Fresh fish is a convenient protein ready for preparation of many dishes, but at the same time it is highly perishable. The main cause of deterioration of fresh fish is the metabolic activity of spoilage seafood microorganisms that provoke degradation of organic molecules and fish tissue, loss of essential fatty acids, fat-soluble vitamins and protein functionality, production of biogenic amines, and formation of off-odors (Gram and Dalgaard, 2002).

Natural antimicrobials and non-thermal treatments such as high-hydrostatic pressure (HP), singly or in combination are attractive candidates for preservation of fish and fish products. Previous studies have addressed possible applications of bacteriocins for food preservation, including seafoods (Galvez et al., 2014; Johnson et al., 2017). Enterocin AS-48 is a broad-spectrum circular bacteriocin with a generally-recognized as safe (GRAS) status (Grande Burgos et al., 2014). Immersion in an enterocin AS-48 solution for 1 min delayed bacterial growth and reduced biogenic amine production in sardines during refrigerated storage (Ananou et al., 2014). Spray-application of an enterocin AS-48 solution reduced viable counts of a cocktail of Listeria monocytogenes strains on raw hake and salmon fillets as well as on smoked salmon, an effect that was potentiated by bacteriophage P100 (Baños et al., 2016).

Essential oils and their antimicrobial compounds have been studied for preservation of different types of foods (Hyldgaard et al., 2012; Patel, 2015; Pandey et al., 2017). Addition of thymol in combination with other hurdles improved the preservation of fresh packed plaice fillets (Altieri et al., 2005), sea bream, fresh cod, and fresh blue fish burgers (Corbo et al., 2008, 2009; Del Nobile et al., 2009). The antibacterial activity of enterocin AS-48 can be potentiated by essential oils and phenolic compounds (Grande Burgos et al., 2014). Thymol in combination with enterocin AS-48 improved inactivation of L. monocytogenes in salads (Cobo Molinos et al., 2009). Combinations of enterocin AS-48 and phenolic compounds such as thymol could also offer new possibilities for fish preservation.

Bacteriocins and essential oils or their antimicrobial compounds have been incorporated on different coating materials with the purpose of preserving fish products, as exemplified by nisin (Neetoo et al., 2008; Lu et al., 2010). However, the only use of natural antimicrobial compounds to inhibit or delay bacterial growth in fish products may require addition of too high concentrations having a negative impact on food flavor. The efficacy of natural antimicrobials incorporated on coating materials could be improved in combination with other hurdles such as HP. HP can be applied on seafood products with several purposes such as inactivation of foodborne pathogens, reduction of biogenic amine production, improving the product shelf life, product texturization, and recovery of fish meat (Murchie et al., 2005; Campus, 2010; Wang et al., 2016). However, pressure treatment of fresh fish flesh formed products whose color deteriorated (cooked appearance) with increasing pressure as well as holding time (Campus, 2010; Wang et al., 2016). Decreasing the intensity of HP treatment reduces the impact on the food properties, but it also decreases its effects on inactivation of microorganisms. Application of HP treatments of low intensity in combination with natural antimicrobials would presumably have a greater effect on inactivation of microorganisms, improving the preservation of fish products.

There is also a growing interest on understanding the complex changes in the food microbiota that may occur during product shelf life as influenced by several factors such as storage temperature, atmosphere, and added antimicrobials. High throughput sequencing (HTS) technology is a powerful tool for studying microbial communities in food systems (Ercolini, 2013; Kergourlay et al., 2015). HTS has proven to be useful in exploring bacterial communities in fish intestine (Ghanbari et al., 2015; Standen et al., 2015; Song et al., 2016) and seafood products (Chaillou et al., 2015; Zhang et al., 2017). In spite of the many studies carried out on the synergistic activities of natural antimicrobials and treatments such as HP, there is still limited information on the effect of such treatments on bacterial populations in fish products. The aim of the present study was to determine the impact of selected treatments (natural antimicrobials thymol and enterocin AS-48 applied on an activated plastic film and HP treatment) on the bacterial load and bacterial diversity of sea bream fillets under refrigerated storage.

