Microbial Indicators and Possible Focal Points of Contamination during Production and Processing of Catfish

The catfish industry is important to the United States economy. The present study determined the levels of microbial indicators and the prevalence of Listeria spp. and Listeria monocytogenes at catfish farms and catfish processing plants. Live fish, water, and sediment samples were analyzed in farms. Fish skin, fillets, chiller water, and environmental surfaces were assessed at the processing plants both during operation and after sanitation. Live fish had 2% prevalence of Listeria monocytogenes, while sediment and water were negative for Listeria. Live fish skin counts averaged 4.2, 1.9, and 1.3 log CFU/cm2 aerobic (APC), total coliform (TCC) and generic Escherichia coli counts, respectively. Water and sediment samples averaged 4.8 and 5.8 log CFU/g APC, 1.9 and 2.3 log CFU/g TCC, and 1.0 and 1.6 log CFU/g generic E. coli counts, respectively. During operation, Listeria prevalence was higher in fillets before (57%) and after (97%) chilling than on fish skin (10%). Process chiller water had higher (p ≤ 0.05) APC, TCC, and Listeria prevalence than clean chiller water. After sanitation, most sampling points in which Listeria spp. were present had high levels of APC (>2.4 log CFU/100 cm2). APC combined with Listeria spp. could be a good approach to understand microbial contamination in catfish plants.


Introduction
Farm-raised catfish is the 8th most consumed fish and seafood in the United States [1]. Catfish production primarily includes channel catfish (Ictalurus punctatus), blue catfish (Ictalurus furcatus), and the hybrid channel catfish female x blue catfish male [2]. The largest producing states are Mississippi, Alabama, and Arkansas with Mississippi as the largest state with respect to production and processing [3].
In the catfish industry, the use of microbial indicators can provide a general assessment of possible fecal contamination, product shelf-life, and sanitation efficacy [4]. This group of microorganisms includes, among others, aerobic bacteria, psychrotrophs, coliforms, and Escherichia coli [5][6][7][8]. Aerobic and psychrotrophic counts indicate fish spoilage and inadequate handling during processing [4,9], whereas the presence of coliforms and generic E. coli is associated with unsanitary working surfaces, fecal contamination, and possible contamination with pathogens [10,11].
In addition to conventional microbial indicators, Listeria spp. could be used as an indicator of a lack of or incomplete sanitation. Prevalence of Listeria spp. in catfish processing plants has been reported as 52-85% during operations [12]. Persistent Listeria spp. have been related to sites with a lack of or inadequate sanitation and surfaces that are difficult to reach during samples, one water sample, and one sediment sample from each pond were sampled for microbiological analysis during the summer season. At each processing plant, fifteen food contact surfaces (FCS) and five non-food contact surfaces (NFCS) were chosen as sampling points during operation (OP) and after cleaning and sanitation (AS). One of the FCS (fillet chiller) could be sampled only in 4 out of 6 plant visits. In addition, 1 clean chiller water sample, 1 process chiller water sample, 5 fish skin at the receiving station, 5 fish fillets before chilling, and 5 fish fillets after chilling were taken at each plant visit. A total of 538 samples (200 samples from catfish farms and 338 samples from processing plants) were evaluated in this study.

Sample Collection and Sample Preparation
Live fish were sampled by rinsing the fish surface with 225 mL of buffered peptone water (BPW). Environmental surfaces (10 × 10 cm) were swabbed with either sterile cotton swabs containing 1 mL of BPW (Difco, Fisher Scientific, Sparks, MD, USA) for FCS and fish skin or sterile sponge-sticks containing 10 mL of Dey-Engley (DE) neutralizing broth (3M Food Safety, 3M™ Sponge-Stick with 10 mL D/E Neutralizing Broth, St. Paul, MN, USA) for NFCS. Sediment (250 g) and liquid samples (250 mL) were collected in sterile bags and plastic containers, respectively. Fillets were collected in sterile bags. All samples were placed on ice and immediately transported to the Mississippi State University Food Microbiology and Safety Lab at the Department of Food Science, Nutrition and Health Promotion for further analysis. Samples were processed within 4 h after arrival to the lab. The amount of sample portion tested for fillet, water, and sediment was 25 g. This amount was aseptically weighed and individually placed in sterile bags with 225 mL of either buffered Listeria enrichment broth (BLEB) for Listeria spp. detection or BPW for APC, TCC, and E. coli counts. Sediment and fillet samples were homogenized for 1 min in a stomacher (Stomacher 400 Circulator, Seward, Cincinnati, OH, USA). Water samples were hand mixed for 1 min.

