Dissection and Modulation of the Four Distinct Activities of Nisin by Mutagenesis of Rings A and B and by C-Terminal Truncation

Bacterial strains and plasmids.Strains and plasmids are listed in Table 1.

Molecular cloning.Constructs coding for mutant prenisin were made by amplifying plasmid pNZnisA-E3 (18) using a phosphorylated downstream sense primer and an upstream antisense primer with a (nonannealing) peptide-encoding tail. The DNA amplification was carried out by using Phusion DNA polymerase (Finnzymes, Finland). The replacement of ISL of ring A of nisin involved primers 5′-NNNNNNNNNACTTGTAATGCGTGGTGATGC-3′ and 5′-PO₄-TGTACACCCGGTTGTAAAAG-3′, in which N is a randomized nucleotide. The replacement of PG of ring B with XX in nisin involved primers 5′-NNNNNNTGTACATAGCGAAATACTTG-3′ and 5′-PO₄-TGTAAAACAGGAGCTCTGATGG-3′. The replacement of PG with PX in ring B of nisin involved primer 5′-NNNGGGTGTACATAGCGAAATACTTG-3′. Self-ligation of the resulting plasmid encoding linear leader peptide was carried out with T4 DNA ligase (Roche). The electrotransformation of L. lactis was carried out as previously described (14) using a Bio-Rad gene pulser (Bio-Rad, Richmond, CA). Nucleotide sequence analysis was performed by BaseClear (Leiden, The Netherlands). Transformation mixtures of the sequenced transformants were directly plated out with or without 0.1 nM nisin in the agar, which contained trypsin to cleave off the leader. After overnight incubation, the transformants were overlaid with top agar containing the indicator strain Lactococcus lactis LL108(pORI80).

Culturing. L. lactis was grown in M17 broth (36) supplemented with 0.5% glucose (GM17) or minimal medium (32) with or without chloramphenicol (5 μg/ml) and/or erythromycin (5 μg/ml). The concentration of the antibiotics was reduced to 4 μg/ml when both selection markers were present simultaneously. Staphylococcus epidermidis K7 was grown on basic medium (30), Lactobacillus johnsonii, Pediococcus pentosaceus, and Leuconostoc mesenteroides on MRS medium, and Micrococcus flavus NIZO B423 on TY medium (Bacto tryptone, 10 g/liter; yeast extract, 5 g/liter; NaCl, 5 g/liter), all at 30°C; Streptococcus thermophilus was grown in GM17 medium at 42°C. Prior to mass spectrometric analyses of peptides produced in the media, cells were cultured as follows. Overnight cultures of L. lactis NZ9000 ΔacmA were grown in GM17 broth containing antibiotic(s) and were diluted 1/50 in minimal medium buffered with 0.12 M 3-(N-morpholino)propanesulfonic acid (MOPS) (pH 7.0) supplemented with 0.1 nM nisin. Incubation was continued for 16 h.

Growth inhibition was measured after 3 h of incubation using microwell plates containing indicator cells at an initial optical density at 600 nm (OD₆₀₀) of 0.1 and a series of twofold dilutions of the nisin mutant. The 50% inhibitory concentration (IC₅₀) was obtained from the midpoint of the sigmoidal growth curve.

Inhibition of the outgrowth of spores. Bacillus subtilis 168 was cultivated at 37°C in Schaeffer sporulation medium (34) in a rotary shaker at 200 rpm for 20 h. The culture was stored at 4°C for one day to lyse. The spores were collected by centrifugation for 10 min at 2,057 × g in a swing-out rotor. The loose pellet that contained the spores was extensively washed with Millipore water and stored at 4°C in Millipore water. The spores were heat activated for 10 min at 80°C just prior to use. Equal amounts of spores and cells (5 × 10⁶) were tested in a growth assay in microwell plates as described above. The outgrowth of spores was followed by measuring the OD₆₀₀. For nisin purified from the supernatant of L. lactis NZ9700, the IC₅₀ values were found to be identical to the values obtained from nisin A isolated after production by using L. lactis containing plasmid-encoded enzymes and plasmid-encoded prepropeptide.

