Overlap of Spoilage-Associated Microbiota between Meat and the Meat Processing Environment in Small-Scale and Large-Scale Retail Distributions

ABSTRACT

Microbial contamination in food processing plants can play a fundamental role in food quality and safety. The aims of this study were to learn more about the possible influence of the meat processing environment on initial fresh meat contamination and to investigate the differences between small-scale retail distribution (SD) and large-scale retail distribution (LD) facilities. Samples were collected from butcheries (n = 20), including LD (n = 10) and SD (n = 10) facilities, over two sampling campaigns. Samples included fresh beef and pork cuts and swab samples from the knife, the chopping board, and the butcher's hand. The microbiota of both meat samples and environmental swabs were very complex, including more than 800 operational taxonomic units (OTUs) collapsed at the species level. The 16S rRNA sequencing analysis showed that core microbiota were shared by 80% of the samples and included Pseudomonas spp., Streptococcus spp., Brochothrix spp., Psychrobacter spp., and Acinetobacter spp. Hierarchical clustering of the samples based on the microbiota showed a certain separation between meat and environmental samples, with higher levels of Proteobacteria in meat. In particular, levels of Pseudomonas and several Enterobacteriaceae members were significantly higher in meat samples, while Brochothrix, Staphylococcus, lactic acid bacteria, and Psychrobacter prevailed in environmental swab samples. Consistent clustering was also observed when metabolic activities were considered by predictive metagenomic analysis of the samples. An increase in carbohydrate metabolism was predicted for the environmental swabs and was consistently linked to Firmicutes, while increases in pathways related to amino acid and lipid metabolism were predicted for the meat samples and were positively correlated with Proteobacteria. Our results highlighted the importance of the processing environment in contributing to the initial microbial levels of meat and clearly showed that the type of retail facility (LD or SD) did not apparently affect the contamination.

IMPORTANCE The study provides an in-depth description of the microbiota of meat and meat processing environments. It highlights the importance of the environment as a contamination source of spoilage bacteria, and it shows that the size of the retail facility does not affect the level and type of contamination.

INTRODUCTION

Meat is a complex niche with chemical and physical properties that allow the colonization and development of a variety of microorganisms, especially bacteria (1, 2). Several factors can influence the occurrence of microbes in meat. After slaughtering, meat can be contaminated by microorganisms from water, air, and soil, as well as from the workers and equipment involved in the processing. In the later processing steps of the fresh meat chain (i.e., handling, cutting, and storage), abiotic factors such as temperature, gaseous atmosphere, pH, and NaCl levels select for certain bacteria, allowing colonization of the meat surface by different spoilage-related species and strains (3, 4).

Microbial growth to large numbers is a prerequisite for meat spoilage that can be considered an ecological phenomenon, encompassing the changes of the available substrata during the proliferation of bacteria (5, 6). Spoilage is the process of food deterioration leading to a reduction in its quality, to the point of not being edible for humans. Signs of spoilage may include a different appearance of the food, compared to its fresh form, and alterations in the sensorial qualities of the product, particularly the aspect (including texture and color) and the presence of an off-odor (6–9). The presence of microorganisms on the surface of cut meat and meat products determines meat spoilage upon their interaction and growth under optimal conditions (2, 8).

Although there are many different types of meat, the main bacterial populations involved in spoilage are common. The most abundant bacteria causing spoilage of refrigerated beef and pork are Brochothrix thermosphacta, Carnobacterium spp., clostridia, Enterobacteriaceae, Lactobacillus spp., Leuconostoc spp., Pseudomonas spp., and Weissella spp., and their metabolic activity can lead to defects such as sour flavors, discoloration, gas or slime production, and decreases in pH (2, 6, 10, 11).

The environmental microbiota from processing plants have often been addressed as sources of microbes that potentially affect the quality attributes of meat (1, 12, 13). Indeed, several studies demonstrated that the microbiota involved in food-processing steps are often found on processing surfaces or tools (11, 12, 14–17), underlying the importance of hygienic practices in influencing the food microbiota. However, no studies have investigated the differences in the contamination types and levels between small-scale retail distribution (SD) and large-scale retail distribution (LD) facilities. Food handling and cleaning practices can be completely different according to the size, level of automation, and organization of specific retail facilities.

