Insights in the Antimicrobial Potential of the Natural Nisin Variant Nisin H

Lantibiotics are a growing class of antimicrobial peptides, which possess antimicrobial activity against mainly Gram-positive bacteria including the highly resistant strains such as methicillin-resistant Staphylococcus aureus or vancomycin-resistant enterococci. In the last decades numerous lantibiotics were discovered in natural habitats or designed with bioengineering tools. In this study, we present an insight in the antimicrobial potential of the natural occurring lantibiotic nisin H from Streptococcus hyointestinalis as well as the variant nisin H F1I. We determined the yield of the heterologously expressed peptide and quantified the cleavage efficiency employing the nisin protease NisP. Furthermore, we analyzed the effect on the modification via mass spectrometry analysis. With standardized growth inhibition assays we benchmarked the activity of pure nisin H and the variant nisin H F1I, and their influence on the activity of the nisin immunity proteins NisI and NisFEG from Lactococcus lactis and the nisin resistance proteins SaNSR and SaNsrFP from Streptococcus agalactiae COH1. We further checked the antibacterial activity against clinical isolates of Staphylococcus aureus, Enterococcus faecium and Enterococcus faecalis via microdilution method. In summary, nisin H and the nisin H F1I variant possessed better antimicrobial potency than the natural nisin A.

The best studied lantibiotic is nisin, which was first discovered in 1928 by Rogers and Whittier and belongs to the class I lantibiotics (Rogers, 1928;Rogers and Whittier, 1928;Arnison et al., 2013). It is used in the food industry since 1953 and obtained the status as generally recognized as safe (GRAS) in 1988 from the Food and Drug Administration (FDA) (Delves-Broughton et al., 1996). The class I lantibiotic nisin contains 34 amino acids and five (methyl)-lanthionine rings. These (methyl)-lanthionine rings require multiple posttranslational modifications (PTMs) which are introduced in the precursor peptide. First, the serine and threonine residues in the core peptide are dehydrated by a specific dehydratase NisB (lantibiotic class I LanB dehydratase) (Kaletta and Entian, 1989;Karakas Sen et al., 1999;Koponen et al., 2002;Ortega et al., 2015;Repka et al., 2017). The next step is a Michael-type condensation of dehydrated residues with the thiol group of a cysteine residue, thereby forming (methyl)-lanthionine rings, guided in a regio-and stereospecific manner by the cyclase NisC (class I lantibiotic cyclase) (Okeley et al., 2003;Li et al., 2006;Li and van der Donk, 2007;Repka et al., 2017). These characteristic (methyl)-lanthionine rings give lantibiotics high heat stability, resistance against proteolytic digestion and are responsible for the nanomolar antimicrobial activity (Gross and Morell, 1967;Rollema et al., 1995;Chan et al., 1996;Lu et al., 2010;Oppedijk et al., 2016).
The sequence of nisin A can be subdivided into three parts (see Figure 1). The N-terminal part with ring A, B, and C is responsible for binding to the cell wall precursor lipid II (Hsu et al., 2004). The hinge region is very flexible and allows reorientation of the C-terminal part to insert into the membrane (van Heusden et al., 2002;Hasper et al., 2004;Wiedemann et al., 2004;Medeiros-Silva et al., 2018), while changes in this region have a strong impact on the target antimicrobial activity (Zhou et al., 2015;Zaschke-Kriesche et al., 2019b). After penetrating the membrane, the C-terminal part with ring D and E forms a stable pore with a stoichiometry of eight nisin and four lipid II molecules, which subsequently leads to rapid cell death (Hasper et al., 2004;Wiedemann et al., 2004;Alkhatib et al., 2014a;Medeiros-Silva et al., 2018).
Unfortunately, some bacteria established resistance mechanism against lantibiotics. For instance lantibiotic producing strains express the immunity system LanI and LanFEG (Alkhatib et al., 2014a,b), which prevent a suicidal effect after the lantibiotic is secreted. In the case of nisin A from Lactococcus lactis these proteins are called NisI and NisFEG. But also non-lantibiotic producing strains showed resistance against lantibiotics like Streptococcus agalactiae COH1, which arises from the expression of the membrane-anchored peptidase SaNSR and the ABC transporter SaNsrFP (Khosa et al., 2013(Khosa et al., , 2016aReiners et al., 2017).