Purification of nisin and nisin mutants.Wild-type nisin A, the ring A KSI, KFI, and VFG mutants, and ring B PT and PH mutants were purified as follows. Cultures were started from an overnight preculture in GM17 medium with antibiotics by diluting them 100-fold into 250 ml minimal medium (32) buffered with 0.12 M MOPS (pH 7.0) buffer, without antibiotics. The supernatant was diluted with an equal volume of 100 mM lactic acid (pH 4), and subsequently, purification proceeded with a single passage of prenisin over a 5-ml fast protein liquid chromatography HiTrap sulfopropyl cation-exchange column (Amersham). Elution was performed at pH 4.0 with 1 M NaCl in 50 mM lactic acid. The fraction containing prenisin was desalted on a PD10 column. The collected desalted fractions were thereafter lyophilized. The leader peptide was completely cleaved off by incubating prenisin with NisP-expressing cells for 16 h. Alternatively, 30 min of incubation at 37°C with 0.01 mg/ml trypsin, which readily cleaves after an arginine, was applied. Mass spectrometric measurement of the high-performance liquid chromatography (HPLC)-purified nisin mutants proved that trypsin had not cleaved after lysines. The nisin and nisin mutants were purified on a C₁₈ column with reversed-phase HPLC with a gradient of 10 to 50% acetonitrile in 0.1% trifluoroacetic acid. The purified peptide was quantified on the basis of a comparison of the 214-nm HPLC peak area with that of a nisin A standard of known concentration. Analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis showed that the peptides were >95% pure.

Induction capacity.The induction capacities of the nisin mutants were determined qualitatively by mass spectrometric measurement of the production of the ITRICK leader peptide by the induced L. lactis NZ9000 ΔacmA containing pTP-ITRICK and pIL3BTC. Quantitative measurement of the induction capacities of the nisin mutants was performed by inducing cells expressing the nisin ring A VFG mutant in the absence of inducing amounts of nisin A. The nisin VFG mutant-expressing cells were incubated for 16 h in GM17 medium containing the appropriate antibiotics with different concentrations of the purified nisin mutant. The trypsin-released activity was determined by measuring growth inhibition, as described above in “Culturing,” with the indicator strain L. lactis LL108(pOri 280). The lowest concentration needed to get full expression of the VFG mutant was taken as the minimal induction concentration. This bactericidal mutant was used because it has no induction capacity when it is expressed as prenisin in liquid cultures but still has full nisin activity when the leader sequence is cleaved off.

Mass spectrometry.Peptides were isolated from culture supernatants in a single step by applying the ziptip (C18 ziptip; Millipore) procedure as previously described (32) and the zipplate (Millipore) procedures according to the protocol. The dried eluent was resuspended in 5 μl of 50% acetonitrile containing 0.1% (vol/vol) trifluoroacetic acid, and 1 μl was applied to the target. Purified peptides were directly applied to the target. Subsequently, 1 μl of matrix (5 mg/ml α-cyano-4-hydroxycinnamic acid in 50% acetonitrile containing 0.1% [vol/vol] trifluoroacetic acid) was added to the target and allowed to dry. The cysteinylation peaks were identified by comparing samples that were either incubated for 15 min at room temperature with 0.05 mg/ml triscarboxyethyl phosphine or not. The thioether ring formation was assessed by an absence of reactivity with 1-cyano-4-dimethylaminopyridinium tetrafluoroborate (CDAP) as described previously (16). Mass spectra were recorded with a Voyager-DE PRO matrix-assisted laser desorption ionization-time-of-flight mass spectrometer (Applied Biosystems). In order to maintain high sensitivity, an external calibration was applied.

Δψ dissipation.The capacity of the nisin mutants and truncated nisin mutants to dissipate the transmembrane electrical potential of L. lactis LL108(pOri 280) was measured by using a DiSC₃(5) probe as described previously (13). L. lactis (25 μg protein/ml) was suspended in a buffer at 30°C and pH 7.0, prepared by adding piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES) to 50 mM NaOH, which buffer contained the Δψ-sensitive probe DiSC₃(5) (6 μM); subsequently, fluorescence was measured at excitation and emission wavelengths of 643 and 666 nm, respectively. By the addition of valinomycin (0.2 μM), a Δψ was induced, resulting in a decrease of the DiSC₃(5) fluorescence. The dissipation of the Δψ resulted in the recovery of the initial level of fluorescence. The fluorescence was measured by using a fluorolog fluorimeter (Jobin Yvon).

