After a century of nisin research - where are we now?

Abstract It is almost a century since nisin was discovered in fermented milk cultures, coincidentally in the same year that penicillin was first described. Over the last 100 years this small, highly modified pentacyclic peptide has not only found success in the food industry as a preservative but has also served as the paradigm for our understanding of the genetic organization, expression, and regulation of genes involved in lantibiotic biosynthesis—one of the few cases of extensive post-translation modification in prokaryotes. Recent developments in understanding the complex biosynthesis of nisin have shed light on the cellular location of the modification and transport machinery and the co-ordinated series of spatio-temporal events required to produce active nisin and provide resistance and immunity. The continued unearthing of new natural variants from within human and animal gastrointestinal tracts has sparked interest in the potential application of nisin to influence the microbiome, given the growing recognition of the role the gastrointestinal microbiota plays in health and disease. Moreover, interdisciplinary approaches have taken advantage of biotechnological advancements to bioengineer nisin to produce novel variants and expand nisin functionality for applications in the biomedical field. This review will discuss the latest progress in these aspects of nisin research.


Introduction
Nisin is one of the oldest known antimicrobial compounds. In 1928, the same year that Alexander Fleming discov er ed penicillin, Rogers and Whittier (Rogers 1928 ) noted that certain lactic streptococci (then called Group N Streptococcus ) were inhibitory to starter cheese cultures. Almost 20 y ears follo wing its discovery, this inhibitory peptide was named nisin, or 'group N Streptococcus I nhibitory S ubstance', the suffix '-in' implying antibacterial pr operties (Hirsc h and Mattic k 1949 ). Although initiall y pr oposed for use as an antibiotic based on its activity a gainst v eterinary and clinical pathogens including Mycobacterium tuberculosis , it was deemed unsuitable due to low aqueous solubility and poor stability at physiological pH (Hurst 1981 ). Ho w e v er, in the 1950s the potency of nisin as a food pr eserv ativ e was realized based on its success in pr e v enting spoila ge of pr ocessed c heese by clostridia (Delv es-Br oughton et al. 1996 ). Since then, this highly modified peptide has gained a ppr ov al by r egulators in over 80 countries, including the Food and Drug Administration (FDA) and the European Food Safety Authority (EFSA) (Chikindas et al. 2018 ). Nisin has been used in a wide assortment of foods including dairy and dairy desserts, canned goods, processed meats, fish, fruit juices, and be v er a ges (Gharsallaoui et al. 2016, Todor ov et al. 2022. Nisin exhibits potent activity against Gram-positive bacteria including Bacillus cereus , Listeria monoc ytogenes , enter ococci, sta phylococci, and streptococci. Nisin has also been used in limited applications in the veterinary industry to prevent or treat bovine mastitis (Roy et al. 2016 ). The commercial success of nisin has made it the most investigated bacteriocin (antimicrobial peptides produced by bac-teria that can kill other bacteria) in terms of its genetic organization, biosynthesis, mode of action and efforts to broaden its food and potential ther a peutic a pplications. Nisin is also a lanthipeptide, a r a pidl y expanding subset of the ribosomall y synthesized and post-tr anslationall y modified peptides (RiPPs) family (Arnison et al. 2013 ). Their defining feature is the presence of unusual thioether amino acids including lanthionine (Lan) and/or methyllanthionine (MeLan) that ar e intr oduced thr ough a series of enzyme-mediated post-translational modifications . Curr entl y, lanthipeptides can be classified into five distinct groups based on the c har acteristics of their lanthipeptide synthetases. For Class I lanthipeptides, of which nisin is the prototypical member, the thioether cross-linked amino acids (Me)Lan ar e gener ated b y tw o distinct enzymes: a dehydratase LanB and cyclase LanC (see later section on nisin biosynthesis). Ho w e v er, in Class II-IV lanthipeptides, multifunctional enzymes are emplo y ed to form (Me)Lan, namely LanM (Class II), LanKC (Class III), and LanL (Class IV). A r ecentl y discov er ed gr oup of lanthipeptides that are generated via a biosynthetically distinct pathway since the biosynthetic gene clusters (BGCs) do not contain genes encoding well-defined Class I-IV (Me)Lan synthase homologues constitutes the newest class (Class V) (the reader is referred to excellent r e vie ws cov ering these aspects in mor e detail (Hegemann and Süssmuth 2020, Montalbán-López et al. 2021, Lee and van der Donk 2022. Lanthipeptides that exhibit antimicrobial activity have been historically termed lantibiotics. Importantly, some lantibiotics, including nisin, exhibit multiple modes of action, which involv e inter actions with highl y conserv ed cell wall synthesis intermediates as well as the ability to form pores in the bacterial membrane (Sahl and Bierbaum 2008 ). In the case of nisin, this dual mode of action and demonstrable high potency against m ultidrug r esistant (MDR) bacteria makes the lantibiotic v ery attr activ e for potential use as a biother a peutic in human and animal health related applications , van Staden et al. 2021. Indeed, recent advances in genome mining tools and next generation sequencing (NGS) technologies (Biermann et al. 2022 ) have facilitated the prediction of novel BGCs from a diverse range of bacteria and sources including human and animal micr obiomes. Consequentl y, the number of natural nisin variants has doubled in the last decade, r epr esenting a v aluable collection of novel peptide structures that exhibit a range of properties and antimicr obial spectr a. Mor eov er, incr easing numbers of meta genomic studies are beginning to shed light on the complex interactions between microbes in the human gastrointestinal tract and the role that antimicrobial peptides might play in this context (Garcia-Gutierrez et al. 2019a, Heilbronner et al. 2021. In fact, r ecent micr obiome-based inv estigations hav e r e v ealed that commensal nisin producing bacteria can inhibit MDR pathogens and sha pe nic he competition in the gastr ointestinal micr obiome, offering exciting prospects for the use of nisin in human therapeutic applications. The k e y features of nisin biosynthesis have been painstakingl y unr av elled ov er decades, including the regulation and functions of the individual biosynthetic proteins and the complex catal ytic pr ocesses involv ed in the formation of the Lan and MeLan rings as well as the export of and immunity to the acti ve pe ptide (de Vos et al. 1995, Lubelski et al. 2008. Ho w e v er, r ecent r eports hav e offer ed a fascinating insight into nisin biosynthesis that also advances our fundamental kno wledge regar ding the cellular localization, spatial configuration, and complex interaction of the post-translational modification and transport machinery as well as a greater perception of producer self-protection (self-immunity) during nisin biosynthesis . Furthermore , the geneencoded nature of nisin makes the bacteriocin accessible to biosynthetic engineering through site-directed mutagenesis or synthetic chemistry approaches. Elaborate expression systems in conjunction with high-throughput screening strategies have gener ated v ast collections of nisin deriv ativ es that exhibit altered functional properties . T his r esearc h has adv anced our a ppr eciation of structure-activity relationships, modification enzymesubstrate specificity, and immunity mechanisms that can be used as the basis for the rational design of next generation nisin deri vati ves with possible applications, which range from activity against Gram-negative bacteria to tools for microbiome editing.