In this study we focused on the natural nisin H variant (Figure 1). We used a standardized workflow for the characterization of lantibiotics, previous described in Lagedroste et al. (2019) to determine the impact on the expression, modification and antimicrobial properties of this nisin variant. We tested further the antimicrobial activity against some pathogen strains from Staphylococcus aureus, Enterococcus faecium, and Enterococcus faecalis using the microdilution method. As a reference we used the wild-type version of nisin A expressed using the same experimental setup. Furthermore, we exchanged the phenylalanine at position one (F 1 ) of nisin H to isoleucine, which is the natural amino acid of nisin A at this position (Figure 1). This position one was previously analyzed in nisin A and showed a major impact on different levels of the characterization (Lagedroste et al., 2019).

Cloning of Nisin H and the F 1 I Variant
Nisin H was created as described in Reiners et al. (2017). The substitution of the phenylalanine at position one (F 1 I) to an isoleucine was performed by site-directed mutagenesis. Here, we used the following primers (forward: 5 -GTGCATCACCACGCTTTACAAGTATTTCGC-3 ; reverse: 5 -GCGAAATACTTGTAAAGCGTGGTGATGCAC-3 ). After sequence analysis a competent L. lactis NZ9000 strain was transformed with the resulting plasmid via electroporation (Holo and Nes, 1989). The L. lactis NZ9000 strain already contain a vector (pil3-BTC) encoding for the modification and secretion proteins (Rink et al., 2005).

Expression, Purification and Activation of Pre-nisin Variants
The precursor of nisin H and its variant were expressed and purified as previously described (Abts et al., 2013;Alkhatib et al., 2014b;Lagedroste et al., 2017Lagedroste et al., , 2019. Briefly, for pre-nisin secretion, the L. lactis strain NZ9000 was grown in minimal medium (Jensen and Hammer, 1993) supplemented with 0.5% glucose and 5 µg/ml of each erythromycin and chloramphenicol at 30 • C. Cells were induced with 10 ng/ml nisin at an OD 600 of 0.4 and further grown overnight at 30 • C without shaking. After harvesting the cells, the 0.45 µm-filtered supernatant was loaded onto a HiTrap SP HP cation exchange chromatography column. After washing with 50 mM lactic acid, the buffer was changed to 50 mM Hepes pH 7 via gradient and the final elution was done with 50 mM Hepes pH 7, 500 mM NaCl buffer. Elution fractions were concentrated in a 3 kDa cutoff filter. With a soluble version of NisP, the activation of all variants was performed overnight at 8 • C . The yield and cleavage efficiency determination was done by RP-HPLC (Agilent Technologies 1260 Infinity II) with a LiChrospher WP 300 RP-18 end-capped column and an acetonitrile/water solvent system (Abts et al., 2013;Lagedroste et al., 2017Lagedroste et al., , 2019.

MALDI-TOF Analysis: Dehydration and (Methyl)-Lanthionine Ring Analysis
With MALDI-TOF analysis we analyzed the modification state of nisin H and its variant. Dehydrations are directly visible in the spectra, due to the loss of mass (−18 Da). For the determination of the presence of (methyl)-lanthionine rings, we used the organic coupling agent CDAP (1-cyano-4 dimethylaminopyridinium tetrafluoroborate) that binds to free cysteine residues (Wu and Watson, 1998). The reaction of the coupling agent to free cysteine side chains would results in an increased mass in the spectra. The analysis was performed as previously descripted (Lagedroste et al., 2019). The samples were analyzed with MALDI-TOF (UltrafleXtreme, Bruker Daltonics, Bremen, Software: Compass 1.4) in positive mode.

Tandem Mass Spectrometric Analysis of Nisin H and Nisin H F 1 I
Nisin H and the nisin H F 1 I variant were purified using solid phase extraction (Oasis HLB, Waters) and finally resuspended in 0.1% trifluoroacetic acid. The samples were first subjected to liquid chromatography on a rapid separation liquid chromatography system (Ultimate 3000, Thermo Fisher Scientific) using an 1 h gradient and C18 columns as described (PMID 24646099) and further analyzed by a QExactive Plus mass spectrometer (Thermo Fisher Scientific) coupled via a nano-source electrospray interface. First, a precursor spectrum was acquired at a resolution of 140,000 (advanced gain control target 3E6, maximum ion time 50 ms, scan range 200-2000 m/z, profile data type). Subsequently, up to four 4-6-fold charged precursors were selected by the quadrupole (2 m/z isolation window), fragmented by higher-energy collisional dissociation (normalized collision energy 30) and recorded at a resolution of 35,000 (advanced gain control target 1E5, maximum ion time 50 ms, available scan range 200-2000 m/z, centroid data type).
