In the United States, an estimated 9 million people fall ill due to food-borne illness each year, resulting in 56,000 hospitalizations and 1,300 deaths (CDC, 2017). From 1998 to 2020, 1,287 outbreaks associated with food have occurred, out of which 960 were caused by Salmonella, 272 due to E.coli O157, and 55 due to Listeria monocytogenes (Interagency Food Safety Analytics Collaboration, 2022). Ensuring food safety is of utmost importance as the incidence of food-borne illnesses continues to rise. Microbial pathogens can thrive in biofilms and lead to foodborne illnesses, making biofilms a concern in food-processing environments (Srey et al., 2013; Dass and Wang, 2022). Biofilms are complex communities of microorganisms that adhere to surfaces and form a protective extracellular matrix, making them resistant to cleaning and disinfection (Van Houdt and Michiels, 2010; Yin et al., 2019). The metabolic potential of microorganisms within biofilms allows them to adapt to nutrient limitations and other environmental stressors, contributing to their persistence (Yin et al., 2019). This leads to the accumulation of pathogenic and spoilage microorganisms that may contaminate food products causing subsequent health and economic issues. Globally, the ability of microorganisms to form biofilms and pathogens to persist within them is a significant challenge for the food industry, underscoring the importance of understanding the underlying mechanisms and developing effective control measures.

To gain a deeper understanding of the complex microbial communities within biofilms in food-processing environments, researchers have increasingly turned to metagenomic approaches such as shotgun metagenomics and whole-genome sequencing (WGS). These methods have become popular because they enable the identification and characterization of the entire microbial community, including microorganisms that are difficult to culture or detect using traditional methods (Yadav et al., 2013). WGS data can predict the presence of genes and their functions within the community and can be used to reconstruct the metabolic pathways present. This entails identifying groups of genes that collaborate to carry out specific biochemical reactions and connecting these groups of genes (Tu et al., 2017). Comparative analysis of WGS data from different biofilm communities can provide insights into the factors impacting their metabolic capabilities. Metagenomic sequencing can potentially improve our understanding of biofilm formation and persistence and aid in the development of control measures. Several recent studies have utilized metagenomic approaches to characterize biofilms in various settings (Ding et al., 2019; Caraballo Guzmán et al., 2020; Rehman et al., 2020; Voglauer et al., 2022; Wang et al., 2022; Zhou et al., 2022; Zhu et al., 2022).

This research aimed to analyze biofilm composition, diversity, and functional potential in three beef-processing plants to further our understanding of the microbiome present in biofilms that can form in food-processing environments. Samples recovered from floor drains located in five distinct areas starting where finished carcasses exit the harvest floor (hot scale), then following the process flow to the hotbox where hot beef carcasses are rapidly chilled under intermittent sprays of cold water (hotbox). Further samples were collected from coolers where chilled carcasses are held and sorted (cooler), before entering fabrication or processing (processing). With the last location sampled being the grind room where trimmings left over from processing are made into ground beef (grind room). Two time points separated by 3 years were examined. By analyzing the microbial communities present in these samples, we gained insights into the factors that contributed to biofilm formation and persistence, which can inform the development of effective intervention strategies to control biofilm formation in meat-processing environments.

Biofilms are a major challenge to food-processing environments because they may harbor food spoilage organisms and foodborne pathogens. Multi-species biofilms, which are commonly found in food-processing environments, may create an ecological niche for pathogens for better survival by conferring resistance to sanitization (Donlan, 2001; Dass and Wang, 2022). Therefore, understanding the microbial diversity present in these multi-species biofilms and assessing their functional potential can provide valuable information for developing effective biofilm control strategies. Biofilm investigation studies have been conducted in the past, but are rare and limited to only specific environments (Voglauer et al., 2022). Metagenomic characterization of biofilms have already been documented in studies in a variety of settings, such as food-processing plants (Caraballo Guzmán et al., 2020; Wang et al., 2022), water-hoses (Voglauer et al., 2022), heritage stone surfaces (Skipper et al., 2022), bioreactors (Rehman et al., 2020; Zhou et al., 2022; Zhu et al., 2022) and in marine environment (Ding et al., 2019). However, there is a paucity of data regarding the biofilms in beef-processing plants.

