Prevalence of Escherichia coli O157:H7 From House Flies (Diptera: Muscidae) and Dairy Samples in North Central Florida¹

Abstract

Efficient detection of enterohemorrhagic Escherichia coli O157:H7 is important to monitor the safety of food products obtained from cattle, and it has been primarily accomplished by analyzing manure samples by selective cultivation techniques, PCR, and ELISA. As each technique suffers from different biases, there may be value in using multiple methods and samples to increase detection efficiency. Difficulties associated with cattle manure sampling can be circumvented by isolation of E. coli O157:H7 from house flies, Musca domestica (L.), which present as an important vector for spreading diseases. Thus, isolation of pathogens directly from house flies provides information about the potential human health impact that house fly dispersal can have because of pathogen distribution. House flies can disperse from dairy farms, where E. coli O157:H7 endemically thrive in cattle, to restaurants where food is prepared and served. Here, we report that detecting E. coli O157:H7 in house flies was 2.7 times more frequent than in manure from nearby dairy farms. Flies appear to offer a promising alternative in efforts to detect E. coli O157:H7 in dairy farms, restaurants, processing plants, and other establishments.

Resumen (español) La detección de Escherichia coli O157:H7 en las lecherías es importante para mejorar la seguridad de los productos lácteos, y se ha llevado a cabo principalmente mediante el aislamiento de las bacterias a partir de las muestras de estiércol. Sin embargo, los componentes biliares presentes en el estiércol complica la identificación genética utilizando la técnica del PCR, y el aislamiento microbiológico se dificulta por la presencia de bacterias competidoras que comparten características microbiológicas similares. El aislamiento de E. coli O157:H7 a partir de la mosca doméstica evita las dificultades asociadas con el estiércol del ganado. El aislamiento de patógenos a partir de las moscas domésticas proporciona información adicional sobre el potencial impacto epidemiológico de la dispersión de la mosca doméstica en la distribución de patógenos, ya que las moscas domésticas se dispersan desde las lecherías donde la E. coli O157:H7 existe en forma endémica en el ganado. En este estudio, se encontró que las moscas domésticas son 2,6 veces más sensibles para la detección de E. coli O157:H7 en las lecherías. Las moscas son más fáciles de capturar y manejar que el estiércol, y deberían ser utilizadas en cualquier ensayo para detectar E. coli O157:H7 en las lecherías y otros establecimientos.

PCR, cattle manure, enteric bacteria, pathogen, molecular isolation

Escherichia coli O157:H7 is present as a commensal microbe in its primary reservoir, cattle, and has been isolated from house flies, Musca domestica (L.), in dairy farms (Alam and Zurek 2004). Many dairy farms in Florida are located in proximity to human population centers because urban expansion has decreased the distance between dairy farms and towns. House flies have been shown to disperse up to 3 km from dairy farms in rural areas to restaurants in nearby urban areas, thus indicating their tremendous potential to serve as vectors for E. coli O157:H7 transmission (Burrus 2010). Although E. coli O157:H7 has been isolated from restaurant dumpsters in Gainesville, FL (Butler et al. 2010), no study has been published to determine the potential for pathogen transmission by house flies from dairy farms to towns.

Because infection with E. coli O157:H7 is a notifiable event in the United States (Mead and Griffin 1998), it is important to use environmental detection methods pointing at potential contamination that are rapid, selective, and sensitive (Ogden et al. 2001). There are many effective methods for detecting Shiga toxin-producing E. coli, including E. coli O157:H7, from food (Fratamico and Bagi 2007), dairy-spilled grains (Callaway et al. 2009), dairy cattle (Hussein and Sakuma 2005), and cattle manure (Toth et al. 2013), including microbiological isolation using various selective enrichments (Fratamico and Bagi 2007), immunological detection systems such as ELISA and immunomagnetic separation (Ogden et al. 2001), and molecular isolation using different PCR assays (DeBoer and Heuvalink 2000). PCR capability has increased with the development of multiplex assays that can successfully test for up to four genes simultaneously (Noll et al. 2015). PCR detected a higher prevalence of Shiga toxin-producing E. coli from dairy products (42%) and meat products (70%) compared with culture methods (36% and 27%, respectively) by Rantsiou et al. (2012). Szalanski et al. (2004) developed a 6-h protocol for detecting E. coli O157:H7 from house flies by PCR targeting Shiga-like toxins. Pava-Ripoli et al. (2015) recently developed a standardized PCR isolation assay for isolation of foodborne pathogens from individual flies including confirmation by microbiological culturing. PCR effectively detects E. coli O157 from many different materials, including human food such as raw beef (Fratamico and Bagi 2007), and swine (Oporto et al. 2008), flies (Pava-Ripoli et al. 2015), and soil and vegetable crops (Islam et al. 2004), in addition to cattle manure (Hu et al. 1999, Noll et al. 2015). In addition, E. coli O157:H7 has also been isolated from many non-livestock mammals and arthropods that may be associated with livestock facilities because of the proximity or with animal movement, including that of deer (Asakura et al. 1998, Dunn et al. 2004a), opossum (Renter et al. 2004), pigeons (Morabito et al. 2001), rabbits (Scaife et al. 2006, Fremaux et al. 2008), house flies (Sanderson et al. 2006), blow flies (Buma et al. 1999, Fotedar et al. 1992), stable flies (Buma et al. 1999), starlings (Swirski et al. 2013), and slugs (Sproston et al. 2006). Thus, when examining dairy farms and other facilities for E. coli O57:H7, it may be useful to obtain samples from a wide range of animals and arthropods. As E. coli is a normal commensal bacteria found in vertebrate digestive tracts (Conway and Cohen 2015), it is important to focus on the E. coli serotypes that produce disease-causing Shiga-like toxins (Cebula et al. 1995), such as E. coli O157:H7.