Mutagenesis of ring A and B amino acid residues of nisin.Ring A of nisin is composed of a lanthionine that encompasses residues Ile4, Ser5, and Leu6. We simultaneously replaced these residues by randomizing their codons in the nisA-carrying plasmid pNZnisA-E3. In a separate experiment, ring B was mutagenized by randomizing the codons for the amino acids in positions 9 and 10 (Fig. 1). We investigated whether the highly conserved (32) Pro9 Gly10 residues are required for the activity of nisin. The plasmid mixtures containing randomized ring A or ring B codons were transformed into L. lactis NZ9000 ΔacmA, containing plasmid pIL3BTC which encodes the ring-forming enzymes NisB and NisC and the transporter, NisT. This host strain does not contain the leader peptidase, NisP. Hence, the leader peptide remains attached to the produced nisin mutants, thereby keeping them inactive. Following secretion, the leader peptide was removed in vitro by a low concentration of trypsin or by NisP-expressing cells, resulting in identical activity. A large number of transformants was obtained (see Tables S1 and S2 in the supplemental material). About 7% of the randomly tested mutants displayed antibiotic activity against the indicator strain, L. lactis LL108(pOri 280). All ring B mutants were also screened qualitatively for their inducing capacities by studying the production of the ITRICK leader peptide. Antimicrobially active mutants and ring B mutants that had inducing capacity were sequenced. Besides wild-type nisin, five mutants were purified for detailed characterization because of particular characteristics as revealed by preliminary experiments: ring A KFI, KSI, and VFG mutants and ring B PT and PH mutants. The KSI mutant was selected for purification because the supernatant of KSI-producing bacteria was about 10 times more potent in inhibiting L. lactis LL108(pOri 280) than wild-type nisin A. Because of the high number of phenylalanine-containing mutants and the high antimicrobial activity found for the KSI mutant, an additional, KFI, ring A mutant was made. A particular further reason for constructing and also purifying this KFI mutant was its mimicking of ring A of epidermin. The VFG mutant combined potent antimicrobial activity with an absence of induction capacity in liquid media. By contrast, the ring B PH mutant was purified because of its induction capacity combined with an absence of antimicrobial activity. The PT mutant was the ring B mutant with the most potent antimicrobial activity. These purified mutants were accurately characterized with respect to their antimicrobial activities against different gram-positive bacteria, induction capacities, inhibition of the outgrowth of spores, and capacity to permeabilize the membrane.

A strikingly high number of phenylalanine-containing ring A mutants was obtained (see Table S1 in the supplemental material). Only four of the mutants did not contain a phenylalanine at position 4, 5, or 6 (KSI, TKI, VGG, and YQI). Two mutants harbored a charged residue, i.e., a lysine (KSI and TKI). Most of the mutant residues were hydrophobic. The numbers of variants with three (FFY), two (FVW, FSY, FVS, FSF, FAF, FFS, FFL, FFY, FFV, and SFF), and one (SFV, VFG, IFS, LFQ, LFA, LSF, and YQI) aromatic residue(s) were 1, 10 and 7, respectively. Interestingly, epidermin and gallidermin contain a phenylalanine at position 5, which in nisin corresponds to a serine that is dehydrated during the posttranslational modification. This serine residue was retained in only 10 transformants, while one mutant (TKI) contained a threonine. The extent of dehydration of all mutants was determined by mass spectrometry (see Table S1 in the supplemental material). Among the mutants that did not lack any dehydrated residue were not only very active ones but also mutants with strongly reduced activity. These data indicate the importance of the amino acid composition for the antimicrobial activity.

Despite the high level of conservation of ring B, several active mutants were obtained (see Table S2 in the supplemental material). Mass spectrometric analysis demonstrated that a significant fraction of the SS, PS, TS, and PT mutants contained an extra dehydrated residue compared to those in wild-type nisin. The PH, PR, PD, PN, PL, and PP mutants lacked a dehydrated residue and showed partial Cys addition, indicating that one cysteine had not undergone cyclization via coupling to a dehydroresidue.