Results presented in this study indicate that reduction of the initial microbial load in sea bream fillets and retardation of its growth during refrigerated storage can be improved considerably by application of combined treatments such as films activated with a mixture of thymol plus the bacteriocin enterocin AS-48 and a mild high-HP treatment. The combined treatment had the advantage of increasing microbial inactivation at the beginning, and still kept viable cell counts in the packed fillets 1.2 log cycles lower than the untreated controls at day 7 of storage. Compared to other methods for bacteriocin addition such as dipping or spraying with bacteriocin solutions, activated films provide the advantage of a slow release of antimicrobials into the medium while at the same time can also provide some protection against bacteriocin degradation by tissue proteases or by complex formation with food components (Gálvez et al., 2007). Although bacteriocins can show synergistic effects with other natural antimicrobials and with non-thermal food processing treatments such as HP (Galvez et al., 2014; Mathur et al., 2017), there are no previous studies on the combined effect of activated packaging with thymol plus enterocin AS-48 and HP treatment. Inability to repair multiple damages caused by the combined action of different antimicrobials most likely results in energy exhaustion and cell death. Furthermore, damage to the outer membrane of Gram-negative bacteria caused by some antimicrobials and/or treatments results in cell permeabilization and faster diffusion of bacteriocins such as enterocin AS-48 to its main cellular target, the bacterial cytoplasmic membrane (Gálvez et al., 1991).

Considering that foods are complex ecosystems containing a mixture of bacterial species that may change in relative abundance during storage depending of different environmental factors such as storage temperature and the presence of antimicrobial substances, among others, it is important to study the changes in the bacterial diversity of foods after application of food preservation methods. The data obtained by HTS can throw light on the behavior of potentially pathogenic and spoilage bacteria in foods during storage. Nevertheless, the influence of the method used to assess the biodiversity on the results needs to be taken into consideration. For example, differences in copy number of the 16S rRNA gene can cause quantification problems. Furthermore, results will also depend on sample variability. Considering these limitations, the results obtained in the present study by HTS revealed that the main bacterial groups detected in the fresh fillets after being packed in vacuum (Listeria, Acinetobacter, Pseudomonas, Chryseobacterium, and members of Enterobacteriaceae) were rapidly displaced by other bacterial groups during storage (mainly members of Vibrio, Photobacterium, and Shewanella together with other groups such Cloacibacterium, Lactobacillales, and Comamonadaceae appearing during late storage). Previous studies reported that Shewanella putrefaciens was found to be very important for spoilage of packed cod (Jorgensen et al., 1988) as well as Photobacterium phosphoreum, Pseudomonas spp., Achromobacter spp., Acinetobacter spp., Flavobacterium spp., and Aeromonas spp. (Dalgaard et al., 1993; Dalgaard, 1995). It is worth mentioning that OTUs for Listeria detected in the present study were assigned to L. seeligeri and that L. monocytogenes was not detected in the sea bream fillets. While isolates of the species L. seeligeri are typically hemolytic, this species is generally considered non-pathogenic (Orsi and Wiedmann, 2016). Furthermore, OTUs assigned to Listeria had very low relative abundance in all treated samples. L. monocytogenes is very sensitive to enterocin AS-48, but it is also known for its high barotolerance and its wide pressure resistance variation between strains (Bruschi et al., 2017). Compared to L. monocytogenes, there are scarce data on pressure resistance variations between strains of L. seeligeri.

Packing in the activated film induced drastic changes on the sea bream microbiota. All relevant groups in control samples (C0) had reduced relative abundances in samples packed in the activated films (AF0), where OTUs assigned to Comamonadaceae and Methylobacterium (and to a less extent also Acidovorax and Shewanella) became predominant. Members of fam. Comamonadaceae have been described as part of the culturable microbiota of zebra fish (Cantas et al., 2012), and Methylobacterium has been described as a commensal bacterium from the skin of the salmonid brook charr (Salvelinus fontinalis) (Boutin et al., 2014). During storage, the bacterial communities of sea bream fillets packed in the activated films seemed to become quite stable, and the main groups detected (Vibrio, Photobacterium, and Shewanella) were the same as in the untreated controls. These results would be expected since the activated films did not prevent bacterial growth during storage. A possible explanation for the observed low efficacy of activated films during storage could be that added antimicrobials decreased below their minimum inhibitory concentrations upon diffusion from the activated film to the fish tissue, together with perhaps partial degradation of bacteriocin molecules by tissue proteases. The results also suggest that these bacterial groups are less sensitive to the antimicrobials employed, thymol and enterocin AS-48 and/or have a greater capacity to survive in the fish fillets under stress conditions. Since enterocin AS-48 acts on the bacterial cytoplasmic membrane (Gálvez et al., 1991), this bacteriocin has much lower activity against Gram-negative bacteria. Nevertheless, large differences in sensitivity to the bacteriocin have also been reported among Gram-negatives (Gálvez et al., 1989). By contrast, thymol can freely cross the cell wall and therefore inhibit the growth of both Gram-positive and Gram-negative bacteria by interacting with the phospholipid bi-layer of cell membranes, resulting in metabolite dissipation (Hyldgaard et al., 2012). The possible species-dependence of the synergistic action of thymol with enterocin AS-48 deserves to be investigated in further studies.

