Nisin J, a Novel Natural Nisin Variant, Is Produced by Staphylococcus capitis Sourced from the Human Skin Microbiota

This study describes the characterization of nisin J, the first example of a natural nisin variant, produced by a human skin isolate of staphylococcal origin. Nisin J displays inhibitory activity against a wide range of bacterial targets, including MRSA. This work demonstrates the potential of human commensals as a source for novel antimicrobials that could form part of the solution to antibiotic resistance across a broad range of bacterial pathogens.

the human skin microbiota that sought to identify novel antimicrobial-producing strains (5). This strain was of particular interest due to its potent activity against the indicator strain Lactobacillus delbrueckii subsp. bulgaricus LMG 6901 and its broad inhibitory spectrum against a panel of Staphylococcus, Streptococcus, and Corynebacterium species and against Cutibacterium acnes. Whole-genome sequencing of this strain revealed a nisin gene cluster of ϳ9.78 kb compared to ϳ13.3 kb for nisin A. The structural gene nisJ encodes a peptide with the following eight amino acid variations compared to nisin A: Ile4Lys, Met17Gln, Gly18Thr, Asn20Phe, Met21Ala, Ile30Gly, Val33His, and Lys34Thr. Nisin J also contains an extra amino acid at the C terminus, making nisin J the longest nisin variant identified to date (Fig. 1A). A dendrogram of the natural nisin variants (Fig. 1B) demonstrates that peptides which have a closer common ancestor are more similar than are peptides than have more distant branching points. Lactococcal nisin variants are structurally distinct from all other nisin variants. Staphylococcal nisin J groups in the middle of the tree and appears to be more similar to streptococcal nisin than to lactococcal nisins. Nisins of Blautia origin appear to be more phylogenetically distinct due to longer branching. Streptococcal nisins H and J are more closely related to lactococcal nisins than to other streptococcal nisins, U, U 2 , and P. The gene order of the nisin J cluster (FEGBTCJP) also differs from that of the nisin A   FIG 1 (A) Visualization of the multiple-sequence alignment from MUSCLE (plotted using http://msa.biojs.net/app/) of all natural nisin (nis) variants aligned with strain origin. The total height of the sequence logo at each position reflects the degree of conservation at that position in the alignment, while the height of each letter in that position is proportional to the observed frequency of the corresponding amino acid at that position. Nisin A (13), nisin Z (48), nisin F (49), nisin Q (50), nisin H (27), nisin J (5), nisins U and U 2 (51), nisin P (52,53), and nisins O 1 to O 4 (54) are shown. L., Lactococcus; S., Staphylococcus; B., Blautia; St., Streptococcus. (B) Dendrogram showing phylogenetic relatedness in primary structures of all known natural nisin variants, suggesting the possible existence of an evolutionary link between the nisin-producing species. The order in which they branch shows the relatedness between them, and the branch length represents phylogenetic distance (0.05 represents a scale for the phylogenetic distance).
in that it contains eight as opposed to the 11 genes within the cluster (Fig. 2). The BAGEL4 bacteriocin genome mining tool predicted that the nisin J prepeptide is composed of 61 amino acids with a leader sequence consisting of 26 amino acids. Overall, the nisin J mature peptide has 62.5% identity to the nisin H structural peptide produced by Streptococcus hyointestinalis (27). The identity and function of features of the nisin J operon are listed in Table 1.
Other genes contained in the S. capitis APC 2923 draft genome. In addition to the nisin J cluster, BAGEL4 and antiSMASH3.0 also highlighted a small gene cluster containing the lanB and lanC genes and a gene encoding a peptide with 93% identity to the gallidermin family in S. capitis APC 2923. These were located on a different contig from that of the nisin J gene cluster, and this mass was not detected from either the colony or purified cell free supernatants.
Purification and predicted structure of nisin J. Nisin J was purified in four steps using Amberlite XAD-16N solid-phase extraction (SPE), SP Sepharose cation exchange, C 18 SPE, and reversed-phase high-performance liquid chromatography (HPLC). Antimicrobial activity correlated with the most dominant peak eluting at 24.5 min in the HPLC chromatogram, and matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) revealed that the corresponding fractions had a mass of 3,458 Da (Fig. 3). This correlates with the predicted mass of the putative nisin J bacteriocin (following subsequent dehydration and ring formation reactions) as calculated from the draft genome sequence. Fractions deemed pure by MALDI-TOF MS were combined and lyophilized to give a yield of 3.00 mg/liter. Given that nisin J is a natural nisin variant with demonstrable conservation between key structural amino acids common to all natural nisin variants, it is predicted that the structure will be in line with those of other lactococcal nisins, as shown in Fig. 4.