This study analyzed drain samples from five different locations in three beef-processing plants at two different time points to determine their metagenomic profiles and microbial diversity. Microbial diversity profiles were determined based on both alpha and beta diversity analysis. Shannon index (Alpha diversity) accounts for both the species richness and evenness of a sample. Shannon index values ranged from 2.08 to 4.62 across all the samples and there were no radical differences in alpha diversity among both the plants and locations (p>0.05) for the two time points T1 and T2. Further, beta diversity analysis was based on Unweighted Unifrac metric. This metric accounts for the incidence of the micro-organism excluding the abundance of the same and detecting the rare lineages (Chen et al., 2012). The PCoA plots of beta diversity did not show any distinct group clustering, which was supported by the PERMANOVA analysis, which did not yield statistically significant results among groups (p>0.05). As a result, this indicates that similar microbiomes may exist in all three beef-processing plants, regardless of how far apart they are in space and time. However, the abundance of bacteria varied, with Pseudomonas, Psychrobacter, and Acinetobacter being common in all three plants but with little variation in their abundance. Based on the core microbiome analysis, these three genera were found to be the core microbiome of beef-processing plants.

The most ubiquitous genus was Pseudomonas which belongs to the phylum Pseudomonadota, a Gram-negative, chemoorganotrophic, rod-shaped bacteria (Iglewski, 1996). It was prevalent across all locations with a prevalence of ≥85% except for hotscale (57%). Pseudomonas is a versatile genus with 191 described species. This study found over 137 different Pseudomonas species in all samples. It has simple nutritional requirements and can use a wide range of organic compounds as carbon and energy sources and colonize a wide range of niches (Madigan et al., 2011). The study also found three dominant Pseudomonas species, including P. fluorescens, P. psychrophila, and an unclassified Pseudomonas sp. LG1D9. ( Figure 1E ). P.fluorescens is an obligate aerobe that can use nitrate instead of oxygen as a final electron acceptor during cellular respiration (Muriel et al., 2015). It has been found to cause food-spoilage, and some strains have the plasticity to grow in higher temperatures. The species is also known to form biofilms in various environments (O’Toole and Kolter, 1998; Bianciotto et al., 2001; Baum et al., 2009) and has been found in meat and dairy-processing environments (Stellato et al., 2017). Another Pseudomonas species found in the samples was P. psychrophila. These are psychrophilic bacteria that can grow in temperatures ranging from -20°C to 20°C (Abraham et al., 2020); it was present in all the locations, with higher abundance in the hotbox, where sample collection temperatures had ranged from 2.9°C to 11.4°C, thus supporting the psychrophilic characteristic of the species. One study has looked into the biofilm-forming ability of P. psychrophila and discovered that the synthesis of the quorum sensing signaling molecule C4-HSL doubled the biofilm-forming ability of the bacterium by increasing its attachment (Bai and Rai Vittal, 2014). In our study, functional annotation showed protein hits for acyl homoserine lactone synthase (AHL), which directs the synthesis of C4-HSL, which in turn aids in swarming motility in the quorum sensing pathway.

Psychrobacter is a gram-negative, psychrotrophic, osmotolerant and aerobic bacteria belonging to the family Moraxellaceae. In this study, it was found to be the second most abundant genus in all plants and locations except the hotscale in T1. However, there was an increase in its abundance in the hotscale in T2. Species-level identification showed the presence of an unclassified Psychrobacter sp. P11F6 that was abundant in the cooler of plant A, along with 12 other species that were not dominant in the samples. Psychrobacter can grow on both complex and simple media, with ammonium as a nitrogen source. It has a wide ecological distribution, from animal bodies to non-host environments, and is mostly associated with colder and saline domains. (Welter et al., 2021). The high abundance of Psychrobacter in the cooler might be related to the temperature range of the local sample collection site in the coolers of the three plants, which were between 2.9°C and 13.3°C at the time of collection. Some species of Psychrobacter are capable of movement by twitching motility (García-López et al., 2014). Unlike Pseudomonas, Psychrobacter does not readily form biofilms, and there is little evidence to support this. However, one species (P.arcticus) was found to be capable of forming biofilms with the help of an adhesin protein in minimal medium under laboratory conditions, when supplied with acetate as the sole carbon source with 1-7% sea salt at temperatures ranging from 4°- 22°C (Hinsa-Leasure et al., 2013). P. arcticus was found in low abundance in majority of the samples in this study, with the highest abundance recorded on a hotscale in plant A in T2. Some species of Psychrobacter produce anti-biofilm agents that disrupt the biofilm formation of other bacterial species. For example, Psychrobacter sp. TAE2020 produces CATASAN, a protein-polysaccharide complex that inhibits the ability of Staphylococcus epidermidis to form biofilms (D’Angelo et al., 2022). Similarly, another Psychrobacter species isolated from a marine sediment produced AHL-lactonase, which degrades the quorum sensing signaling molecule N-acylhomoserine lactones and thus controlled biofilm formation by Pseudomonas (Packiavathy et al., 2021).