How is nisin biosynthesized?
To produce the acti ve pe ptide, nisin is first ribosomally synthesized as an unmodified 57 amino acid precursor peptide (pre-NisA) consisting of a 23-amino acid N-terminal leader peptide and a 34-amino acid C-terminal core peptide (Fig. 1 ). The leader peptide serves as a signal sequence for export by NisT and as a r ecognition motif r ecognized by the modification enzymes NisB and NisC (Siegers et al. 1996, Kuipers et al. 2006 ). The first phase in nisin maturation involves the dehydration of selected serine and thr eonine r esidues in the cor e peptide, a process mediated by a dimer of the membrane-associated dehydratase NisB (Fig. 1 ). This process involves the transfer of glutamate to specific serine and threonine side chains follo w ed b y subsequent elimination reactions to generate dehydroalanine and dehydrobutyrine, r espectiv el y (Gar g et al. 2013 ). In the next phase, these dehydr ated r esidues ar e coupled to nearb y c ysteines via intramolecular addition reactions mediated by the cyclase NisC to form five cyclic bridges composed of a lanthionine (Lan, where an initial serine is linked to a neighbouring cysteine) and 4 methyllanthionine (MeLan, where a threonine is linked to cysteine) rings (Koponen et al. 2002 , Li andvan der Donk 2007 ). In the final step, the modified 57 amino acid peptide is exported from the cell via the dedicated ABC-type transporter NisT (Kuipers et al. 2004 ) and the matur e bioactiv e peptide is r eleased following r emov al of the leader region by a dedicated serine protease, NisP (Lagedroste et al. 2017, Montalbán-López et al. 2018 (Fig. 1 ). The producer cell is protected from the now active peptide by two distinct immunity systems composed of a lipoprotein NisI and an ABC transporter NisFEG (Khosa et al. 2016a ). To ensure a proper equilibrium between production and immunity, nisin expression is autoregulated by the mature nisin peptide via a two-component signal transduction system (TCS) composed of a sensor histidine kinase NisK, and r esponse r egulator NisR   (Fig. 1 ). Binding of mature nisin to NisK stimulates autophosphorylation. The phosphate group is then tr ansferr ed to activ ate NisR, whic h then induces transcription from two of the four promoters in the nisin gene cluster, P * nisA and P * nisF (Fig. 1 ). T hus , nisin functions as both an antimicrobial peptide and a peptide pheromone that plays an essential role in quorum sensing control of its own biosynthesis (Kleerebezem 2004 ). A multitude of studies with a particular focus on the individual biosynthetic pr oteins hav e been instrumental in elucidating the intricate catalytic processes and stereoc hemistry involv ed in formation of the Lan and MeLan residues (For compr ehensiv e r e vie ws see Repka et al. 2017, Montalbán-López et al. 2021. For example, the pr ecise mec hanism of the NisB dehydratase action remained elusive due to the inability of r esearc hers to r econstitute NisB activity in vitro . This was finally ac hie v ed when it was established that NisB r equir ed glutamyl-tRNA to bring about the dehydr ation r eaction (Gar g et al. 2013 ). Later studies identified the domains within NisB that catalyse the transfer and subsequent elimination of glutamate from the tRNA to the core peptide to form Dha or Dhb, as well as elucidating the importance of the highl y conserv ed -FNLD-box within the leader peptide for recognition and binding of NisB and NisC to pre-NisA (Abts et al. 2013, Ortega et al. 2015. Furthermor e, the r equir ement of Zn2 + for NisC catalytic activity provided further mechanistic insight into how the cyclase guides the formation of the lanthionine rings (Li and van der Donk 2007 ). Another important br eakthr ough r e v ealed the details of the formation of the modification complex which was shown in vitro to have a stoichiometry of 2:1:1 (NisB:NisC:pre-NisA) (Reiners et al. 2017a ). Additionally, the purification and in vitro ATPase activity of NisT was recentl y demonstr ated (La gedr oste et al. 2020 ). Importantly, numerous reports concerning the organization and cellular location of the nisin modification machinery have alluded to the existence of a membr ane-associated m ultimeric synthetase pr otein complex consisting of NisB, NisC, and NisT (for a compr ehensiv e r e vie w see Lubelski et al. 2008 ).
Ho w e v er, while these pr e vious studies r epr esent extr aordinary advances in their own right, a recent seminal w ork emplo y ed a combination of m uta genesis and fluor escent micr oscopy to not only confirm the existence of this enzymatic complex, but also to r e v eal the natur e of its assembly and cellular location in vivo and the highly co-ordinated series of spatio-tempor al e v ents that unfold to produce and export active nisin (Chen et al. 2020 ). Using an elegantly designed suite of green fluorescent protein (GFP)labelled NisA fusions as well as sfGFP and mCherry fusions to Figur e 1. Nisin biosynthesis , r egulation, and imm unity. Nisin is first synthesized as an unmodified precursor (NisA) consisting of a leader peptide and core peptide. NisA is processed by the dehydratase NisB and cyclase NisC and transported by NisT (NisABCT complex) at the old pole in L. lactis . The matur e bioactiv e peptide is r eleased by leader peptide cleav a ge performed by NisP. Imm unity fr om the activ e peptide is pr ovided b y tw o distinct immunity systems composed of a lipoprotein NisI and an ABC transporter NisFEG. To ensure a proper balance between production and immunity, nisin expression is regulated via a two-component signal TCS composed of a sensor NisK, and response regulator NisR that activate the nisin promoter (P * ).
the N-and/or C-termini of NisB, NisC, and NisT, r espectiv el y, it was r e v ealed that pr e-NisA binds NisB and NisC to form a complex, i.e. localized to the poles of the Lactococcus lactis host (Fig. 1 ). Further analysis established that NisB and NisC were preferentially located at one pole in single cells, later identified as the 'old' pole through time-course experiments. Surprisingly, the transporter NisT was shown to be e v enl y distributed in the cell periphery and not part of a NisBCT complex as predicted, suggesting that NisT is recruited to the pole localized NisBC complex only when r equir ed to tr ansport of the full y modified pr e pe ptide. Indeed, a NisT mutant incapable of secretion and dissociation confirmed this when it was visualized together with NisB and NisC, also at a polar position. Finally, the order of assembly was clarified and r e v ealed that NisB plays a central role in the initial recruitment of the other components (Chen et al. 2020 ). A short domain (NisB 750-769 ) targets dimeric NisB to the pole of the cell. NisC is recruited to form a NisBC complex that interacts with nisin precursor and both enzymes act in succession to perform the dehydration and cyclization reactions and create each lanthionine ring in turn. NisT is drafted to the cell poles from the cytoplasmic membr ane onl y when the modifications are complete and all five rings are in place . T he fully modified pre-NisA is then released from NisBC and transferred to NisT. Following export of the fully modified pre pe ptide, the complex dissociates and NisT again becomes e v enl y distributed ar ound the cell. It is thought that this polar-localized synthesis and secretion of nisin prevents access of the peptide to its target lipid II, given that peptidoglycan synthesis would be significantly lo w er at the old pole, thereby limiting the possibility of any self-killing action (Chen et al. 2020 ). This remarkable insight into the mechanistic details and co-oper ativ e spatiall y contr olled actions during nisin biosynthesis could potentially lead to enhanced nisin pr oduction and ultimatel y enable the expression of other rationally designed and novel lanthipeptides.
Indeed, the promiscuity of the nisin enzymatic complex to modify and transport a broad range of substrates attached to the nisin leader sequence, including medically relevant nonlantibiotic pe ptides (Klusk ens et al. 2005, Moll et al. 2010, validates the broader potential applications for the bioengineering of novel compounds other than nisin. A perfect example of this describes the expression of more than 30 novel lantibiotics from almost 60 promising candidate genes identified from genome mining of publicl y av ailable pr okaryotic sequences (v an Heel et al. 2016 ). The genes were synthesized with the nisin leader peptide sequence and introduced into a nisin expression system. Notably, five modified lantibiotic peptides from a variety of organisms including Corynebacterium lipophiloflavum DSM 44291, Streptococcus agalactiae ATCC 13813, and Streptococcus suis R61, were produced and found to be activ e a gainst se v er al pathogenic bacteria including v ancomycin r esistant enter ococci (VRE) and methicillin resistant Staphylococcus aureus (MRSA)  ). Indeed, efforts to further expand the substrate flexibility of biosynthetic enzymes pr ovides e v en mor e tantalizing pr ospects for nov el and bioengineered peptides with unique functionalities. For example, a recent study emplo y ed a r andom m uta genesis a ppr oac h to generate mutant libraries of the dehydratase NisB (Zhao et al. 2020a ). A high throughput selection strategy based on cell surface display of modified and cyclized peptides identified a NisB deri vati ve exhibiting broader substrate flexibility with a capacity to modify substrates normally incompatible with the wild type NisB.

