Popope C., Harris L., Lammerding A. M. and Odumeru J. A. (1997) Analysis of whole-cell fatty acid profiles of verotoxigenic Escherichia coli and Salmonella enteritidis with Microbial Identification System. Appl. Environ. Microbiol. 63, 757–760. Tamplin M. L., Jackson J. K., Buchrieser C., Murphree R. L., Portier K. M., Ganger V. V., Miller L. J. and Kaspar C. W. (1996) Pulsed-field gel electrophoresis and ribotype
386
Salina Parveen et al.
profiles of clinical and environmental Vibrio vulnificus isolates. Appl. Environ. Microbiol. 62, 3572–3580. Vauterin L., Yang P., Hoste B., Pot B., Swings J. and Kersters K. (1992) Taxonomy of Xanthomonads from cereals and grasses based on SDS-PAGE of proteins, fatty acid analysis and DNA hybridization. J. Gen. Microbiol. 138, 1467–1477. Welch D. (1991) Application of cellular fatty acid analysis. Crit. Microbiol. Rev. 4, 422–43837.
Wiggins B. A. (1996) Discriminant analysis of antibiotic resistance patterns in fecal Streptococci, a method to differentiate human and animal sources of fecal pollution in natural waters. Appl. Environ. Microbiol. 62, 3997–4002. Wilson R. A. and Francis D. H. (1986) Frimbriae and enterotoxins associated with Escherichia coli serogroups isolated from pigs with collibacillosis. Am. J. Vet. Res. 47, 213–217.
Wat. Res. Vol. 35, No. 2, pp. 379–386, 2001 # 2000 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/00/$ - see front matter
PHENOTYPIC AND GENOTYPIC CHARACTERIZATION OF HUMAN AND NONHUMAN ESCHERICHIA COLI SALINA PARVEEN1,2*, NANCY C. HODGE3, ROBERT E. STALL3, SAMUEL R. FARRAH1M and MARK L. TAMPLIN4 1
Departments of Microbiology and Cell Science, P.O. Box. 110700, University of Florida, Gainesville, FL 32611-0700, USA; 2 Food Science and Human Nutrition, P.O. Box. 110700, University of Florida, Gainesville, FL 32611-0700, USA; 3 Plant Pathology, P.O. Box. 110700, University of Florida, Gainesville, FL 32611-0700, USA and 4 Water Examination Technologies Inc, Gainesville, FL 32609, USA (First received 1 December 1999; accepted in revised form 1 May 2000)
Abstract}Estuarine waters receive fecal pollution from a variety of sources, including humans and wildlife. Escherichia coli is one of several fecal coliform bacteria that inhabit the intestines of many warmblooded animals that sometime contaminate water. Its presence does not specifically implicate human fecal input, therefore it is necessary to differentiate contamination sources to accurately assess health risks. E. coli were isolated from human sources (HS) and nonhuman sources (NHS) in the Apalachicola National Estuarine Research Reserve and analyzed for fatty acid methyl ester (FAME), O-serogroup, and pulsedfield gel electrophoresis (PFGE) profiles. For FAME and PFGE analyses, there was no relationship between profile and isolate source. Human source PFGE profiles were less diverse than NHS isolates, and conversely for FAME. In contrast, O-serogrouping showed less diversity for HS vs. NHS isolates, and the predominant HS O-serogroups differed significantly (P50:01) from those of NHS isolates. # 2000 Elsevier Science Ltd. All rights reserved Key words}fecal pollution, human and nonhuman sources, Escherichia coli, FAME, O-serogroup, PFGE
INTRODUCTION
Fecal pollution can be a problem for estuaries which are associated with wildlife and human populations. Such entities can introduce fecal pollution that not only degrades water quality, but also restricts its use for harvesting seafoods and recreational activities. Escherichia coli is a fecal coliform that has been extensively used to indicate the presence of human enteric pathogens in water (Geldreich, 1966). In more recent years, it has been established that E. coli is widely distributed in the intestines of numerous warm-blooded animals (Orskov and Orskov, 1981). Therefore, since the presence of E. coli in water does not always indicate human fecal input, methods are needed to identify the source as being human (HS) or nonhuman (NHS). Recently, we reported that multiple antibiotic resistance (MAR) and ribotype (RT) profiles can discriminate HS and NHS E. coli (Parveen et al., 1997, 1999). In an effort to investigate other potential differentiating methods, we have studied fatty acid methyl ester (FAME), O-serogroup, and pulsed-field gel electrophoresis (PFGE) profiles of E. coli.
