INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Understanding the mechanisms of microbial attachment and adherence to various surfaces is critical to finding new methods for inactivating or removing attached microorganisms from those surfaces. This problem includes microbial attachment to the surfaces of meat animal and poultry carcasses and food-processing surfaces.

Microbial attachment to or association with animal and poultry carcasses has been studied by using a variety of techniques, most of which involve surface sampling, culturing, and back-extrapolation of the resulting counts to the original surface area. This approach and the data derived from it are predicated on the efficiency of the sampling method used, as well as the recovery of culturable organisms. Both destructive and nondestructive sampling methods are inherently variable in terms of the efficiency of removing organisms from food surfaces (11). Alternatively, direct microscopic observation by light microscopy of sample sections and by electron microscopy of fixed and stained specimens has been used (36). Scanning confocal microscopy of nonfixed samples has revealed the actual spatial arrangement of the natural menstruum and has permitted optical sectioning below the surface (18). All three microscopic approaches allow observation of only a small percentage of the total surface area and rely on some form of sampling.

Ideally, microbial attachment processes are best observed under natural or undisturbed conditions without sampling, fixation, or any treatment detrimental to the biological or environmental integrity of the specimen. Nondestructive, in situ investigations of microbial associations have been performed by using bioluminescent and fluorescent reporter gene systems with a myriad of biological systems (5, 6, 8, 20, 29, 38), as well as food contaminated with microbes (1, 3, 10, 12, 14, 16, 19, 28, 31, 34, 35, 37). The biochemical basis of bioluminescence has been reviewed previously; examples of applications abound in the literature and are cited but not reviewed in this paper (15, 21, 22, 23, 24, 26).

The power of bioluminescent tagging of pathogens was demonstrated by an approach in which the infection process could be monitored in living mammalian hosts (4). Using an intensified charge-coupled device (ICCD) camera, workers visualized light from viable bacteria directly through the viscera, skin, and fur of intact, live animals as the infection spread, and thus this technique provided both spatial and temporal information (5, 6). Antibiotic inhibition of bacterial disease progression in the gastrointestinal tract was demonstrated in situ by the loss of luminescence. This noninvasive, in situ approach for studying microbial processes provides important real-time information that could not otherwise be obtained and has direct applications to the study of food contamination.

A similar approach for assessing the association or attachment of food-borne pathogens, such as Escherichia coli O157:H7, to beef carcass surface tissue in situ would have several advantages. First, assessing the presence of inoculated bacteria on a carcass surface would not rely on any type of sampling that would only partially estimate the microbial population. Second, larger, complete, contiguous areas of surface tissue could be examined simultaneously, thereby eliminating the need for multiple sampling of smaller areas of surface tissue and extrapolation of the results to larger areas. Third, the presence or partial removal of the inoculum, as well as the physical location of the inoculum, could be monitored in real time, eliminating the need for repeated sampling and retrospective culture data. And finally, specific gene expression could be monitored and potentially quantified in situ if the proper promoter lux gene fusions were used (13, 26).

Association of the enterohemorrhagic E. coli serotypes, including the most prevalent serotypes of this class (O157:H7 and O157:nonmotile), with products of cattle origin (17, 25) has led to implementation of zero-tolerance rules for visible fecal matter on beef carcasses and to legislation classifying E. coli O157 as an adulterant of ground beef in the United States. The source of E. coli O157 and other gram-negative enteropathogens is feces. Regardless of whether the feces come directly from an animal's gastrointestinal tract (e.g., from a ruptured gut, which occurs infrequently) or are part of the hide and hoof soil load, fecal sources have been implicated in postharvest carcass contamination. Decontamination research has focused on antimicrobial treatment of the carcass surface, which either inactivates or removes pathogenic E. coli and other pathogens (36). Such research has traditionally relied on inoculating carcass surface tissues, treating them, sampling them, and then counting the pathogens by a quantitative culture enumeration method requiring from 24 to 48 h. While this approach is valuable for carcass antimicrobial intervention research, it has not been able to address the localization of contamination on the superficial fascia of the beef carcass surface or to determine preferential binding to different tissue surface types on the carcass. Thus, there are still questions concerning possible preferential attachment or association of bacterial pathogens to the carcass surface.