Detection and Isolation of Listeria spp.
Samples were enriched and isolated according to the FDA-BAM protocol [37] with some modifications. Buffered Listeria enrichment broth (BLEB) was added to the samples prior to incubation for 24 h at 30 • C. Aliquots from turbid samples were streaked on Modified Oxford medium agar plates (Difco, Fisher Scientific) and incubated at 30 • C for 24 h. Non-turbid samples were re-incubated at 30 • C for an additional 24 h. Five presumptive Listeria colonies (black colonies surrounded by a halo) were selected from each plate to be confirmed by Multiplex PCR. The selected colonies were transferred to trypticase soy agar (30 • C for 24 h) and then to trypticase soy broth (30 • C for 24 h).

Identification of Listeria spp. by Multiplex PCR
The Multiplex PCR procedure was based on the protocol used by Chen et al. [12] with some modifications. DNA of presumptive isolates was extracted using boiling lysis. First, 1 mL aliquot was transferred to a 1.5 mL microcentrifuge tube and centrifuged at 10,000 rpm × 2 min (Eppendorf, Centrifuge 5415C, Westbury, NY, USA). The obtained supernatant was removed, and the resulting pellet was rehydrated with 50µL distilled water. The sample was then boiled for 5 min and centrifuged again for 2 min. The resulting supernatant was used as the DNA template for Multiple PCR. This technique allowed for the differentiation of Listeria spp. into L. monocytogenes, L. inocua, L. grayi and a group of Listeria seeligeri-Listeria welshimeri-Listeria ivanovii [38]. The reaction mixture contained  [12]. The amplifications were carried out by a thermocycler (Eppendorf, New York, NY, USA). The PCR products were separated by electrophoresis using 1.4% agarose gel and photographed under UV light (BioDoc-itTM Imaging System (UVP, Upland, CA, USA)).

Determination of Aerobic Plate (APC), Total Coliform (TCC), and Generic Escherichia coli Counts
Aerobic plate (APC), total coliform (TCC), and generic E. coli counts were determined in all samples using 3M™ Petrifilm™ Plates (3M Food Safety, 3M™ Petrifilm™ Aerobic Count plates, 3M™ Petrifilm™ E. coli/Coliform Count Plates, St. Paul, MN, USA), according to the manufacturer's instructions. Serial dilutions (10 −1 to 10 −5 ) were carried out for all the samples using 0.1% BPW. An aliquot of 1 mL for each dilution was placed in each petrifilm plate and incubated for 48 h at 35 • C ± 2 • C. After incubation, colonies were counted and reported as log CFU/cm 2 for live fish and fish skin, log CFU/100 cm 2 for food contact and non-food contact surfaces, and log CFU/g for fillet, water, and sediment samples.

Experimental Design and Statistical Analysis
For farm samples, a randomized complete block design with three treatments (sample type: live fish skin, pond water, and pond sediment) and five replications (farms as blocks) was utilized to compare APC, TCC, and generic E. coli log counts. For overall levels in the processing environment, a randomized complete block design with two treatments (sampling time: during operation and after cleaning and sanitation) and six replications (plants as blocks) was utilized to compare APC, TCC, and generic E. coli log counts. For chiller water samples at the processing plants, a randomized complete block design with two treatments (sample type: process chiller water and clean chiller water) and six replications (plants as blocks) was utilized to compare APC, TCC, and generic E. coli log counts. For fish samples at the processing plants, a randomized complete block design with three treatments (sample type: fish skin, fillet before chilling, and fillet after chilling) and six replications (plants as blocks) was utilized to compare APC, TCC, and generic E. coli log counts. When significant (p ≤ 0.05) interactions existed, data were analyzed using Fisher's Protected least significant difference test to separate treatment means (SAS Version 9.4). In addition, a paired t-test was performed for each sampling point (FCS and NFCS) to find the difference (p < 0.05) for APC, coliforms, and E. coli log counts between when the plant is in operation and after cleaning and sanitation. The data were reported as mean ± standard deviation (SD) for microbial counts and prevalence for Listeria spp. and Listeria monocytogenes.

Indicators in Catfish Pond Samples (Farm)
The presence of Listeria spp. (L. monocytogenes, L. inocua, L. grayi or a group of Listeria seeligeri-Listeria welshimeri-Listeria ivanovii) and three different microbial indicators (APC, TCC, and generic E. coli) were assessed in 150 samples of live fish and 25 samples of both pond water and pond sediment ( Table 1). Two percent of live fish (3/150) were positive for Listeria spp. that were confirmed as L. monocytogenes. Pond water and sediment samples were negative for Listeria spp. Pond sediment had the highest (p ≤ 0.05) APC counts, followed by pond water and live fish. No significant differences (p > 0.05) occurred in TCC and generic E. coli counts between live fish, pond water, and sediment (Table 1).
With no significant change in generic E. coli load, APC and TCC for chiller water were higher (p ≤ 0.05) during processing (process chiller water) than at the start of the day (clean chiller water). In addition, higher Listeria spp. prevalence was detected in process chiller water (83%) when compared to clean chiller water (33%) ( Table 2). The Listeria species that were isolated from chiller water during operation include L. monocytogenes (50%) and the group Listeria seeligeri-Listeria welshimeri-Listeria ivanovii (33%).