Autoinduction and induction.In an overlay assay with sensitive L. lactis cells and trypsin, a variety of halo sizes was observed on the plates when nisin was omitted (see Fig. 2, left plate). This implies that even without the inducer nisin, many of the nisin variants were produced and autoinduced. The presence of inducing amounts of nisin in the plates resulted in an increase in halo size (Fig. 2, right plate). When the assay was performed in the absence of trypsin but using NisP-expressing cells for the overlay, the same halo sizes were observed as in the presence of trypsin. Combining trypsin with NisP-expressing cells led neither to larger nor to smaller halo sizes, which suggests that NisP-mediated leader cleavage was complete and demonstrates that the trypsin-mediated removal of leader peptide did not result in inactivation by cleavage in nisin itself. In the absence of trypsin or NisP, no halos were observed except around the wild-type nisin A-producing strain NZ9700, indicating that endogenous leader cleavage is limited. These data indicate that many mutants are equipped with an autoinducing capacity and that all tested mutants could be processed by NisP irrespective of the amino acid composition of ring A and ring B.

• Open in new tab
• Download powerpoint

FIG. 2.

Growth inhibition by nisin ring A and B mutants. Two amino acids correspond to positions 9 and 10 of the nisin mutant, and three amino acids correspond to positions 4, 5, and 6 of the nisin mutant. The L. lactis ΔacmA mutant containing pIL3BTC and the nisin mutant-encoding plasmid was grown overnight on the plates and subsequently covered with a layer of L. lactis LL108(pOri 280) and trypsin to cleave off the leader peptide in the presence (right plate) or absence (left plate) of inducing amounts of nisin. Plates were incubated overnight. Halos correspond to zones in which inhibition of the growth of the overlaid strain occurred. In the absence of trypsin, no halos were observed except for the NisP-expressing wild-type strain NZ9700 (wt1 and wt2).

In a liquid medium, most leader peptide-containing mutants were unable to self induce. Only prenisin A itself and the ring B PT and PS mutants were able to self induce. Interestingly, after leader cleavage, the ring B PH mutant, which is devoid of antimicrobial activity, also displayed induction capacity (see Table 3). No self-induction was observed when the ring B PH and PR mutants were produced as prenisin. The PR mutant was treated with trypsin and analyzed by mass spectrometry, showing that this nisin mutant is cleaved after the R site (Fig. 3). The N-terminal fragment lacked one dehydrated residue, indicating that Thr at position 8 was not dehydrated. Hence, the observed masses of 1,038, 2,453, and 3,471 Da correspond to IDhbLanIDhaLLanTPR, fragment C11-K34, and the complete nisin PR variant, respectively. Together, these data demonstrate that ring B was not closed, likely because it is not formed in the PR mutant nor, likely, in the PH mutant. Apparently, bulky mutations in position 10 may prevent dehydration at position 8. Hence, it is not only the residues that directly flank the serine or threonine residue that affect dehydration but also residues more remote from these positions. Previously, we observed that ring formation is needed for cleavage by NisP (18). Now, we show that the cleavage reaction does not require a closed ring B. Also, the ring B PD, PN, PL, and PP mutants, which lack one dehydrated residue (see Table S2 in the supplemental material) and concomitantly show only partial Cys addition, were inducers. The presence of induction capacity of the above open-ring-B mutants indicates that ring B is not needed for the activation of NisK. The fully modified prenisin form of the ring A VFG mutant had no capacity to autoinduce in liquid medium, but after cleavage of the leader, the VFG mutant has significant bactericidal activity. Hence the strain that produced this fully modified prenisin form of the ring A VFG mutant could be induced and the induction capacity could be quantified by measuring the bactericidal activity of the produced VFG mutant after leader cleavage. Therefore, the induction capacity in liquid medium of the purified mature forms, without leader peptide, of nisin and its mutants was measured using the strain coding for the VFG mutant. Table 3 shows the minimal concentrations for the variants to obtain maximal induction. The ring B PT mutant was as good an inducer as nisin A, while the induction capacities of the other mutants decreased, with capacities ranging in descending order from KFI to KSI to VFG to PH.

• Open in new tab
• Download powerpoint

FIG. 3.

Trypsin digestion of nisin G10R mutant analyzed by mass spectrometry. The observed average masses correspond to that of IDhbLanIDhaLLanTPR (observed, M + H⁺, 1,038 Da; theoretical value of three-times-dehydrated peptide, 1,037 Da): fragment C11-K34, observed, 2,453 Da; theoretical value, 2,454 Da for four-times-dehydrated peptide; whole G10R nisin, observed, 3,471 Da; theoretical value of seven-times-dehydrated peptide, 3,472 Da.