Nisin resistance and immunity mechanisms
Given the remarkable ability of microbes to adapt to their environments, it is not surprising that persistent bacterial exposure to bacteriocins can lead to resistance development (Kumariya et al. 2019 ). Resistance to lantibiotics has been noted, with the most fr equent mec hanisms involving physiological adaptations to the cell-envelope including cell wall thickening, alterations to phospholipid and membrane fatty acid composition, and the ov er all cell wall c har ge (The r eader is dir ected to compr ehensiv e r e vie ws describing these mechanisms in greater detail; Bastos et al. 2015, Dr a per et al. 2015, Barbosa et al. 2021. Over the last decade, bacterial resistance mediated by transporters has gained significant attention due to the high degree of protection these systems provide against cell wall-active drugs, including antimicrobial peptides (Gebhard 2012 ). Although a wide variety of these integral membr ane pr oteins hav e been c har acterized as imm unity systems in producing organisms, they have also been identified in the genomes of nonproducing and often pathogenic organisms. To understand how these transporters confer resistance to nisin, it is first necessary to a ppr eciate the mode of action of the peptide. Nisin exerts its antimicrobial action both by pore formation and by inhibition of cell wall synthesis through specific binding to lipid II, an essential precursor in peptidoglycan biosynthesis (Breukink et al. 1999, Wiedemann et al. 2001 ). This dual functionality is mediated by two structural domains, an N-terminal domain composed of rings A, B, and C, linked to the C-terminal rings (D and E) by a short thr ee r esidue hinge r egion ( Fig. 2 A). Studies hav e r e v ealed that the N-terminal lanthionine rings form a pyr ophosphate ca ge ar ound the head-gr oup of lipid II, thus inhibiting cell wall synthesis . T he interaction primarily occurs via five hydrogen bonds formed between the amide backbone of rings A and B of nisin and the pyrophosphate moiety of lipid II (Hsu et al. 2004 ). This binding enhances insertion of rings D and E in a transmembrane orientation, facilitated by the flexible hinge, forming a stable por e. Notabl y, lipid II is also the target of the glyco-and lipogl yco-peptide antibiotics v ancomycin, teicoplanin, telavancin, dalba vancin, and orita vancin that serve as first line antibiotics to tr eat MDR Gr am-positiv e pathogens . T his is also true for the depsipeptides ramoplanin and teixobactin, though they bind different segments of the highly conserved lipid II molecule than nisin, whic h has r ele v ance for both mode of action and mode of resistance (Ulm and Schneider 2016 ). In the case of nisin, binding to lipid II also facilitates pore formation leading to the rapid efflux of ions and small cytoplasmic compounds from the target organism (Bierbaum and Sahl 2009 ). Initially, it was believed that the pores consisted of eight nisin and four lipid II molecules (Hasper et al. 2004 ), though r ecent e vidence suggests the pore complex gr ows continuousl y thr ough the r ecruitment of mor e and mor e nisin-lipid II a ggr egates leading to immense membr ane dama ge (Sc her er et al. 2015 ). The producer strain escapes the killing action of nisin by expressing a set of nisin-specific immunity proteins consisting of a membr ane-anc hor ed lipopr otein NisI and a m ultipr otein ABC-type export complex NisFEG (Stein et al. 2003 ). One line of evidence suggests that NisI intercepts nisin and blocks the peptide from reaching its molecular target lipid II (Stein et al. 2003, Koponen et al. 2004, thereby preventing pore formation (Geiger et al. 2019 ). Furthermor e, r ecent r eports indicate that NisI-nisin interactions also promote cell clustering that acts as a shield to inhibit the action of nisin (AlKhatib et al. 2014 ). On the other hand, the ABC transporter NisFEG functions by ejecting nisin from the cell membrane before a pore can be formed (Stein et al. 2003 ). The specificity of NisFEG appears to reside in the C-terminal region of nisin, since deletion of the terminal six amino acids and Ring E reduced the degree of immunity provided by NisFEG (AlKhatib et al. 2014 ). Although it is postulated that NisI and NisFEG act co-oper ativ el y, the r ole of NisI in self-imm unity of the producer appears to be more critical than the transporter since a deletion of nisI results in greater susceptibility to nisin compared to a nisFEG knockout (Siegers and Entian 1995 ). While the manner of NisI-mediated immunity is not yet fully understood, r ecent NMR (Hac ker et al. 2015 ) and molecular modelling studies (Jeong and Ha 2018 ) hav e r e v ealed a C-terminal cleft and gr oov e r egion that may r epr esent important sites for NisI-NisA interactions.
While suc h imm unity or self-r esistance systems ar e typically found in almost all lantibiotic producing strains with individual specificity, it has become a ppar ent that some nonproducers, including pathogenic bacteria, can harbour gene clusters encoding functional proteins and ABC transporters linked to two-component signal TCS (Dr a per et al. 2015, Clemens et al. 2018, Barbosa et al. 2021. Despite the diversity in their genetic organization, these 'orphan' gene clusters confer demonstr able r esistance to one or mor e lantibiotics. Suc h systems often bear a resemblance to the immunity systems found in lantibiotic gene clusters (LanFEG) or resemble BceAB-type transporters, so-called after the B a c itracin e fflux (Bce) transporter system from Bacillus subtilis (Ohki et al. 2003 ). For example, Streptococcus mutans UA159 harbours two systems, one being the LcrSR-LctFEG system that provides resistance to lacticin 481 and nukacin ISK-I and another, NsrFE 1 E 2 G-XRK, that affords resistance to nisin (Kawada-Matsuo et al. 2013 ). Clostridioides difficile (formerly Clostridium difficile ; Lawson et al. 2016 , and fr om her ein termed C. difficile ) harbours the cprABCK-R ( cationic antimicrobial peptide resistance ) operon that provides protection from several lantibiotics including nisin, mutacin 1140 and subtilin (Suárez et al. 2013 ) and in L. monocytogenes the V irSR TCS/V irAB and AnrAB ( abc transporter involved in nisin resistance ) system imparts resistance to multiple antimicrobials including nisin (Kang et al. 2015, Jiang et al. 2019. Mor eov er, S. aureus harbours a complex network of different TCS (BraRS (also known as NsaRS), and GraRS) linked to ABC transporters (VraDE and BraDE) that confer nisin resistance (Blake et al. 2011, Hiron et al. 2011, Randall et al. 2018. Recentl y, a r esistance oper on has been identified in S. agalactiae composed of a TCS NsrRK and an ATP-binding cassette transporter NsrFP, but unlike the pr e viousl y described systems, an additional membrane-associated serine protease termed the nisin r esistance pr otein (NSR) is pr esent (Khosa et al. 2013 ) that inactivates nisin through proteolytic cleavage at its C-terminus (Sun et al. 2009 ). The resulting shortened peptide (nisin 1-28 ) is up to 100-fold less active and exhibits a considerably reduced ability to form pores (Sun et al. 2009 ). Furthermore, computational modelling of the protease/nisin complex revealed the importance of the C-terminus of nisin for NSR specificity (Khosa et al. 2013 ).
The latest studies regarding NsrFP has provided fresh insights into the mechanism of BceAB-type transporters (Gottstein et al. 2022 ). An unusual feature of these systems is that both transporter permease and the histidine kinase component of the TCS are thought to be mutually indispensable for both sensing of and resistance to the antimicrobial, forming a sensory complex in which the transporter represents the actual sensor (Dintner et al. 2011, Clemens et al. 2018. Ho w e v er, heter ologous expr ession of NsrFP alone (i.e. without its cognate TCS) in L. lactis conferr ed r esistance to nisin (Reiners et al. 2017b ) and bacitracin, as well as a number of other lipid II targeting compounds (Gottstein et al. 2022 ). Mor eov er, compar ativ e pr oteomic anal ysis of L. lactis cells expressing NsrFP with a nonfunctional mutant (NsrF H202A P) suggests that NsrFP may also act by shielding lipid II cycle intermediates from the antimicrobial compound (Gottstein et al. 2022 ), and ther eby pr ovide additional r esistance thr ough tar get Despite its widespread use by the food industry, detection of stable and transmissible resistance to nisin has yet to be reported. Ho w e v er, the pr esence and distribution of resistance genes as described above across multiple species including important human pathogens poses a significant challenge to potential use of nisin and other bacteriocins in ther a peutic a pplications. For example, the nsr operon has been detected in animal and human pathogenic streptococci (including S. agalactiae , Streptococcus dysgalactiae , S. suis , Streptococcus canis ), staphylococci (including Staph ylococcus capitis , Staph ylococcus h yicus , Staph ylococcus epidermidis ), and in Enterococcus faecium (Khosa et al. 2013, Simões et al. 2016. It is worth noting that these genes are often positioned on transmissible elements such as plasmids (Froseth andMcKay 1991 , Liu et al. 1997 ), highlighting the potential for NSR-associated resistance transfer to other microbes. Indeed, the pr e v alence of NSR amongst lactococci was r ecentl y emphasized when whole genome sequencing of 710 dairy-associated L. lactis strains found that an impr essiv e 270 (38%) harboured an nsr gene (van Gijtenbeek et al. 2021 ). Furthermore, a singular nis I immunity gene located on a 50-kb plasmid w as sho wn to provide nisin resistance to the non-nisin producer L. lactis NCDO712 (Tar azanov a et al. 2016 ). These studies highlight the potential for r esistance tr ansfer to other gener a or species within a shar ed envir onmental nic he. Ultimatel y, NGS tec hnologies will facilitate r esistance-guided genome mining, whic h in combination with sequence-based functional metagenomics will not only assist in establishing how pr e v alent nisin and lantibiotic r esistance determinants are within microbiome populations, but also impr ov e the prospects for the discovery of genetic variants or new structurally related homologs of known resistance mechanisms. For example, a whole-genome analysis and evaluation of clinical isolates of C. difficile established a link between strains exhibiting high or low nisin resistance to genetic variants in the cpr ABC nisin resistance module (Ide et al. 2023 ). Such knowledge will be invaluable to accelerate the development of strategies that could eventually counteract or avoid nisin resistance action. Several approaches ha ve been in vestigated in this regard including the application of nisin in combination with other antimicrobials including different bacteriocins and antibiotics, particularly those with different modes of action or that inter act syner gisticall y in a bid to target MDR pathogenic targets more effectively (for comprehensive reviews see Mathur et al. 2018, Soltani et al. 2021a. Bioengineering str ategies ar e also being exploited and se v er al engineer ed v ariants have been described that can effectively evade some nisin immunity and resistance mechanisms (see later). If nisin is to ac hie v e mor e widespr ead ther a peutic use, it is critical that r esistance be taken into consideration at every stage of development.

Genome mining and new nisin variants
The exponential increase in genomic data derived from metagenomic sequencing of microbial communities has led to the availability of vast amounts of genetic information to probe for novel lantibiotic operons of interest. The highly conserved sequences inher entl y found within lantibiotic biosynthetic genes can be utilized as driver sequences to identify areas of a genome that may contain other novel lantibiotic-like BGCs. For example, an in silico screen for BGCs with homology to the nisin A biosynthetic genes nisB and nisC resulted in the identification of more than 49 novel lantibiotic clusters across a range of bacterial species, genera, and phyla not previously linked with lantibiotic production (Marsh et al. 2010 ). Ho w e v er, a wide variety of genome mining tools have since been de v eloped (BAGEL, AntiSMASH, RiPP-Miner, RiPP-PRISM, and RODEO) that have become the pr eferr ed and fastest means for the discovery of novel RiPP BGCs (Russell and Truman 2020 ). Consequently, the availability of high-quality genome sequence data combined with these po w erful bioinformatic softwar e pac ka ges has gr eatl y expanded our knowledge of the variety and distribution of BGCs, particularly those from human and animal gut microbiomes (Drissi et al. 2015, Walsh et al. 2015. Indeed, giv en the str engthening association between the gut microbiota and human health and disease (Bull and Plummer 2014 ), bacteriocin producing strains have attracted significant attention as potential micr obiome-sha ping tools that could be used in the pr e v ention or tr eatment of diseases associated with gut pathogens (Mousa et al. 2017, Heilbronner et al. 2021. Genomic mining has doubled the number of natural nisin variants now c har acterized with the majority of these having been sourced from human, animal, and insect microbiomes. Indeed, only a decade ago, just seven natural variants were known: nisin A, nisin Z, nisin Q, and nisin F produced by L. lactis species, nisin U and nisin U2 produced by Streptococcus uberis , and a putative nisin P cluster was identified in Streptococcus pasteurianus (Field et al. 2015a ).