Recorded spectra were analyzed by the MASCOT search engine (version 2.4.1, Matrix Science) and searches triggered by Proteome Discoverer (version 2.4.0.305, Thermo Fisher Scientific). A database was generated for the searches including 1000 randomly generated sequence entries each 34 amino acid long) and the sequences of nisin H and nisin H F 1 I. Methionine oxidation and dehydration of serine and threonine residues were considered as variable modifications and the precursor mass tolerance set to 10 ppm and the fragment mass tolerance set to 0.02 Da. For peptide validation, the Fixed Value PSM validator was used (1% false discovery rate) and the IMP-ptmRS node for site validation (PMID 22073976). No random sequences were found by the search.

Determination of the Antimicrobial Activity by Growth Inhibition Assay
The determination of the antimicrobial activity of the different nisin variants was tested using a growth inhibition assay. The used strains were described in Alkhatib et al. (2014a,b), Reiners et al. (2017), and Lagedroste et al. (2019).
Briefly, the L. lactis NZ9000 strains were grown in M17 medium (Terzaghi and Sandine, 1975) at 30 • C supplemented with 0.5% glucose (GM17 and the appropriate antibiotics) overnight with 1 ng/ml nisin. In a 96-well plate, a serial dilution of the nisin variant was applied and incubated with the test strains at a final OD 600 of 0.1 for 5 h at 30 • C. Later on, the optical density was measured at 584 nm via 96-well plate reader BMG. The normalized optical density was plotted against the logarithm of the nisin concentration and the resulting inhibitory concentration (IC 50 ), represents the value where 50% of the cells died in the presence of the different nisin variants. By dividing the IC 50 values of strains expressing the immunity or resistance proteins from the IC 50 value of the sensitive strain we calculated the fold of immunity/resistance, which is an indicator for the recognition of nisin H or its variant by the immunity or resistance proteins.

Minimum Inhibitory Concentration Determination of Nisin Variants
Nisin variants were tested for antibacterial capabilities against B. subtilis and different strains from S. aureus, E. faecium, and E. faecalis using the microdilution method, according to the recommendations of Clinical and Laboratory Standards Institute (2012). Briefly, fresh cultures prepared from overnight cultures were incubated until exponential phase (OD ∼ 0.6) and seeded at 5 × 10 4 CFU/well in 96-well round-bottom microplates, in a total volume of 100 µL containing twofold serially diluted test peptides. Moxifloxacin was used as a positive control. Plates were incubated statically and aerobically for 24 h at 37 • C. MICs were determined macroscopically by identifying the least concentration of peptides that resulted in complete inhibition of bacterial visual growth.

SYTOX Green Nucleic Acids Binding Assay
The cells of NZ9000Cm were grown overnight in GM17 supplemented with 5 mg/ml chloramphenicol. The overnight culture was diluted to an OD 600 of 0.1 in fresh media and the cultures were grown until the OD 600 reaches 0.3. The SYTOX green dye was added at a final concentration of 2.5 mM according to the manual of the manufacturer (Invitrogen). After reaching a stable baseline (∼200 s) we added 100 nM of the nisin variants. The fluorescence signal was measured at an excitation wavelength of 504 nm and emission wavelength of 523 nm, respectively (using a fluorolog Horiba III). After a stable baseline was reached, the nisin variant was added and the fluorescence was monitored over an additional time period. The measurement was performed at 30 • C.

RESULTS
O'Connor et al. (2015) described a new natural nisin variant from S. hyointestinalis DPC 6484 and named it nisin H. In following, we compared nisin A and its natural variant nisin H, both heterologously produced in L. lactis, following the protocol of lantibiotic characterization (Lagedroste et al., 2019). We also included the nisin H F 1 I variant.