In meat handling environments, the presence of resident microbiota, possibly contributing to the occurrence of spoilage (3), can lead to economic losses (7, 8) and/or safety issues (12, 18, 19). Various microbial contamination sources can be identified in a butchery, including chopping boards, refrigerators, operators' hands, cloths, and knives and other tools (1, 12). The availability of organic residues on surfaces can lead to the growth and aggregation of microorganisms and represents a significant source of cross-contamination (16, 20–22). Good cleaning and sanitization practices for surfaces and equipment are thought to solve the problem of food contamination, since low hygiene standards in food processing plants are the major cause of contamination of raw meat and meat products (12). The most abundant species present on processing tools were also found at high levels on meat, suggesting the establishment of an equilibrium between food and the environment that affects the quality of the final product (1, 12, 23). However, the effects of retail size and organization have never been investigated as possible variables affecting the microbiological quality of meat. In this study, we describe the microbiota in environmental swabs and meat samples collected in small-scale and large-scale retail distribution facilities, in order to explore the influence of the microbiota in meat handling environments on the initial microbiological quality of meat and to assess the effect of the type of retail facility on the extent of microbial contamination.

MATERIALS AND METHODS

Sampling.

Samples were collected from 20 butcheries, including 10 small-scale retail distribution (SD) facilities and 10 butcher counters in large-scale retail distribution (LD) facilities, located in the Campania region (southern Italy), all operating under a certified food safety management system (i.e., hazard analysis and critical control points [HACCP]). Sample collection was replicated twice, with a 3-week interval. The sampling of the surfaces took place at least 1 h after routine cleaning and before the start of sales. Meat samples collected included fresh beef (n = 40) and pork (n = 40) cuts, while surface samples were taken from the knife (n = 40), the chopping board (n = 40), and the operator's hand (n = 40). A description of the samples analyzed in this study is presented in Table S1 in the supplemental material. The surface sampling was carried out using sterile sponges (Whirl-Pak Speci-Sponge; Nasco, Fort Ankinson, WI, USA) premoistened with 25 ml sterile peptone buffer solution. Sponges were rubbed vertically, horizontally, and diagonally across the meat chopping board surface (100 cm²), both sides of the knife, and the palm of the butcher's hand. After collection, samples were cooled at 4°C and analyzed within 3 h. All samples were collected with the permission of the butchers. No animals were involved in the present study, only animal products.

Microbiological analysis.

Prior to analysis, 25 g of each meat sample was homogenized in 225 ml sterile quarter-strength Ringer's solution (Oxoid, Basingstoke, United Kingdom), in a stomacher (Stomacher400 circulator; Seward Medical, London, United Kingdom), for 1 min at 230 rpm at room temperature. The homogenized meat and surface samples were used to perform 10-fold serial dilutions, using sterile Ringer's solution as the diluent. Pour plating was used to determine total psychrotrophic counts, numbers of lactic acid bacteria, and numbers of Enterobacteriaceae, by using plate count agar, de Man-Rogosa-Sharpe (MRS) agar, and violet red bile glucose agar (VRBGA), respectively (all from Oxoid). Spread plating was used to determine numbers of Pseudomonas spp. and Brochothrix thermosphacta, by using a Pseudomonas agar base with a cetrimide-fucidin-cephalosporin (CFC) supplement and streptomycin sulfate-thallium acetate-actidione (STAA) agar with STAA selective supplement SR0151E, respectively (all from Oxoid). All the media were incubated at 20°C for 48 h. Plate counts were determined in triplicate. Data were analyzed using analysis of variance (ANOVA) and the Tukey post hoc test, using a significance level of 0.05 for sample comparisons. The statistical analysis was performed using IBM SPSS Statistics software (version 16.0).

DNA extraction.

Total DNA extraction from sponges and meat samples was carried out by using a Biostic bacteremia DNA isolation kit (Mo Bio Laboratories, Inc., Carlsbad, CA). The extraction protocol was applied to the pellet (12,000 × g) obtained from a 10-fold dilution in sterile Ringer's solution for meat samples and from 20 ml of sponge buffer for swabs.