Comparing the activities of purified nisins A, Z, and J. The spectrum of activity of pure nisin A, nisin Z, and nisin J, by means of a well diffusion assay (WDA), was performed on several target indicator strains. Nisin J was more active than nisin A against 12 of the 13 strains tested, while nisin J was more active than nisin Z for 7 of the target strains tested, including Corynebacterium xerosis, MRSA, Streptococcus uberis, and S. aureus (Table 2). However, in an MIC assay using L. lactis HP as the indicator, no difference was observed between nisins A, Z, and J, with all exhibiting MICs of 32 nM.
The nisin J-producing strain is cross-immune to nisin A and H but not to nisin U producers. Cross-immunity assays were performed to investigate whether the nisin J-, A-, H-, and U-producing strains were cross-immune to one another (Table 3). No zones were observed between nisins A, H, and J, indicating that these producing strains are all cross-immune. However, a zone was observed from the nisin J-producing strain against the nisin U producer (S. uberis strain 42), demonstrating that the strain is sensitive to nisin J. Not all S. capitis strains contain a nisin-like gene cluster. The nisJ structural gene was amplified from nine antimicrobial-producing S. capitis strains isolated from human skin in a previous study by our group (5). Two of the nine S. capitis strains (APC 2918 and APC 2934) did not contain the nisJ structural gene. The other seven S. capitis strains tested positive for the nisJ structural gene, correlating with findings from our earlier study which found these strains to be cross-immune and to possess the same pulsotype, indicating that they were the same strain or very closely related strains and were therefore most likely producing the same bacteriocin (5). These 7 strains were isolated from 4 different subjects, indicating that the same pulsotype is shared across a number of individuals, implying that the ability to produce nisin J may be a dominant feature and thus an ecological advantage for this S. capitis strain. The nisin J gene cluster resides on a plasmid. Analysis of the S. capitis APC 2923 contig harboring the nisin J gene cluster identified the presence of a plasmid replication protein A (RepA) and other plasmid replication-associated proteins, suggesting that it was of plasmid origin. Plasmid DNA was readily obtained from S. capitis APC 2923 using a commercially available plasmid maxi kit (data not shown). Short-read sequencing was performed on the plasmid DNA using the Illumina MiSeq platform to approximately 200-fold coverage. De novo assembly resulted in four contigs (Fig. 5), with a combined size of 49,951 bp. A plasmid map of pJOS01 (GenBank accession number MN602039) shows all of the genes encoding immunity and the biosynthetic machinery for nisin J (nsjFEG, nsjB, nsjT, nsjC, nisJ, and nsjP) reside on one of the contigs, supporting   Table 1). Restriction digestion with EcoRI yielded a profile comparable to the virtual digestion of the generated plasmid sequence, supporting the predicted size of ϳ50 kb (data not shown). Subsequent analysis revealed a GC content of ϳ28%, which is considerably lower than that of S. capitis chromosomal DNA (32 to 33%), a characteristic that has been observed for plasmids of many Gram-positive species (28). Nisin J exhibits resistance to NSR. Deferred antagonism assays using L. lactis subsp. diacetylactis DRC3 (nisin resistance protein positive [NSR ϩ ]) as a target indicator strain revealed that nisin J is partially resistant to NSR (result not shown). To establish if nisin J had increased inhibitory activity against NSR compared to that of nisin A, further WDAs were conducted using the NSR ϩ and NSR Ϫ strains L. lactis MG1614/ pNP40 and L. lactis MG1614, respectively. While the inhibition zone of the nisin J producer is slightly decreased against the NSR-positive strain compared to the NSRnegative strain, it appears that nisin J is more active than nisin A and may be less susceptible to the proteolytic effects of NSR (Fig. 6A), which was also demonstrated in agarose assays (Fig. 6B). The analysis revealed a significant difference in the zones of inhibition between nisin A and nisin J against an NSR ϩ strain (MG1614/pNP40), with a P value of 0.0001 compared to zone sizes against an NSR Ϫ strain (MG1614), where no statistical difference (P ϭ 0.1701) was observed (these data support Fig. 6).