Nisin H
In 2015, Nisin H was the first variant isolated from a mammalian gastr ointestinal tr act, in this case, that of a pig (O'Connor et al. 2015 ). Genome analysis of Streptococcus hyointestinalis r e v ealed that it differs from nisin A by five amino acids (Fig. 2 B). Despite this, nisin H retains k e y features of the lactococcal peptides. Another pr ominent featur e was the absence of a corresponding nisI immunity gene within the gene cluster (O'Connor et al. 2015 ).

Nisin P
Although a nisin P gene cluster was pr e viousl y noted in the genome of S . pasteurianus (Zhang et al. 2012 ), more recent studies have demonstrated production of nisin P by a clinical isolate of Streptococcus gallolyticus (AB39) and the purified peptide displayed antibacterial activity towards se v er al drug-r esistant bacteria, including MRSA, VRE, and penicillin-resistant Streptococcus pneumoniae (Aldarhami et al. 2020 ). A nisin P gene cluster was also shown to be present in both a porcine isolate of Streptococcus suis (Wu et al. 2014 ) as well as a strain of S. agalactiae isolated from human faeces and was the first such example of a Group B str eptococcal str ain to pr oduce a nisin v ariant (Garcia-Gutierr ez et al. 2019b ).

Nisin O
In silico screening of human gut bacterial genomes identified a nisin BGC in Blautia obeum A2-162, a dominant species in the human colon (Hatziioanou et al. 2017 ). Nisin O is unusual in that it is the first nisin cluster to encode 4 peptides, the first 3 (NsoA1-3) are indistinguishable and resemble nisin U, while the fourth exhibits the gr eatest de viation fr om nisin A. Mor eov er, the gene cluster contains two sets of nis RK genes but sur prisingl y no corresponding nis P gene could be identified. The presence of aromatic residues at the first position, as is the case with NsoA1-3 ( Fig. 2 B), has been shown to significantly reduce cleavage efficienc y b y the protease NisP  ). The NsoA peptides exhibited str ong antimicr obial activity a gainst C. difficile and Clostridium per-fringens following heter ologous expr ession in L. lactis , but only in the presence of trypsin (Gherghisan-Filip et al. 2018 ).

Nisin BP SCSK
A m ultipeptide nisin oper on was also pr esent in a Blautia species obtained from the faecal microbiota of mice. Blautia producta BP SCSK produces a nisin-like peptide variant with similarity to nisin O. In contrast ho w ever, BP SCSK encodes five lantibiotic precursor genes ( lan A1-lan A5) (Fig. 3 ). The first four are identical while the fifth is more divergent. Notably, the antimicrobial activity of BP SCSK w as sho wn to be comparable to nisin A a gainst str ains of VRE and other Gr am-positiv e nosocomial pathogens but displayed reduced potenc y to w ar ds other gut commensals (Kim et al. 2019 ). Nisin BP SCSK differs from nisin A at residues corresponding to I4K, K12V, A15I, G18Dhb, N20P, M21V, K22Q, H27G, and a five-amino acid tail consisting of 29 QIDhbGK 33 (Fig. 2 B). Se v er al of these modifications are located at positions corresponding to bioengineered nisin variants with altered antimicrobial activity and spectrum of antibacterial activity and/or target sites for the digestive enzymes trypsin/ α-chymotrypsin including I4, K12, N20P, M21, K22, and S29Q (see later sections on nisin bioengineering/modulation of gut microbiota).

Nisin J
Nisin J r epr esents the first nisin v ariant to be pr oduced by a sta phylococcal species (O'Sulliv an et al. 2020 ). Identified from the toe web space during a screen of the human skin microbiota, Staphylococcus capitis APC 2923 produces a nisin variant with nine residues differing from those in nisin A, as well as having an extra residue at the C-terminus, making it a 35-residue peptide (Fig. 2 ). Inter estingl y, some of these amino acid alter ations corr espond to bioengineered nisin A variants with enhanced activity (see later section on bioengineering), most notably those pertaining to ring A (I4K and L6I) and ring C (M17Q and G18Dhb). Indeed, nisin J was shown to exhibit greater activity against staphylococcal species compared to its A and Z counter parts, pr ompting the suggestion that the nisin J producer may have naturally evolved to produce a peptide with enhanced activity against other skin-associated sta phylococci (O'Sulliv an et al. 2020 ). Unusuall y, the gene or ganization of the nisin J cluster also differs from that of other nisin clusters in that the two-component regulatory system ( nisRK ) and the nisin immunity gene nisI are absent (Fig. 3 ).

Nisin E
The most recent stre ptococcal-deri ved nisin variant is nisin E, produced by multiple Streptococcus equinis str ains, originating fr om a canine oral cavity as well as sheep gut (Christophers et al. 2023, Sugrue et al. 2023. Nisin E differs from nisin U b y tw o residues (I15A and L21I), but also possesses an extra C-terminal asparagine residue (Asn32). Despite the relatively high homology between the peptides , nisin E displa yed r educed activity a gainst a wide panel of tar get or ganisms, but most especiall y a gainst se v er al str eptococcal species ( Streptococcus mitis , Streptococus gordonii , and Streptococcus anginosus ) compared to its nisin U counterpart, prompting the authors to speculate that the I15A and additional aspar a gine residue could be responsible (Christophers et al. 2023 ). Notably, the residues of the hinge region of nisin E ( 20 PIK 22 ) represent a novel hinge sequence compared to the other closel y r elated nisin U, nisin U2, and nisin P variants (Fig. 2 B). Additionally, while the genetic organization of the nmd locus essentially matches the nisin U ( nsu ) cluster, the nisin E structural gene ( nmd A) is situated

Nisin G
Adding to the growing list of nisin-producing streptococci ( S. uberis , S. hyointestinalis , S. gallolyticus , S. suis , S. agalactiae , and S. equinis ) is Streptococcus salivarius DPC6487, which was sourced from a neonatal faecal sample and produces nisin G. Although S. salivarius strains have been shown to produce another lantibiotic, salivaricin D (Birri et al. 2012 ), this varies greatly from nisin, distinctly lacking the last MeLan ring. The nisin G pe ptide di v er ges fr om nisin A with respect to seven amino acids (Fig. 2 B) and was shown to exhibit a more limited spectrum of activity when compared to nisin A, with activity confined to other streptococci but more notably to Fusobacterium spp , including Fusobacterium nucleatum . Fusobacterium nucleatum is an emerging human pathogen linked to se v er al gut-associated disorders including colorectal cancer (CRC) de v elopment. The desir e for intestinal-deriv ed bacteriocin pr oducing str ains that can kill specific target organisms without causing collateral damage to host bacterial populations makes S. salivarius DPC 6487 a potential candidate for probiotic development (Lawrence et al. 2022 ).

Kunkecin A
The nisin variant Kunkecin A (Zendo et al. 2020 ) was identified from a honeybee isolate Apilactobacillus kunkeei FF30-6, a fructophilic lactic acid bacterium r ecentl y c har acterized as one of the major components in the gastrointestinal tract of honeybee queens and larvae (Endo and Salminen 2013 ). Kunkecin A rep-resents the longest natural variant described to date, possessing fiv e extr a amino acids at the C-terminus compar ed to nisin A (Fig. 2 B). Despite displaying an ov er all narr ow antimicr obial spectrum, kunk ecin A was re ported to exhibit superior acti vity over nisin A against several bacteria originating from honeybees, including Melissococcus plutonius , the causative agent of European foulbrood, a global honeybee brood disease. Crucially, many other honeybee commensals including lactobacilli and bifidobacteria a ppear ed less sensitive to this nisin variant, prompting the authors to suggest the kunkecin A producer as a potentially useful probiotic to inhibit honeybee pathogens in apiaries.
There is mounting evidence that bacteriocins produced by micr obial r esidents of the gut impart a competitive adv anta ge and play a major role in shaping niche competition among intestinal bacteria (Dobson et al. 2012 ). While some bacteriocins display a narr ow r ange of activity, tar geting onl y closel y r elated members of the same species, others like nisin display a broader spectrum of activity. Recent microbiome in vestigations ha ve begun to elucidate the impact of nisin on the gastr ointestinal micr obiota (outlined below) and r e v eal in more detail members of the microbiome that are susceptible or resistant to its action. Given the contrasting antimicrobial activities of the natural nisin variants when compared to nisin A as discussed abo ve , it is tempting to speculate that these variants have evolved because of localized competition with specific microbes in their respective en vironments . Indeed, the potential for more precise targeting of individual pathogens whilst at the same preserving the integrity of the microbial composition are very desirable properties in light of the important role of the microbiome in overall human health.

