To go back to the first bacteriocin descriptions amounts to studying the first works concerning bacterial antagonism. Such bacterial antagonism was described by the pioneers of microbiology during the last decades of the 19^th century. At that time, the molecular basis of bacterial inhibition was abstruse, so it was difficult to distinguish antagonism due to bacteriocins from that provoked by other compounds such as antibiotics, organic acids or hydrogen peroxide, except on the basis of their spectrum of activity, usually narrower than that of the other ones. Although Cornil and Babès suggested a very narrow antagonism within the genus Staphyloccoccus (“le staphylocoque empêche surtout le staphylocoque”) in their 1885 treatise of bacteriology [32], the scientific community acknowledges the Gratia et al. findings [33] in 1925 as the first documented bacteriocin activity. Indeed, it was named colicin V by the same team in 1949 [34]and later microcin V [35].

The term bacteriocin did not appear until the fifties [34]. This bacteriocin definition is based on the properties of the colicins, that is to say, a lethal biosynthesis, a very narrow spectrum of activity limited to the same species as the producer bacteria and a receptor-mediated mechanism of action [36]. In those days, during the fifties and sixties, the bacteriocin world was mainly made up of bacteriocins from Gram negative bacteria [37,38]. Three genera of Gram positive bacteria were studied for bacteriocin production: Bacillus sp., Listeria sp. and Staphylococcus sp., but it should be noted that during the first half of the 20th century, two lantibiotics, one of the most famous bacteriocins to date, were described. Indeed, the first observations of nisin activity could be those of Roger et al. [39], while subtilin was identified in 1944 from Bacillus subtilis [40]. The exotic amino acid sequences of nisin and subtilin were only elucidated in the early seventies [41,42].

The eighties saw an increase in the number of publications on bacteriocin for both colicin type- and non colicin bacteriocins (Figure 1). But the attribution of nisin GRAS-status by FDA in 1988 [43] would unleash interest in the bacteriocins produced by lactic acid bacteria. Indeed, the industrial applications and the medical and veterinary potential of these microorganisms considered as technological ones are enormous [44–48]. These bacteriocins have aroused a keen interest which has resulted in an exponential increase in the number of publications, while scientific publications about colicins, which may represent the most extensively studied bacteriocins to date, seem to be stabilizing (Figure 1).

Such interest in LAB bacteriocins has resulted in applications as food preservatives, eg antimicrobial ingredients [45–50]. Over the last 20 years, 706 patents based on LAB bacteriocins activity have been recorded around the World, 421 of which were linked to food preservation, and 124 to animal probiotics (http://www.freepatentsonline.com). The non LAB bacteriocins are not devoid of application fields. Applications have also been suggested for plant protection [12,51,52], to prevent local infections in humans [53] and recently in aquaculture [11]. Two dedicated freely available bacteriocin online databases have been assembled: BACTIBASE [54] and BAGEL [55]. Moreover, bacteriocins are part of antimicrobial peptides and on this account, are referenced in various antimicrobial peptide databases such as APD2 [56,57] or CyBase [58].

A new category of bacteriocins has emerged over the last two decades: that of the microcins (Figure 1). These may be considered as the “little sisters” of colicins since they exhibit low molecular weight and are produced by enterobacteriae (for reviews see [35,59,60]). Besides, most microcins exhibit intensive post translational modifications yielding exotic amino acids [61]. In a way, microcins are counterparts of lantibiotics in Gram negative bacteria [61].

Only a few publications are dedicated to bacteriocin production by marine bacteria. Only a few BLIS have been described from marine bacteria and a unique bacteriocin has been fully characterized (see below). In light of marine bacterial biodiversity and the urgent requirement for antibiotic alternatives, we can assume that the marine bacteriocin category will grow exponentially in the near future.