The characterization starts with the expression, secretion and purification of the lantibiotic and its comparison to nisin A. The heterologously expressed and secreted nisin A and the variants nisin H and nisin H F 1 I can be purified with high purity as observed on the Tricine-SDS-PAGE gel (Figure 2A). Nisin A was purified with a yield of 6.0 ± 0.3 mg/L of cell culture (Lagedroste et al., 2019), nisin H was expressed and purified with a yield of 5.3 ± 0.6 mg/L of cell culture, which is identical within experimental error. The nisin H F 1 I variant displayed a slightly reduced yield of 4.9 ± 0.2 mg/L of cell culture ( Figure 2B and Table 1).
An important step prior to the activity assay is the cleavage of the leader sequence form the pre-nisin variants, resulting in biologically active compounds. For the cleavage reaction, we used the peptidase NisP and monitored the cleavage efficiency via RP-HPLC (Figures 2C-E).
Intriguingly, the natural variant nisin H showed only a low cleavage efficiency of 15.6 ± 1.4%, compared to nisin A with 94.6 ± 2.0% ( Figure 2C and Table 1). In comparison to nisin A, nisin H contains a phenylalanine at the first position (O'Connor et al., 2015), which apparently leads to a significant reduction in cleavage efficiency. The first residue of nisin A is an isoleucine, and as demonstrated before (Lagedroste et al., 2019), the introduction of aromatic residues at position one results in a clearly reduced cleavage efficiency. To counteract the lower cleavage efficiency of nisin H by NisP, we created a mutant of nisin H, in which the phenylalanine was substituted by isoleucine and termed it nisin H F 1 I. For this variant, nisin H F 1 I, the cleavage efficiency of the pre-lantibiotic was restored with an efficiency of 91.8 ± 0.8% ( Figure 2C and Table 1), which corresponds to levels previously observed for nisin A. We monitored the cleavage via RP-HPLC, the pre-nisin elutes normally between 18 and 22 min (shown as blue dashed lines, Figures 2D,E). After cleavage by NisP, the leader peptide can be detected at 14.5 and 15.5 min in the chromatogram (shown as black lines, Figures 2D,E). For nisin H there was a high amount of uncleaved nisin H visible (eluting from 18 to 21 min) and only a small amount of cleaved product at 23-24 min (black lines, Figure 2D). For the nisin H F 1 I variant, high amounts of leader peptide and cleaved product could be detected in the chromatogram, indicating high cleavage efficiency (black lines, Figure 2E). This efficiency was similar as observed for nisin A and in line with previous results that the position one is important for the cleavage reaction (Lagedroste et al., 2019). Thus, we assume, that the four other mutations naturally occurring in nisin H (compared to nisin A) do not interfere with cleavage, however the isoleucine at position one does.
The next step was to determine the modification state of the heterologous produced nisin H and its F 1 I variant. As the natural variant nisin H contains ten possible residues in the core peptide that can be dehydrated, we were curious to determine if the modification machinery of nisin A was able to modify the peptide as efficiently (O'Connor et al., 2015). In the MALDI-TOF spectra, a dominant species of eightfold dehydrated residues was observed for nisin H, followed by a minor species containing nine-and sevenfold dehydrations. The possible 10-fold dehydrated species however was not observed ( Figure 3A and Table 1). Furthermore no CDAP-coupling products were observed, which indicates that all cysteine residues are linked in (methyl)-lanthionine rings. We proved the functionality of the assay with unmodified prenisin A as demonstrated in Lagedroste et al. (2019). Thus, the modification enzymes were able to modify nisin H proving the promiscuity of the nisin modification machinery. Interestingly, the nisin H F 1 I variant showed a dominant ninefold dehydrated species in comparison the nisin H wild-type ( Figure 3B and Table 1) and also showed no CDAP-coupling