*Author to whom all correspondence should be addressed. Tel.: +1-352-392-1885; fax: +1-352-392-5922; e-mail: [email protected] 379
Gas-liquid chromatography of bacterial cellular FAME has been used in clinical microbiology as a primary, or adjunctive method, for identifying medically important bacteria (Kotilainen et al., 1991; Mukwaya and Welch, 1989). It has been well established that the fatty acid composition of a microorganism is an important taxonomic character, and it can be quantitatively analyzed to provide useful taxonomic information at the species level and in some cases, sub-species (Kotilainen et al., 1991; Mukwaya and Welch, 1989; Vauterin et al., 1992). O-serogrouping of microorganisms is based on the presence or absence of somatic (O) antigenic determinants and their reaction with specific antisera. Several investigators have used this method for discriminating E. coli from different sources (Crichton and Old, 1979; Gonzalez, 1989). The distribution of different E. coli serotypes appears to be different for humans and animal isolates (Bettelhiem et al., 1976; Hartly et al., 1975; Orskov and Orskov, 1981), although many human serotypes can also be associated with isolates from nonhuman sources (Orskov and Orskov, 1981). Pulsed-field gel electrophoresis (PFGE) is a genotypic method that directly detects variations in nucleotide sequences of chromosomal DNA. It has been shown to resolve genomic restriction fragments ranging from microorganisms responsible for
380
Salina Parveen et al.
nosocomial infections (Allardet-Servent et al., 1989), Vibrio species colonizing oysters (Buchrieser et al., 1995; Tamplin et al., 1996), Listeria species in vegetables (Del Rosario et al., 1995), to coliforms isolated from water distribution systems and source water (Edberg et al., 1994). From the literature, it is apparent that PFGE has been mainly used in epidemiological studies, especially for E. coli O157:H7 (Barrett et al., 1994; Johnson et al., 1995). Herbein et al. (1996) have found an association between E. coli PFGE profiles and isolate source. Therefore, the purpose of this study was to determine whether FAME, O-serogroup, and PFGE profiles could be effectively used to characterize HS and NHS E. coli.
MATERIALS AND METHODS
Isolates E. coli were selected from among 700 isolates previously described in a publication of multiple antibiotic resistance within the Apalachicola Bay National Estuarine Research Reserve. HS and NHS isolates were collected from sewage treatment plants and a northeast region of Apalachicola Bay, respectively. The source of NHS isolate was influenced by wildlife species and not by any HS pollution. In the present study, 104 (53 HS and 51 NHS) isolates were selected from the original collection, in proportion to the number of isolates associated with specific MAR clusters. E. coli were isolated and identified using standard procedures. (Parveen et al., 1997). FAME extraction Whole-cell fatty acids were extracted and analyzed as methyl ester derivatives by the method of Mukwaya and Welch (1989). In brief, Bacteria were cultured on trypticase soy broth agar (TSBA, BBL, Cockeysville, MD) for 24 2 h at 288C. The collected bacteria were saponified with 1.0 ml of 1.2 N NaOH in 50% methanol at 1008C for 30 min. Saponified fatty acids were methylated by adding 2.0 ml of 6.0 N HCl in methanol and heating to 808C for 10 min, followed by rapid cooling. FAMEs were extracted with 1.25 ml hexane/methyl-t-butyl ether (1 : 1 v/v), discarding the acidic aqueous phase. The extract was washed and neutralized with 3.0 ml of 0.3 N NaOH. Extracts were then transferred to a chromatography vial and capped. With each batch of extractions, a reagent blank was included along with a control strain of Stenotrophomonas maltophilia grown on TSBA for 24 h at 288C. The profile of this well characterized bacterium was included to ensure reproducibility among different batches. Gas liquid chromatography Fatty acid methyl esters were separated using a HewlettPackard 5890A gas liquid chromatograph fitted with a 25 m 0.2 mm phenyl methyl silicone fused silica capillary column, and a flame ionization detector, a model 3392A integrator, a model 7673A automatic sampler, and a model 310 computer. Initial GC oven temperature was 1708C, and increased at a rate of 58C/min to a final temperature 2708C. The carrier gas was H2 at a flow rate of 20 ml/min. The peaks were automatically integrated, and fatty acid identities and percentages calculated by comparing their retention times to those of a bacterial fatty acid standard (bacterial acid methyl ester mix CP, 4-7080; Supelco, Inc., Bellefonte, PA).