To date, about two hundred bacteriocins have been characterized (BACTIBASE, BAGEL). Bacteriocin classification is not well-established yet and is still the subject of debate. Although dating back to 1993, the bacteriocin classification defined by Klaenhammer is still the most cited one [62]. An update was proposed by Cotter et al. in 2005 [63] and debated by Heng and Tagg in 2006 [64,65]. Bacteriocins are usually classified combining various criteria. The main ones being the producer bacterial family, their molecular weight and finally their amino acid sequence homologies and/or gene cluster organization. An overview of bacteriocins known to date, proposed in Table 1, shows two main categories: the protein-bacteriocins mainly produced by Gracilicutes, mostly enterobacteriae and the peptide-bacteriocins from Firmicutes, chiefly from LAB. Even so, this statement needs to be qualified since enterobacteriae and LAB were the main bacteria studied for bacteriocin production. Our feeling is that peptide bacteriocins from Gracilicutes such as microcins are no exceptions.

Colicins are protein-bacteriocins containing about 500–600 amino acid residues [66]. They are organized in three specific domains. Binding to a specific receptor of the target cell, which is the first step of colicin cytotoxic action is governed by the central domain of colicins. The N-terminal and C-terminal domains are respectively responsible for colicin translocation and antibacterial activity (for a review see [67]). They have been classified in two sub-classes, based on cross resistance [68], translocation system, mechanism of release from the producing cell, and size of encoding plasmids [69]. Group A, translocated by the Tol system and encoded by small plasmids, is composed of colicins A, E1 to E9, K, L, N, S4, U, and Y while group B, translocated by the TonB system and encoded by large plasmids, are made up of colicins B, D, H, Ia, Ib, M, 5, and 10.

Both groups act on sensitive cells by targeting either the inner membrane by pore formation or an intracellular target using enzymatic activity such as DNAse or RNAse [67]. Bacteriocins of such molecular weight are exceptions in Firmicutes compared with the colicin family. Only two have been described in LAB [79,80]. Such protein-bacteriocins produced by LAB have been named class III bacteriocins. The others are specific of Bacillus megaterium [93], Enterococcus faecalis [81] or Staphylococcus aureus [82].

The peptide-bacteriocin group is produced by Gracilicutes and Firmicutes as well. Until 2007, the microcin group was composed of two classes, based on their post-translational modifications [94]. According to their gene cluster organization, this classification has recently evolved [35,59] to give birth to two main classes and two sub-classes. Class I comprises the smallest microcins with molecular masses ranging from 1.1 kDa to 3 kDa (Table 1). They display drastic post-translational modifications leading to exotic structures such as thiazole and oxazole rings in MccB17 (Figure 2). This class acts on sensitive cells by interaction with an intracellular target such as DNA gyrase inhibited by MccB17 [95]. The second microcin class is divided into two sub-classes. The microcin class IIa bridges the gap between colicin and microcin since these peptides are bigger (about 8 kDa) than a typical microcin and exhibit no modifications with the exception of a single disulfide bond formation. One of them, Microcin V (MccV) was previously called colicin V [35], the first documented bacteriocin [33]. Nevertheless, its gene cluster organization connects them undoubtedly to the microcin family [35]. Unlike previous microcins, class IIb microcins are chromosomally encoded, lacking disulfide bond, exposing a conserved serine-rich C-terminal and carrying for some of them a siderophore-type part (MccE492). MccE492 carries out its antibacterial activity by membrane permeabilization. But it was shown to target inner membrane proteins belonging to the mannose permease family [96].

The other main peptide bacteriocins family is the LAB one. Indeed, of the two hundred or so bacteriocins described to date, almost 90% are from LAB. With the exception of Helveticin J [79] and Milletricin [80], which are members of class III bacteriocins, they all are of peptidic nature. They have been divided into two main classes: class I and class II, the latter in turn containing three sub-classes (Table 1).

Lantibiotics have been defined as class I. Lantibiotic peptides undergo drastic posttranslational modification leading to unusual amino acid residues such as lanthionine. In a way, they are the counterpart of microcins in Firmicutes. To date, about 50 different lantibiotics have been described in LAB and non LAB bacteria such as Staphylococcus aureus [97]. Overall, lantibiotics are divided on the basis of their topology, that is to say their lanthionine bridge arrangements. Type-A lantibiotics such as nisin (Figure 2) are linear and cationic peptides, while type-B ones are globular [86,98]. The former exerts its antibacterial activity by membrane permeabilization by pore formation in a torroid manner [98] after binding to lipid II, while the latter targets intracellular enzyme function [98]. Another emerging lantibiotic class is the two-component lantibiotics such as haloduracin [99–101].