However, >100 E. coli distinct serotypes (Cebula et al. 1995) can produce Shiga-like toxins. Even with multiplex PCRs that successfully target multiple genes simultaneously, isolation and presumptive identification of E. coli O157:H7 using culture-based methods can be useful to ensure accurate E. coli O157:H7 identification and quantification from cattle feces (Noll et al. 2015) or flies (Pava-Ripoli et al. 2015). Therefore, isolation and presumptive identification of E. coli O157:H7 improve specificity of PCR-based E. coli O157:H7 identification.

Problems encountered when sampling farms and urban areas for the prevalence of E. coli O157:H7 include competing microorganisms and bacterial metabolic products and organic acids in feces and soils that inhibit the growth of E. coli O157:H7 impact the reliability of DNA extractions (Shin et al. 2002). USDA:APHIS:VS (2007) reported that cattle herd prevalence in the United States can range from 22–100%; however, Hancock et al. (1997) determined that prevalence rates in individual cattle are typically 0–9.5%, with adult cattle exhibiting lower rates (0.4%) than weaned heifers (1.8%). Detection of E. coli O157:H7 from dairy samples is typically very low, with rates that range from 0.28% (Hancock et al. 1994) to 1.93% (Murinda et al. 2002) for isolation from dairy cattle feces, and with rates of 0.75% from bulk tank milk (Murinda et al 2002) on dairy farms. Higher rates have been reported by Fernández et al. (2012), as they noted that 38/808 (4.7%) of dairy cattle rectal swabs tested positive for Shiga toxin-producing E. coli. Similarly, Stenkamp-Strahm et al. (2016) report a 3.0% prevalence of enterohemorrhagic E. coli. Prevalence can differ in one animal, based on the location from where the sample is obtained, and differences between rectoanal-junction mucosal swabs and feces in dairy cattle (heifers) have been observed (Davis 2006). Detection of prevalence can also be affected by the type of selective and differential agars used (Wallace and Jones 1996).

Therefore, it is critical to obtain an adequate sample for accurate improved E. coli O157:H7 detection specificity and sensitivity. The objective of this project was to study the prevalence of E. coli O157:H7 in adult house flies, spilled grain, and manure samples from two dairy farms and monitor the dispersal by adult house flies collected from two restaurant garbage dumpsters in a nearby town in northern Florida.

Materials and Methods

Microbiological Detection

Throughout the entire study period from 14 June 2008 to 16 September 2008, samples for microbiological examination were collected from two dairy farms (Dairy A and Dairy B), with 400-600 cattle each and from two restaurants (Restaurant C and Restaurant D) located within 4 km of both dairy farms in North Central Florida. Samples collected from dairy farms included adult house flies, spilled feed grains, and fresh manure. Restaurant dumpster collections consisted of adult house flies. All collection materials were sterilized before use.

Fly samples were collected between 0900 and 1600 h, in one to four sweeps of aerial sweep-net, transferred to sterile 120-ml clear polypropylene specimen cups (Samco Scientific Corp., San Fernando, CA), placed on ice in a large cooler (10–20°C), and transported to the laboratory, as ambient temperature prevented adequate chilling and manipulation of flies in the field. Two exceptions to the time of adult house fly collection occurred on 20 July 2008 and 28 July 2008 when fly samples were collected between 1600 and 1900 h, as mid-day temperatures averaged 37°C, greatly reducing fly movement. At the laboratory, specimen cups were maintained at −20°C for 1 min to permit subsequent transfer of individual flies into individual microcentrifuge tubes. In the laboratory, three to four flies at a time were gently shaken onto a chilled metal pan lined with aluminum foil and identified to species level. This action was repeated until 10 house flies were obtained in this manner, or until it was determined that no additional house flies remained in the specimen cups owing to low capture during the hot summer. Foil was changed after each sampling. Specimens identified as M. domestica were placed individually into sterilized 1.5-ml snap cap microcentrifuge tubes (Fisher Scientific Co., Waltham, MA) using sterilized forceps, placed inside a sealable plastic bag, and maintained at 10–20°C in a cooler for 2–6 h to keep the flies alive and the bacteria viable while also reducing bacterial interactions between flies until use. After 2–6 h, these individually contained flies were removed from the 10-–20°C cooler for enumeration of aerobic bacteria in unenriched buffered peptone water (BPW; Oxoid Ltd., Basingstoke, Hampshire, England) broth; each entire fly was placed individually in a sterile 15-ml polypropylene centrifuge tube (Becton Dickinson Labware, Franklin Lakes, NJ) containing 9 ml of BPW broth. Flies were not homogenized. These individual fly samples were incubated at room temperature with one fly in 9 ml of BPW broth for 2–6 h.

Flies that were not removed from the specimen cups during sorting were kept in the specimen cups (no liquid broth) and placed in a − 20°C freezer for up to 30 min to reduce the potential of flies escaping during processing using the laboratory storage protocol in use at that time (Fratamico and Bagi 2007) in the food laboratory where microbiological analysis was conducted.

Solid samples were collected from fresh manure droppings (manure) obtained within 30 s after animal defecation and from spilled feed grain samples (grain) from feed troughs or from spilled grains below feed augers. In addition, swab samples were obtained from the manure or grain sources by inserting two sterile plastic-shaft, polyester-tipped swabs (Fisherbrand, Thermo Fisher Scientific, Pittsburgh, PA) premoistened with BPW broth into either the manure or the grain to a depth of 2.5–5.0 cm. The two swabs were rolled gently until completely coated with either manure or grain particulates and then placed together as one sample in a sterile 15-ml polypropylene centrifuge tube (Becton Dickinson Labware) containing 9–25 ml of BPW broth. Swab samples were transported to a Biosafety Level 2 laboratory within 6 h of collection, and were then processed within 24 h of collection using microbiological methods. Before processing, fly samples consisting of only adult flies that remained in the specimen cups (no liquid broth) were placed in a −20°C freezer for up to 30 min to reduce the potential for flies escaping during processing; manure and grain samples remained in the cooler at 10–20°C, and the swab samples were maintained at room temperature (25°C) until processed.