PCR amplifications, 16S gene amplicon library preparation, and sequencing.

The bacterial diversity was studied by pyrosequencing of the amplified V1 to V3 region of the 16S rRNA gene, amplifying a fragment of 520 bp (24); 454 adaptors were included in the forward primer, followed by a 10-bp sample-specific multiplex identifier (MID). PCR conditions were as described previously (1). After agarose gel electrophoresis, PCR products were purified twice with an Agencourt AMPure kit (Beckman Coulter, Milan, Italy) and quantified using a PlateReader AF2200 (Eppendorf, Milan, Italy), and equimolar pools were obtained prior to further processing. The amplicon pools were used for pyrosequencing on a GS Junior platform (454 Life Sciences, Roche), according to the manufacturer's instructions, using titanium chemistry. The same DNA templates were also PCR screened for the presence of Toxoplasma gondii by using the 18S rRNA gene as the target (25); the test results were negative for all samples.

Bioinformatics and data analysis.

Raw reads were first filtered according to the 454 processing pipeline. Sequences were then analyzed and further filtered by using QIIME 1.8.0 software (26) and a pipeline described previously (27). In order to avoid biases due to different sequencing depths, the operational taxonomic unit (OTU) table was rarefied at the smallest number of reads per sample. Alpha diversity and beta diversity were studied by using QIIME, as described previously (27). Core microbiota were defined as microbial genera/species present in at least 80% of the samples. Statistical analysis and plotting were carried out in the R environment (http://www.r-project.org), by using the packages vegan, stats, psych, corrplot, and made4. Permutational multivariate analysis of variance (MANOVA) (nonparametric MANOVA) based on Jaccard and Bray-Curtis distance matrices was carried out by using 999 permutations to detect significant differences in the overall microbial community composition, as affected by the type of sample or the type of retail facility. Pairwise Wilcox tests were used in order to determine significant differences in alpha diversity parameters, in OTUs, or in predicted pathway abundances between environmental and meat samples. Correction of P values for multiple testing was performed when necessary (28). Principal-component analysis (PCA) was carried out on logarithmically transformed abundance tables by using the dudi.pca function in the vegan package. Venn diagrams were obtained by using the Bioinformatics and Evolutionary Genomics software (29), in order to describe the microbial community shared by different sets of samples.

PICRUSt (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) (http://picrust.github.io/picrust) was used to predict the potential functional profiles of the microbial communities in environmental swabs and meat samples. For this analysis, OTUs were picked against the Greengenes database (version 13_5) using QIIME 1.8. The abundances of the predicted metagenomes were normalized with respect to 16S rRNA gene copy numbers. KEGG orthologs were identified from the inferred metagenomes and collapsed at hierarchy level 3. Subsequent analyses were carried out in R as described above.

Nucleotide sequence accession number.

The 16S rRNA gene sequences are available at the Sequence Read Archive (SRA) of the National Center for Biotechnology Information (NCBI), under accession number SRP072347.

RESULTS

Enumeration of bacterial populations.

The viable counts on appropriate media of the target meat spoilage groups in meat and environmental samples are reported in Tables S2 to S6 in the supplemental material. Mean log counts were not significantly different (P > 0.05) between beef and pork samples (e.g., lactic acid bacteria loads were 3.99 ± 0.92 and 3.96 ± 0.98 log CFU/g and Pseudomonas counts were 4.68 ± 1.23 and 4.72 ± 1.30 log CFU/g for beef and pork cuts, respectively). The mean log counts for hand and knife samples were significantly lower than the chopping board results (P < 0.05), while knife and hand results did not differ significantly (P > 0.05) (e.g., PCA counts were 1.58 ± 1.21 and 1.40 ± 1.03 log CFU/cm² and Enterobacteriaceae loads were 0.24 ± 0.86 and 0.41 ± 1.07 log CFU/cm² for hand and knife samples, respectively). With grouping of the samples into small-scale retail distribution (SD) and large-scale retail distribution (LD) groups, the effect of butchery type was also not significant (see Table S7 in the supplemental material).