Acinetobacter, a Gram-negative genus of aerobic, non-fermentative coccobacilli belonging to the Gammaproteobacteria class, was the third most prevalent genus found in the drains of the three beef-processing plants. A total of 33 different Acinetobacter species were found in all samples, with A. johnsonii, A. lwoffii, and an unclassified Acinetobacter sp. TTH00_4 being the most dominant. A.baumannii was abundant in the cooler of plant B in T1 but decreased in T2, and was also found in hotscale and hotbox. Some Acinetobacter species, primarily A.baumannii and A. calcoaceticus, have been found to form biofilms, and produce biofilm associated protein (BAP) that aids in the formation of biofilms and cell adhesion (De Gregorio et al., 2015). In one study, A. calcoaceticus improved the biofilm formation ability of Shiga toxin producing E.coli (STEC) in a meat-processing environment by increasing the surface colonization of the latter (Habimana et al., 2010).

Other genera found in various plants and locations include Moraxella, Yersinia, and Stenotrophomonas. Moraxella osloensis, a Gram-negative bacterium, was highly abundant in the hotbox of plant C in T2, while Stenotrophomonas was particularly abundant in the hotscale in T2. A biofilm formation study in drinking water distribution system pipes exposed to long-term high chlorine levels identified M.osloensis and Stenotrophomonas sp. biofilm community structure in all pipe materials (Zhu et al., 2014). Yersinia was slightly more abundant in the hotbox in T1 and processing in T2. These bacteria may participate in biofilm formation and coexist in the beef-processing environment through synergistic interactions with one another.

Thermus parvatiensis, found through metagenome binning, belonged to the phylum Deinococcota. The phylum Deinococcota was particularly abundant in plant A in T2. It can withstand temperatures ranging from 41° to 122°C and was previously isolated from a hot water spring located atop the Himalayan ranges in India (Dwivedi et al., 2015). Interestingly, in this study, it was found to be present in coolers with temperatures ranging from 2.9° to 13.3°C. Another microorganism identified through binning was Hydrogenophilus thermoluteolus, which was also abundant in the cooler of plant A. H. thermoluteolus was also moderately thermophilic, but evidence suggests that it was isolated from colder environments. DNA from H.thermoluteolus has been discovered in accretion ice in Antarctica’s subglacial Lake Vostok (Lavire et al., 2006).

Biofilm formation is a complex process involving the synthesis and secretion of a wide range of proteins by the microorganisms that comprise the biofilm. These microorganisms, including bacteria, fungi, and others, produce extracellular polymeric substances (EPS) containing polysaccharides, proteins, lipids, and nucleic acids, which are crucial for the structural integrity and adhesion of the biofilms (Limoli et al., 2015; Costa et al., 2018). Attachment is a crucial step in the colonization of an environment or host by bacteria. Adhesins are commonly found on bacterial pili, which are hair-like projections on the surface of bacterial cells. Adhesins may also be found on bacterial cell walls or membranes and are distributed differently depending on the type of bacteria, mediating the attachment of microorganisms to surfaces or neighboring cells (Kline et al., 2009). This study found that all the de novo assemblies contained numerous adhesins, with plant B having the most in T2. Previous studies have shown that adhesins play an important role in biofilm formation in bacteria (Pratt and Kolter, 1999; Nobile et al., 2006; Absalon et al., 2011).