Indicators in Food Contact (FCS) and Non-Food Contact (NFCS) Surfaces at the Processing Plant
Prevalence of Listeria spp. and L. monocytogenes and the levels of three different microbial indicators (APC, TCC, and generic E. coli) were evaluated in 118 environmental samples during operation and after cleaning and sanitation (Table 2). Overall, APC, TCC, and E. coli counts were less (p ≤ 0.05) after cleaning and sanitation than during operation. Likewise, the prevalence of Listeria spp. and L. monocytogenes was lower after sanitation (14% and 10%, respectively) than during operation (37% and 21%, respectively) ( Table 2). There were 11, 8, and 5 sampling points for APC, TCC, and E. coli counts, respectively, that had less (p < 0.05) bacteria after sanitation (Tables 3 and 4). During operation, Listeria spp. were isolated from 17 (85%) processing points with 15 (75%) isolates identified as L. monocytogenes. With respect to other microbial indicators, APC values were above 2.4 log CFU/100 cm 2 at 18 sampling points (90%), and generic E. coli counts were below 2.0 log CFU/100 cm 2 for all sampling points, regardless of sampling time (Tables 3 and 4). After sanitation, Listeria spp., including L. monocytogenes, were not detected in 10 locations (50%). Listeria was present at 7 locations. Coliforms and generic E. coli were reduced to undetectable levels in 5 (25%) and 16 (80%) locations, respectively. APC were above 2.4 log CFU/100 cm 2 at 8 points (40%) and reduced below this value at 10 points (50%). Ten environmental samples (FCS: 6; NFCS: 4) were positive for Listeria spp. after cleaning and sanitation, including the skinner, fillet chiller, tray in freezer with fish/gray lug, fish holding tray for graded fillets, tote with whole fish, belt before injection, ice container/ice pipe, waste belt, floor, and drain (Tables 3 and 4).