Sequencing data analysis and alpha and beta diversity.

A total of 658,572 reads passed the filters applied through the QIIME split_library.py script, with an average length of 454 bp. The diversity indices varied among the samples, and there was a significant association between the sample type and the microbial diversity (Fig. 1). Interestingly, the chopping board samples showed significantly greater diversity, compared to the other surface swabs (false discovery rate [FDR], <0.05), with an average number of 581 ± 303 OTUs and an average Chao1 index of 1,371 ± 700. No difference was found between small-scale retail distribution facilities and butcher counters in large-scale retail distribution facilities (FDR, >0.05). The principal-coordinate analysis based on a weighted UniFrac distance matrix showed that samples from the two samplings did not cluster separately (see Fig. S1 in the supplemental material) and the microbial composition did not differ significantly between the two samplings (P < 0.001).

FIG 1.

Box plots showing numbers of observed OTUs (A and B) and Chao1 diversity index values (C and D) for environmental swabs (red) and meat samples (blue) from SD (A and C) and LD (B and D) establishments. Boxes, interquartile ranges (between the first and third quartiles); lines inside boxes, medians (second quartiles); whiskers, lowest and highest values within 1.5 times the interquartile range (from the first and third quartiles, respectively); circles, outliers beyond the whiskers. *, significant difference obtained with the pairwise Wilcox test (FDR, <0.05).

Bacterial diversity in meat and processing environments.

The microbial diversity at the species level in small-scale and large-scale retail distributions is shown in Fig. 2, where the average values for the two samplings are reported. Streptococcus spp., Pseudomonas spp., Brochothrix spp., Psychrobacter spp., and Acinetobacter were part of the core microbiota, inasmuch as they were abundant in both types of butcheries and occurred in the 99% of the samples, although with different distributions. The highest levels of Pseudomonas were observed in the SD environment (an average of 84% for butcher S) (Fig. 2A) and in meat samples from LD facilities (an average of 60% for all meat samples) (Fig. 2C). Brochothrix occurred in all of the samples (average of 20%) but showed a remarkable occurrence in the hand samples. Psychrobacter showed a homogeneous distribution among all of the samples, with remarkable relative abundances in the environmental samples from both SD and LD facilities (averages of 35% and 40%, respectively) and the greatest abundance in pork meat from LD facilities. Finally, some OTUs were characteristic of specific SD samples, although with low abundance, such as the case of Acinetobacter in the beef samples from retail facilities E, F, S, and T (average of 8%) and Leuconostoc in the hand samples from butcheries O and N (average of 4%).

FIG 2.

Abundance of bacterial species in meat (A and C) and environmental (B and D) samples from SD (A and B) and LD (C and D) facilities. Only OTUs showing relative abundances of ≥2% and occurring in >5 samples are reported. Other, all OTUs that failed to reach the cutoff value. Samples are coded as follows: 1, knife; 2, chopping board; 3, hand; 4, pork; 5, beef. Capital letters, different butcheries, as reported in Table S1 in the supplemental material.

In Fig. 3, the genera shared among the samples are represented. With meat and environmental samples grouped separately for LD and SD facilities, 31 genera were common to all samples (Fig. 3A); these genera included Streptococcus, Brochothrix, Pseudomonas, Acinetobacter, and Psychrobacter spp., which were also the most abundant in the core microbiota (average relative abundances of >10%) (see Table S8 in the supplemental material). Forty-eight genera were shared by meat and environmental samples for both type of retail facilities (Fig. 3B).

FIG 3.

Venn diagrams showing the numbers of shared genera between groups of samples, as determined by 16S rRNA gene pyrosequencing analysis. Samples were grouped as meat versus environmental samples for the LD and SD groups (A) and for the LD and SD groups combined (B).