Aerobic plate counts (APC) of unenriched background microbial organisms for individual fly, manure, and grain samples in BPW broth were performed using six serial dilutions of the nutrient broth containing the respective samples. Each fly, manure, and grain sample was vortexed for 30–60 s to suspend bacteria in the BPW broth, and then maintained at room temperature for 1 min to permit the debris to settle. Flies were not homogenized. A 1-ml aliquot of the liquid BPW broth containing suspended bacteria from each unenriched background microbial sample was pipetted into a 15-ml borosilicate glass screw-capped culture tube (No. 9825-16X, Corning Pyrex Inc., Lowell, MA) containing 9 ml of fresh BPW broth. Serial dilutions were prepared in fresh BPW broth, with dilutions of 10⁻¹–10⁻⁶. Two APCs using Petrifilm APC plates (3M, St. Paul, MN) were performed for each dilution tube. First, tubes were vortexed for 30–60 s and were allowed to settle for 5–10 s. Then 1-ml aliquots were pipetted onto APC plates and dispersed over a 20-cm² surface, providing colony-forming unit (CFU/g) estimates. Plates were incubated at 37°C for 6–18 h, and CFUs were counted the next day. Plates with 14–300 CFUs were used to calculate the average values of total aerobic bacteria CFUs per gram for each sample.

To improve selectivity for E. coli O157:H7 from high concentrations of competing background microorganisms, tryptic soy broth (TSB; enrichment broth) was modified by the addition of novobiocin (20 mg/l; mTSB + N; FDA-CFSAN 2007). Bacteria species used as positive and negative controls in this study were obtained from the Food and Environmental Toxicology Laboratory, University of Florida. A nalidixic acid-resistant E. coli O157:H7 strain 204P was used as the positive control. Negative controls used in this study were Shigella dysenteriae (American Type Culture Collection [ATCC] 49550) and Salmonella enterica serovar Thompson (ATCC 8391).

Frozen fly samples containing intact multiple fly species (not in liquid) were removed from the freezer, and 4–25 M. domestica were randomly selected and placed as a pooled sample in fresh mTSB + N. Adult house flies were incubated in pools of 4–25 flies. For pooled samples with 4–9 flies, M. domestica adults were added to 9 ml of mTSB + N (4–9 flies) or to 10 ml of mTSB + N (10–15 flies) in a 15-ml polypropylene centrifuge tube (Model No. 35-2096, Becton Dickinson Labware), or to 25 ml mTSB + N (16-25 flies) in a 50-ml polypropylene centrifuge tube (Model No. 43089, Corning Incorporated, Corning, NY). Fly samples containing pooled flies in mTSB + N broth were incubated at 37°C for 24 ± 2 h. After incubation, pooled fly samples were vortexed for 30–60 s (flies were not homogenized) and allowed to settle for 1 min in preparation for enumeration on cefixime tellurite sorbitol MacConkey (CT-SMAC) and for isolation on both CHROMAgar and CT-SMAC. Both enumeration and isolation procedures conducted on fly samples used the nutrient broth vice the fly carcasses.

Excreted manure and spilled grain swab samples were processed using a swab sample technique (Rice et al. 2003). Sample tubes containing two substrate-inoculated swabs in 9 ml of BPW broth were processed by vortexing each tube for 30–60 s. One swab and 1 ml of supernatant were transferred to a 15-ml centrifuge tube containing 9 ml of mTSB + N. Transferred samples were incubated at 37°C for 24 ± 2 h. The second swab and remaining BPW supernatant (8 ml) were stored at −20°C for future analysis. Following selective enrichment in mTSB + N, samples were removed from the incubator, vortexed for 30–60 s, and allowed to settle for 1 min in preparation for enumeration and isolation procedures.

Enumeration of E. coli O157:H7 was performed using SMAC plates supplemented with cefixime (15 μg/l) and potassium tellurite (1.25 μg/l; CT-SMAC; Alam and Zurek 2004). Further, 1-ml aliquots of each unenriched vortexed sample liquid for each tested fly (unhomogenized), grain, and manure samples were transferred to culture tubes containing 9 ml of sterile BPW broth and vortexed for 30–60 s. Six serial dilutions were prepared for each sample, as previously described, and 100-µl aliquots were pipetted onto each of two CT-SMAC plates with a range from 10⁻² to 10⁻⁷ and spread evenly using a glass rod. CT-SMAC spread-plates were incubated at 37°C for 24 ± 2 h. Spread-plate counts of E. coli O157:H7 were obtained by averaging the number of CFUs for plates that contained 14–300 colonies.

Isolation of E. coli O157:H7 was performed by immunomagnetic separation (IMS) followed by streak-plating of IMS products consisting of 100-μl bacteria-bead complexes onto CHROMAgar (BBL CHROMAgar O157, Beckton Dickinson Co.) plates (Dynal 2007). Plates were incubated at 37°C for 24 ± 2 h.

After incubation, both CT-SMAC enumeration spread-plates and CHROMAgar isolation plates were examined for presumptive E. coli O157:H7 colony growth, as seen by colorless colonies with or without a light smoky center (sorbitol-negative) on CT-SMAC plates and light-violet to violet colonies on CHROMAgar plates (Dynal 2007). Up to five presumptive E. coli O157:H7 colonies from CHROMAgar plates were selected and streaked onto CT-SMAC plates that were incubated at 37°C for 24 ± 2 h. Presumed E. coli O157:H7 colonies were subsequently loop-streaked onto non-selective, general growth trypticase soy agar with yeast extract (TSAYE) plates and incubated at 37°C for 24 ± 2 h.