Class II bacteriocins are lightly modified peptides. These peptides are 20 to 70 amino acid residue-long. Extensive studies have been carried out about their mechanism of action. It has appeared that they use a common global procedure targeting a membrane-embedded domain of an integrated membrane protein [91]. The conformational modifications resulting from membrane protein–bacteriocin interactions lead to membrane perturbations, permeabilization and finally bacterial cell death [102]. It was divided into four sub classes on the basis of their activity. Class IIa was also named pediocin-like or anti-Listeria bacteriocins since all of them displayed antibacterial activity against Listeria spp. [62]. These bacteriocins are peptides sharing a highly conserved N-terminal part harboring a consensus sequence: -Y-Y-G-N-G-V-X-C-x-x-x-x-C (Figure 2) where C residues are involved in a disulfide bridge [48]. Their more variable C-terminal part has been used for their segregation in four sub-groups [63,102]. They act on target cells by a pore-forming mechanism of action [48,87,102]. This class constitutes the bacteriocin success story of the last twenty years. Class IIb is an original antimicrobial peptide class because it is made up of two independent peptides, each being active but both being required for optimal activity [102]. Around twelve such two-component bacteriocins have been described in LAB. Each time, the most active mix was obtained with equivalent concentration of each peptide [88]. LAB bacteriocin group IIc are real cyclic peptides since their N- and C-termini are covalently connected (for review, the reader is referred to [63,89]). Their mechanism of action when explored was permeabilization of the inner membrane of target cells leading to cell death. Finally, unmodified and non-pediocin-like peptides and single peptide active bacteriocins form class IId. To date, about 32 different class IId peptides have been described [102].

Bacteriocins are unique antimicrobial peptides. Indeed, the producing strain has to protect itself from its own peptides, so bacteriocin-producing bacteria have to develop some sort of immunity strategy. In addition to a structural gene, post-translational gene and export machinery, the gene cluster organization of bacteriocin encodes as well for an immunity protein. The latter ensures bacteriocin protection in various ways, depending on the bacteriocin mechanism of action.

Immunity to pore forming colicins is mediated by a 11 to 18 kDa small membrane protein. A direct and specific interaction within the inner membrane between the immunity protein and the C-terminal part of colicin achieves cell protection. Transmembrane helices have been shown to be the main motifs recognized by immunity proteins. Colicins targeting intracellular enzymes such as nuclease are inactivated by direct binding of the immunity protein (about 10 kDa) to the active domain of colicin leading to a 71-kDa heterodimer.

Microcin immunity still remains opaque, while that towards lantibiotic has been recently reviewed [103,104]. Lantibiotic immunity is conferred by lipoprotein intercepting lantibiotic at the cytoplasmic membrane and/or ABC transporter–type membrane protein complex. Immunity to class II bacteriocins produced by LAB has recently been cleared up [91]. It implies that components of the mannose phosphotransferase system are receptors for both bacteriocin and the imunity protein [105]. To define the role of bacteriocins in producing bacteria is still a challenge. Its production entails advantages in colonizing or defending ecological niches for producing bacteria.

Vibrio species are ubiquitous in the marine environment and are commonly isolated from fish and shellfish specimens [109]. Some species may be pathogenic to marine life, but some do not appear to affect them. Due to their capability to occupy this ecological niche they have been studied for their capacity to produce bacteriocin-like inhibitory substances (BLIS). Zai et al. [110] have isolated and identified fifty strains of the genus Vibrio isolated from the gills and gut region of healthy and infected catfishes (Arianus thalassinus). BLIS was detected and called Vibriocin AVP10 (Table 2).