products, indicating that all cysteine residues are closed to (methyl)-lanthionine rings. The difference in the dehydration pattern for the nisin H F 1 I variant indicates a different accessibility of the serine and/or threonine residues in the core peptide for at least the dehydratase NisB. To validate which serine or threonine residues is dehydrated, we performed a tandem mass spectrometric analysis of nisin H and the F 1 I variant. Here we found that the Thr 2 partially escape the dehydration in nisin H. Only in the small amount of the ninefold dehydrated species the Thr 2 is dehydrated, in all other species we found a mix in the dehydration pattern, where Thr 2 partially escape the dehydration. For example in the eightfold species we have a mix of dehydrated Thr 2 or Ser 33 . In the nisin H F 1 I variant the Thr 2 was in all species dehydrated, which suggests that, the phenylalanine at position one in nisin H is critical for the dehydratase NisB. Ser 29 was never dehydrated in the found species. Lantibiotics are very potent and possess an antimicrobial activity in the nanomolar range (Gross and Morell, 1967;Rollema et al., 1995;Chan et al., 1996;Lu et al., 2010;Oppedijk et al., 2016). To verify this potential for nisin H and the nisin H F 1 I variant we used a standardized growth inhibition assay and first screened against the nisin sensitive L. lactis strain NZ9000Cm. Here, Cm stands for chloramphenicol resistance, which arises from the empty plasmid, which was transformed. In comparison to nisin A (IC 50 value: 4.8 ± 0.7 nM), the heterologous expressed variant nisin H possessed a comparable IC 50 value (5.3 ± 1.0 nM) (Figure 4 and Table 2). Both values are in line with previously determined IC 50 values for the strain NZ9000Cm (Reiners et al., 2017). The effect of the nisin H F 1 I variant was more pronounced. For the NZ9000Cm sensitive strain we calculated an IC 50 value of 14.2 ± 0.2 nM, approximately threefold lower than the wild-type nisin H (Figure 4 and Table 2).
To test the effect of the nisin variants on the immunity proteins NisI (Alkhatib et al., 2014a) and NisFEG (Alkhatib et al., 2014b), as well as the resistance proteins SaNSR (Khosa et al., 2016a) and SaNsrFP (Reiners et al., 2017), we expressed each of them in a plasmid-based system in a L. lactis NZ9000 strain. We termed these strains NZ9000NisI, NZ9000NisFEG, NZ9000SaNSR and NZ9000SaNsrFP, respectively.
Nisin A displayed an IC 50 value of 48.6 ± 6.3 nM against strain NZ9000NisI and 53.0 ± 4.5 nM against strain NZ9000NisFEG. For the resistance strains NZ9000SaNSR and NZ9000SaNsrFP nisin A displayed IC 50 values of 73.0 ± 3.6 and 82.1 ± 3.1 nM, respectively (Figure 4 and Table 2). By comparing theses values, we calculated the fold of immunity/resistance to 10.1 ± 2.0 for NZ9000NisI, 11.1 ± 1.9 for NZ9000NisFEG, 15.2 ± 2.5 for NZ9000SaNSR and 17.1 ± 2.7 for NZ9000SaNsrFP ( Table 2). After the first screen against the sensitive strain NZ9000Cm, nisin H and its variant were used to screen against the strains expressing the immunity or resistance proteins.
Nisin H revealed an IC 50 value of 43.2 ± 8.7 nM against the NZ9000NisI strain, similar to nisin A, and a fold of immunity of 8.1 ± 2.2. Against the NZ9000NisFEG strain we determined an IC 50 value of 23.4 ± 3.3 nM for nisin H which displayed a fold of resistance of 4.4 ± 1.0. Against the NZ9000SaNSR strain we obtained an IC 50 value of 52.4 ± 9.9 nM, resulting in a fold of resistance of 9.8 ± 2.6. Nisin H showed an IC 50 value of 86.4 ± 4.0 nM for the NZ9000SaNsrFP strain, resulting in a fold of resistance of 16.2 ± 3.1 (Figure 4 and Table 2). NZ9000SaNsrFP showed the highest fold of resistance for nisin H [in-line with a previous report (Reiners et al., 2017)]. Intriguingly, strain NZ9000NisFEG showed a reduced immunity and consequently we propose that nisin H is not recognized as efficiently as nisin A. Even NZ9000SaNSR had a reduced resistance. That could be due to the exchange of His 31 against lysine in the C-terminal part of nisin H (Figure 1).