DISCUSSION
As the burden of antibiotic resistance increases globally, there is an urgent need for novel therapeutic options. In addition to the well-established use of nisin as a food  preservative, many studies have focused on using nisin against drug᎑resistant pathogens in clinical or veterinary settings due to its high potency and multiple mechanisms of action (10 -12). Nisin J is a novel nisin variant and the first such variant reported from a Staphylococcus species. A combination of whole-genome sequencing of S. capitis APC 2923 and peptide purification resulted in the identification of this broad-spectrum lantibiotic. The nisin J-producing S. capitis strain was isolated from the toe web space, an area associated with high microbial load. This suggests that the production of a broad-spectrum bacteriocin confers an advantage on this strain over competing commensal skin flora, as was also observed by O'Sullivan and colleagues (5) when four of the twenty subjects screened in the study exhibit the same pulsotype. The residence of the nisin J gene cluster on a plasmid is significant in that it may facilitate its dissemination to other skin microbes. As mentioned previously, nisin J has eight amino acid changes and one extra amino acid near the C-terminal end compared to nisin A. Interestingly, six of the eight changes are unique compared to natural nisin variants. Natural nisin variants are tolerant to some amino acid changes at the N terminus, with Ile4 being the most commonly . At position 20, nisin J has a highly hydrophobic residue, phenylalanine, compared to the polar asparagine in nisin A. Li et al. (30) found that extending the C terminus of nisin improves both its ability to permeate membranes and its inhibitory potential against Gram-negative bacteria. Therefore, nisin J's longer C terminus (compared to other nisin variants) could be more attracted to negatively charged cell membranes resulting in enhanced membrane insertion, which may be responsible for its broader host range. The skin origin of this nisin J producer suggests that its exposure to many competitors from the external environment may be responsible for the greater variation in the structure of nisin J. Analysis of the nisin J gene cluster identified several key features associated with bacteriocin operons. These include a structural gene (nisJ), 2 genes associated with enzymatic modification (nsjB and nsjC), a gene involved in transport (nsjT), and immunity genes (nsjFEG) ( Table 1 lists the identity and functions of features of the nisin J gene cluster). The arrangement of genes in the nisin J gene cluster differs from that of other nisin operons. Interestingly, the only conservation of gene order throughout all operons of natural nisin variants is lanBTC. Similarities in the structural peptides of different nisin variants from different origins indicate the possibility that an evolutionary link exists between lactococcal, streptococcal, Blautia, and now, staphylococcal species, a link previously mentioned by O'Connor et al. (27) with reference to streptococcal and lactococcal species. A dendrogram based on the primary structures of all known natural variants highlights the genetic relatedness between the nisin-producing species and further suggests the likelihood of this evolutionary link. The FEG locus is present in lantibiotic systems other than nisin, including subtilin (31) and epidermin (32), and has been linked to transport, immunity, and defense (33). Inactivation of these genes in the nisin A gene cluster decreased nisin production and immunity, confirming their role in immunity (34). Although the nsjFEG genes are present in the nisin J gene cluster, the absence of a specific immunity gene, nsjI, as well as the absence of an expression regulatory system, nsjRK, could explain why nisin J immunity mechanisms appear to be less able to protect the cell. It also further supports the finding that the producing strain was more sensitive to its own purified nisin J peptide than was a nisin A producer with a specific nisin immunity determinant.
The production of lantibiotics such as gallidermin and epidermin is associated with increased release of lipids and ATP and protein excretion, which are indicators of cell membrane damage (35). Thus, the production of these lantibiotics has been deemed a "burden" to staphylococci that produce them; therefore, the incomplete lantibiotic gene cluster, having only the lanB and lanC genes present, may be either an evolutionary feature of S. capitis genomes or may be an incomplete cluster of lantibiotic biosynthetic genes previously shown to occur in many microbes (35).
As previously discussed, the nisin J gene cluster resides on a plasmid, inviting the speculation that S. capitis acquired its antimicrobial ability through horizontal gene transfer. Indeed, residence on mobile genetic elements is a feature of natural nisin variants, as observed with nisins A and H, and may explain their presence in many different species.
Purification of nisin J resulted in a peptide with a mass of 3,458 Da. The mass of nisin J was predicted to be 3,622 Da, where the difference between predicted and observed masses can be accounted for by 9 dehydration reactions (Ϫ18 Da per loss of water residue) involved in the formation of lanthionine and ␤-methyllanthionine bridges (36). The predicted peptide structure was based on the nisin A template, with a lanthionine bridge likely to occur between Ser3 and Cys7 and four ␤-methyllanthionine bridges between Thr8 and Cys11, Thr13 and Cys19, Thr23 and Cys26, and Thr25 and Cys28.