Nisin bioengineering to modulate antimicrobial activity and spectrum of antibacterial activity
The inexorable proliferation of MDR pathogens has significantly impacted on the effectiveness of commonly used antibiotics (Gupta and Datta 2019 ). Consequentl y, ther e is an urgent need for ne w antimicr obial compounds as well as nov el deriv ativ es of curr ent antimicr obials that specificall y tar get MDR bacteria and/or other pr oblematic or ganisms. A consequence of the gene encoded nature of nisin is the r elativ e ease with whic h ne w structur al variants can be created through genetic manipulation. Moreover, the diversity of the natural variants and related homologues emphasizes the tolerance to changes of particular residues and regions within the peptide. Indeed, their activity and physicochemical properties would suggest they hav e e volv ed to kill specific or a narro w er range of targets, and thus could be viewed as templates for new, tailor-made and specific targeting peptides.
Over the last two decades, banks of engineered nisin deri vati ves have been generated and characterized. Simple single or multiple modifications and chimeric molecules are beginning to furnish a blueprint of those residues and domains essential for structureactivity r elationships-not onl y r elating to nisin biosynthesis, but also in terms of antimicrobial activity and spectrum, solubility, heat stability, challenges to immunity or resistance proteins as well as sensitivity to pr oteol ytic enzymes . T he following section will focus on r ecent adv ances r egarding the implementation of bioeng ineering strateg ies to enhance the functional c har acteristics of nisin and the most notable successes ac hie v ed as a consequence of employing these v arious str ateg ies. For prior bioeng ineering studies, we direct the reader to a number of comprehensiv e r e vie ws (Lubelski et al. 2008, Field et al. 2015a, Shin et al. 2016. Nisin can be structur all y divided into an N-terminal region (composed of a lanthionine ring A and the methyllanthionine rings B and C, a hinge region, and a C-terminal region; with the intertwined methyllanthionine rings D and E) (Fig. 2 A). The impact of m uta genesis on each of these regions will be discussed in terms of impr ov ed functional c har acteristics.

N-terminus
A systematic satur ation m uta genesis a ppr oac h at the N-terminal isoleucine (Ile1) and analysis of the resulting 20-generated analogues r e v ealed that although the cor e peptide was completel y modified, impacts on production, leader peptide cleav a ge, and antimicrobial activity varied drastically and correlated with the nature of the amino acid substituted in each case (i.e . aliphatic , aromatic, c har ged, and so on) (La gedr oste et al. 2019 ). Results revealed that aromatic amino acids at position one (I1W and I1F) gave rise to superior antimicrobial activity (Fig. 4 ), particularly a gainst lactococcal str ains expr essing the nisin imm unity and r esistance proteins nisI, nisFEG, NSR, and NsrFP. In contrast, polar or c har ged amino acids triggered a reduction in activity, highlighting the influence small changes in the peptide structure can have on activity and ability to circumvent nisin resistance systems. Moreover, the majority of Ile1 variants were subject to NisP processing albeit with lo w er efficienc y , the exception being I1P , whic h r emained intractable to NisP activity.
As one of the earliest locations targeted for mutagenesis (K uipers et al. 1995 , W iedemann et al. 2001 ), position two variants provided the first evidence that nisin activity could be impr ov ed when the threonine at position 2 in nisin Z (a residue , i.e . normall y conv erted to dehydr obutyrine) was c hanged to a ser-ine/dehydroalanine and displayed enhanced activity against two nonpathogenic target organisms . More recent in vestigations ha ve highlighted the importance of Thr2 within the lipid II binding motif of nisin when a bioengineer ed c himeric lantibiotic found to be more potent against VRE was rendered inactive following a T2D substitution (Zhao et al. 2020b ). Notably, in a tar geted a ppr oac h to identify nisin peptides more suited for bo vine mastitis applications , a threonine to leucine variant (T2L) (Fig. 4 ) w as sho wn to exhibit exce ptional antimicrobial acti vity against a selection of bovine mastitis-associated staphylococci  ), but were noticeably less active against many of the commensal lactic acid bacteria found in milk such as lactococci and lactobacilli.

Rings A and B
Early engineering attempts targeted at the N-terminal ring A concerned a S3T substitution (which would change the first Lan residue to MeLan) that led to a dramatic loss of activity (Wiedemann et al. 2001 ). The amino acid composition of ring A appears quite variable as is evident upon inspection of the contrasting ring arrangements in several natural nisin variants (Fig. 2 B) and nisinlik e pe ptides . T his diversity at positions 4-6 is in a region, i.e. at the border of the pyrophosphate cage involved in the mechanism of action of nisin (Hsu et al. 2004 ), underscoring the large mutational freedom available and may thus represent suitable targets for m uta genesis. Indeed, two v ariants gener ated via satur ation m uta genesis at positions 4-6 corresponding to 4 KSI 6 and 4 KFI 6 ( Fig. 4 ) displayed impr ov ed activity a gainst se v er al nonpathogenic indicator strains (Rink et al. 2007 ). Notably, several natural nisin variants as well as novel nisin-lik e pe ptides (agalacticin, flavucin, moraviensicin, and maddinglicin) (van Heel et al . 2016 ) possess a lysine at position 4 (I4K) (Fig. 2 B). Ad ditionally, a 4 VFG 6 deri vativ e r etained str ong antimicr obial activity but suffered a dramatic loss in its autoinduction ca pacity. Mor eov er, these v ariants had the ability to escape the self-immunity proteins of NisI and/or Nis-FEG, proving to be toxic to a nisin pr oducing str ain (Rink et al. 2007 ). Mor e conserv ed r esidue c hanges suc h as the nisin I4V v ariant (Fig. 4 ) demonstrated improved antimicrobial as well as potent antibiofilm activity a gainst se v er al str ains of Staphylococcus pseudintermedius (Field et al. 2015b ). These enhanced antimicrobial properties were further extended to include S. aureus when I4V was combined with conventional antibiotics (Field et al. 2016 ).
The highl y conserv ed ring B is composed of a MeLan and the amino acids proline (Pro9) and glycine (Gly10). Mutational analysis has underpinned its importance in both antimicrobial and induction activity (Rink et al. 2007, Ge et al. 2016. For example, the 9 PT 10 v ariant r etained compar able induction ca pacity and antimicrobial activity as nisin A, while 9 PH 10 , 9 PR 10 , 9 PD 10 , 9 PN 10 , and 9 PL 10 , display ed significantly lo w er induction and antimicrobial activities due to incomplete ring formation (Rink et al. 2007 ). More r ecentl y, a ring B variant where both residues are replaced with threonine ( 9 TT 10 ) (Fig. 4 ) and designated nisin M, retained full autoinduction capability but exhibited up to more than 16-fold less activity against several genera and species of bacteria (O' Connor et al. 2020a ), emphasizing the lack of a direct correlation between lipid II binding and induction capacity.
The residue Lys12 is located between rings B and C and a sitesatur ation a ppr oac h at this location led to the identification of a small number of bioengineered deriv ativ es with impr ov ed activity (Molloy et al. 2013 ) (Fig. 4 ). Indeed, more recent NMR analyses confirmed the importance of Lys12 as a 'pharmaceutical hotspot' by acting as a flexible region that permits nisin to adopt a

Ring C
Although the precise function of ring C has not yet fully been elucidated, this MeLan ring has been shown to be critical for the biological activity of nisin (van Kraaij et al. 2000 ) and suggests that this region is involved in very specific interactions. For example, converting the thioether bond of ring C to a disulphide bond was found to abolish activity completely (van Kraaij et al. 2000 ). Recent studies have revealed the unique membrane-interacting properties of ring C and have implied a structural link between the arrangement of the hinge in tandem with ring C conformations in the ability to form por es (Medeir os-Silv a et al. 2018 ). Mutational anal ysis of r esidues within ring C has been beneficial with respect to enhanced functional deri vati ves as was observed by L16A, L16H, and L16V (Fig. 4 ) that displayed a slight increase in both antimicrobial activity and induction capacity (Ge et al. 2016 ). More r ecent r eports hav e demonstr ated the enhanced specific activity of other ring C variants against pathogenic strains, as when nisin M17Q pr ov ed to be mor e effectiv e than nisin A at reducing Staphylococcus epidermidis biofilms from medical de vice-r elated materials as well as significantly reducing viable cells in simulated wound fluid experiments (Twomey et al. 2020 ). Nisin M17Q (Fig. 4 ) was also shown to exhibit enhanced activity against some strains of bovine mastitis-associated S. aureus  and displayed antilisterial activity when used in combination with other bioengineered nisin deri vati ves (Nyhan et al. 2021 ), which was sustained in model food experiments.