O-serogrouping The E. coli isolates were serogrouped with 181 specific Oantisera using standard methods at the E. coli Reference Center, Pennsylvania State University (Wilson and Francis, 1986). In brief, isolates were grown on Veal Infusion Yeast Extract agar (veal infusion, 25 g; yeast extract, 2 g; dextrose, 1 g; agar, 15 g; dH2O, 1 l) and incubated at 378C for 18 to 24 h, suspended in 0.6% formal saline (NaCl, 6 g; 37% formaldehyde, 6 ml; dH2O 1 l) and then heated at 1008C for 2 h. Slide agglutinations were performed by using O-group antisera; the reactions were confirmed by tube titration. When an ‘‘O’’ reaction was negative, the antigen was reheated at 1208C for 60 min and the slide procedure repeated. PFGE PFGE analysis was performed as described by Buchrieser et al. (1995). A single colony of each isolate was inoculated into 10 ml TSB (Difco), and incubated overnight at 378C. Cells were harvested, washed and suspended in agarose plugs. To lyse cells and degrade cellular proteins, agarose plugs were incubated at 508C overnight in proteinase K solution (Fisher Biotech, 2 mg/ml in 0.5 M EDTA and 0.5% N-Lauryl sarcosine; Sigma). Plugs were then treated with phenylmethylsulfonyl fluoride (Sigma) and washed three times in TE buffer. The genomic DNA embedded within agarose plugs was digested with SfiI restriction endonuclease as recommended by the manufacturer (New England Biolabs, Beverly, MA; Stratagene, La Jolla, CA; or Promega Corp.). High-molecular-weight restriction fragments were resolved with a CHEF-DR II pulsed-field system (Bio-Rad Laboratories, Richmond, CA) using 1.0% pulsed-field certified agarose gel (Bio-Rad). An electrophoresis regimen of 200 V for 24 h at a temperature of 188C, and a switching time from 1 to 40 s was used to separate fragments. Low-range PFGE and lamda ladder PFGE (New England Biolabs) markers were used as standard. Gels were stained in ethidium bromide solution (Sigma, 5 mg/ml) and photographed under short-wave UV light. The size of DNA fragments was compared visually. Statistical analysis The quantitative data obtained from the FAME profiles were used as the basis for numerical analysis. Peak-area values for each fatty acid were calculated as percentages of the total-peak area to eliminate the effect of sample-size variation. Differences in the amount of the fatty acids between HS and NHS isolates were evaluated statistically using the two-sided test of binomial proportion (P50:05). Similarities were calculated with the generalized similarity coefficient of Gower (1971) and the coefficient based on the Euclidian distance between the pairs of bacteria. Clustering of strains was achieved by the unweighted pair group method for arithmetic averages with a program provided in the MIDI Library Generation Software (MIDI, Newark, DE), resulting in a dendrogram. Principal-component analysis of the quantitative fatty acid data was performed with the above mentioned statistical program, and the results were plotted graphically in two dimensions. The significant differences between the predominant Oserogroups of HS and NHS isolates were also determined by a two-sided test of binomial proportion (p50:01).