Growth inhibiting activity.The levels of activity of nisin mutants against L. lactis and other bacteria were compared with the level of activity of wild-type nisin A (Table 2). The KFI mutant was more active than nisin A against L. johnsonii (10-fold-lower IC₅₀) and L. mesenteroides (ninefold-lower IC₅₀) (Table 2). The purified KSI mutant (Table 2) showed 1.6- and 2-fold-higher activities than nisin A against indicator strains L. lactis LL108(pOri 280) and L. johnsonii but was 2.6-fold less active against S. epidermidis and P. pentosaceus. This mutant contains a conserved mutation, L6I, and a charge alteration, I4K. The TKI mutant has similar amino acids but in a different order, and hardly showed antimicrobial activity. Interestingly, epidermin, gallidermin, and the recently discovered nisin U (42) also contain a lysine at position 4, while epidermin also contains an isoleucine at position 6. These data demonstrate that the strain-specific bactericidal activity of nisin can be enhanced by altering the composition of ring A.

Media containing the ring B TS, SS, AA, or GS mutants showed lower growth-inhibiting activities than medium containing nisin A. Therefore, further mutagenesis was performed by randomizing only position 10 of nisin A, and transformants were analyzed for the production of growth-inhibiting nisin mutants. Mutant PA contained all five thioether rings and showed a modest growth-inhibitory activity (see Table S2 in the supplemental material). The PT mutant, in which the threonine was completely dehydrated, was about 1.5- to 3-fold less active than nisin A (Table 2). These data show that ring B can be composed of other amino acids than P and G but, for antimicrobial activity, small amino acids are required. With the preferred proline at position 9, these amino acids likely favor ring formation.

Nisin mutants that escape the nisin self-protection systems.Striking differences were observed in the comparison of the antimicrobial activities of the mutants when using L. lactis NZ9700, a nisin producer, as the indicator strain (Table 2). Ring A VFG (IC₅₀, 2.7 nM), KFI (IC₅₀, 10 nM), and KSI (IC₅₀, 18 nM) mutants showed efficient growth inhibition of L. lactis NZ9700 (Table 2), while significant activities were also observed with a range of mutants: in decreasing order, FFS, FVS, IFS, FFS, FFY, FVW, LFA, LFQ, FLI, and FFV (not shown). Apparently these mutants circumvent or inhibit the self-protection mechanisms of NisI and/or NisEFG of strain NZ9700.

Inhibition of the outgrowth of spores.The capacities of the purified nisin A and nisin PT, KSI, KFI, and VFG mutants to inhibit the outgrowth of Bacillus subtilis strain168 spores were determined. In parallel, the growth inhibition of B. subtilis 168 cells was also measured. The respective IC₅₀ values are shown in Fig. 4. While nisin A inhibited the growth of B. subtilis 168 cells with an IC₅₀ of 7.9 ± 0.6 nM (mean ± standard deviation), much lower concentrations sufficed to inhibit the outgrowth of spores, i.e., an IC₅₀ of 0.3 ± 0.2 nM. The tested mutants showed a similar pattern (Fig. 4). Since the KFI and VFG mutants efficiently inhibit the outgrowth of spores, we conclude that Dha5 is not required for this activity. The KFI and KSI mutants even showed an enhanced capacity to inhibit the outgrowth of spores compared to that of nisin A (Fig. 4).

• Open in new tab
• Download powerpoint

FIG. 4.

IC₅₀ values for the inhibition of Bacillus subtilis 168 cells and spores by nisin A and derived mutants. Cells (filled bars) and spores (open bars) were incubated for 4.5 h with nisin or indicated mutants, after which the OD₆₀₀ was measured. For experimental details, see Materials and Methods. The levels of pore inhibition by nisin KSI and KFI mutants that differed significantly (P < 0.05) from that of the wild type (wt) by a paired t test are indicated with an asterisk. Error bars indicate standard deviations.