In this study, we used a strain of bioluminescent E. coli O157:H7 to study the association of this organism with beef carcass surface tissue macroscopically by using a very sensitive ICCD camera. Images of bacterial inocula in buffer suspensions, as well as in bovine feces, were obtained for a variety of beef carcass surface tissue types to determine the level of detection by this method. Our data indicate that real-time monitoring of the duration of attachment of a bioluminescent bacterial strain during simulated carcass washing is possible without exogenously sampling the tissue surface and that there are differences in the adherence of this organism to different carcass surface types.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. E. coli O157:H7 strain B6-914 (= ATCC 43888) was used in this study. This parent strain was screened to determine its stx1, stx2a, and stx2b gene content by PCR (Life Technologies PCR reagent system) by using a previously described primer set (27) and the method described by Nancy Strockbrine, Centers for Disease Control and Prevention, Atlanta, Ga. (35a); E. coli O157:H7 strain B6-914 was negative for the stx toxin genes as determined by this assay (data not shown).

Imaging. As described previously (4), photons emitted from inoculated specimens and cultures were collected and imaged with an ICCD camera (model C2400-32; Hamamatsu, Hamamatsu City, Japan) fitted with a 50-mm f-1.2 Nikkor lens (Nikon, Tokyo, Japan). For all experiments, spatial reference images (grayscale) were collected in dim light immediately before photon emission data were collected in total darkness. Bioluminescent signals were represented as pseudocolor images, where color was indicative of light intensity (red represented the most intense signal, and blue represented the least intense signal). The pseudocolor images were overlaid on the reference images to reveal the spatial distribution of the bioluminescent bacteria and the signal intensities on tissue surfaces. Five-minute integration times were used unless otherwise noted. The images were processed with an Argus 20 image processor (Hamamatsu) and were transferred to a Macintosh Power PC (model 8100-100; Apple Computer, Cupertino, Calif.). Images were superimposed by using Adobe Photoshop, version 4.0 (Adobe Systems, Inc., Seattle, Wash.), and annotations were introduced by using Canvas, version 5.0 (Deneba Systems Software, Miami, Fla.). The bit range settings for pseudocolor images varied between 0 to 3 and 2 to 9 for appropriate display.

Beef carcass surface tissues. Each type of tissue section (lean or adipose) used in these experiments was excised from several different carcasses. The tissues used in each triplicate experiment were from the same tissue section. The tissues collected were in similar states of hydration.

Tissue inoculation and washing. E. coli O157:H7 strain L-2 was propagated as described above. For larger tissue samples (6 by 6 cm), 5 ml of a diluted culture in Luria broth containing 100 µg of ampicillin per ml was placed in a plastic weigh boat (7 by 7 cm), and the beef carcass surface tissue was placed surface side down in the inoculum. Following incubation at room temperature for 15 min, the tissue section was lifted, allowed to drain and drip directly onto a sterile gauze pad, and then placed face up in a clean dry weigh boat for imaging. For fecal inoculation, a diluted aliquot of a strain L-2 culture was mixed with a 1:10 slurry of bovine feces that had been strained through a glass fiber filter in a filtered stomacher bag (Spiral Biotech, Bethesda, Md.) and was inoculated directly onto the tissue surface. For comparisons of surface tissues obtained from different carcass sites, 100-µl portions of a bacterial suspension were applied to the surfaces of tissue sections (2 by 6 cm) and spread with the round fire-polished end of a solid glass spreading rod over the entire surface. After 15 min of incubation at room temperature, any remaining liquid was allowed to drip or drain off, and the inoculated section was placed in a clean dry weigh boat.

Enumeration of bacteria on tissues. Tissue sections were placed in a filtered stomacher bag along with 25 ml of buffer consisting of 0.1% (wt/vol) Tween 20 in buffered peptone water (Difco Laboratories, Detroit, Mich.). After 2 min of treatment with a Colworth stomacher at the normal or midrange setting, samples from the filtered side of the bag were diluted and plated by the track dilution method (32) onto Luria agar containing 100 µg of ampicillin per ml. The plates were incubated for 24 h at 37°C before enumerating and calculating the number of CFU per square centimeter of tissue surface.