RESULTS
FAME A total of 104 isolates were analyzed for FAME profile. All isolates contained significant amounts of hexadecanoate (30%), cis-7-hexadecenoate (13%),
Phenotypic and genotypic characterization of human and nonhuman Escherichia coli
381
Table 1. Major fatty acids of human source (HS) and nonhuman source (NHS) isolates Name
Dodecanoate Tetradecanoate Hexadecanoate Cis-7-hexadecenoate Cycloheptadecanoate Cyclononadecanoate Summed feature 3 Summed feature 7
Fatty acida
12 : 0 14 : 0 16 : 0 16 : 1 cis 7 17 : 0 cyclo 19 : 0 cyclo
% of fatty acidsb
No. of isolates HS
NHS
HS
NHS
53 53 53 53 53 52 53 53
51 51 51 51 50 47 51 51
4 7 30 13 11 3 8 21
4 8 30 13 13 4 9 18
a
The number before the colon refers to the number of carbon atoms, the number following the colon refers to the number of double bonds. Cyclo indicates a cyclopropane ring, summed feature: a group of isomers which elute so closely in time that they are not differentiated. Summed feature 3 is comprised of 12:0 aldehyde, 16:1 ISO I/14:0 3 OH, 14:0 3 OH/16:1 ISO I, summed feature 7 is comprised of 18:1 w7cis/w9trans/w12trans, 18:1 w9cis/w12trans/w7cis, 18:1 w12trans/w9trans/w7cis. b The P-value for HS and NHS isolates was determined by two-sided binomial test. For all fatty acids P-value was >0.05.
cycloheptadecanoate (>11%) and summed features 7 (>18%). No significant differences (p > 0:05) were observed among type of FAME profile, quantity of a specific FAME, and whether the isolate was HS and NHS (Table 1). The interrelationships of HS and NHS E. coli FAME profiles are depicted by dendrogram in Fig. 1, and are based on the coefficient of similarity generated by the Euclidian distance between pairs of bacteria. On the basis of the MIDI dendrogram software, at a Euclidian distance of approximately 13.0, all isolates were grouped as a single cluster. At a Euclidian distance of approximately 4.0 (the lowest distance that included all isolates), 18 clusters were well separated. Clusters 1, 2, 7, 8, 11, 13, and 17 comprised only HS isolates, whereas clusters 3, 5, 6, 9, 10 and 12 comprised only NHS isolates. Clusters 4, 14, 15, 16, and 18 contained isolates from both sources. The number of isolates in each cluster was too low to indicate significance differences. Based on less than 10% variance for 85–95% of the FAME profiles, 87 of the 104 isolates were divided into 10 subgroups. It is interesting that subgroups 7, 8, and 10 contained only HS isolates and subgroup 9 only NHS isolates. The remaining subgroups comprised isolates from both sources. The relationship of FAME profiles is also illustrated by principal-component analysis of the data in two dimensions (Fig. 2). HS isolates were more widely distributed throughout the plot compared to NHS isolates. No well-separated HS and NHS groups were observed.
O-Serogrouping A total of 100 HS and NHS isolates were tested for O-serogroup; 77% could be typed. The predominant O-serogroups are shown in Table 2. Human source isolates exhibited 19 serogroups, with 48% belonging to seven serogroups (O2, O20, O28, O79, O148, O153, O159) and the remaining isolates to 12
Fig. 1. Dendrogram of FAME profiles of human source (HS) and nonhuman source (NHS) E. coli isolates. Bottom line indicates number of clusters.
382
Salina Parveen et al.
Fig. 2. Two-dimensional plot of FAME profiles of E. coli isolates generated by principal-component analysis.
Table 2. O-Serogroups of human source (HS) and nonhuman source (NHS) isolates O-Serogroups
2 8 11 13 19 20 28 79 148 150 152 15/143 153 159 Othersa MR NT a
Number (%) of E. coli isolates HS (n ¼ 50)
NHS (n ¼ 50)
3(6) 1(2) 1(2) 0 0 2(4) 2(4) 7(14) 3(6) 0 0 0 2(4) 5(10) 10(20) 4(8) 10(20)
1(2) 3(6) 2(4) 2(4) 5(10) 0 0 1(2) 0 2(4) 2(4) 2(4) 0 0 17(34) 0 12(24)
10 and 17 serogroups of HS and NHS isolates, respectively, that contained only one isolate are not shown. MR: multiple reaction. NT: nontypeable. : not accepted by World Health Organization. One NHS isolate was autoagglutinable.