Where sufficient samples were available, additional biochemical tests were conducted. These tests included phenotypic expression of characters typical for E. coli O157:H7 as follows: 1) hydrolyzation of tryptophan; 2) the presence of a brilliant green sheen on Levine’s eosin methylene blue (L-EMB) agar; and 3) the lack of fluorescing 4-methylumbelliferyl-β-D-glucuronide (MUG; FDA-CFSAN 2007).

Hydrolyzation of tryptophan was assessed by placing a filter paper wetted with Kovac’s Reagent (Ricca Chemical Co., Arlington, TX) on each positive-growth plate. Colonies that are presumptive-positive for E. coli O157:H7 typically turn pink. Presumptive E. coli O157:H7 colonies were plated concurrently onto the following two different media: L-EMB (Oxoid) agar and fresh TSAYE agar plates. Incubation of L-EMB plates, which inhibit gram-positive growth, occurred at 37°C for 24 ± 2 h. Following incubation, plates were examined for a distinctive bright metallic green sheen on dark-blue to black nucleated colonies. This brilliant green sheen is characteristic of E. coli O157:H7, and it differentiates E. coli O157:H7 from non-pathogenic E. coli that turn dark green without the brilliant green sheen.

Isolates transferred to TSAYE plates were tested for the presence of glucoronidase, an enzyme that hydrolyzes MUG to yield fluorescing 4-methylumbelliferone, by the addition of one ColiComplete (Bothell, WA) disk to the plate’s heaviest bacterial streak. After incubation at 37°C for 24 ± 2 h, colonies presumptive for E. coli O157:H7 were considered negative if they lacked blue fluorescence around the disks when examined under 365-nm ultraviolet (UV) light.

Aerobic bacteria were enumerated using Petrifilm APC plates containing 14–300 CFUs/g. Aerobic bacteria counts were obtained for 19 selected samples that consisted of nine house flies, seven grain, and three manure samples that were collected from 31 May 2008 to 26 August 2008. APCs of the two grain and two house fly samples obtained from Dairy A on 26 August 2008 were averaged for each type. Enumeration data for the remaining 17 samples were based on single samples. When multiple dilution plates containing 14–300 CFUs/g were obtained from individual samples, the range of those plates was recorded.

Presumptive E. coli O157:H7 colonies were counted using CT-SMAC spread plates containing 14–300 CFUs on 31 May 2008 and 14 June 2008, with samples of 14 June 2008 submitted for microbiological analysis after enumeration. Colonies that tested positive on CHROMAgar plates were classified as presumptive-positive. However, because of failure of the CHROMAgar on 26 August 2008 and 16 September 2008, isolates that were presumptive-positive on CT-SMAC agar plates on these dates were also submitted to PCR. Thus, isolates of 24 samples of the 57 dairy-collected samples were submitted to PCR, with isolates from 11 samples originating from CHROMAgar plates and those from 13 samples originating from CT-SMAC plates.

Multiple isolates were obtained from several samples because of sequential transfer of individual colonies during the microbiological testing. Therefore, prevalence rates were determined for the number of both the samples tested and the isolates that were sub-cultured from the samples.

Sub-cultures of each isolate were prepared for long-term storage by aseptic loop-transfer of one loop of logarithmic growth phase E. coli O157:H7 from TSAYE culture plates to 850 µl of sterile lysogeny broth (LB) (Oxoid). The LB and bacteria were mixed by gentle pipetting and then added to 150 µl of sterile glycerol in a sterile 2-ml cryogenic vial (P/N 10-500-26, Thermo Fisher Scientific, Pittsburgh, PA). The LB, bacteria, and glycerol mixtures were stored as a 15% glycerol stock in a −20°C non-thawing freezer. Presumptive-positive isolates were prepared for DNA extraction and PCR analysis ∼1.5 yr after placement in the freezer.

Molecular Detection of E. coli O157

In total, 10 µl of glycerol (20% wt/vol) frozen culture (each culture having been obtained from liquid mTSB + N broth of vortexed but unhomogenized intact pooled flies after initial incubation immediately following collection of intact adult house flies) were transferred into a 15-ml culture tube containing 2-ml sterile LB, followed by 37°C for 24 ± 2 h with shaking (250 RPM). Total genomic DNA was extracted from incubated stock using the QIAquick DNeasy Blood and Tissue Kit (QIAgen, Valencia, CA). Extracted DNA was suspended in 100-μl AE buffer and stored at 4°C. DNA concentration was measured by absorbance at 260–280 nm, using a NanoDrop 1000 mini-spectrophotometer (Thermo Fisher Scientific, Waltham, MA) by comparison of band intensity on agarose gels against a standard 100-base pair (bp) ladder (Invitrogen, Carlsbad, CA; GenVault 2010).

PCR was performed using one multiplex and two uniplex PCR assays (Table 1) to amplify gene fragments using two primer pairs (20 pmol/µl each) designed to target rfbE_H7 and fliC_O157 (Hu et al. 1999, Cagney et al. 2004, Szalanski et al. 2004). The fliC_O157 primer pair amplifies a 625-bp E. coli O157 serotype gene fragment (Gannon et al. 1997), and the rfbE_H7 primer pair amplifies a 259-bp E. coli H7 gene fragment (Paton and Paton 1998). Because of low amplification of both gene fragments simultaneously in the multiplex PCR, uniplex PCR assays were also conducted using the original DNA sample.

Table 1.