Statistical analyses. Bioluminescent signal data was normalized by calculating the signal intensity (I) on a per-pixel-area (A) basis (I/A) by using the quantitation feature of the Argus 20 image processor (Hamamatsu). The fraction of the bioluminescent signal (I/A) or the number of CFU per square centimeter remaining after the wash procedure was determined (means of triplicate determinations were used when replicate data could be obtained) and analyzed by using the InStat, version 3.00, statistical analysis package (GraphPad Software, San Diego, Calif.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Correlation between bioluminescence and cell numbers. The relationship between the signal intensity (I/A) and viable cell numbers, as determined in a solid-white, flat-bottom microtiter plate by using serial dilutions of cells in phosphate-buffered saline, indicated that there was a high degree of linear correlation in a 30-min exposure experiment conducted at 37°C (r² = 0.97) (Fig. 1) or a 5-min exposure experiment conducted at room temperature (r² = 0.93). As determined with microtiter plates, the minimum number of strain L-2 cells with observable luminescence that was at least two times the background luminescence was approximately 50 CFU.

View larger version (15K): [in this window] [in a new window]   FIG. 1.   Correlation of luminescent signals (log I/A) and viable plate counts of bioluminescent E. coli O157:H7 strain L-2(pCGLS1) in microtiter plate wells. Twofold serial dilutions of strain L-2 were prepared in phosphate-buffered saline in triplicate with 50 µl (final volume) per well. The total luminescence in each well was determined by using the ICCD camera with a 30-min integration time. Samples from each well were plated, and the number of CFU per well was determined. The correlation between viable cell number and light intensity is linear.

View larger version (86K): [in this window] [in a new window]   FIG. 2.   Beef carcass surface tissue inoculated with bioluminescent E. coli O157:H7 strain L-2 by inversion in 5 ml of an ~107-CFU/ml culture. (a) Grayscale reference image. (b) Bioluminescence. The arrows indicate areas of concentrated residual bioluminescence along tissue striations of the overlying fascia.

View larger version (144K): [in this window] [in a new window]   FIG. 3.   Growth of bioluminescent E. coli O157:H7 strain L-2 on beef carcass surface tissue inoculated by inversion in 5 ml of an ~106-CFU/ml culture (resulting in approximately 103 to 104 CFU/cm2). (Top panels) Grayscale alone (left) and bioluminescence superimposed on grayscale (right), obtained immediately after inoculation. (Bottom panels) Grayscale alone (left) and bioluminescence superimposed on grayscale (right), obtained following 16 h of incubation at 37°C.

View larger version (140K): [in this window] [in a new window]   FIG. 4.   Inoculation of beef carcass surface tissue with bioluminescent E. coli O157:H7 strain L-2 suspended in buffered peptone water. Lean (top) and adipose (bottom) tissue sections were inoculated by spreading 100-µl diluted culture aliquots, which resulted in approximately 1.6 × 105 CFU/cm2 (1.9 × 106 total CFU over the entire surface area), and water washed. Washing reduced the signal intensities obtained for both lean and adipose tissues, but the inoculum in buffered peptone water was removed more efficiently from adipose tissue than from lean tissue. (a) Grayscale prewash. (b) Bioluminescence prewash. (c) Bioluminescence postwash.

View larger version (129K): [in this window] [in a new window]   FIG. 5.   Effects of bovine feces on adherence of E. coli O157:H7 strain L-2 to lean and adipose beef carcass surface tissues. Three lean tissue sections (top) and three adipose tissue sections (bottom) were inoculated with a dilution of E. coli O157:H7 strain L-2 in a bovine feces slurry (approximately 107 total CFU over the entire surface area). (a) Grayscale prewash. (b) Bioluminescence prewash. (c) Bioluminescence postwash.

View larger version (139K): [in this window] [in a new window]   FIG. 6.   Inoculation of mixed surfaces. Bioluminescent E. coli O157:H7 strain L-2 was inoculated (9.2 × 107 total CFU) onto carcass surface tissues that included lean and adipose surfaces across the tissue sections. (Images i) Prewash grayscale. (Images ii) Prewash bioluminescence superimposed on grayscale. (Images iii) Bioluminescence superimposed on grayscale after a single wash. (Image iv) Bioluminescence after a second wash. A transitional tissue section with the overlying fascia tissue intact (a) and a transitional tissue section with the overlying fascia tissue removed by scalpel trimming (b) (indicated by the box in panel b, image i) were inoculated. In panel b inoculation included the section that had intact fascia and the section from which the fascia was removed (indicated by the box in image i). The arrows in panel a indicate the line of tissue topography transition. The arrows in panel b indicate the excision line between the intact fascia and the excised fascia.