additional serogroups. In contrast, NHS isolates displayed 26 serogroups, with 36% in seven serogroups (O8, O11, Ox13, O19, O150, O152, O15/143), and the remaining isolates in 19 additional serogroups. Only 2 of 50 NHS isolates shared the seven most frequently detected serogroups for HS isolates (P50:01). This indicated that there was a association between O-serogroups and the isolate source. PFGE A total of 32 E. coli isolates were analyzed for PFGE profile. Preliminary experiments examined several restriction enzymes (SfiI, NotI, and XbaI) with a panel of E. coli isolates to determine the appropriate enzyme for this study. Results showed that SfiI yielded 12–18 discernable bands ranging from 9 to 528 kb, and thus was used for subsequent studies. Patterns were considered to be unique when differentiated by one or more bands. HS and NHS isolates formed 9 and 18 PFGE profiles, respectively. There was no association between PFGE profile and
Phenotypic and genotypic characterization of human and nonhuman Escherichia coli
383
Fig. 3. Representative PFGE profiles of human source (HS) and nonhuman source (NHS) E. coli isolates digested with SfiI restriction endonuclease. Lanes 1 and 20: Low-range PFG marker in kb; 2: SPS 35, 3: SPS 80; 4: SPS 42, 5: SPS 45; 6: SPS 118, 7: SPS 205; 8: SPS 466, 9: SPS 493;10: SPS 601, 11: SPS 602;12: SPS 561,13: SPS 573;14:SP 5; 15: SP 50, 16: SP 93;17: SP 6, 18: SP 112.
source of isolate. In addition, we used PFGE to determine whether isolates which shared RT and MAR profiles were from the same clonal origin. Out of 32 isolates only four pairs shared the same MAR, RT and PFGE profiles (Fig. 3), indicating the existence of multiple E. coli strains rather a few clones in the ANERR. DISCUSSION
When attempting to identify specific health risks associated with polluted water, it is necessary to understand input sources. This approach implies that methods exist that can discriminate the pollution source. Because human feces can carry various human enteric pathogens, such as Salmonella spp., Shigella spp., E. coli, hepatitis A and Norwalk group viruses. In contrast, most of these pathogens do not colonize nonhuman species (Guzewich and Morse, 1986; Orskov and Orskov, 1981). For microbiological hazards, early attempts to differentiate fecal pollution included the ratio of fecal coliform to fecal streptococci, and fecal streptococci species associated with humans, livestock and wildlife (Geldreich and Kenner, 1969). After much research, these methods have proven unreliable (Pourcher et al., 1991). More recently, research has investigated multiple antibiotic resistance (MAR) profiles of E. coli, and fecal streptococci (Parveen et al., 1997; Wiggins, 1996). This approach has shown some promise, and
is certainly a simple and cost-effective method for most laboratories. Concerns about the method include the stability of resistance (i.e. plasmids), geographical variation in MAR, and lack of specificity due to the use of similar antibiotics among humans and other animals. Our laboratory has investigated the ability of ribotype profiling to discriminate E. coli sources, and found that, like MAR, specific ribotype patterns are associated with HS and NHS E. coli isolates (Parveen et al., 1999). However, research has yet to establish the broad geographical application of ribotyping and other possible limitations. More than likely, the combined use of phenotypic and genotypic methods will produce the highest level of discrimination, compared to basing conclusions on a single microbial trait. For this reason, we examined the applicability of FAME, O-serogroup and PFGE profiles as potential discriminatory tools. A FAME profile is a stable phenotypic expression of a bacterial genotype when cells are grown under controlled conditions (Welch, 1991). Fatty acid metabolism is constitutive and directed by the chromosome, not plasmids. The presence of certain bacterial fatty acids has been shown to correlate with certain taxonomic conventions. Hence, fatty acids profiles can be used to produce bacterial ‘‘fingerprints.’’ This observation has led to the use of FAME profiling for chemotaxonomic classifications (Drucker, 1976; Lechevalier, 1977).
384
Salina Parveen et al.