True to all nisin variants, nisin J is a broad-spectrum lantibiotic with inhibitory activity similar to that of nisins A and Z, as can be seen in Table 2, inhibiting a wide range of bacterial genera with greater inhibition of staphylococcal targets than with nisins A and Z. This suggests that the nisin J-producing S. capitis strain may have naturally evolved to produce a nisin peptide with enhanced activity against other staphylococci in the skin microbiota (Table 2). Nisin J-, A-, and H-producing strains are immune to nisin peptides J, A, H, and U; however, the nisin U-producing strain is not immune to nisin J (Table 3). This may be due to the lack of the nsjI immunity gene in the nisin J cluster.
The nisin resistance protein (NSR) is a protease which cleaves nisin A at Ser29, significantly reducing the activity of the peptide. Employing a bioengineering strategy, Field et al. (37) demonstrated that the substitution of residues 29 and 30 with proline and valine, respectively (derivative designated S29PV), rendered the peptide resistant to proteolytic digestion by NSR. In this study, we found that the nisin J producer displays a higher resistance to NSR proteolytic enzymes than does nisin A, which is possibly due to a glycine residue at position 30 instead of the isoleucine as found in nisin A. Interestingly, a study carried out by Simões et al. (38) involving a multidrugresistant S. capitis clone, NRCS-A, a major pathogen involved in sepsis in preterm neonates, demonstrated the presence of an NSR-encoding gene. PCR analysis failed to detect the presence of any nsr gene in any nisin J-producing S. capitis strain from our previous study (5).
Nisin J may have evolved to be more potent against specific competing organisms in a particular niche environment such as the skin. Employing a bioengineering strategy, Rink et al. (39) demonstrated that the replacement of residues I, S, and L at positions 4, 5, and 6 in nisin A with the residues K, S, and I, respectively, resulted in enhanced bioactivity. Notably, the residues K-S-L are naturally present in nisin J at the same positions. In a separate bioengineering study, Kuipers et al. (29) generated a novel nisin variant (M17Q/G18T) exhibiting enhanced bioactivity. It is interesting that both of these mutations are naturally present in nisin J. Furthermore, Field et al. (40) reported that a nisin A derivative, M21A, demonstrated enhanced bioactivity. Remarkably, alanine is naturally present at position 21 in nisin J.
In conclusion, we have identified a new natural nisin variant, nisin J, produced by S. capitis APC 2923, which was isolated from the human skin microbiota. Nisin J represents the first nisin variant isolated from Staphylococcus species and the first to demonstrate partial recalcitrance to NSR. Indeed, the enhanced activity of nisin J compared to that of nisin A and Z as observed against all staphylococcal strains utilized in this study is notable. The production of bacteriocins such as nisin J from skin bacteria highlights the potential of bacterial strains of skin origin to be used as live biotherapeutics.

MATERIALS AND METHODS
The antimicrobial-producing strain S. capitis APC 2923 was isolated in a previous screening study of the human skin microbiota by our group (5).
Bacterial strains and culture conditions. The growth conditions of the bacterial strains used in this study are listed in Table 4. Anaerobic conditions, where appropriate, were attained using anaerobic jars and Anaerocult A gas packs (Merck, Darmstadt, Germany).
Draft genome sequence of S. capitis APC 2923 and in silico analysis of the nisin J gene cluster. Bacterial DNA was extracted using the GenElute kit, as described by the manufacturer (Sigma-Aldrich Ireland Limited, Arklow, County Wicklow, Ireland), and was prepared for sequencing following the Nextera XT DNA library prep reference guide (Illumina, Inc.). A Qubit 3.0 fluorometer (Thermo Fisher Scientific, MA) was used for DNA quantification. Sequencing was performed at the Teagasc/APC Microbiome Ireland Sequencing facility, Teagasc Food Research Centre, Moorepark, Fermoy, County Cork, Ireland. In total, 94 contigs, including 16 large contigs, were revealed by de novo assembly using SPAdes (version 3.10.0). A total of 2,453 open reading frames (ORFs) and 60 tRNAs were detected and subsequently annotated using Prokka (version 1.11). The online tools Bacteriocin GEnome mining tooL (BAGEL4) and antiSMASH 3.0 were employed to identify bacteriocin operons/gene clusters in the genomes of interest, and by combining these software programs with the ARTEMIS genome viewer, the presence of the nisin J gene cluster was confirmed.
Evolutionary links between natural nisin variants. The European Bioinformatics Institute toolkit (https://www.ebi.ac.uk/services) was used to investigate the evolutionary relationships between the nisin structural variants. A multiple-sequence alignment was generated using MUSCLE (version 3.8) and visualized on a neighbor-joining tree without distance corrections. This tree was visualized using the ggtree package (version 1.10.5) in R (version 3.4.4).