The HP-treated samples had much lower relative abundances of OTUs assigned to Vibrio and Photobacterium compared with controls (C2, C5, C10) and with activated film packaging samples (AF2 to AF10). This may suggest a higher sensitivity of these bacterial groups to HP treatment. Compared to the single HP treatment, the combined treatment (activated film-HP) decreased the relative abundance at time 0 of Comamonadaceae and Methylobacterium and increased that of Firmicutes (mainly Enterococcus and Lactobacillaceae). However, samples from both treatments had similar bacterial compositions during storage. These results could be explained because the activated film alone only had limited effects on bacterial diversity during storage as discussed above. It is also worth mentioning that the main bacterial group detected during early storage of samples treated by HP (singly or in combination with the activated film) was Acinetobacter. Genus Acinetobacter has been reported as a member of the bacterial community of sea bream (S. aurata) larvae early after hatching (Califano et al., 2017) and also as one of the dominant genera of the intestinal microbiota of Antarctic fish (Song et al., 2016). Acinetobacter was also predominant in salted bighead carp (Aristichthys nobilis) refrigerated fillets (Liu et al., 2017), in farmed sea bream (S. aurata) (Parlapani et al., 2013) and in grass carp (Ctenopharyngodon idellus) packed in sealed polyvinyl chloride bags during the first 6 days of chill storage (Wang et al., 2014). Acinetobacter was also detected in common carp (Cyprinus carpio) fillets packed in air or under vacuum and stored under refrigeration (Zhang et al., 2015). In the present study, most OTUs included in this group were assigned to the environmental species A. guillouiae and to A. johnsonii, which rarely causes human infections (Montaña et al., 2016). Nevertheless, further studies need to be carried out in order to determine the potential risks of Acinetobacter in the sea bream fillets as well as resistance of the bacterium to inactivation by HP. Acinetobacter, together with Pseudomonas and Shewanella seemed to be the main spoilage bacteria during early storage in the HP-treated samples and also in the samples from the combined treatment (AF-HP). The observed higher relative abundances of these bacterial groups could be explained by several factors such as higher capacity to repair cellular damage induced by HP treatment, a higher capacity for utilization of available nutrients, ammensalism, or inactivation of competitors by HP treatment.

Carnobacterium and Lactobacillales were detected at high relative abundances in the HP-treated samples during late storage. Lactobacillales were also the most relevant group during late storage of samples from the combined treatment (AF-HP7, AF-HP10). These results would suggest that HP had selected lactic acid bacteria. Similar results have been reported by high-throughput sequencing studies in mango pulp (Pérez Pulido et al., 2017) and by culture-dependent methods in meat products treated by HP (Garriga et al., 2004; Diez et al., 2009; Han et al., 2011).

Altogether, results from the present study reveal the complexity of bacterial populations in sea bream fillets and how these can be influenced by application of different preservation methods and storage time. The activated film and the HP treatment distinctively selected for different bacterial communities during food storage. While the HP treatment (applied singly or in combination) was most clearly associated with a predominance of OTUs assigned to lactic acid bacteria toward the end of storage, the high relative abundance of OTUs assigned to Acinetobacter in the HP-treated samples deserves further investigation.

IO carried out sample preparation, microbiological analysis and DNA extraction for biodiversity studies. MG participated in supervision of the experimental work and interpretation of data. RP-P contributed with high-HP processing, data analysis and preparation of graphical material for the manuscript. RL was responsible for planning and supervision of the study. RL and AG carried out global analysis of the results and wrote the manuscript.