For this study, we used FAME profiles to characterize HS and NHS E. coli. We found that there were no significant differences between FAME profiles of HS and NHS isolates. At a Euclidian distance of 10.0, two major groups were formed, indicating that all isolates were likely members of the same species (Sasser, 1990). It was determined that HS isolates were more widely distributed throughout the plot than NHS isolates. This suggested a high level of phenotypic diversity within the E. coli population. Havelaar et al. (1992), studying Aeromonas strains from patients with diarrhea and from drinking water, observed that FAME profiles could be used to compare strains from different sources. However, they could not identify FAME profiles that discriminated human and drinking water isolates. Similarly, other investigators have found that FAME analysis did not provide a high level of discrimination. For example, FAME profiles could not differentiate pathovars of Xanthomonas campestris, Streptomyces scabies and members of Enterobacteriaceae at the intraspecies level. The treatment of results by principal-component analysis or by other methods of classification did not improve the separation of Xanthomonas campestris and Saccharomyces scabies strains (Chase et al., 1992; Steele et al., 1997). Kotilainen et al. (1991) used this method to identify and type coagulase-negative Staphylococcus, and concluded that FAME could identify isolates at the species level. In addition, FAME was able to differentiate among multiple coagulase-negative staphylococcal blood isolates. We conclude that FAME analysis was unable to differentiate HS and NHS E. coli. However, it may depend on the species examined. For example, some species may exhibit specific fatty acids that are associated with certain microbial characteristics. Similarly, PFGE profiles of E. coli were not able to differentiate HS and NHS isolates. This was not surprising since this method detects small sequence differences that may not associate with a specific bacterial characteristic, such as host source. This technique has also been used by Edberg et al. (1994) to differentiate Enterobacter cloacae in a county drinking water distribution system, hospital source waters, and associated clinical samples. A single PFGE pattern was found in the water distribution system, which differed from that of the isolates found in hospital source water and the clinical E. cloacae isolates. Herbein et al. (1996) reported PFGE profiles for three racoons, one otter, one goose, three humans, and one muskrat. Within this limited sample size, they found that PFGE profiles of racoons differed significantly from those of the otter and goose. There was no significant difference between PFGE profiles of racoons, humans and a muskrat.
From this study, it was determined that the Oserogroups of HS isolates were rarely found in NHS isolates, similar to previously published studies (Orskov and Orskov, 1981). The majority of Oserogroups that we observed for NHS isolates were reported for cattle, chicken, and swine (Orskov and Orskov, 1981). These results also agree with studies by Hartley et al. (1975) who compared the prevalent O-serogroups of diseased animals and humans, and found that most strains associated with diarrhea in pigs and humans were different, while infantile and piglet diarrhea have many common traits. Bettelheim et al. (1976) reported that although the distribution of different E. coli strains identified by serotype appears to differ in human and animals, many serotypes are shared among humans and other animals. This situation was illustrated by analyzing the O : H serotypes of 13,139 strains of E. coli isolated from humans, and 1076 strains isolated from animals, 689 of the latter obtained from cow-pats sampled at 22 sites in England and Wales. Typing with a virtually complete set of 154 O and 53 H antisera led to the identification of 708 distinct O : H serotypes, among which 520 were only found in humans, 130 only in animals, and 58 in both. The great diversity of O : H serotypes is illustrated by the finding that, of 78 serotypes identified in cow-pat specimens, 65 were found only in one site and 12 were found only at both sites. Moreover, approximately half of the isolates from animals could not be typed in spite of the large number of antisera used. In the present study, we found that 22% of the isolates were nontypeable, which is consistent with Blanco et al. (1994), who showed that 19% isolates could not be typed. Moreover, the percentage of untypeable isolates from both HS and NHS was very similar (20 and 24%). This result is supported by other investigators who noted that the frequency of untypeable animal isolates was approximately the same as the frequency of untypeable human isolates obtained in a study of the remote Yanomama Indians of South America (Eveland et al., 1971). Serogrouping may be a useful tool for differentiating HS and NHS of E. coli. Drawbacks include (1) the need for a large bank of antisera, many of which are not commercially available (2) many serotypes are shared among HS and NHS isolates, and that some isolates cannot be typed.
CONCLUSIONS
From this study we conclude that (1) there were no significant differences among FAME profiles, quantity of a specific FAME, and whether the isolate was HS and NHS, (2) PFGE profiles did not show any correlation with source of fecal pollution, (3) HS PFGE profiles were less diverse than NHS isolates, and conversely for FAME, (4) O-serogrouping showed less diversity for HS vs. NHS isolates, and
Phenotypic and genotypic characterization of human and nonhuman Escherichia coli
the predominant O-serogroups differed significantly from those of NHS isolates, and (5) O-serogrouping may be useful tool for differentiating sources of fecal pollution in conjunction with MAR and RT profiling. Acknowledgements}This publication was supported by the U.S. Department of Commerce, National Oceanographic and Atmospheric Administration, Sanctuaries and Reserves Division, grant #NA370R0166 and Engineering Research Center (ERC) for Particle Science and Technology at the University of Florida. We thank Lee Edmiston and Chip Bailey for assistance with sample collection, Dr. J. K. Jackson for technical assistance and K. Robinson for statistical analysis of data.