View larger version (47K): [in this window] [in a new window]   FIG. 7.   Sites of sampling for various beef carcass surface tissues. The diagram is representative of a typical grain-fed beef carcass and indicates the locations of tissues used in this experiment. The numbers indicate sampling sites (see text).

View larger version (20K): [in this window] [in a new window]   FIG. 8.   Effects of washing on the adherence of E. coli O157:H7 strain L-2 to inoculated tissues from different carcass regions. The fractions of signals (bioluminescence or viable count) remaining on various beef carcass surface tissues obtained from six different sites (Fig. 7) following a flow water wash are shown. The data are the means of triplicate values obtained for tissue sections (2 by 6 cm) that were inoculated with an average of 2.1 × 107 CFU of strain L-2.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Spatiotemporal analyses of the interaction between E. coli O157:H7 and beef carcass tissue surfaces were conducted by using transformed bacteria which expressed the complete lux operon of P. luminescens (13) as an optical tag. The interaction of the tagged bacterium with tissues was monitored by using bioluminescence as a real-time indicator. The complete lux operon includes genes that encode the biosynthetic enzymes for the substrate, and thus, exogenous substrate addition is not required. Although addition of an exogenous aldehyde, such as decanal, may increase the output of the luxAB genes compared to the output when the substrate is supplied through expression of all of the lux genes (luxCDABE) (2), a system that does not require exogenous substrate addition removes any variability due to substrate availability and streamlines the assay. Moreover, any potential effects of exogenous substrate addition on the physical or temporal location of bacterial cells on the surface tissues are eliminated. The P. luminescens luciferase genes have the advantage of functioning at a higher temperature (up to 45°C) than other luciferases that have been characterized (2), a trait which may be useful in future in vivo applications of the bioluminescent strain of E. coli O157:H7.

The utility and validity of using luminescent bacterial strains to study biological processes have been reviewed previously (33). Data from this study and elsewhere indicate that there is a strong correlation between the viable counts of bioluminescent bacteria and light emission (3).

We observed that as few as 50 cells of E. coli O157:H7 strain L-2 could be detected by bioluminescence and that there was an approximately 10-fold decrease in sensitivity when bacteria were suspended in bovine fecal slurries compared with bacteria suspended buffer or broth. The slopes of the curves in plots of intensity versus cell number indicated that the sensitivity of detection was probably affected more by the optical properties of the fecal slurry than by direct effects of feces components on light output per cell. Bioluminescence from mouse fecal suspensions inoculated with Yersinia enterocolitica containing luxAB (16) and luxCDABE-bearing Salmonella typhimurium (4) and bioluminescence from bovine feces from cows fed luxAB-bearing Y. enterocolitica have been reported previously (16).

Initial inoculation of E. coli O157:H7 strain L-2 onto tissues revealed that bacteria could be visualized directly on a large area of beef carcass surface tissue. The luminescence appeared to be especially concentrated along muscle striations on the surface of the tissue, which differed in appearance, composition, and topography (Fig. 2). In a separate experiment, increases in bioluminescence were observed after incubation on a section of beef tissue (Fig. 3). A similar observation was reported by Chen and Griffiths for luxAB-bearing Salmonella enteritidis inoculated onto chicken meat following incubation (3).

Water washing of beef tissues inoculated with bioluminescent E. coli O157:H7 strain L-2 by the immersion-shaking method or by using a bottle to squirt water onto the inoculated surface revealed that the effects of both decontamination methods could be visualized in real time, whether the surface tissue was lean or adipose. Although these washing procedures were not entirely comparable in terms of pressure and volume to the spray washing procedures used by the meat industry or during pilot-scale research (7), our results suggest that imaging bioluminescent bacterial reporter strains can be used to assess carcass decontamination processes.