Primer nucleotide sequences used to amplify target genes in PCR assay

Primera	Sequence (5′–3′)	Expected PCR amplicon (bp)	Target gene	Specificity
Assay 1 (multiplex: O157 somatic and H7 flagellar antigen)
fliCH7F	GCGCTGTCGAGTTCTATCGAGCc	625	flic	H7 flagellar gene
fliCH7R	CAACGGTGACTTTATCGCCCATTCCc
rfbEO157-F	CGGACATCCATGTGATATGGc	259	rfb	nt 393-651 of rfbO157:H7
rfbEO157-R	TTGCCTATGTACAGCTAATCCc
Assay 2 (uniplex: O157 somatic antigen)
rfbEO157-F	CGGACATCCATGTGATATGGb	259	rfb	nt 393-651 of rfbO157:H7
rfbEO157-R	TTGCCTATGTACAGCTAATCCb
Assay 3 (uniplex: H7 flagellar antigen)
fliCH7F	GCGCTGTCGAGTTCTATCGAGCc	625	flic	H7 flagellar gene
fliCH7R	CAACGGTGACTTTATCGCCCATTCCc

a

F, forward primer; R, reverse primer.

b

Reference: Paton and Paton 1998.

c

Reference: Gannon et al. 1997.

Master Mix reagents for each assay consisted of deionized sterile water, 10X DNA polymerase PCR buffer (-MgCl₂), deoxynucleotide (dNTP) Mix (10 mM each), 50 mM MgCl_2, oligonucleotide primers (20 pmol/µl each; Eurofins MWG Operon, Huntsville, AL), and recombinant DNA polymerase Taq enzyme (Invitrogen, Carlsbad, CA) (Table 1). For the multiplex and uniplex assays, the PCR program consisted of an initial denaturing at 94°C (2 min), 35 cycles of denaturing at 94°C (45 s), annealing at 56°C (multiplex) or at 65°C (uniplex) (45 s), and extending at 72°C (1 min). A final extension was performed at 72°C (5 min). A nalidixic acid-resistant E. coli O157:H7 strain 204P was used as the positive control. Negative controls were sterile distilled water, S.dysenteriae (ATCC 49550) and S.enterica serovar Thompson (ATCC 8391). In addition, to ensure that PCR-product gene fragments were from bacterial DNA, a broad-range 16S rDNA PCR assay was performed on 46 selected isolates.

Amplified PCR gene fragments were visualized by ethidium bromide staining in 1% agarose gel electrophoresis. Visualized PCR products were photographed under UV light. Isolates were considered positive if both fliC and rfbE fragments were amplified. Because multiple isolates corresponded to each dairy-collected sample, samples were considered positive if at least one isolate was positive.

Numerical Analysis

CFUs of aerobic bacteria and E. coli O157:H7 were enumerated for plates containing 14-300 CFUs/plate. Prevalence rates were calculated as the percentage of tested samples that contained at least one positive isolate.

Results

Microbiological Detection

Throughout the entire study period from 14 June 2008 to 16 September 2008, 35 house fly, 24 spilled grain, and 9 manure samples were collected (Table 2). Of these 68 samples, 14 fly, 17 grain, and 6 manure samples were obtained from Dairy A, and 10 fly, 7 grain, and 3 manure samples were obtained from Dairy B (Table 3). During the same period, seven and four fly samples were collected from restaurants C and D, respectively (Table 3).

Table 2.

Prevalence by sample-type: prevalence (%) and number of E. coli O157:H7 CHROMAgar-positive samples and number of CT-SMAC- and PCR-positive samples relative to the number of samples tested for each sample type within each site after collection from two dairy farms and two restaurant dumpsters in North Central Florida in 2008

Compare	No. collected	CHROMAgar		CT-SMAC	PCRa
		No. tested	No. positiveb (%)	No. positive (%)	No. positive (%)
Flies	35	33	8 (24%)	11 (33%)	8 (24%)
Grain	24	17	3 (18%)	8 (47.1%)	5 (29%)
Manure	9	7	0 (0%)	3 (42.9%)	1 (14%)
Total	68	57	11 (19.3%)	22 (38.6%)	14 (24.6%)

a

Samples submitted to PCR consisted of all CHROMAgar-positive samples except on two dates when CHROMAgar failure resulted in submission of subsequent CT-SMAC-positive samples to PCR.

b

Percent positive was calculated per dairy farm by dividing the number of positive samples by the number of samples tested in the laboratory. Overall positive percentages are the average of all three testing methods combined.

Table 3.

Prevalence by site and sample type: prevalence (%) and number of E. coli O157:H7 CHROMAgar-positive samples and number of CT-SMAC- and PCR-positive samples relative to the number of samples tested for each sample type within each site after collection from two dairy farms and two restaurant dumpsters in North Central Florida in 2008

Compare	Site/Sample	No. collected	CHROMAgar		CT-SMAC	PCRa
			No. tested	No. positiveb (%)	No. positive (%)	No. positive (%)
Farm A	Fly	14	13	4 (30.8%)	4 (30.8%)	2 (15.4%)
	Grain	17	11	2 (18.2%)	5 (45.5%)	4 (36.4%)
	Manure	6	5	0 (0%)	2 (40%)	1 (20%)
Farm B	Fly	10	9	1 (11.1%)	1 (11.1%)	2 (22.2%)
	Grain	7	6	1 (16.7%)	3 (50%)	1 (16.7%)
	Manure	3	2	0 (0%)	1 (50%)	0 (0%)
Restaurant C	Fly	7	7	2 (28.6)	4 (57.1%)	2 (28.6%)
Restaurant D	Fly	4	4	1 (25.0%)	2 (50%)	2 (50%)
Total	All	68	57	11 (19.3%)	22 (38.6%)	14 (24.6%)

a

Samples submitted to PCR consisted of all CHROMAgar-positive samples except on two dates when CHROMAgar failure resulted in submission of subsequent CT-SMAC-positive samples to PCR.

b

Percent positive was calculated per dairy farm by dividing the number of positive samples by the number of samples tested in the laboratory. Overall positive percentages are the average of all three testing methods combined. Samples tested (n = 57) included all samples tested (fly, grain, and manure) from Dairy A (n = 29), from Dairy B (n = 17), from Restaurant C (n = 7), and from Restaurant D (n = 4).

