The goal of this study was to identify phage exolytic lysins that can be used as antimicrobials toward Gram positive LAB known to contaminate fuel ethanol fermentations. From seven putative lysin genes, we have identified four phage lysins (LysA, Lysa2, LysgaY and λSa2 endolysin), that show high activity against LAB contaminants from fuel ethanol facilities. These enzymes have broad exolytic activity in vitro towards numerous Gram positive LAB including several fermentation isolates of L. fermentum. Although LysA, LysA2 and LysgaY, and λSa2 endolysin all demonstrated exolytic activity against lactobacilli, the lactobacilli lysin LysA and the streptococcal λSa2 phage endolysin showed the greatest efficacies to reduce populations of L. fermentum. Interestingly, λSa2 endolysin also exhibited the broadest lytic activity towards the LAB and other Gram positive bacteria tested here.
It is virtually impossible to avoid LAB contaminations in fuel ethanol fermentations, and therefore, the risk of reduced ethanol yield is a major concern. The most common commercially available products used to control contamination in fuel ethanol facilities are based on the antibiotics virginiamycin and penicillin [2, 9] with recommended dosages ranging in fuel ethanol fermentations between 0.25–2.0 ppm  which makes the fuel ethanol industry one of the largest consumers of antibiotics. However, the emergences of antibiotic resistant lactobacilli have occurred in fuel ethanol production facilities [9, 17]. Phage lysins can avoid many resistance pitfalls associated with antibiotic use. Typically antibiotic resistance is a consequence of a bacterial mutation or acquisition of genes that improve the fitness of the recipient bacterium allowing it to evade the action of antibiotics. These adaptations generally occur inside the bacterial cell and employ three general strategies; modification of the drug, alteration of the target (or its level of expression), or decreased accessibility of the drug to its target (reviewed by Bischoff et al. ). Whereas, phage lysins target the PG, which is located outside the cytoplasmic membrane and reduce the number of possible known mechanisms by which bacterial resistance typically emerges .
Lysins are currently being used as disinfectants in industrial settings. Lysozyme which is isolated from hen egg albumen is also a PG hydrolytic enzyme similar to phage encoded lysins. It has been found to be useful in controlling unwanted bacteria in wines at concentrations of 250–500 mg l−1. Although lysozyme has been found to inhibit undesirable malolactic fermentation by Oenococcos oeni, strains of pediococci and lactobacilli, which are usually blamed for serious defects in musts and wines, were resistant . Bacteria are also known to produce various PG hydrolases, for example, a Streptomyces strain produces a mixture of muramidases and proteases that are secreted into the medium. When collected, this mixture has shown broad lytic activity against a variety of wine-relevant LAB, however lytic activity on L. fermentum was not tested  therefore it is uncertain if this approach could be used in a fuel ethanol fermentation system. Other examples for the potential use of lysin-based environmental disinfectants include lysins PlyC and Lysostaphin . The streptococcal lysin PlyC was found to be 1,000 fold more active on a per weight basis than a commercially available oxidizing disinfectant and was shown to retain effectiveness when tested in the presence of non-ionic detergents, hard water, and organic material . The staphylococcal lysin lysostaphin, a PG hydrolase bacteriocin, has been shown to be effective in killing methicillin resistant S. aureus (MRSA) on solid surfaces . Therefore the study of phage-encoded lytic enzymes with activity against problematic LAB in fuel ethanol fermentations is highly relevant and needed.
The differences we observed in exolytic activity of the four lysins against different species Table 1, are possibly reflected in compositional changes in the cell walls. Of the three lactobacilli lysins tested, the exact cut site has only been experimentally determined for LysA2. LysA2 cleaves the bond established between the bridging aspartic acid and the final D-alanine of one of the tetrapeptides involved in the binding of adjacent PG chains of L. casei (Figure 1b) . This particular architecture is typical of lactobacilli as well as many other LAB, including the pediococci . However, some lactobacilli were only moderately exolysed by LysA2 (L. fermentum, L. reuteri, and L. gasseri) while others (L. delbrueckii, L. malefermentans, L. paracasei) were not sensitive to the lysin. A plausible explanation for this lysin specificity could be derived from the cell wall binding domain. Certain Listeria monocytogenes lysins have cell wall binding domains that specifically recognize and bind to teichoic acids before degradation of the peptidoglycan can occur . For the lysins tested here, it is not known if and how the cell wall binding domains are interacting with the cell wall or the prevalence of potential cell wall binding epitopes on the cell surface of the LAB tested in this study. Similarly, ethanol stresses on the cell wall (Figure 3b) or phase of growth (logarithmic, stationary, or biofilm) may contribute significantly to sensitivity to lysins, although lysins are generally believed to be active on all phases of cell growth.