The hinge region
The hinge consists of a 3-amino acid linker region, i.e. critical for antimicrobial activity by providing conformational flexibility between the N-and C-termini of nisin. Following interaction between the two N-terminal rings with lipid II, the flexible hinge region facilitates insertion of the nisin C-terminal domain into the bacterial membrane to form a por e (Br eukink et al. 1999, Wiedemann et al. 2001. Bioengineering of the hinge r egion pr ovided the first reports of deri vati ves with improved activity against Gram-negative pathogenic targets (Yuan et al. 2004 ), which was followed shortly ther eafter by impr ov ed v ariants a gainst Gr am-positiv e or ganisms (Field et al. 2008 ). An a ppr oac h involving simultaneous randomization of all three hinge residues in nisin A ( 20 NMK 22 ) generated a suite of peptides whereby a preference for small, chiral amino acids was linked to increased bioactivity (Healy et al. 2013 ). Additionally,s a nisin peptide incorporating a hinge region composed of 20 HTK 22 (Fig. 4 ) r epr esents a nov el v ariant  ) to add to those pr e viousl y identified as having impr ov ed antimicr obial activity ( 20 AAK 22 , 20 NAI 22 , and 20 SLS 22 ) (Healy et al. 2013 ).
Furthermore, the impact of expanding or reducing hinge length on the antimicrobial activity and target spectrum of nisin has also been explored. Zhou et al. ( 2015 ) revealed that both shortened hinge peptides ( −1 amino acid) and extended hinge peptides ( + 2 amino acids) displayed impr ov ed efficacy as determined by growth inhibition assays against several target strains including L. lactis , Enterococcus faecalis , L. monocytogenes , and B. cereus , but were tar get or ganism-and temper atur e-dependent due to v ariations in bacterial membr ane composition. Similarl y, Zasc hke-Kriesc he et al. ( 2019b ) examined the adv anta ge of decreasing ( 21 MK 22 ) or extending the hinge region ( 20 NMKIV 24 and 20 NIVMK 24 ). These hinge v ariants wer e impacted in their ability to form pores when compared to wild type nisin A, in particular the truncated 21 MK 22 peptide, and when assessed a gainst str ains expr essing the nisin immunity (NisI and NisFEG) and nisin resistance determinants (SaNSR and SaNsrFP), the v ariant 20 NMKIV 24 displayed incr eased activity, possibly as a result of reduced substrate recognition (Zasc hke-Kriesc he et al. 2019b )

C-terminal rings and tail
The C-terminus of nisin A is essential for pore formation and consists of the highly conserved rings D and E follo w ed b y a six-amino acid tail ( 29 SIHVDhbK 34 ). NMR studies have revealed Ser29 within this region as an important flexible region for the orientation of nisin within the membrane pore (Medeiros-Silva et al. 2018 ). Indeed, a site-satur ation a ppr oac h r e v ealed that bioengineer ed m utants at this location extended nisin activity to w ar ds some Gramnegative species (Field et al. 2012 ).
Recent investigations have highlighted the importance of ring D and E for recognition by the NSR (Khosa et al. 2016a ). Specificall y, the r esistance pr ovided b y NSR w as over come b y nisin variants lacking rings D and E or only ring E. This was further emphasized by molecular simulations of the NSR/nisin complex, which r e v ealed the r ole of the C-terminal rings for substrate specificity to ensure the exact coordination of the nisin cleavage point (Ser29) at the enzymatic active site (Khosa et al. 2016b ). Considering this, Field et al. ( 2018 ) used saturation mutagenesis such that serine at position 29 was replaced with all other 19 amino acids . T he results identified one deri vati ve, S29P (Fig. 4 ), as having comparable activity to nisin A but crucially exhibited a 20-fold increase in specific activity a gainst se v er al NSR pr oducing str ains due to the inability of NSR to cleave the mutant peptide . Furthermore , another vari-ant termed nisin PV (S29P and I30V substitutions)) pr ov ed to not only be as active as S29P, but to be more stable by virtue of being less prone to oxidation . Alternativ el y, r eplacement of Cys 28 with pr oline (C28P) r esulted in a nisin A peptide that was lacking ring E and retained a more structurally rigid and smaller ring (Zasc hke-Kriesc he et al. 2019a ). Nisin C28P (Fig. 4 ) was as active as wild type nisin and retained the ability to form pores in the membrane, but was notably more effective against an NSR pr oducing str ain due to the inability of NSR to cleave this variant efficientl y (Zasc hke-Kriesc he et al. 2019a ).
Ther e ar e just a handful of nov el antibiotic compounds in curr ent de v elopment that tar get Gr am-negativ e bacteria (Imai et al. 2019 ). The outer membrane of Gram-negative organisms is impermeable to nisin, pr e v enting access to the inner membrane and its target lipid II (Nikaido and Vaara 1985 ). Consequently, nisin exhibits poor activity to w ar ds Gr am-negativ e species. Ho w e v er, disruption of the outer membrane with chelating agents such as EDTA can r estor e susceptibility to nisin (Ste v ens et al. 1991 ). T hus , the main impediment for nisin to kill Gr am-negativ e bacteria appears to be its inability to tr av erse the outer membrane. A r ecent a ppr oac h to assist nisin in penetr ating the outer membr ane involv ed the fusion of se v er al small anti-Gr am-negativ e peptides tails including apidaecin 1b, oncocin, and drosocin to the C-terminal end of nisin (Zhou et al. 2016 ). One such fusion, containing an eight amino acid (PRPPHPRL) tail from apidaecin 1b attached to full length nisin (Fig. 4 ), exhibited 2-fold greater activity against Escherichia coli CECT101, highlighting that the strategy of combining a lipid II binding peptide with the penetr ativ e ability of eukaryotic antimicrobial peptides extends the activity of nisin to w ar ds Gr am-negativ e bacteria (Zhou et al. 2016 ). Subsequent studies involved a greater selection of peptide tails and reengineer ed v ariants and identified se v er al of them with gr eater activity against a range of important pathogenic Gram-negative organisms including E. coli , Klebsiella pneumoniae , Acinetobacter baumannii , Pseudomonas aeruginosa , and Enterobacter aerogenes when compared to nisin alone (Li et al. 2018 ). Moreover , bioengineering a ppr oac hes to include differ ent linker domains as well as a C-terminal supplementary lysine residue (to mimic many lantibiotics that possess a positiv el y c har ged amino acid at the Cterminus) pr ov ed beneficial in mor e effectiv el y tar geting Gr amnegative pathogens (Fig. 4 ).