REFERENCES
Allardet-Servent A., Bouziges N., Carles-Nurit M. J., Bourg G., Gouby A. and Ramuz M. (1989) Use of lowfrequency-cleavage restriction endonucleases for DNA analysis in epidemiological investigations of nosocomial bacterial infections. J. Clin. Microbiol. 27, 2057–2061. Barrett T. J., Lior H., Green J. H., Khakhria R., Wells J. G., Bell B. P., Greene K. D., Lewis J. and Griffin P. M. (1994) Laboratory investigation of a multistate food borne outbreak of Escherichia coli 0157 : H7 by using pulsed-field gel electrophoresis and phage typing. J. Clin. Microbiol. 32, 3013–3017. Bettelheim K. A., Ismail N., Shinebaum R., Shooter R. A., Moorhouse E. and O’Farrell S. (1976) The distribution of serotypes of Escherichia coli in cow-pats and other animal materials compared with serotypes of E. coli isolated from human sources. J. Hyg. 76, 403–406. Blanco J. E., Blanco M., Blanco J., Rioja L. and Ducha J. (1994) Serotypes, toxins, and antibiotic resistance of Escherichia coli strains isolated from diarrhoeic and healthy rabbits in Spain. Vet. Microbiol. 38, 193–201. Buchrieser C., Ganger V. V., Murphree R. L., Tamplin M. L. and Kaspar C. W. (1995) Multiple Vibrio vulnificus strains in oyster as demonstrated by clamped homogenous electric field gel electrophoresis. Appl. Environ. Microbiol. 61, 1163–1168. Chase A. R., Stall R. E., Hodge N. C. and Jones J. B. (1992) Characterization of Xanthomonas campestris strains from Aroids, using physiological, pathological, and fatty acid analyses. Phytopathology 82, 754–759. Crichton P. B. and Old D. C. (1979) Biotyping of Escherichia coli. J. Med. Microbiol. 12, 473–486. Del Rosario L., Barth S. S., Rutter M. and Ray B. (1995) Molecular Analysis of Listeria monocytogenes Isolates Associated with the Consumption of Listeria Contaminated Mixed Vegetables. Abstract from 95th General Meeting of ASM. Washington, DC. Drucker D. B. (1976) Gas liquid chromatography. Methods Microbiol. 9, 51–125. Edberg S. C., Patterson J. E. and Smith D. B. (1994) Differentiation of distribution systems, source water, and clinical coliform by DNA analysis. J. Clin Microbiol. 32, 139–142. Eveland W. C., Oliver W. J. and Neel J. V. (1971) Characterization of Escherichia coli serotypes in the Yanomama, a primitive Indian tribe of South America. Infect. Immun. 4, 753–756. Geldreich E. E. (1966) Sanitary Significance of Fecal Coliforms in the Environment. Water Pollution Control Research Series, Publication WP-20-3. Federal Water Pollution Control Administration, U.S. Department of the Interior, Cincinnati, OH.