Discussion
In this study, Listeria spp. was not found in pond water and sediment and the prevalence was 2% on live fish (Table 1). Low contamination on live fish might be related to the nature of the skin mucus, which is considered the first physical barrier to pathogens in the fish immune system. Fish skin mucus contains antimicrobial substances, such as proteases, antimicrobial peptides, lectins, lysozyme, immunoglobulin, and transferrins [39]. A study by Miettinen and Wirtaneny [40] reported that the number of samples of aquaculture fish contaminated with Listeria spp. was typically greater after rainy periods. River waters as well as other runoff waters seemed to be the main contamination source at the farm studied. The farmed fish originally found to carry L. monocytogenes become gradually Listeria-free (in the pond/farm). Thus, the positive samples could have been from ponds sampled after a rain/runoff event (ponds sampled were filled from water from streams for the most part).
Pond water and sediment had APC ≥ 4.8 log CFU/g and TCC and generic E. coli levels ≥1.0 log CFU/g ( Table 1). The temperature in summer is commonly greater than 30 • C (90 • F) in Mississippi, resulting in favorable growth conditions for microorganisms, such as aerobic bacteria, coliforms, and E. coli. In addition, fish ponds can accumulate organic matter that comes from animal feed, fecal matter, decaying plankton, and external debris [41]. These sources of nutrients increase the probability that microorganisms survive and grow for long periods in sediment or water. Total coliforms and generic E. coli are commonly used as indicators for microbial water quality and fecal contamination [42]. Although these indicators are not used as a direct measure of pathogens, fecal matter may contain foodborne pathogens. Catfish ponds are in an open environment that is exposed to fecal contamination from animals.
At the processing plant, the prevalence of Listeria spp. and Listeria monocytogenes was low on fish skin and high in fillets after chilling (Table 2). In the present study, a discriminatory typing method, such as whole genome sequencing (WGS), was not performed on the Listeria isolates in order to determine a definitive source. However, other authors have found that live fish coming from farm ponds are not a major source of Listeria contamination in the processing environment and that the contamination occurs mainly at the processing plant [12,43,44].
Fish skin and fillets, both before and after chilling, had APC and generic E. coli levels below the recommended microbiological limits for fresh fish (APC < 5.7 log CFU/g for good quality products and <7.0 log CFU/g for marginally acceptable products, E. coli ≤ 1 log CFU/g for good quality products and 2.7 log CFU/g for marginally acceptable products) [9] ( Table 2). Removal of heat during chilling from live fish flesh in a relatively short time plus rapid handling during filleting and chilling help maintain the highest possible quality through slowing microbial growth. In addition, APC levels above 7 log CFU/g are commonly reached after 6-7 days of storage [45,46]. Since the catfish were collected and analyzed the same day of processing, the microbial loads were low.
Process chiller water had high APC and TCC counts. These TCC levels are above the EPA Maximum Contaminant Level Goal (MCLG) for drinking water (MCLG = zero mg/L) [47]. Chiller water also had high prevalence of Listeria spp. and L. monocytogenes ( Table 2). The chiller water is recycled and used all day, and this could have caused cross-contamination and the spread of bacteria on some surfaces (Table 3). Moreover, levels of TCC for AC fillets (1.8 log CFU/g) might be associated with cross-contamination from process chiller water since TCC was greater than it was in clean chiller water. Chen et al. [12] identified L. monocytogenes in 55.6% (5/9) of chiller water samples, which suggests that this water might be a vehicle for cross-contamination in fresh fillets. Processing live fish to fillets or chilling of fillets was associated with possible contamination.
Listeria spp., including L. monocytogenes, did not totally disappear on environmental surfaces after sanitation, showing the ability to survive on surfaces that received inadequate/incomplete sanitation. These results are similar to those reported by previous authors [30,36,48]. There were 6 out of 10 positive sites with the presence of Listeria spp. after sanitation (tray in freezer with fish/gray lug, fish holding tray for graded fillet, tote with whole fish, belt before injection, ice container/ice pipe, and waste belt) that are made of some type of plastic material (Tables 3 and 4). Although an attachment strength study for Listeria spp. was not performed during this study, different authors have reported that Listeria cells can attach better to rubber and plastic than to glass and stainless steel [49,50]. For that reason, it is important to have an effective removal of organic waste and cleaning prior to the application of sanitizers on those sites. In addition, biofilms are very different to remove and can harbor L. monocytogenes, making it more difficult to remove this pathogen from processing plants.
Listeria spp. and L. monocytogenes can survive in the tray in the freezer with fish and in the ice container/ice pipe (Tables 3 and 4). These bacteria can be isolated from several areas in food processing plants due to their ability to adhere to any contact surfaces at different levels of bacterial attachment. In addition, several authors have considered that non-food contact surfaces are potential sources of contamination with Listeria spp. and L. monocytogenes in the processing environment and on final food products [12,23]. The drain, floor, and waste belt are points where high loadings of organic waste and water converge from all areas in the food plants.
Seven environmental surfaces were prevalent for Listeria spp. after cleaning and sanitation, including the skinner, fillet chiller, fish holding tray with graded fillets, tote with whole fish, waste belt, floor, and drain. Listeria prevalence was accompanied with APC levels above the microbiological limits (APC < 2.4 log CFU/100 cm 2 ) [51,52]. The high levels of APC on some surfaces might suggest that there was an incomplete removal (inadequate cleaning) of organic matter that was accumulated during processing.
The highest counts of generic E. coli were on the fish holding tray and injector tank (Table 3), although they were within the recommended microbiological limits for environmental surfaces (2 log CFU/100 cm 2 ) [53]. Montville, Chen, and Schaffner [54] indicated that during food handling, fecal coliforms can be transferred from contaminated hands to food products and subsequently to other surfaces. Reij and Den Aantrekker [55] attributed the incidence of E. coli to the lack of hygiene, specifically improper handwashing. Siberio-Pérez [6] reported1.3 and 1.4 log CFU/100 cm 2 generic E. coli on food contact and non-food contact surfaces in catfish plants during processing. This is greater than the counts in this study. E. coli was not detected on 7 of the 10 sample sites where Listeria spp. was found after cleaning and sanitation. In this study, E. coli absence does not indicate the absence of Listeria. These results agree with some previous findings, since E. coli is an indicator of possible fecal contamination [10], and Listeria spp. presence may or may not derive from fecal source [15][16][17][18].

Conclusions
The data obtained in this investigation indicated that the prevalence of Listeria spp. at the farm is low. Listeria spp., including L. monocytogenes, was not detected in these samples and was low on live fish (2%). APC counts were >4.5 log CFU/g for water and sediment samples and TCC counts >2.0 log CFU/g for sediment. At the processing plant, the prevalence of Listeria spp. and levels of indicator microorganisms were higher during processing than after cleaning and sanitation. Once the processing plants were sanitized, there was a lower prevalence of these microorganisms, providing a general idea of how effective and important hygiene and disinfection procedures are at the plant. It is important to state that areas in the processing environment that were positive for non-pathogenic Listeria spp. are also considered critical points because they could easily serve as a reservoir for L. monocytogenes. In addition, chiller water could be a vehicle of cross-contamination during processing since it contained a higher prevalence of Listeria spp., APC, and TCC when compared to clean chiller water. For this reason, it is important to take control measures, such as more frequent water changes, maintaining a low temperature, using antimicrobial treatments, and/or having an adequate water:fish ratio.