Differences in the attachment of food-borne bacterial pathogens to various tissue surfaces are an important issue in food microbiology and food processing. Attachment to tissue-specific molecules or polymers (30) and attachment to or detachment from carcass surface tissues and excised sections of meat are typically determined by sampling and by performing plate count experiments (9). The ability to monitor the process of attachment of pathogens to or association of pathogens with carcass surfaces in real time without exogenous sampling should accelerate such studies and has specific advantages. Unlike using physical biochemical indicators, using luciferase as the reporter ensures that the signal observed is from viable, metabolically active cells which can be detected as they exist in or on a food matrix.

Data presented in this paper indicate that bacterial association with the beef carcass surface may be influenced by the physical topography and structure of the surface. We observed that following a water wash, tissues that were rougher retained more bioluminescence (which correlated with the viable counts) than tissues that were relatively smooth and homogeneous. It has been reported that bacterial removal from or decontamination of beef carcass surface tissue is influenced by the level of lean or adipose tissue in the underlying tissue menstruum (7, 9). In these previous studies, excised samples of treated tissues were more likely to retain bacterial inocula if the surface tissue was lean than if it was adipose. Our bioluminescence data are in agreement with these observations; however, our bioluminescence method generated both spatial and quantitative information in real time. Although bioluminescence monitoring will not entirely replace culture methods in the study of attachment and decontamination, obviating the need for excision sampling, it has distinct advantages as a screening tool for selecting decontamination protocols for validation.

The observation that increased bioluminescence indicative of bacterial growth occurs on tissue surfaces after incubation has broad implications for studying decontamination processes. Since small numbers of tagged organisms (approximately 50 cells) can be detected on tissue surfaces and since luciferase-based photon counting imaging has a broad dynamic range of many logs, low levels of microbial contamination can be monitored over time. Moreover, spatially resolved analyses of bacterial survival and growth patterns after antimicrobial treatments could be rapidly performed by using the molecular biophotonic method described here, which would eliminate the need for exogenous sampling and culturing.

Eliminating back-extrapolation of sample plate count data would greatly improve the validity of carcass surface microbial count data. Biophotonics eliminates the need for the assumption that bacterial loads are homogeneously distributed across the surfaces of carcasses. Extrinsic factors that influence attachment of microorganisms to fascia-covered animal carcasses, such as the degree of hydration and composition, as well as putative intrinsic attachment factors, such as cell surface receptors, host tissue components (e.g., hyaluronan, collagen, and chondroitin sulfate [36]), could be systematically examined in situ over significantly larger areas of carcass tissue surface than the areas previously examined to determine their roles in attachment by using the biophotonic system described in this study.

Cloning promoters of interest upstream of a complete lux operon could provide indicator strains that could report metabolic, regulatory, or gene expression activity under different environmental conditions found in food processing. A similar approach has been described by Dodd et al., who used a spvA-lux fusion in S. enteritidis to monitor levels of the RpoS product in cells in the presence and absence of a competitive microflora (10).

Since bioluminescence has been used as a real-time genetic reporter in live animal models (4-6), the growth and survival of E. coli O157:H7 and other enterohemorrhagic E. coli organisms could be evaluated in animal models involving human infection and carcass processing. Specifically, expression of genotypic traits that influence survival in animals, such as the factors which determine the minimum infectious dose in human hosts, and transfer from the animal or gastrointestinal tract to the carcass during processing could be studied by this biophotonic procedure.

Our findings demonstrate that real-time macroscopic imaging of bacteria on beef carcass tissue surfaces is possible if bacterial strains with a lux gene reporter are used. We present evidence showing that different tissue surfaces, especially lean and adipose surfaces, retain different levels of contaminating bacteria. The magnitude of the differences is consistently less than 1 log₁₀ unit of inoculated population per unit of area. Therefore, any practical implications of this observation are yet to be determined. Previously, however, Thomas and McMeekin (36a) documented the role of the water microlayer on poultry skin, which has a very significant effect on microbial attachment. It is not unlikely that in our experiments adipose tissues retained less water on their surfaces (due to their inherent hydrophobicity) and thus retained fewer bacteria after simulated spray washing. The clear differences between the prewashed and postwashed tissues indicate that a molecular biophotonic approach to studying microbial association with carcass tissue surfaces offers a truly in situ strategy for understanding a problem previously approached only retrospectively (i.e., through sampling, culturing, and subsequent back-calculation of microbial densities on tissue surfaces).