Truncated nisin mutants.To investigate whether the pore formation and lipid II abduction mechanisms can be separated, C-terminally truncated variants of nisin A and its PT, PH, KSI, KFI, and VFG mutants were constructed. Herein, rings D and E were removed, resulting in nisin A(Δ23-34) mutants. The peptides were purified and subjected to mass spectrometry to determine whether rings A, B, and C were correctly formed. In mutant PH(Δ23-34), two thioether bridges were lacking and a disulfide was present instead (see Table S3 in the supplemental material). For the other truncated mutants, the measured masses agreed with the expected masses for the correct formation of all three thioether rings. All cysteine residues were part of lanthionines, as demonstrated by the lack of CDAP addition (see Table S4 in the supplemental material). To investigate whether the bridging patterns of the truncated mutants were identical to those of full-length mutant peptides, postsource decay was applied and the fragments were analyzed (see Table S5 in the supplemental material). Since only a few fragments were observed, conclusions about the bridging patterns could be drawn. The observation of fragments y10 and y11, which correspond to ring C (Fig. 1) indicates that this ring is correctly formed. The y14 ion is also found for all mutants, indicating that Cys11 of ring B is not connected to the alternative dehydroamino acid Dha5. Thus, rings B and C are correctly formed. With the exception of the PH mutant, fragments that confirm normal bridging patterns were obtained, indicating that ring A is also properly formed (see Table S5 in the supplemental material).

The truncated nisin A and derived mutants showed strongly decreased induction capacities, but truncated nisin A and mutant PT remained more-effective inducers than the other mutants (Table 3). The truncated mutants had retained some antimicrobial activity against L. lactis LL108(pOri 280) (Table 2), but this varied among the mutants. Truncated nisin A (IC₅₀, 3.6 nM) and the truncated KFI mutant (IC₅₀, 6.1 nM) showed significant levels of activity: their IC₅₀ values were only 9- and 15-fold higher than that of full-length nisin. Truncation led to larger increases for the PT (77-fold increase in IC₅₀), VFG (96-fold increase in IC₅₀), and KSI (268-fold increase in IC₅₀) mutants (Table 2). This relatively larger decrease in activity for the latter mutants might indicate that pore formation by the corresponding full-length mutants might contribute more to the antimicrobial activity than the inhibition of cell wall synthesis. Next, we determined whether the truncated mutants were still capable of permeabilizing the membrane. Therefore, L. lactis LL108(pOri 280) cells were suspended in a sodium buffer and a Δψ was induced by adding the potassium ion ionophore valinomycin. The resulting Δψ caused by the electrogenic efflux of potassium ions was monitored by the decrease in the fluorescence of the Δψ-sensitive probe DiSC₃(5). The addition of full-length nisin A (Fig. 5, dotted line) or the KSI, PT, KFI, or VFG mutant (not shown) caused a rapid dissipation of the Δψ, as shown by the recovery of the DiSC₃(5) fluorescence. In contrast, neither truncated nisin A (Fig. 5, solid line) nor any of the other truncated mutants affected the Δψ. Hence, the removal of the C-terminal region containing rings D and E resulted in a complete loss of the membrane permeabilization activity. This implies that the antimicrobial activity observed for the truncated peptides likely results from the remaining capacity to inhibit cell wall synthesis. To validate this suggestion, the ability to inhibit the outgrowth of B. subtilis spores was also monitored. This activity was diminished to an extent similar to that of the ability to inhibit the growth of B. subtilis cells (not shown), although the outgrowth of spores was inhibited at lower concentrations than those resulting in the inhibition of cells, as discussed before for inhibition by full-length peptides. These data indicate that the inhibition of cell wall synthesis contributes significantly to the antimicrobial effect of nisin A and the KFI mutant but less to the antimicrobial effects of the KSI, VFG, and PT mutants.

• Open in new tab
• Download powerpoint

FIG. 5.

Nisin A(Δ32-34) is defective in Δψ dissipation in L. lactis. L. lactis strain LL108(pOri 280) (25 μg protein/ml) was suspended in 50 mM NaOH-PIPES, pH 7.0, at 30°C, containing the Δψ-sensitive probe DiSC₃(5) (6 μM). Arrow 1, addition of valinomycin (0.2 μM); arrow 2, addition of nisin A (3 nM) (dotted line) or nisin A(Δ23-34) (7 nM, solid line). a.u., arbitrary units.