Permutational MANOVA based on both Bray-Curtis and Jaccard distance matrices showed a significant difference in the overall microbiota between swabs and meat samples (P < 0.001). In contrast, no effect of the type of retail establishment was observed (P > 0.05) (Fig. 1). The hierarchical clustering in Fig. 4 shows a certain degree of separation between meat and environmental samples, mostly driven by the abundance of OTUs within the Proteobacteria phylum, which was significantly higher in meat samples than in swab samples (FDR, <0.05). Pseudomonas and several Enterobacteriaceae members were significantly more abundant in meat samples, while Staphylococcus, Streptococcus, Lactococcus lactis, Leuconostoc, Brochothrix, and Psychrobacter showed higher levels in environmental samples (FDR, <0.05). Accordingly, a principal-component analysis based on the composition of the microbiota showed no clustering of the samples according to the retail type (Fig. 1; also see Fig. S2 in the supplemental material) and, even when SD or LD samples were analyzed separately, the clustering was consistently driven by the sample type (see Fig. S2 in the supplemental material).

FIG 4.

Hierarchical average linkage clustering of the samples based on the Pearson's correlation coefficient for the abundance of genera present in ≥20% of the samples. The color scale indicates the scaled abundance of each variable, denoted as the Z-score; red, high abundance; blue, low abundance. Column bars are colored according to the type of sample (meat or environmental swab) and the type of retail facility (SD or LD), and the row bar is colored according to the classification at the phylum level. Samples are coded as follows: 1, knife; 2, chopping board; 3, hand; 4, pork; 5, beef. Capital letters, different butcheries, as reported in Table S1 in the supplemental material.

The OTU cooccurrence was investigated by considering the genus-level taxonomic assignment and including OTUs with at least 0.1% relative abundance in at least 50% of the samples. Significant correlations (FDR, <0.05) are plotted in Fig. S3 in the supplemental material. Basfia showed strong positive correlations with Bordetella and Streptococcus. Gammaproteobacteria cooccurred with OTU core members such as Acinetobacter and Moraxellaceae, while Lactococcus showed weak cooccurrence with Lactobacillus.

Predicted metabolic activities.

Potential metabolic activities of the samples were predicted by using PICRUSt software. A consistent grouping of the samples on the basis of the sample type (meat versus environment) was achieved also when the predicted microbial pathways were considered (Fig. 5). Pathways related to carbohydrate metabolism were increased in environmental swabs, while amino acid metabolism and lipid metabolism were more abundant in meat (FDR, <0.05). In particular, arginine, proline, and aromatic amino acid metabolism, as well as fatty acid metabolism, demonstrated higher levels in meat (FDR, <0.05). Spearman's correlations between predicted pathways and OTUs are presented in Fig. 6, in which only Proteobacteria and Firmicutes phyla are shown. Proteobacteria OTUs, particularly for Pseudomonas, several Enterobacteriaceae members, and Psychrobacter, were positively correlated with lipid metabolism and amino acid metabolism, while Firmicutes members, such as Brochothrix and lactic acid bacteria, cooccurred with carbohydrate-related pathways (FDR, <0.05).

FIG 5.

Hierarchical average linkage clustering of the samples based on the Pearson's correlation coefficient for the abundance of predicted KEGG orthologs collapsed at hierarchy level 3, filtered for sample prevalence of ≥20%. The color scale indicates the scaled abundance of each variable, denoted as the Z-score; red, high abundance; blue, low abundance. Column bars are colored according to the type of sample (meat or environmental swab) and the type of retail facility (SD or LD), and the row bar is colored according to the higher hierarchy level in the KEGG classification. Only KEGG orthologs related to carbohydrate, amino acid, or lipid metabolism are reported. Samples are coded as follows: 1, knife; 2, chopping board; 3, hand; 4, pork; 5, beef. Capital letters, different butcheries, as reported in Table S1 in the supplemental material.

FIG 6.

Heatplot showing the correlations between Firmicutes and Proteobacteria members and predicted KEGG orthologs collapsed at hierarchy level 3, both filtered for sample prevalence of ≥20%. Rows and columns are clustered by Euclidean distance and Ward linkage hierarchical clustering. The intensity of the colors represents the degree of association between the OTUs and the KEGG orthologs, as measured by Spearman's correlations. The row bar is colored according to the OTU classification at the phylum level, and the column bar is colored according to the higher hierarchy level in the KEGG classification. Only KEGG orthologs related to carbohydrate, amino acid, or lipid metabolism are reported.