Bioengineering modular nisin analogues and semisynthetic hybrids
More than 100 lanthipeptides have now been identified that exhibit enormous structural diversity; especially in terms of posttranslational modifications including thioether-based intramolecular rings, unusual dehydroamino and unsaturated amino acids as well as a variety of flexible linker regions that when combined, are critical for stability and biological activity (Dischinger et al. 2014 ). Despite this diversity, a common feature of lantibiotics is the ability to bind lipid II and form pores in the membrane, a blueprint that encompasses different functional elements or modules. Importantly, lantibiotic peptides vary substantially in terms of activity to w ar ds target strains. While bioengineering str ategies hav e been crucial for adv ancing our knowledge with respect to structure-activity relationships and have pr oduced deriv ativ es with enhanced properties, in the main these studies have focused on the diversification of individual lantibiotic peptides which can deliver only limited structural novelty. A recent study sought to investigate the consequences of the molecular shuffling of functional modules (i.e. a lipid II binding domain, a flexible domain, and a pore-forming domain) from 12 different and well-c har acterized class I and class II lantibiotic peptides including nisin, gallidermin, Pep5, Lacticin 481, mersacidin, and cinnamycin that also included synthetic linker modules to imitate or replace the hinge region of nisin (Schmitt et al. 2019 ). A library of over 6000 combinatorial variants was generated and attached to the leader peptide of nisin and expressed using the nisin biosynthetic machinery NisBTC in L. lactis . Following library screening, 11 ne w to natur e peptides wer e found to exhibit impr ov ed antimicr obial activity compared to their wild-type counter parts against a panel of Gr am-positiv e pathogens including S. pneumoniae , MRSA, VRE ( E. faecalis and E. faecium ) as well as L. lactis strains harbouring the nisin immunity cluster ( NisIFEG ) and the NSR. Notably, peptides consisting of combinatorial modules of gallidermin and nisin joined by a synthetic hinge were most active against the majority of pathogenic strains tested as well as being able to bypass the nisin immunity system (Fig. 5 ). Ad ditionally, pe ptide deri vatives containing atypical hinge and C-terminus modules (such as those found in Pep5, lactocin S, paenibacillin, and so on) pr ov ed to be more active against NSR clea vage , highlighting the importance of the C-terminus of nisin for NSR recognition (Schmitt et al. 2019 ) (Fig. 5 ).
The targeting of lipid II, crucial for bacterial cell wall synthesis and r eadil y accessible on the outer surface of the cell membr ane, is an effectiv e and ancient antimicr obial str ategy (Ulm and Schneider 2016 ) and an important target for potent antibiotics including vancomycin, teixobactin, and several lantibiotics. Rings A and B contain the lipid II binding motif of nisin, i.e. conserved among other lantibiotics including subtilin, epidermin, gallidermin, microbisporicin, and mutacin 1140, and have been the focus of se v er al str ategies to gener ate nov el lipid II tar geting compounds (Koopmans et al. 2015, Deng et al. 2020, Zhao et al. 2020b. One suc h a ppr oac h involv ed the c hemical coupling of functional moieties to N-terminal nisin ring fr a gments (AB, ABC ring systems) using click chemistry. The addition of membrane active lipids to the nisin AB ring fragment yielded semisynthetic analogues with enhanced stability (Koopmans et al. 2015 ) while the conjugation of synthetic hydrophobic polyproline peptides to rings ABC produced nisin hybrids that displayed increased antimicr obial activity a gainst E. f aecium when compar ed to the nisin ABC fr a gment alone, though they pr ov ed to be 8-fold less active than full length nisin (Deng et al. 2020 ). Mor eov er, two nisin hybrids could bypass the NSR and were less prone to pr oteol ytic degr adation.
In another a ppr oac h, nov el hybrid peptides were designed consisting of two different lipid II binding regions, one composed of the N-terminal region fragments of nisin (1-10, 1-11, 1-17, 1-25, and so on) linked to the C-terminal lipid II binding regions of the two-component lantibiotics lacticin 3147 or haloduracin (Zhao et al. 2020b ). A suitable linker region was included between each of the domains. Of 20 such hybrid peptides that were expressed using the nisin biosynthetic machinery, two showed potent antimicr obial activity a gainst Micrococcus flavus . Further anal ysis of one such hybrid, TL19 (composed of nisin + haloduracin lipid binding motifs), displayed 64-fold higher potency against E. faecium compared to the single lipid II binding component of nisin1-22, highlighting the potential of a single molecule with two lipid II binding motifs to enhance antimicrobial activity (Zhao et al. 2020b ).
Recentl y, a cell-fr ee pr otein synthesis (CFPS) system using E. coli cell extracts was employed for the production of novel variants of nisin Z (Liu et al. 2020 ). This r equir ed some adjustments such as increasing the levels and ratio of NisB and NisC in conjunction with the precursor peptide for optimal efficiency. Fol-lo wing a sear c h of all publicl y av ailable genomes for nisin analogues , 18 no v el cor e peptide v ariants wer e subsequentl y linked to the nisin leader peptide and expressed in the CFPS system. In total, four variants displayed antimicrobial activity and following further purification exhibited good activity a gainst E. f aecalis , S. aureus , and MRSA, with the RL14 variant exhibiting superior activity compared to nisin Z against E. faecalis . Notably, the residue c hanges acr oss the four peptides included I4K and I4V , K12V , A15V and A15I, M17Q, and G18Dhb, a shortened hinge ( 20 AL 21 , 20 AI 21 , and 20 PI 21 ) and a varied C-terminal tail. An additional library of 3000 variants was generated via saturation mutagenesis targeting positions 4, 12, 15, 24, and 29 of nisin Z and expressed in the CFPS system. Subsequent screening identified a further two variants with impr ov ed activity a gainst Gr am-negativ e bacteria compared to parental nisin Z (Liu et al. 2020 ).
In a bid to identify novel peptides that specifically target C. difficile , a Clostridium genome mining a ppr oac h w as emplo y ed that scr eened ov er 600 publicl y av ailable genomes. A total of 10 putati ve lanthipe ptide genes wer e identified in Clostridium beijerinc kii , Clostridium ihumii , and C. perfringens and expressed in L. lactis using the nisBCT synthetase (Cebrián et al. 2019 ). When all but one failed to pr oduce antimicr obial activity, a synthetic biology a ppr oac h was undertaken whereby a new series of hybrid lantibiotics were designed and expressed that were composed of Ntermin us or C-termin us modules of the putative clostridial peptides fused to the N-terminus or C-terminus modules of nisin. An active hybrid peptide with good antimicrobial activity against a C. difficile strain was isolated but was notably less active against other clostridial str ains. Inter estingl y, the peptide was not active a gainst C. beijerinc kii , the str ain fr om whic h the N-terminus module used in the hybrid was composed (Clos AB + nisin CDE). The authors note that the specificity and activity observed for the hybrid peptide makes it an interesting potential candidate in the treatment of C. diffic ile infections , a voiding side effects and protecting the normal gut microbiota (Cebrián et al. 2019 ).