Interestingly, the streptococcal enzyme, λSa2 prophage endolysin, was our strongest candidate antimicrobial for lactobacilli. The most likely explanation for this cross-genera activity is that the λSa2 endolysin glycosidase lytic domain targets the conserved sugar backbone common to all PG. This C-terminal N-acetyl-glucosaminidase cleaves the glycan component of the PG on the reducing side of GlcNAc (Figure 1a) . The N-terminal λSa2 endolysin catalytic domain harbors a D-glutaminyl-L-lysine endopeptidase, which cleaves the peptide bonds between the two amino acids D-glutamine and L-lysine . This exact sequence is present in most lactobacilli PG, but is lacking in the PG of L. fermentum (Figure 1a), the species where λSa2 endolysin is apparently most active (Table 1). L. fermentum has an L-ornithine substituted for L-lysine at the third position of the tetrapeptide (Figure 1a). Although we have no biochemical data to indicate which domain (glycosidase or endopeptidase) is responsible for the lysis of the LAB strains, it is possible that the λSa2 endolysin endopeptidase domain functions to cleave this bond in LABs, when considering the similarity between ornithine and lysine (lysine harbors 4 carbons, rather than 3 for ornithine in the amino terminal alkyl side chains). In support of this possibility is the fact that LysA2 endopeptidase activity [hydrolyzes the bond between the terminal D-alanine of the PG tetrapeptide and the D-aspartic acid residue that forms the bridge with the L-lysine of a neighboring PG chain] was reported to function on species that harbor either an L-lysine or an L-ornithine at position three in the neighboring tetrapeptide . Although not a definitive or even a direct comparison, this suggests a degree of flexibility in the recognition sequences surrounding the cut site of these PG hydrolase lytic domains.
It is interesting to speculate on whether or not sensitivity to ethanol will be a significant factor in the efficacy of the lytic enzymes we have tested. Although, final fermentation conditions yield ethanol concentrations that may be greater than 5%, the starting feedstock is a major culprit for introducing LAB contamination to the fermentation system , a point at which the fermentation is highly sensitive to contamination . At this initial aerobic stage in the fermentation there is no significant ethanol concentration that would be ideal for lysin treatment therefore ethanol levels might not be a factor. Certainly enzymes that have a broad range of activity regardless of ethanol concentration might provide a longer-lived protection during the fermentation process, especially if they are designed to be secreted from recombinant fermentative yeast throughout the fermentation process. Yet to be determined is whether a broader species-target range or biochemical resilience under ethanolic fermentation conditions is more important for the optimal antimicrobial when considering the complex environment of a lignocellulosic fermentation. The increased activity of LysA2 and LysgaY against L. brevis with increasing ethanol concentration is intriguing and suggests these might be preferred enzymes if treating L. brevis contaminants in late fermentations.
From this study, LysA, LysA2, LysgaY and λSa2 endolysin were shown to be excellent candidate antibacterial agents. By homology screening of these lysins to other known PG hydrolase lytic and cell wall binding domains there is no shortage of phage lysins for future consideration as public datasets contain numerous putative lytic PG hydrolases from both bacterial (prophage) and phage genome origins. Due to the high interest in phages that impinge on yogurt production, there are hundreds of known lactobacilli phage, with nine complete Lactobacillus phage genomes and 11 Lactobacillus poly-lysogenic bacterial genomes with sequence available on the NCBI website . For example, lactobacilli lysins from Φadh , and Φg1e [44, 45] have been shown to be functional lytic enzymes, although their ability to exolyse cells has not been reported. There is also a report of an amidase domain from the PL-1 that infects L. casei and a muramidase (Mur-LH) that have shown a broad species activity (. However, due to limitations in the species range of lytic activity each candidate will need to be tested empirically against target LAB.
Our most broadly effective enzymes, λSa2 endolysin and LysA, were tested in mock fermentations and shown to effectively reduce the LAB load by up to 2 orders of magnitude. However promising these results may seem, these initial trials do not necessarily reflect normal fermentation conditions. In order to easily detect changes in target contaminant profiles, the mock trial was performed under pre-sterilization conditions, such that the only contaminant was L. fermentum, a scenario not likely in industrial fermentations. Also, neither the yeast nor the LAB were adapted for growth on hydrolysate, rather, they were both grown in rich broth and then added to the hydrolysate. This assay was designed to test the enzymes for activity under hydrolysate conditions, the results of which are encouraging, but activity on the contaminant might be affected by the cell wall of the contaminant during hydrolysate growth conditions. Thus, the usefulness of these enzymes in large scale fermentations remains to be determined.
Due to the absence of PG in yeast cell walls, none of these lysins including λSa2 endolysin had any catalytic activity towards S. cerevisiae when applied externally and should not adversely affect the fermentation process. In addition, the heat of distillation and the heat of drying the distiller’s grains should denature the lysins, minimizing any potential impact on gut microflora of animals fed the ethanol co-products. Therefore, lysins appear to share qualities worth considering for future works to protect fuel ethanol fermentations from LAB contaminants.