Surprisingly, all strains displayed a reducing resistance/immunity effect for the nisin H F 1 I variant in comparison to nisin A and also, with exception of NZ9000FEG for nisin H. Against the NZ9000NisI strain we determined an IC 50 value of 34.1 ± 0.3 nM for the nisin H F 1 I variant, with a fold of resistance 2.4 ± 0.1, which is nearly threefold lower than for nisin A. We obtained an IC 50 value of 32.1 ± 0.8 nM against the ABC transporter NZ9000NisFEG, with a fold of resistance 2.3 ± 0.1 (Figure 4 and Table 2). Nisin H F 1 I showed for the resistance strain NZ9000SaNSR and NZ9000SaNsrFP an IC 50 value of 44.2 ± 1.3 and 50.2 ± 1.6 nM, respectively. The calculated folds of resistance were 3.1 ± 0.1 and 3.5 ± 0.1, respectively (Figure 4 and Table 2) and both are fivefold less than the observed fold of resistances for nisin A.
Since a similar activity for nisin H and nisin A was observed it became obvious that both exhibit the same mode of action. In  the case of nisin A this combines growth inhibition with pore formation in the membrane with subsequent cell death. To test this we performed a SYTOX assay previously used for nisin A (Reiners et al., 2017). Here the SYTOX dye was added to L. lactis cells and displayed an increased fluorescence signal upon binding of DNA which is released from the cell upon pore formation (Roth et al., 1997). We use 100 nM of nisin A, nisin H and nisin H F 1 I variant respectively and observed an almost instant fluorescence increase similar to the signal increase observed for nisin A (Figure 5). This shows that nisin H as well as its F 1 I variant form pores in the membrane of L. lactis cell.
The nisin variants were further tested for antibacterial capabilities against B. subtilis and different pathogenic strains from S. aureus, E. faecium, and E. faecalis using the microdilution method, according to the recommendations of CLSI (2012). Here we found that nisin H and the F 1 I variant performed almost identically or in most cases even better than the natural nisin A. Especially against the MSSA and MRSA strains, nisin H had significant lower MIC values of 0.19 and 0.78 µM, in comparison to 0.78 and 6.25 µM for nisin A, respectively (

DISCUSSION
We focused in this study on the natural variant nisin H and the nisin H F 1 I mutant. Nisin H was discovered from the gutderived strain S. hyointestinalis DPC6484 in 2015 by O'Connor et al. (2015). Here, we showed the heterologous expression of nisin H and the F 1 I variant with the NICE-system in L. lactis (Eichenbaum et al., 1998;Mierau and Kleerebezem, 2005;Rink et al., 2005;Zhou et al., 2006;Lagedroste et al., 2019) and extended the characterization in terms of cleavage efficiency by the protease NisP and the antimicrobial activity against FIGURE 5 | Nisin mediated pore formation, visualized with the SYTOX green assay. The NZ9000Cm strain incubated with the SYTOX dye. After a stable baseline (∼200 s), one of the nisin variants (100 nM) was added (indicated with an arrow). The fluorescence signal was measured using a fluorolog (Horiba III). The rapid increase in fluorescence indicates pore formation. The black line represents the addition of nisin A, the red line nisin H, and the blue line nisin H F 1 I. As a control we added buffer shown as green line.
Both lantibiotics, nisin H and the F 1 I variant were purified in high amounts and purity with 5.3 ± 0.6 mg/L of cell culture for nisin H and 4.9 ± 0.2 mg/L of cell culture for nisin H F 1 I variant, respectively ( Figure 2B and Table 1). In comparison, the homologous expression of nisin H in S. hyointestinalis and nisin A in L. lactis NZ9700 results in a very low amount of 0.15 mg/L of cell culture and 0.50 mg/L of cell culture (O'Connor et al., 2015), respectively, which demonstrates the enormous potential of the NICE-system, for lantibiotic and even non-lantibiotic expression (Eichenbaum et al., 1998;Mierau and Kleerebezem, 2005;Rink et al., 2005;Zhou et al., 2006;Lagedroste et al., 2019).