Bacterial motility is also a crucial step in biofilm formation as it facilitates surface exploration, cell-to-cell communication, nutrient acquisition, and biofilm expansion. Flagella, a complex apparatus made up of 20 different proteins; and pili (also known as fimbriae): proteinaceous, filamentous polymeric organelles, both aid in bacterial motility, particularly type IV pili. One of the most important flagellar proteins is flagellin, a polymeric protein that makes up most of a bacterial flagellum and is located in the hollow cylinder of the flagellum with a helical structure. Motility proteins A and B, encoded by the genes motA and motB, respectively, are integral membrane proteins that form components of the flagellar motor. These proteins harness energy from forming pores and help the mechanical movement of flagella across the membrane through proton motive force (PMF) (Minamino et al., 2011). This study found abundant flagellin proteins across the three plants in T1 and T2, with the most abundant in plant A in T2. Although not as numerous as flagellin, there were fewer gene hits for motA and motB across the three plants in two time points. FleQ, a gene required for flagellar assembly in the cAMP/Vfr signalling pathway, had fewer gene hits in our datasets, with a maximum of 10 hits recorded for plant A in T1 and plant C in T2. Pilins are fibrous proteins that are found in the pilus of bacteria. Type IV pilins are structural modules that help with motility, attachment, electrical conductance, and DNA acquisition (Giltner et al., 2012). This study found several gene hits for pilins, with plant B in T2 having the most hits. The presence of these proteins suggests that bacteria are actively moving around the biofilm architecture, acquiring nutrients, and communicating with other bacteria.

A complex network of genes and proteins governs polysaccharide biosynthesis in biofilms. Psl genes (PslA) and Pel genes (PelA, PelB, PelC, PelD, PelE, PelF, and PelG) play critical roles in polysaccharide biosynthesis in the c-di-GMP signalling pathway (Colvin et al., 2012). The gene Alg44 (mannuronan synthase) is involved in the biosynthesis of alginate (Franklin et al., 2011). However, these genes were present in relatively low numbers in most of the samples, with the exception of plant A in T2, which had no hits for both the Psl and Pel genes.

Enzymes are also important in forming biofilms, particularly glycosyl transferases (GTs), which are responsible for synthesizing exopolysaccharides (EPS) that make up the biofilm matrix (Wang et al., 2020). These enzymes transfer sugar molecules from activated nucleotide sugars to growing polysaccharide chains, helping to maintain the structural stability of biofilms. Several studies have shown that inactivating or deleting the GTs reduced biofilm formation (Tao et al., 2010; Theilacker et al., 2011; Rainey et al., 2018) emphasizing the importance of these enzymes. In this study, GTs were identified using the CAZy database (Cantarel et al., 2009), and several hits were found for GTs in all the assemblies, with the highest being 736 hits in plant B in T2.

An interesting discovery from the functional analysis is that, despite observing differences in alpha diversity and varying microbial signatures across the datasets, the functional potential remained consistent ( Figure 4 ). This phenomenon can be explained by an ecological property known as functional redundancy (Tian et al., 2020). Numerous studies across different niches, such as the gut microbiome, macroalgae, and plants, have demonstrated that while the taxonomic composition of microbiomes varies significantly, the functional genes remain conserved (Turnbaugh et al., 2008; Burke et al., 2011; Louca et al., 2016). The persistence of functional redundancy, even following changes in the microbiome, may be attributed to the emergent properties of microbial systems resulting from biotic interactions, as well as other environmental and spatial processes (Louca et al., 2018).