To test for microbial presence of E. coli O157:H7, 57 samples consisting of 33 fly, 17 grain, and 7 manure samples were plated on CHROMAgar plates for initial presumptive-positive analysis (Table 2). Of the 57 tested samples, presumptive-positive prevalence rates for E. coli O157:H7 on CHROMAgar and CT-SMAC and by PCR were 11 (19.3%) and 22 (38.6%) and 14 (24.6%), respectively (Table 2, Supp. Table 1 [online only]). Microbiological processing of these 57 samples on CHROMAgar plates produced 103 fly, 61 grain, and 33 manure isolates for a total of 197 isolates that were tested further on CT-SMAC and by PCR (data not shown for breakdown of isolates). Across the study, overall E. coli O157:H7 sample prevalence for all sample types (flies, grain, and manure) for the combined dairy farms was 24.6% (Table 4).

Table 4.

Prevalence by dairy farm: prevalence (%) and number of E. coli O157:H7 CHROMAgar-positive, CT-SMAC-positive, and PCR-positive samples relative to the number of samples tested after collection from two dairy farms in North Central Florida in 2008

Compare	CHROMAgar	CT-SMAC	PCR	Overall
	Positivea %	Positive %	Positive %	Positiveb %
Dairy A	20.7	37.9	24.1	27.6
Dairy B	11.8	29.4	17.7	19.6
Dairy average	17.4	34.8	21.7	24.6

Samples submitted to PCR consisted of all CHROMAgar-positive samples except on two dates when CHROMAgar failure resulted in submission of subsequent CT-SMAC-positive samples to PCR.

a

Percent positive was calculated per dairy farm by dividing the number of positive samples by the number of samples tested in the laboratory.

b

Overall positive percentage is the calculated average of all three testing methods’ positive percentages at each dairy farm. Samples tested (n = 46) included all samples tested (fly, grain, and manure) from both Dairy A (n = 29) and Dairy B (n = 17).

Overall prevalence (including all fly, manure, and grain samples per farm) was higher at Dairy A (27.6%) than at Dairy B (19.6%) (Table 4). Each isolation method also provided higher prevalence at Dairy A than at Dairy B, regardless of the testing method used (Table 4). Dairy A prevalence rates ranged from 20.7% (CHROMAgar) to 37.9% (CT-SMAC), whereas those at Dairy B were lowest on CHROMAgar (11.8%) and highest on CT-SMAC (29.4%) (Table 4). The overall prevalence of E. coli O157:H7 from flies at the both restaurants combined (39.4%) was nearly double the prevalence from flies at both dairy farms combined (21.2%) (Table 5).

Table 5.

Prevalence (%) from house flies only: comparing prevalence from house flies at dairy farms vs house flies at restaurants and the number of E. coli O157:H7 CHROMAgar-, CT-SMAC-, and PCR-positive samples relative to the number of samples tested at each dairy farm and each restaurant after collection from two dairy farms and two restaurants in North Central Florida in 2008

Compare flies from sites	CHROMAgar	CT-SMAC	PCR	Overall
	Positivea %	Positive %	Positive %	Positiveb %
Diary A flies	18.2	18.2	9.1	15.2
Diary B flies	4.5	4.5	9.1	6.1
Average, dairy flies	22.7	22.7	18.2	21.2
Restaurant C flies	18.2	36.4	18.2	24.2
Restaurant D flies	9.1	18.2	18.2	15.2
Average, restaurant flies	27.3	54.5	36.4	39.4
Average, all flies	23.9	37.3	29.0	30.1

Some samples were positive by all three testing methods, whereas some that tested positive by one method may have tested negative by either one or both other testing methods.

a

Percent positive was calculated per dairy farm by dividing the number of positive samples by the number of samples tested in the laboratory.

b

Overall positive percentages are the average of all three testing methods combined. Flies tested in this study (n = 33) were from Dairy A (n = 13), Dairy B (n = 9), Restaurant C (n = 7), and Restaurant D (n = 4).

Escherichia coli O157:H7 was presumptively isolated and identified from house flies at all four sites and from grain at both dairy farms using CHROMAgar plates. All manure samples tested negative on CHROMAgar, although one manure sample was positive by PCR and three were positive by CT-SMAC. The CT-SMAC agar provided higher isolation prevalence results for fly (n = 11), grain (n = 8), and manure (n = 4) samples than either CHROMAgar (8, 3, and 0, respectively) or PCR (8, 5, and 1, respectively) (Table 2, Supp. Table 1 [online only]). Of the 22 samples that were positive on CT-SMAC agar (six fly, five grain, and two manure samples), 13 were positive only on CT-SMAC agar and 9 additional samples were clones of CHROMAgar-positive samples and tested positive on both media: 6/9 samples from Dairy A, 3/4 from Dairy B, 2/2 from Restaurant C, and 1/2 from Restaurant D were presumptive-positive on both CHROMAgar and CT-SMAC agar (data not shown).

Total bacteria prevalence is likely underreported in this study because of the use of selective media for enumeration of both aerobic bacteria and E. coli O157:H7. Enumeration of aerobic bacteria using Petrifilm APC plates varied across sample types and dates, and CFUs obtained from preliminary samples obtained between 14 June 2008 to 26 August ranged from 1.3 × 10³ CFUs/g (fly, 14 June 2008) to 9.8 × 10⁷ CFUs/g (grain, 14 June 2008) (Table 6).

Table 6.