Fresh and frozen seafood were studied by Carraturo et al. [111]. They have isolated three non-pathogenic (for humans) species of Vibrio (V. mediterranei 1, V. mediterranei 4 and V. fluvialis) displaying antagonistic activity on solid agar medium against pathogenic V. parahaemolyticus and V. mediterranei. A partial purification of a BLIS produced by V. mediterranei 1 was reported. Its proteinaceous nature was revealed by enzymatic degradation by proteinase K. Thanks to size exclusion chromatography, Carraturo et al. [111] have purified an antimicrobial fraction whose molecular mass was determined by SDS-PAGE to be 63–65 kDa corresponding to a mixture of unrelated polypeptides, including the bacteriocin.

Furthermore, V. harveyi is a serious pathogen of many vertebrate and invertebrate marine animals [112,113]. McCall and Sizemore [114] have reported for the first time the production of a bacteriocin in a strain of Beneckea harveyi (V. harveyi). The bacteriocin, ‘harveyicin SY’, with an estimated molecular mass of 24 kDa, was lethal to two strains of V. harveyi, KN96 and BBP8 (Table 2). Harveyicin SY was susceptible to proteolytic enzymes, and is apparently plasmid associated [114,115].

Prasad et al. [112], whilst screening various V. harveyi isolates from their culture collection have recognized a possible BLIS production by a strain of V. harveyi (VIB 571). Interestingly, this strain has been demonstrated to be pathogenic to rainbow trout (Oncorhynchus mykiss) and Atlantic salmon (Salmo salar) [113].

Inter-strain and inter-species inhibition mediated by a bacteriocin-like inhibitory substance (BLIS) from V. harveyi VIB 571 was demonstrated against four isolates of the same species and V. fischeri, V. gazogenes and V. parahaemolyticus (Table 2). The crude BLIS, which was obtained by ammonium-sulphate precipitation of the cell-free supernatant of a 72 h broth culture, was inactivated by lipase, proteinase K, pepsin, trypsin, pronase E and SDS. Incubation for 10 min at more than 60 °C resulted in loss of activity. On the other hand, antibacterial activity was not affected by pH. Anion-exchange chromatography, gel filtration, SDS-PAGE and two-dimensional gel electrophoresis revealed the presence of a single major peak, comprising a protein with a pI of ~5.4 and a molecular mass of ~32 kDa (Table 2). The N-terminal sequencing of the ~32 kDa protein yielded: D-E-Y-I-S-X-N-K-XS-S-A-D-I where ‘X’ may be cystein or modified amino acid residues.

Other vibriocins were isolated by Shehane and Sizemore [116]. Their aim was to identify bacteriocins effective against V. vulnificus in seafood. Isolates from estuaries near Wilmington (NC, USA) containing plasmids were checked for antimicrobial activity which was not due to lytic bacteriophage or small, non specific molecules. Three bacteriocin producers of V. vulnificus were detected and their inhibitory spectra determined (Table 2). Strain IW1 inhibited few strains of V. vulnificus; BC1 inhibited several strains of V. vulnificus, V. cholerae and V. parahaemolyticus and BC2 inhibited all tested Vibrio spp, Plesiomonas shigelloides and E. coli. Loss of inhibitory activity coincided with loss of the bacteriocinogenic plasmid. The molecular weights of the bacteriocins were estimated to be 9.0 kDa for IW1, 7.5 kDa for BC1 and 1.35 kDa for BC2 thanks to size exclusion chromatography. IW1 was heat labile, while BC1 was moderately stable except at extreme temperatures. BC2 was very stable and maintained its activity when frozen, autoclaved or exposed to extreme pH values [116]. The authors suggested that these bacteriocins might provide a tool for the removal of V. vulnificus from seafood.

Strain Vibrio sp. NM 10 was isolated from spotnape ponyfish (Leiognathus nuchalis) collected in coastal regions of Enoshima Island, Kanagawa, Japan. This strain exhibited high activity against P. piscicida K-III, but was also able to inhibit E. coli IAM 1264, V. vulnificus RIMD 2219009 and Enterococcus seriolicida YT-3 [117]. The antibacterial substance produced by Vibrio sp. NM 10 is a proteinaceous heat-labile substance with a molecular mass of less than 5 kDa. These facts strongly suggest that the antibacterial substance is either a bacteriocin or a bacteriocin-like substance [117].