An important step in the maturation of a lantibiotic is the cleavage of the leader peptide to become biological active. The cleavage efficiency of the natural substrate nisin A was determined with 94.6 ± 2.0%. For nisin H the cleaving efficiency was drastically reduced to 15.6 ± 1.4%. The first residue of nisin A is an isoleucine, while the corresponding residue in nisin H is phenylalanine and as demonstrated before (Lagedroste et al., 2019), aromatic residues prevent efficient cleavage likely by interfering with the S1 binding pocket of NisP. With the nisin H F 1 I variant, the cleavage efficiency was restored to 91.8 ± 0.8% (Figures 1, 2C and Table 1). This indicated that the other point mutations naturally occurring in nisin H did not affect cleavage by NisP. With respect to other natural nisin variants, NisP cleavage could be a critical step. For example nisin O1 to O4 from B. obeum A2-162 (Hatziioanou et al., 2017) has a tyrosine or a threonine, respectively, at position one, which should also result in a low NisP cleavage efficiency. Natural variants such as nisin U (Wirawan et al., 2006), nisin J (O'Sullivan et al., 2019, 2020, nisin Q (Zendo et al., 2003), nisin Z (Mulders et al., 1991) andnisin F (de Kwaadsteniet et al., 2008) have an isoleucine and nisin U2 (Wirawan et al., 2006) and nisin P (Zhang et al., 2012;Wu et al., 2014) a valine at position one, which should result in high NisP cleavage efficiency.
Furthermore, we made a sequence alignment with Clustal Omega (Madeira et al., 2019) for the NshP (the natural protease for the nisin H cleavage) from S. hyointestinalis and NisP from L. lactis to see if there is any difference in the active site, which could be the reason for the reduced cleavage efficiency ( Figure 6A and Supplementary Figure 1). Here the three important residues His 306 , Asp 259 , and Ser 512 which build up the catalytic triad in NisP are conserved in NshP. We also calculated a homology model of NshP based on the known NisP structure (PDB code 4MZD) using Phyre2 (Kelley et al., 2015; Figure 6B and Supplementary Figure 2). Here, no significant differences are found within the overall fold as well as the active site which would explain the difference in the cleavage site. This is intriguing since the recognition site within the leader peptide differs between nisin A (sequence is ASPR) and nisin H (sequence is ASTR) (see Supplementary Figure 3). This suggests that the proteases NisP and NshP recognize their substrate by small difference in their active site.
Nevertheless, for an efficient cleavage, (methyl)-lanthionine rings have to be present. This even holds true in light of the presence of all (methyl)-lanthionine rings, which is generally assumed as the prerequisite for fast and efficient conversion of the pre-nisin to modified nisin (Plat et al., 2011; Lagedroste et al., 2017). To check the amount of dehydrations, necessary for (methyl)-lanthionine ring formation we applied MALDI-TOF analysis. The loss of water molecules within the peptide is directly visible in the reduced molecular weight, but not the (methyl)-lanthionine ring formation. Here we used 1-cyano-4 dimethylaminopyridinium tetrafluoroborate (CDAP) (Wu and Watson, 1998), which binds to free cysteine residues, indicating that these cysteines are not involved in a (methyl)-lanthionine ring. For both lantibiotics, nisin H and the nisin H F 1 I variant no CDAP coupling products were found, indicating that no (methyl)-lanthionine ring is lacking (Figure 3 and Table 1). Nisin H has 10 possible dehydration sites and is predicated to be ninefold dehydrated when expressed homologous (O'Connor et al., 2015). A minor species of nine dehydrations was found, but the dominant species was eightfold dehydrated. The dehydration pattern of the F 1 I variant is changed in comparison to nisin H. Here we determined a dominant ninefold species (Figure 3 and Table 1). This provides a hint, that position two of wildtype nisin H might not have been previously dehydrated, due to steric hindrance of the phenylalanine. To validate which serine or threonine residues is dehydrated, we perform a tandem mass spectrometric analysis of nisin H and the F 1 I variant. Here we found that the Thr 2 partially escape the dehydration in nisin H. In the nisin H F 1 I variant the Thr 2 was in all species dehydrated, which gives a hint that, the phenylalanine at position one in nisin H is critical for the dehydratase NisB. This is in line with previous data from the I 1 F variant of nisin A, where the dominant species was sevenfold dehydrated and not eightfold as wild-type nisin A (Lagedroste et al., 2019). It has also been reported for the natural nisin Z (Mulders et al., 1991), that the I 1 W mutation showed a partial inhibition of dehydration of the Thr 2 (Breukink et al., 1998), which could also be the case for nisin H with the aromatic phenylalanine at position one, resulting in eight dehydrations. A dehydration of position Ser 29 normally goes in line with the lack of ring E (Lubelski et al., 2009), which drastically reduces the antimicrobial activity of nisin A against the sensitive NZ9000Cm strain (Alkhatib et al., 2014b;Khosa et al., 2016a;Reiners et al., 2017). Since the activity was high for nisin H and the F 1 I variant, we expected that Ser 29 was not dehydrated and tandem mass spectrometric analysis supported this. Nisin H showed nearly the same activity as nisin A against the sensitive NZ9000Cm strain but the nisin H F 1 I variant is roughly threefold less active. For the immunity protein NisI, it was revealed that nisin H has an identical activity like nisin A within experimental error. However, the nisin H F 1 I variant exhibited a lower IC 50 value of 34.1 ± 0.3 nM and due to the weaker wild-type activity more than a threefold lower fold of resistance (2.4 ± 0.1 compared to 8.1 ± 2.2) (Figure 4 and Table 2). NisI recognizes the N-terminus of nisin (Wiedemann et al., 2001) and the lower IC 50 could be due to the fact that Thr 2 is dehydrated in the nisin H F 1 I variant in contrast to nisin H. An additional change is the leucine at position 6 against methionine in nisin H and the nisin H F 1 I variant, which could be responsible for the better recognition by NisI.