Lateral gene transfer, also referred to as horizontal gene transfer, is a significant driver of microbial evolution and niche adaptation (Song et al., 2021). Lateral gene transfer and biofilms are linked processes because biofilms provide plasmid stability, which enhances the transfer of mobile genetic elements (Madsen et al., 2012). Predicting the presence of LGT events in the genome of an organism is a direct analysis, but identifying LGT events in a sequence data from a community is a challenge. The study did not detect any LGT events for plant B in T1, while they were putatively present in the other assemblies. The highest LGT events were observed in plant A in T2, with a functional annotation of the transferrable genes revealing a diversity that included those related to flagellar proteins, transposases, polymerase, kinases, integrases, and ABC transporters. ABC transporters are integral membrane proteins that play a crucial role in transporting substrates across membranes with the aid of ATP (Rees et al., 2009). They are essential in the transition from reversible to irreversible attachment during biofilm formation (Hinsa et al., 2003) and also helps cope with desiccation tolerance. Transposases are enzymes that recognize and bind to the inverse terminal repeats in DNA, thereby cleaving DNA transposons (Hickman and Dyda, 2015). These enzymes facilitate the movement of DNA segments between nonhomologous sites, which plays a critical role in lateral gene transfer (Craig, 2002; Craig et al., 2002). The presence of transposases in biofilms, particularly in high abundance, suggests that the communities are primed to transfer genetic material and adapt to changing conditions. Previous studies have reported 3,735 transposases in biofilm formed in hydrothermal chimneys covering more than 8% of the metagenomic reads (Brazelton and Baross, 2009). Higher abundance of transposases were observed in Zn-associated marine biofilms (Ding et al., 2019), acid mine drainage biofilms (Tyson et al., 2004), and upregulated transposases in Treponema denticola, a bacterium that exists as part of a dense biofilm in subgingival dental plaque accumulated in teeth (Mitchell et al., 2010). This study did not find those levels, with only more than 800 transposases identified in all samples, with the highest number observed in plant A-T2 (n=1573) ( Figure 5 ).

Antimicrobial resistance genes (ARGs) were minimally detected in beef-processing plants, and while this study detected a number of different antimicrobial resistance genes being present in the drains, ( Supplementary Table S5 ). The most notable were quaternary ammonium compound (QAC) resistance genes. QACs are cationic surfactants used in disinfectants and biocide. These compounds are widely used as sanitizers in the food processing industry. The structure of QAC consists of a quaternary ammonium nitrogen linked by four alkyl or aryl groups and an anionic ion like chloride or bromide. One of the alkyl groups is a long hydrophobic chain containing more than eight hydrocarbons, which provides antimicrobial properties to QACs (Sun, 2011). This study revealed the presence of five QAC resistance genes (QacL, QacE, Qac-pB8, QacF, and QacG2). A previous study has shown that exposure to QAC (cetyltrimethylammonium bromide (CTAB)) can enhance the expression of antimicrobial resistance genes in O₂-based membrane biofilm reactors (O₂-MBfRs) (Luo et al., 2021). Therefore, the detection of QAC resistance genes in the drains of beef-processing plants suggests that the microbial communities present in these environments may have the potential to develop resistance to QACs. The prevalence of QAC resistance in biofilms is therefore not surprising.

This study demonstrates the microbial diversity and functional potential of the microbiome in a beef-processing environment. Our findings indicate a high level of microbial alpha diversity across all three beef-processing plants. The study also identified the necessary protein arsenal possessed by microbial assemblages, enabling them to withstand environmental stresses. This study provides a comprehensive snapshot of the microbial profile of biofilms found in the meat-processing environment. To gain a better understanding of the dynamic microbial community in beef-processing environmental biofilms, transcriptomic studies should be conducted in the future, and focus on identifying gene expression profiles in real-time, providing a more comprehensive understanding of the behavior of microbial communities.

The data analyzed in this study is subject to the following licenses/restrictions: The authors declare that the data supporting the findings of this study are presented within the article and the supplementary information files . The fastq.qz files of the Illumina paired-end sequencing and metadata originate from meat processing plants, and authors are obliged to maintain confidentiality, preventing the public deposition of these sequences. However, in case of a reasonable request, the authors are fully capable and willing to make these data available through file sharing. The data was generated under a non-disclosure agreement with meat processors, and we can share data with those willing to be bound by the same non-disclosure agreement. Requests to access these datasets should be directed to the corresponding author. https://github.com/dasslab/ShotgunTAMUData.

VP, JB, and SD contributed to the conception and design of the study. VP, DB and SV performed the metagenomic analysis and VP wrote the first draft of the manuscript. JB and SD revised the manuscript critically for important intellectual content. All authors contributed to the article and approved the submitted version