Purification of the antimicrobial produced by S. capitis APC 2923. To purify the antimicrobial produced by S. capitis APC 2923, the culture was grown in a shaking 37°C incubator overnight in 1,800 ml Comparison of the inhibitory spectra of nisins A, Z, and J. Pure nisins A, Z, and J were resuspended in RNase-free water to a final concentration of 1 mg/ml and subsequently assayed by WDA against a range of target indicator strains (Table 2). Zone diameters were measured in millimeters using Vernier calipers (DML-Digital Micrometers Ltd., Sheffield, United Kingdom) and recorded in Table 2 as area of the zone (r 2 ) minus the area of the well (r 2 ) in millimeters.
MIC determinations. MICs were determined in triplicate from pure nisins A, Z, and J against approximately 1 ϫ 10 5 CFU/ml of the target indicator strain Lactococcus lactis subsp. cremoris HP using 96-well microtiter plates (Sarstedt, Co. Wexford, Ireland) and using a Libra S2 colorimeter (Biochrom Ltd., Cambridge, United Kingdom) to measure the optical density at 600 nm (OD 600 ) of the indicator strains. Peptide concentrations of 4ϫ the test concentration (2,048 nM) were prepared in 400 l RNase-free and DNase-free water. One hundred microliters of growth medium was added to all wells of the 96-well plate. One hundred microliters of 4ϫ concentration was added to the first well, and subsequently, 2-fold serial dilutions were carried out. MIC readings were taken after 16 h at 30°C. The MIC was recorded as the lowest concentration of lantipeptide where no growth of the indicator was observed (42).
Cross-immunity of nisin J-producing S. capitis APC 2923 to other nisin-producing strains. To investigate if the nisin J-producing S. capitis APC 2923 strain was immune to other nisin-producing cultures (L. lactis NZ9700 producing nisin A, Streptococcus hyointestinalis DPC 6484 producing nisin H, and S. uberis strain 42 producing nisin U), cross-immunity assays were performed based on the WDA method, whereby each strain was tested as an indicator and a producer (43).
Determining if the nisin J structural gene is unique to S. capitis APC 2923. To determine if the nisin J structural gene was present in other S. capitis strains isolated from the study by O'Sullivan et al. (5), oligonucleotide primers designed to specifically amplify the nisin J structural gene (nisJ F, 5=-ACTT TATAACTAAGATTAGC-3=, and nisJ R, 5=-TCGCTTTATTATTTAGTATGCACG-3=) were used in a PCR under the following conditions: initial denaturation, 94°C for 5 min; 35 cycles of 94°C for 40 s, 52°C for 30 s, and 72°C for 1 min; and a final extension 72°C for 10 min. Sequencing was conducted by Genewiz (Essex, United Kingdom). Sequencing data were analyzed employing the Lasergene 8 software (DNAStar, Inc., Madison, WI) and subsequently input into the ExPASy online translate tool (https://web.expasy.org/translate/) to translate the nucleotides into amino acid sequences.
Sequence analysis of the nisin J plasmid pJOS01. To confirm that the nisin J gene cluster was plasmid associated, the plasmid DNA of S. capitis APC 2923 was extracted using the Plasmid maxi kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions following an adapted userdeveloped protocol specific to staphylococcal species (https://www.qiagen.com/ie/resources/ resourcedetail?idϭ82ddd661-fbab-4d35-819c-defd6269fc64&langϭen), using lysostaphin (Sigma-Aldrich Ireland Limited, Arklow, County Wicklow, Ireland). The resulting DNA extract was sequenced by Illumina MiSeq technology (2 ϫ 250-bp paired-end reads; GenProbio, Parma, Italy). De novo sequence assemblies and automated gene calling were performed using the MEGAnnotator pipeline (44) and assessed for predicted tRNA genes via transcend-SE version 1.2.1 (45). Predicted open reading frames (ORFs) were determined via Prodigal version 2.6 and Genemark.hmm (46). A BLASTP (47) analysis was performed to assign functional annotations to the predicted ORFs (https://blast.ncbi.nlm.nih.gov/Blast .cgi) ( Table 1). PlasmidFinder (version 2.0) was employed to confirm that the generated assembled contigs were plasmid sequences based on the identification of Rep proteins. SnapGene version 2.3.2 was employed to generate a map of the plasmid harboring the nisJ gene cluster (designated pJOS01 here). In addition to the sequence data analysis to confirm the plasmid association of the nisin J cluster, PCR-based analysis was undertaken using the plasmid DNA extract as the template. Oligonucleotide primers designed to specifically amplify the nisin J structural gene (nisJ F, 5=-ACTTTATAACTAAGATTAG