Nisin as a modulator of the gut microbiota
It is now becoming incr easingl y clear that the gut microbiome is integral to human health (Mousa et al. 2017, Garcia-Gutierrez et al. 2019a. Indeed, maintaining a healthy gut microbiome could be considered as a new therapeutic target since any significant imbalance could contribute to the de v elopment or pr ogr ession of diseases such as inflammatory bo w el disease , diabetes , obesity and infection by intestinal pathogens such as C. difficile (Duan et al. 2022 ). Enteric diseases in farm animals are also a cause of significant economic losses in the agribusiness sector . Furthermor e, tr eating suc h diseases accounts for considerable antibiotic use in both humans and animals that not only has the potential to adv ersel y affect the composition and function of the gut microbiome, but also encourages the selection of antibioticr esistant pathogens. It would, ther efor e, be adv anta geous to be able to modulate the gut microbiota to selectiv el y tar get or deplete undesir able micr obes without impacting the beneficial micr obes. Bacteriocins ar e gaining credibility as precise modulators of the human microbiome (O'Connor et al. 2020b ). Indeed, sever al r ecent studies pr ovide compr ehensiv e e vidence that bacteriocins and bacteriocin-producing bacteria can be used to modify the gut microbiota, making them an attr activ e str ategy to addr ess gut-related diseases and disorders (Guinane et al. 2016, Umu et al. 2016, Lawrence et al. 2022. Bacteriocins could address some of the issues associated with conventional antibiotics and significant r esearc h is ongoing to elucidate an y r ole they could play within the gastrointestinal tract, their ability to inhibit pathogens as well as any potential functions in safeguarding host health (Heilbronner et al. 2021 ). Several recent studies have investigated the capacity of nisin, either by direct application or secreted by in situ bacteria, to influence the gut microbiota and/or to control specific pathogens associated with c hr onic intestinal diseases. For example, it was evident that nisin, when supplemented in the diet of poultry, can beneficially modulate the gastrointestinal ecology and enhance growth performance (Józefiak et al. 2013, Kieronczyk et al. 2017, 2020. Similar effects wer e observ ed in rabbit models (Lauková et al. 2014 ). Likewise, nisin treatments elicited significant changes in the gastrointestinal microbiota in a bacterial diarrhoea mouse model by incr easing favour able species such as Lactobacillus , Bacteroides , and Bifidobacterium while reducing pathogenic strains of E. coli and Enterococcus spp (Jia et al. 2018 ). Mor e r ecentl y, when nisin was used as a feed ad diti ve in aquaculture it w as sho wn to alter the diversity and composition of the intestinal microbiota of common carp (Ke et al. 2021), bream (Moroni et al. 2021, and flounder (Nguyen et al. 2018 ).
Nisin has also been considered for its therapeutic potential in targeting the gut pathogen C. difficile . Antibiotic use is a major risk factor for de v eloping a C. difficile infection as a result of changes to the gut microbiota that allow C. difficile to proliferate and cause r ecurr ent C. difficile -associated diarrhoea and intestinal inflammatory disease (collectiv el y designated C. difficile infection or CDI). The standard treatment for CDI is use of the antibiotics fidaxomicin, metr onidazole, or v ancomycin, although none of these antibiotics are fully effective (Czepiel et al. 2019 ). In compar ativ e studies carried out with nisin, v ancomycin, and metr onidazole a gainst > 60 clinical C. difficile isolates, the results r e v ealed that nisin was the most effective of the three antimicrobials tested. Nisin was more potent as observed by its overall MIC90 of 0.256 mg/l while both vancomycin and metronidazole had MIC90s of 0.8 mg/l ( Bartoloni et al. 2004 ). Furthermore, nisin A w as sho wn to significantly reduce C. difficile spore viability in liquid suspension following 1 hour of treatment (Lay et al. 2016 ).
To e v aluate nisin in conditions that mor e closel y mimic the GIT and as a consequence observe the effects on the total micr obiota, r esearc hers hav e explor ed its ability to inhibit or kill C . difficile in model colon systems (Le Lay et al. 2015, O'Reilly et al. 2022. Nisin A (in the form of Nisaplin ® ) proved to be an effective inhibitor of C. difficile, achieving over a 100-fold reduction in cell numbers when used at 76 μmol/l (equivalent to 20X MIC) in a simulated human colon system (consisting of a cell immobilized and continuous fermentation single-stage reactor to simulate the pr oximal, tr ansv erse, and distal colons; Cinquin et al. 2004, Le Lay et al. 2015. The treatment also brought about significant alterations to the microbial composition whereby Grampositive bacteria were notably disturbed. Indeed, Ruminococcaceae were most impacted in that they underwent an almost 4-log r eduction, while Lac hnospiraceae , Lactobacillaceae , Leuconostocaceae groups as well as bifidobacteria were less affected (Le Lay et al. 2015 ). Although an increase in the Gram-negative population ( Bacteroidetes and Enterobacteriaceae) was also noted, the initial balance was r estor ed 24 hours after nisin ad dition. In the same stud y, a L. lactis nisin pr oducer, whic h was shown to persist and pr olifer ate in the model colon, was ineffective against C. difficile , most likely due to the inability of the strain to generate a sufficient quantity of peptide in these conditions to inhibit the pathogen (Le Lay et al. 2015 ).
A r ecent r eport emplo y ed the micro-Matrix 24-w ell parallel controlled cassette minifermentation system as a batch colon model (as described by O'Donnell et al. 2018 ) to investigate the efficacy of pure nisin A to kill C. difficile in a dose-dependent manner. Importantly, it was established that a concentration of 50 μM was sufficient to completely eradicate C. difficile whilst simultaneously eliciting the least impact on the ov er all micr obial composition (O'Reilly et al. 2022 ).
Enterococcus spp are also commonly found in the human GIT (Vankerc khov en et al. 2004 ). Worryingly, some enterococci, in particular E. faecalis and E. f aecium ar e a leading cause of hospitalacquired infections, as a result of developing resistance to many frontline antibiotics including vancomycin (Miller et al. 2014 ). Antibiotic administration disrupts the normal gut microbiota balance, leading to expansion of VRE in the intestinal tr act, pr edisposing vulnerable patients to bloodstream infections . T he situation is so challenging that VRE have been earmarked by the World Health Organization as a critical target for the identification and de v elopment of ne w antimicr obials and novel approaches to combat infections (AbdelKhalek et al. 2018 ). Notably, a recent study r e v ealed that administr ation of a four str ain coc ktail consisting of Clostridium bolteae , Bacteroides sartorii , B. producta , and Parabacteroides distasonis could reduce the density of enterococci in the colon of infected mice and r estor e r esistance to VRE infection (Caballero et al. 2017 ). One of the consortia, B. producta (BP SCSK ) was found to inhibit VRE by production of the nisin variant, nisin BP SCSK . Mor eov er, this v ariant exhibited r educed activity a gainst intestinal commensal bacteria including Bifidobacterium longum and Pediococcus acidilactici (Kim et al. 2019 ). Notably, substitution of the nisin BP SCSK producer for a L. lactis nisin pr oducing str ain in the consortium failed to inhibit VRE in vivo , most likely as a result of its inability to colonize the gut. This study elegantly demonstrates that nisin expression by a commensal microorganism can influence niche competition in the gastrointestinal tract by preventing pathogen colonization and with minimum disruption to the microbiota.
A major limitation with respect to the therapeutic use of nisin is its sensitivity to pr oteol ysis by intestinal enzymes. Nisin A has been shown to be degraded by pepsin, trypsin, and chymotrypsin (Heinemann and Williams 1966, Jarvis and Mahoney 1969, Slootweg et al. 2013 ). More recent investigations involving sim ulated or al, gastric, and small intestinal digestion trials also r e v ealed the pr oteol ytic degr adation of nisin A (Gough et al. 2017 ) and nisin Z (Soltani et al. 2021b ). Indeed, despite the low pH environment (pH 3), nisin Z remained mostly intact under gastric conditions with only minor degradation fragments detected, but the peptide was significantly degraded after 2 hours under conditions r epr esenting the small intestine (pH 7). Mor eov er, in addition to some of the expected breakdown products of proteolysis, oxidized forms of nisin were detected in the untr eated, or al and gastric fractions (Soltani et al. 2021b ), most likely from oxidation of the methionine residue (M21) located in the hinge region of the peptide (Rollema et al. 1996 ). Pr e vious findings have shown that the degree of oxidization in nisin peptides leads to a reduction in antimicrobial acti vity (Yone yama et al. 2008 ), highlighting other potential factors that could play a role in the efficacy of nisin.
In order to bypass the sensitivity of nisin to pr oteol ytic cleava ge, enca psulation has been investigated as a means of facilitating tr ansit thr ough the GIT. Indeed, in vitro studies hav e found that a pectin/HPMC (hydr oxypr opyl methyl cellulose) env elope is suitable for delivery of nisin to the colon (Ugurlu et al. 2007 ). Similarly, encapsulation of nisin using two different resistant starch-based matrices was investigated for delivery to the lo w er GIT and its re-lease in a controlled manner. Analysis of faecal pellets of mice fed the encapsulated nisin revealed the presence of fully intact and functional peptides, though the concentration detected was matrix dependent (Gough et al. 2018 ). Finall y, a v ariety of silica-based mesopor ous matrices wer e examined for nisin enca psulation and found to protect nisin A from the pr oteol ytic action of pepsin in bior ele v ant media, highlighting the suitability of this a ppr oac h to shield nisin from any degrading enzymes in the intestinal tract (Flynn et al. 2019 ). Other novel means of providing proteolytic resistance could take the form of the chemical coupling of specific synthetic and nonproteinaceous moieties to nisin that can provide the desired resistance to specific proteases (Deng et al. 2020 ).
Inter estingl y, nisin A contains three trypsin sensitive residues; K12, N20, and K22 while N20, M21, and H31 ar e c hymotrypsin sensitiv e r esidues (Slootweg et al. 2013 ). T hus , it is notable that many of the amino acids corresponding to trypsin/ α-chymotrypsin 'target' sites i.e. K12, N20, M21, and H31 hav e been r eplaced in the gut-deriv ed natur al nisin v ariant nisin O, and ar e completel y r eplaced with nontar get r esidues in nisin BP SCSK , (K12V, N20P, M21V, K22Q, and H31Dhb), a response that suggests adaption to their environment in the gut. Moreover, the reduced activity of nisin BP SCSK against some beneficial commensal bacteria suggests these r esidues ar e excellent tar gets for bioengineering not onl y to maintain protection from disintegration in the gastrointestinal tract but also to enhance activity to w ar ds specific pathogenic organisms whilst having minimum impact on the commensal microbiota resident in the gastrointestinal tract.

Perspectives
In the face of the antimicrobial resistance crisis, researchers are struggling to identify new antibiotic classes. Nisin has been studied since the earliest days of the antibiotic era, and although it has found global success as a biopr eserv ativ e, its m ultiple modes of action, potent activity against MDR microbes and long safety record has meant that the focus on nisin-related research is shifting from food preservation to w ar ds therapeutic use for the treatment of bacterial infections.
The modular nature of nisin and other lantibiotics and the dev elopment of expr ession systems to r eorder the extensiv e r ange of thioether-derived ring structures at random (akin to a plug 'n' play system with provision for an almost inexhaustible array of structur al perm utations) is a thrilling ne w pr ospect for nisin and lantibiotic r esearc h. T his , and adv ances in widening the substr ate specificity of existing modification enzymes will undoubtedly lead to ne w structur es with enhanced functional c har acteristics (specific activity, target spectrum including Gram-negative targets, diffusion, solubility, and impr ov ed gastr ointestinal stability) and with them a portfolio of potential ther a peutic a pplications.
T he disco v ery and c har acterization of ne w natur al v ariants fr om differ ent envir onments, including the gastrointestinal tract of humans and animals, is welcome and likely to accelerate with the advent of cheaper and faster genome sequencing technologies and finely tuned genome mining tools. Indeed, the current rate of discovery emphasizes the broad distribution of nisin-related BGCs across bacterial species, implying a strong role in Grampositive bacterial competition within a variety of microbiomes. Furthermor e, the mounting e vidence of the beneficial role of the gut microbiome in human and animal health will ensure that greater efforts will focus on the ability of nisin and bioengineered nisin deri vati ves to eliminate specific pathogens and pathobionts in a predictable and beneficial manner that could steer antimicr obial r egimens to w ar ds mor e personalized and pr ecise medical methodologies and pr e v ent indiscriminate microbiome damage.
We can expect that r esistance de v elopment will follow any introduction of bacteriocins as new therapeutic tools. Although bioengineering pr ogr ammes hav e alr eady identified nov el nisin v ariants or peptide hybrids able to overcome some resistance mechanisms, sensible stew ar dship, as w ell as limiting and tailoring applications will ensure the therapeutic success of nisin and minimize future resistance development. Despite almost a century of nisin-related research, no significant investment on the part of Pharma companies is evident in relation to the development of nisin as a drug suitable for human use. Ho w e v er, the fact that nisin has r eac hed a point wher e it is under consider ation for use as a veterinary pharmaceutical in the treatment of bovine mastitis may provide the impetus for incr eased inv estment for human clinical applications . T he successful clinical development of nisin will r equir e cr eativ e impr ov ements in its bioav ailability , stability , solubility under physiological conditions, and other parameters including pharmacokinetics and pharmacodynamics. Man y knowledge ga ps ar e still e vident. Futur e studies will aim to provide a more complete picture of nisin biosynthesis and the membr ane-associated m ultimeric complex, including the interactions between the transporter and protease components that may aid in the expression of a broader range of substrates. Furthermore, a better understanding of the nisin imm unity mec hanisms and their as yet unconfirmed co-oper ativ e natur e in providing full immunity will be required in efforts to increase nisin production, as will a more in-depth analysis of the cell wall precursor target and membrane interactions with nisin to rationalize the membr ane disruptiv e a ggr egation behaviour r ecentl y described.
Given the immense strides and tec hnological impr ov ements in nisin-and bacteriocin-related research in recent y ears, w e have already passed a significant threshold, and it is anticipated that the tremendous potential for the de v elopment and a pplication of tailor-made and highly specific nisin deri vati ves for in vivo antibiotic use will finally be realized at the beginning of the second century of nisin r esearc h.

Funding
This w ork w as supported b y the Science Foundation Ireland under grant number SFI/12/RC/2273.