DISCUSSION

In this study, the microbiota in 20 butcheries were studied by rRNA gene-based culture-independent high-throughput sequencing, in order to identify the relationships between the microbial diversity of processing environments and meat samples and to compare the microbiota occurring in small-scale retail distribution (SD) facilities and butcher counters in large-scale retail distribution (LD) facilities. Results showed no significant effect of the butchery type on bacterial counts, in agreement with previous reports (30, 31). The microbiota of the environments were complex, including more than 500 taxa at the genus/species level, as were those of fresh meat cuts and environmental samples. Meat microbial complexity usually decreases sharply after storage, as a consequence of the effects of abiotic factors such as storage temperature and the type of packaging used, which select a few species to become dominant and to spoil the meat (1, 32–34). Significantly greater microbial diversity and higher viable counts were observed for the chopping board samples than for the knife samples. These differences suggest that surface contamination is strongly affected by the surface material, which represents an important factor to take into account in order to maintain acceptable levels of hygiene in food processing plants (1, 12, 17, 35). Tools made of porous materials, such as wooden chopping boards, are less adequate for thorough cleaning and increase the possibilities of adherence of bacteria and establishment of resident microbiota (16, 35, 36). Moreover, ecological factors (pH, water activity, redox potential, nutrient availability, and matrix composition), the capabilities of microbes to develop biofilms and to adhere to surfaces, cleaning procedures, and staff hygiene training all have important effects on the microbiological quality of fresh meat (2, 37, 38). The microbial community composition across surfaces in meat processing plants is reported to be highly variable, and most of the OTUs identified in meat samples (raw, spoiled, and processed) originate from the processing environment (1, 12, 20). In the present study, Pseudomonas spp., Brochothrix spp., Psychrobacter spp., Streptococcus spp., and Acinetobacter spp. were identified as the core microbiota occurring in all of the samples analyzed; they were previously reported as contaminants in food processing environments (39–41). Our results indicate that they are part of a resident microbiome, but we showed that their prevalence is not influenced by the type of retail establishment considered. The microorganisms found in the processing facilities sampled here occur frequently on freshly cut and aerobically stored meat (2, 12), and they are recognized as undesirable bacteria in food processing environments (2) and as main contributors to meat spoilage (5, 7, 9, 12). In particular, Pseudomonas spp. are recognized as able to form biofilms (36, 42–44), adhering to surfaces and improving their resistance to sanitation and cleaning procedures (20, 45, 46).

Predicted metagenomes highlighted the remarkable abundance of amino acid metabolism and lipid metabolism in meat samples and their strong correlation with Proteobacteria, such as Pseudomonas and Enterobacteriaceae OTUs. Pseudomonas fragi was reported previously to be lipolytic and proteolytic, as were species belonging to the Enterobacteriaceae family, and they can contribute to spoilage through the production of volatile organic compounds and other undesirable metabolites, such as biogenic amines (47–52). In contrast, B. thermosphacta and lactic acid bacteria were mainly correlated with carbohydrate metabolism. Accordingly, none of the strains of B. thermosphacta or Carnobacterium maltaromaticum tested previously was found to be proteolytic and lipolytic, while producing off-flavors arising from sugar catabolism (53, 54). Lactic acid bacteria are reported to be producers of polysaccharidic ropy slime (55–57). However, spoilage-related activities must be considered strain specific and may be influenced by abiotic factors such as pH, NaCl concentration, or temperature, as well as interactions with other components of the microbial community (2, 10, 55).

Our results supported the importance of environmental microbiota in influencing the quality and safety of meat, and they highlight the lack of differences in the distributions of microbiota in small-scale versus large-scale meat processing environments. Meat contamination is strongly dependent on the environment in which the meat is handled and processed. The initial levels of microbial contamination and the community composition influence the potential shelf-life of meat, depending on storage conditions. Therefore, adequate choices of surface materials and extremely accurate cleaning procedures are necessary in order to avoid spreading of bacteria that can contaminate the meat and potentially cause spoilage.