385
Geldreich E. E. and Kenner B. A. (1969) Concepts of fecal streptococci in strem pollution. J. Water Pollut. Control Fed. 41, R336–R352. Gonzalez E. A. and Blanco J. (1989) Serotypes and antibiotic resistance of verotoxigenic (VTEC) and necrotizing (NTEC) Escherichia coli strains isolated from calves with diarrhoea. FEMS Microbiol. Lett. 60, 31–36. Gower J. C. (1971) A general coefficient of similarity and some of its properties. Biometrics 27, 857–874. Guzewich J. J. and Morse D. L. (1986) Sources of shellfish in outbreaks of probable viral gastroenteritis: implications for control. J. Food Prot. 49, 389–394. Hartly C. L., Howne K., Linton A. H., Linton K. B. and Richmond M. H. (1975) Distribution of R plasmids among O-antigen types of Escherichia coli isolated from human and animal sources. Antimicrob. Agents Chemother. 8, 122–131. Havelaar A. H., Schets F. H., Silfhout A. V., Jansen W. H., Wieten G. and Kooij D. V. D. (1992) Typing of Aeromonas strains from patients with diarrhea and from drinking water. J. Appl. Microbiol. 72, 435–444. Herbein S. A., Simmons Jr. G. M. and Myers S. L. (1996). Use of Pulsed-Field Gel Electrophoresis (PFGE) to investigate nonpoint fecal coliform sources to tidal inlets on the Eastern shore of the Chesapeake Bay. 96th General Meeting of the American Society for Microbiology, p. 467, May 19–23. New Orleans, LA. Johnson J. M., Weagant S. D., Jinneman K. C. and Bryant J. L. (1995) Use of pulsed-field gel electrophoresis for epidemiological study of Escherichia coli O157 : H7 during a food-borne outbreak. Appl. Environ. Microbiol. 61, 2806–2808. Kotilainen P., Huovinen P. and Eerola E. (1991) Application of gas liquid chromatographic analysis of cellular fatty acids for species identification and typing of coagulase-negative staphylococci. J. Clin. Microbiol. 29, 315–322. Lechevalier M. P. (1977) Lipids in bacterial taxonomy. A taxonomist’s view. Critical Rev. Microbiol. 5, 109–210. Mukwaya G. M. and Welch D. F. (1989) Subgrouping of Pseudomonas cepacia by cellular fatty acid composition. J. Clin. Microbiol. 27, 2640–2646. Orskov F. and Orskov I. (1981) Enterobacteriaceae. In Medical Microbiology and Infectious Diseases, ed. A. I. Broude, pp. 340–352. W.B. Saunders Co., Philadelphia. Parveen S., Murphree R. L., Edmiston L., Kaspar C. W., Portier K. M. and Tamplin M. L. (1997) Association of multiple-antibiotic resistance profiles with point and nonpoint sources of Escherichia coli in Apalachicola Bay. Appl. Environ. Microbiol. 63, 2607–2612. Parveen S., Portier K. M., Robinson K., Edmiston L. and Tamplin M. L. (1999) Discriminant analysis of ribotype profiles of E. coli for differentiating human and nonhuman Sources of fecal pollution. Appl. Environ. Microbiol. 65, 3142–3147. Pourcher A. M., Devriese L. A., Hernandez J. F. and Delattre J. M. (1991) Enumeration by a miniatured method of Escherichia coli, Streptococcus bovis and enterococci as indicators of the origin of fecal pollution of waters. J. Appl. Bacteriol. 70, 525–530. Sasser M. (1990) ‘‘Tracking’’ a Strain Using the Microbial Identification System. Technical Note No. 102 MIDI, Newark, Delaware. Steele M., McNab W. B., Read S., Popope C., Harris L., Lammerding A. M. and Odumeru J. A. (1997) Analysis of whole-cell fatty acid profiles of verotoxigenic Escherichia coli and Salmonella enteritidis with Microbial Identification System. Appl. Environ. Microbiol. 63, 757–760. Tamplin M. L., Jackson J. K., Buchrieser C., Murphree R. L., Portier K. M., Ganger V. V., Miller L. J. and Kaspar C. W. (1996) Pulsed-field gel electrophoresis and ribotype
386
Salina Parveen et al.
profiles of clinical and environmental Vibrio vulnificus isolates. Appl. Environ. Microbiol. 62, 3572–3580. Vauterin L., Yang P., Hoste B., Pot B., Swings J. and Kersters K. (1992) Taxonomy of Xanthomonads from cereals and grasses based on SDS-PAGE of proteins, fatty acid analysis and DNA hybridization. J. Gen. Microbiol. 138, 1467–1477. Welch D. (1991) Application of cellular fatty acid analysis. Crit. Microbiol. Rev. 4, 422–43837.
Wiggins B. A. (1996) Discriminant analysis of antibiotic resistance patterns in fecal Streptococci, a method to differentiate human and animal sources of fecal pollution in natural waters. Appl. Environ. Microbiol. 62, 3997–4002. Wilson R. A. and Francis D. H. (1986) Frimbriae and enterotoxins associated with Escherichia coli serogroups isolated from pigs with collibacillosis. Am. J. Vet. Res. 47, 213–217.