Enumeration of aerobic bacteria (CFU/g) using Petrifilm APC plates inoculated with 1 ml of the unenriched sample after collection from two dairy farms and two restaurant dumpsters in North Central Florida in 2008

Collection Date	Site, Source	Aerobic bacteria count range (CFU/g)a
14 June	A, Grain	9.8 × 107
	A, Manure	2.8 × 107 − 4.6 × 107
	B, Grain	4.7 × 106
	B, Manure	2.5 × 106 − 6.7 × 106
	C, House fly	1.3 × 107 − 3.0 × 107
23 June	A, Grain	1.6 × 105 − 2.8 × 106
	A, House fly	1.4 × 107 − 2.0 × 107
	B, Grain	1.8 × 105 − 3.0 × 105
	B, House fly	2.2 × 105 − 4.0 × 106
	C, House fly	1.3 × 103
26 Aug.	A, Grain	4.8 ×106 − 8.2 × 107
	A, House fly	3.4 × 104 – 1.5 × 105
	B, Grain	2.9 × 106 – 3.5 × 106
	B, House fly	1.1 × 105 – 1.4 × 105
	C, House fly	1.5 × 104
	D, House fly	2.1 ×104

These data provide information about the background aerobic microbial community population level that existed in each tested sample.

a

Range (CFU/g) presents only one value for samples types comprising only one sample. Where value ranges are shown, multiple samples of that sample type were enumerated. Sample enumeration was conducted approximately every month, so that changes in enumeration for each sample type at each site could be observed over the period from 14 June 2008 to 26 August 2008. Samples were collected from two dairy farms and from dumpsters at two restaurants in North Central Florida. A = Dairy A, B = Dairy B, C = Restaurant C and D = Restaurant D.

Molecular Detection

Multiplex and uniplex PCR assays confirmed the presence of both genes in 12 (50%) and 14 (58%) of the 24 bacterial samples that were submitted to PCR (data not shown for breakdown of multiplex vs uniplex PCR assays). Multiplex PCR confirmed 58% (7/12) of house fly, 56% (5/9) of grain, and 0% (0/3) of manure samples as positive. In contrast, uniplex PCR confirmed 67% (8/12) of house fly, 56% (5/9) of grain, and 33% (1/3) of manure samples as positive. Multiplex and uniplex assays typically were positive for identical samples, although a few were positive on either the multiplex assay only or on the uniplex assay only. Across the study, the combined PCR multiplex and uniplex assays determined presumptive-positive E. coli O157:H7 results for 24% fly (8/33), 29% grain (5/17), and 14% manure (1/7) samples that were tested by all methods, including CHROMAgar, CT-SMAC, and PCR (Table 2).

Samples obtained in week 6 (5 August 2008) were not available for PCR assays because of freezer malfunction that killed the corresponding isolates. NanoDrop spectrophotometry analysis resulted in 260:280 ratios that ranged from 1.82 to 1.90. The 16S rDNA amplification efficiency of the 46 selected isolates was 89% (41/46 isolates). Eleven percent of the samples did not contain detectable levels of DNA.

Discussion

We report a higher prevalence by combined IMS and CHROMAgar isolation of E.coli O157:H7 from house flies (24%) than from grain (18%) or from manure (0%) (Table 2), and we report a similar prevalence of E. coli O157:H7 by PCR from house flies (24%) and from grain (29%) that is much greater than that from manure (14%) (Table 2). These data support the findings of Lahti et al. (2003) and Pao et al. (2005), and suggest that E. coli O157:H7 on dairy farms may be more accurately and easily detected by testing only house flies. Escherichia coli O157:H7 was isolated from house flies at all four locations. Isolation of E.coli O157:H7 from house flies could be underreported in our study, as the house flies were not homogenized, so that only exterior bacteria were obtained from house flies. Only seven manure samples were tested, whereas 17 grain and 33 house fly samples were tested (Table 2, Supp. Table 1 [online only]). If target pathogen numbers are low, then detection can be difficult without increased numbers of samples (Brichta-Harhay et al. 2007). Zero or very low recovery rates are not uncommon in cattle manure (Galland et al. 2001), often because of high densities of competing background microorganisms present in manure (Pao et al. 2005), and because cattle manure typically does not contain a high number of E. coli O157:H7 (10⁵ CFU) per gram of substrate (Ogden et al. 2004).

It is interesting to note that the overall prevalence of E. coli O157:H7 was higher at Dairy A (27.6%) than at Dairy B (19.6%) for each testing method used (Table 4). This is made even more interesting by noting that Dairy A is located between Dairy B and the town center, within the flight range between the restaurants and the two dairy farms. It is possible that its central location between multiple attractive sites plays a role in the higher prevalence of E. coli O157:H7 at dairy Farm B; this should be considered in any sanitation and/or mitigation efforts aimed at reducing the E. coli O157:H7 prevalence.

Escherichia coli O157:H7 was recovered from 30.1% of the overall house fly samples in this study (Table 5), which is similar to the 32.5 ± 1.3% reported by Fotedar et al. (1992). This confirms that the use of a non-selective enrichment broth and unmodified CHROMAgar plates (lacking antibiotics) might be optimal for isolation of E. coli O157:H7. In contrast, studies that used selective enrichment broths and selective CHROMAgar plates reported only 6.2 ± 1.0% isolation of E. coli O157:H7 (Agui 2001, Keen et al. 2006). This suggests that antibiotic stress may be inhibitory for isolation of E. coli O157:H7 from house flies, although not as inhibitory as that from either manure or feces.