The immunity protein NisFEG, in comparison to nisin A, showed a strong reduction in immunity in the presence of nisin H and the nisin H F 1 I variant. NisFEG recognizes the C-terminus of nisin (Alkhatib et al., 2014b), which indicates that the point mutations of nisin H affect NisFEG. So, we suppose that nisin H and the nisin H F 1 I variant are not recognized and subsequently transported out of the membrane like nisin A.
The resistance protein SaNSR also recognizes the C-terminus of nisin (Khosa et al., 2016a), and cleaves nisin between the positions 28 and 29. Other studies showed that mutations in this area of the nisin molecule, e.g., S 29 P or C 28 P strongly reduce the efficiency of SaNSR (Field et al., 2019;Zaschke-Kriesche et al., 2019a). We assume that the exchange of His 31 against lysine in nisin H and the nisin H F 1 I variant (Figure 1) has the same effect on SaNSR thereby lowering the resistance efficiency to an IC 50 value of 52.4 ± 2.6 nM for nisin H and 44.2 ± 1.3 nM for the nisin H F 1 I variant, respectively (Figure 4 and Table 2).
For the resistance protein SaNsrFP, we observed an activity for nisin H identical to nisin A. SaNsrFP recognizes the N-terminus of nisin (Reiners et al., 2017), which is affected due to the different dehydration pattern in wild-type nisin H in comparison to the nisin H F 1 I variant. The nisin H F 1 I variant showed a lower IC 50 value of 50.2 ± 1.6 nM, compared to 86.4 ± 4.1 nM for nisin H. This effect is even more pronounced when comparing the fold of resistances of 16.2 ± 3.1 for nisin H to 3.5 ± 0.1 for the nisin H F 1 I variant, respectively (Figure 4 and Table 2).
This study demonstrated again that only a complete characterization of a lantibiotic reveals the full antimicrobial potential. Based on the IC 50 value of the sensitive NZ9000Cm strain the F 1 I variant might be classified as weakly antimicrobial active, but with respect to the immunity and resistance proteins, it becomes more interesting, due to its high activity even against the immunity proteins NisI and NisFEG from L. lactis and the nisin resistance proteins SaNSR and SaNsrFP from S. agalactiae COH1. Against the tested pathogenic bacteria, we found that nisin H and the nisin H F 1 I variant performed almost identically or in the most cases even better than the natural nisin A. Nisin H displayed high antimicrobial potential against both methicillin-resistant and -susceptible the S. aureus strains, both vancomycin-resistant and -susceptible, E. faecium strains, as well as B. subtilis.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
LS and SS conceived and directed this study. JR and ML conducted the expression, purification, MS-analysis, and the growth inhibition experiments. JG performed the SYTOX experiments. GP and KS performed the tandem mass spectrometric analysis. EA and RK performed the MIC experiments. JR, ML, SS, and LS wrote the manuscript. All authors read and approved the manuscript.

FUNDING
This work was supported by the Deutsche Forschungsgemeinschaft (DFG, grant Schm1279/13-1 to LS). The Center for Structural studies is funded by the DFG (Grant number 417919780 to SS). Project number 270650915 (Research Training Group GRK2158, TP2a to RK and TP4a to SS).