Authors Moro et al. [118] and Messi et al. [119] have shown their interest in evaluating BLIS production in Aeromonas hydrophila. All strains of Aeromonas hydrophila in these two studies demonstrated inhibitory activities against several strains of Staphylococcus aureus (Table 2). Messi et al. [119] have demonstrated further inhibitory effect against Listeria species, Streptococcus agalactiae and Lactobacillus sp. No inhibition was observed against all Gram-negative strains assayed, including related species (Aeromonas sobria ATCC 43979, A. caviae ATCC 13137). Such an inhibitory spectrum is not compatible with the bacteriocin definition.

Longeon et al. [121] investigated bacteria collected from different substrates on the littoral of Brittany and they focused their attention on a Pseudoalteromonas sp. named X-153 that exhibited high antimicrobial activity. Purification of the active protein P-153 from the bacterial cells was achieved. This antibacterial protein was evaluated by size exclusion chromatography to be of 280 kDa size. This antibacterial protein was shown to be active against both gracilicutes (ichthyopathogenic Vibrio) and firmicutes (Staphylococcus epidermidis, Propionibacterium acnes and P. granulosum) (Table 2). Such a broad spectrum of activity is not consistent with the definition of a bacteriocin.

It is generally considered that Gram-positive bacteria, including lactic acid bacteria, are numerically dominant members of the normal microbiota in the gastrointestinal tract of endothermic animals at an early stage of their lives [122]. The gastrointestinal microbiota of healthy fish is usually composed of lactic acid bacteria belonging to the genera Streptococcus, Lactobacillus, Carnobacterium, Leuconostoc [122]. Divercins and piscicocins have been fully characterized from Carnobacterium isolated from fish intestine (Table 3). These two bacteriocins belong to class IIa of bacteriocins produced by LAB (see Table 1, for review the reader is referred to [123]).

In 2004, Pirzada et al. [120] isolated and studied a bacteriocinogenic strain ZM81, a Gram positive pleomorphic rod, which was isolated from the open sea region of Karachi. The proteinaceous nature of the cell-free supernatant of marine strain ZM81 was defined by enzyme degradation with pronase and trypsin. Fractionization of the crude bacteriocin thanks to a molecular weight cut-off membrane showed an enrichment of activity in the fraction containing >10 kDa bacteriocin-like inhibitory substance. BLIS produced by Marine Bacterium ZM81 is heat labile and exhibits activity within a wide pH range of 4–12 [120].

While most small peptides found in Cyanobacteria are biosynthesized by nonribosomal peptide synthetases [131], a microcin-like pathway for the biosynthesis of a family of cyclic peptides, the patellamides (Figure 2), has been recently reported in Prochloron didemni, a cyanobacterial symbiont of tropical ascidians [92]. The patellamides are moderately cytotoxic and composed of a pseudosymmetrical, cyclic dimer, with each substructure having the sequence thiazole-nonpolar amino acid-oxazoline-nonpolar amino acid. Despite these unusual features, patellamide biosynthesis is ribosomal [132]. The discovery of patellamides has provided first insight into the biosynthesis of microcin-like peptide distribution and versatility in Cyanobacteria [133].

The patellamide family are cyclic octapeptides (Figure 2) characterized by the presence of thiazole and oxazole moieties. Although nonribosomal biosynthesis was anticipated for the formation of these peptides, heterologous expression of a microcin-like gene cluster discovered in the genome of the cyanobacterium Prochloron didemni unambiguously showed that these peptides are produced by a ribosomal pathway [92,133,134]. An increasing number of other cyclic peptides containing heterocyclic amino acids has recently been isolated from planktonic and other animal-associated cyanobacteria, including nostocyclamide [135], tenuecyclamides [136], venturamides [137], dendroamides [138], and microcyclamides [139]. The variety of structures is reflected in an equally large variety of bioactivities, such as antibacterial, cytotoxic, and antimalarial activities [133].