The negative impact of the highly selective detection method owing to the combined use of IMS and CHROMAgar is potentially confirmed by observing that in our study, CHROMAgar results of IMS bead-bacteria complexes consistently achieved the lowest prevalence rates of E. coli O157:H7 from fly (n = 8), grain (n = 3), and manure (n = 0) compared with either CT-SMAC (n = 11, 8, and 4, respectively) or PCR (n = 8, 5, and 1, respectively) (Table 3). This is likely because of the fact that up to five presumptive-positive isolates were transferred from the CHROMAgar plates onto fresh CT-SMAC plates. In contrast, only one presumptive-positive colony was submitted to PCR. It is likely that PCR prevalence results would increase if more isolates underwent PCR assays. In addition, because of failure of some CHROMAgar plates on 26 August 2008 and all CHROMAgar plates on 16 September 2008, presumptive-positive isolates from CT-SMAC plates were submitted to PCR where CHROMAgar plates were not usable, resulting in higher numbers of presumptive-positive E. coli O157:H7 for both CT-SMAC and PCR, whereas CHROMAgar data for those dates are zero (Supp. Table 1 [online only]). As almost half (10/22) of the presumptive-positive CT-SMAC isolations occurred on 16 September 2008, a comparison of the overall prevalence rates between CHROMAgar and CT-SMAC in this study should take this into consideration, and note that isolation of E. coli O157:H7 may otherwise be approximately equal between the two microbiological methods (Supp. Table 1 [online only]).

Because of the slow growth of colonies on CHROMAgar plates, prevalence of E. coli O157:H7 initially may have been underreported in this study when characterization occurred at 18 h. Subsequent evaluations included examinations at 24 and 48 h to improve accuracy. Positive E. coli O157:H7 fly samples were obtained routinely throughout the test period, at both dairy farms and both restaurants, even though low numbers of flies were available at restaurants.

Background aerobic bacteria counts at the dairy farms were highest in manure and lowest in flies, corresponding to highest isolation of E. coli O157:H7 from house flies and no isolation of E. coli O157:H7 from manure. These results agree with those obtained by others for house flies (Alam and Zurek 2004) and for cattle manure (Vold et al. 2000, Brichta-Harhay et al. 2007). In addition, results of this study suggest that decreased detection of E. coli O157:H7 from manure corresponds either to the high background counts of competing microorganisms or to the low presence of the target pathogen. Our data indicate that CHROMAgar does not appear to be ideal for isolating E. coli O157:H7 from manure, as all manure samples tested negative on CHROMAgar, and only one manure sample was positive by PCR and three were positive by CT-SMAC, indicating that the CHROMAgar provided false negative results on at least one to three samples (Table 3).

PCR assays were used to confirm direct-culture presumptive isolation and identification of E. coli O157:H7 colonies. CHROMAgar testing of 57 samples provided 32 samples for CT-SMAC testing; 24 samples were submitted to PCR following microbiological processing with both CHROMAgar and CT-SMAC. However, only 58% (14/24) of the presumptive-positive samples were confirmed with PCR (Supp. Table 1 [online only]). This suggests that the remaining 42% were either false positives on the culture plates or that the fresh cultures lacked adequate quantities of DNA for successful PCR amplification following an extended time in the freezer.

It is interesting that 33% (1/3) of the manure samples tested positive by uniplex PCR because all three samples were obtained from isolates that initially appeared negative using CHROMAgar media. These results suggest either an increased sensitivity of PCR over direct culture (Fratamico et al. 2005) or detection of sorbitol-fermenting E. coli strains (Cebula et al. 1995).

Confirmation by PCR of the microbiological results may have been reduced because the purified isolates were stored long term at −20°C (Fratamico and Bayles 2005), and thawing and refreezing of cultures may have contributed to mortality (Mennigmann 1979), gene loss (Achá et al. 2005), and/or contamination. The influence of freezing may have been a minor factor, given that bacterial cultures were grown before PCR analysis. In this study, an 8.3% reduction in viable counts was observed over 1.5 yr, which is similar to the 7% mortality over 9 mo reported by Doyle and Schoeni (1984). Mortality of individual isolates likely impacted the PCR confirmation rate, which underscores the importance of obtaining multiple isolates from each presumptive-positive colony. Because cultures in this study were provided with minimal nutrients for 24 ± 2 h incubation, they may have been either dead or stressed and nonviable before their placement in the freezer. Survival of cultures in −20°C storage was recently shown to be increased by restricting the exponential growth phase to ∼3 h or increasing the nutrient availability before storage (Sezonov et al. 2007).

Samples positive for fliC_H7 segments could include E. coli of serotypes other than O157 (Cebula et al. 1995, Mead and Griffin 1998). Because the combination of fliC_H7 and rfbE_O157 is unique to E. coli O157:H7 (Bilge et al. 1996), a multiplex PCR assay including these two fragments was informative. In several samples, multiplex PCR produced only one of the two target PCR products, suggesting that one primer pair worked more efficiently than the other. This may have been strengthened by successful amplification of both bands when samples were submitted to both uniplex PCR assays.

Flies provided the most reliable source of E. coli O157:H7 from dairy farms using either microbiological or molecular (PCR) techniques. Escherichia coli O157:H7 has been shown to survive in the house fly for up to 4 d (Sasaki et al. 2000). House flies have a 3-km flight range (Burrus 2010); thus, successful control of fly populations at any particular farm will necessitate close attention to fly populations at neighboring farms, in town, and at farms where controlling the fly population is desired.

Acknowledgments

We thank the Florida dairy producers and restaurants for access to the study sites. We also thank W. Y. Hsu, A. Maruniak, L. Wood, H. Furlong, and M. Doyle for their assistance with this study. This study was supported by the University of Florida Agricultural Experiment Station federal formula funds, Project No. FLA-04598, received from Cooperative State Research, Education, and Extension Service, U.S. Department of Agriculture. The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, or the U.S. Government. Roxanne G. Burrus is a military service member. This work was prepared as part of his official duties. Title 17 U.S.C. 105 provides that “Copyright protection under this title is not available for any work of the United States Government”. Title 17 U.S.C. 101 defines a U.S. Government work as a work prepared by a military service member or employee of the U.S. Government as part of that person’s official duties.

Supplementary Data

Supplementary data are available at Journal of Medical Entomology online.

Author notes

Subject Editor: Timothy Lysyk