C terminus of NisI provides specificity to nisin

Lantibiotics are a group of ribosomally synthesized and post-translationally modified peptide antibiotics containing special amino acids and lanthionine ring structures (Chatterjee et al., 2005). They are produced by a variety of Gram-positive bacteria, including nisin producers Lactococcus lactis and Streptococcus uberis (Wirawan et al., 2006). Nisin kills Gram-positive bacteria by binding to membrane-bound cell wall precursor lipid II, leading to membrane pore formation and inhibition of peptidoglycan synthesis (Breukink et al., 1999; Breukink & de Kruijff, 2006). An additional killing action of nisin is the release of cell wall hydrolysing enzymes, leading to autolysis of the cell (Bierbaum & Sahl, 1985). Nisin also reduces the thermal resistance of Bacillus spores (Beard et al., 1999), and prevents the germination of Bacillus (Nissen et al., 2001) and Clostridium (Thomas et al., 2002) spores.

Nisin-producing bacteria protect their own cell membranes against nisin by a specific immunity mechanism consisting of the ABC transporter complex NisFEG, and the membrane-bound lipoprotein NisI (Siegers & Entian, 1995; Immonen & Saris, 1998; Kuipers et al., 1993; Qiao et al., 1995). LanFEG transporters function as specific exporters of their cognate lantibiotics, as first shown with epidermin immunity proteins EpiFEG (Otto et al., 1998). Similarly, the function of the nisin transporter NisFEG is to decrease the concentration of cell-associated nisin by exporting the bacteriocin from the cell surface to the external environment (Stein et al., 2003). Pre-NisI (245 aa) carries a 19 aa lipoprotein signal peptide, and the site for lipid modification (Cys1 in mature NisI) (Sutcliffe & Russell, 1995). Lipid-modified pre-NisI is secreted through the cytoplasmic membrane, then the signal peptide is cleaved by signal peptidase II, and the mature NisI (226 aa) is anchored to the extracellular side of the membrane by its N-terminal lipid. In addition to the membrane-bound form, a fraction of the produced pre-NisI escapes the lipid modification machinery, and, instead of anchoring to membrane, it is secreted to the external environment in a lipid-free form (Koponen et al., 2004). The lipid-free NisI has been shown to slightly increase the nisin immunity level of L. lactis, indicating that the secretion of lipid-free NisI is potentially a part of the entire nisin immunity mechanism (Takala et al., 2004). However, the exact mode of action of NisI is not fully understood. It has been suggested that the NisI immunity function involves co-operation with the NisFEG transporter complex, since the nisin immunity level achieved when both NisI and NisFEG are expressed is higher than their immunity provided separately (Ra et al., 1999). On the other hand, based on the studies with Bacillus subtilis, it has been proposed that NisFEG and NisI are actually two independent immunity systems, with no co-operative function (Stein et al., 2003). In either case, it is evident that the immunity function of NisI involves specific interaction with nisin (Stein et al., 2003; Takala et al., 2004). NisI is a negatively charged protein (net charge 7), which, by forming a labile complex with cationic nisin, protects the cell against the killing action of nisin (Stein et al., 2003). The mechanism by which NisI binds nisin is not known, nor is it known if there are specific domains in NisI for nisin binding. The interaction between nisin and NisI could partly be based on the opposite charges of the two polypeptides. However, the subtilin immunity lipoprotein SpaI has an even higher negative net charge (8) than NisI, but SpaI does not interact with nisin (Stein et al., 2005). Therefore, NisI must contain specific regions to recognize nisin.

In this study, to examine a site for specific nisin interaction in NisI, C-terminally truncated NisI proteins, and a C-terminal fragment of NisI, were expressed in nisin-sensitive L. lactis. Analysis of the nisin immunity phenotype of these strains showed that the C terminus of NisI confers specificity to protection against nisin. This is the first example of determining a specific region of a lantibiotic immunity protein important for recognition of its cognate bacteriocin.

Bacterial strains, plasmids and growth conditions.
The bacterial strains and plasmids used are presented in Table 1. Escherichia coli and B. subtilis ATCC 6633 were grown at 37 °C in LuriaBertani medium (Sambrook & Russell, 2001). When needed, ampicillin or erythromycin (Erm) was added to a final concentration of 100 or 250 µg ml¹, respectively. L. lactis strains were grown in M17 medium (Oxoid) supplemented with 0.5 % (w/v) glucose (M17G), at 30 °C. When needed, Erm was used at a final concentration of 5 µg ml¹. For determination of the nisin immunity levels, L. lactis strains were grown overnight, and a 2 % inoculum was added to fresh medium, which was then incubated for 5 h. To induce the expression of the nisFEG genes in NZ9840 derivatives, 25 ng nisin ml¹ was added to the medium. From the 5 h exponential-phase cultures, 3 µl was inoculated into 300 µl medium containing different concentrations of nisin (Sigma) in Bioscreen microtitre plates (Labsystems). The plates were grown in the Bioscreen C apparatus at 30 °C for 24 h. Every hour, the plates were shaken moderately for 10 s, followed by optical density measurement with a wideband filter (420580 nm). For determination of subtilin resistance, pasteurized and sterile filtered supernatant of B. subtilis ATCC 6633 culture was used.

Table 1. Bacterial strains and plasmids used in this study

Determination of nisin immunity.
The nisin immunity levels were estimated from the growth curves of the Bioscreen C cultures. An increase of 0.4 optical density units in 24 h was used as the definition of survival.

Secondary structure prediction.
NisI and SpaI secondary structures were predicted using the consensus prediction method on the NPS@ web server (network protein sequence analysis; ) of the Pôle BioInformatique Lyonnais (Combet et al., 2000).

DNA techniques.
Standard procedures were used for agarose gel electrophoresis and PCR amplifications of NisI fragments (Sambrook & Russell, 2001). The accession numbers of the nisI and spaI sequences used for designing PCR primers are Z18947 and U09819, respectively. L. lactis and E. coli were transformed by electroporation, as described by others (Holo & Nes, 1989; Zabarovsky & Winberg, 1990). Restriction enzymes, DNA ligase, T4-polynucleotide kinase and shrimp alkaline phosphatase were used as recommended by the manufacturers (New England Biolabs; Promega; MBI Fermentas).

Cloning of nisI, spaI and truncated nisI fragments.
The genes nisI and spaI, and the four truncated nisI fragments, were cloned as PCR products from L. lactis N8 (nisI) and B. subtilis ATCC6633 (spaI) chromosomal DNA into the E. coli T/A cloning vector pCR2 (Invitrogen). The cloned fragments were cut from the T/A plasmids with BamHI. The native and truncated nisI genes were ligated into the BamHI-restricted vector pLEB565. The spaI gene was ligated into the BamHI-restricted vector pLEB580. In both vectors, the genes were cloned under the control of the constitutive lactococcal promoter P45 (Koivula et al., 1991). The nisI gene in pLEB580 was replaced by the cloned spaI, resulting in the spaI-expression plasmid pLEB626.

Construction of spaI''nisI hybrid.
The spaI''nisI hybrid was constructed via two sequential PCRs, using pLEB626 as a primary template. PCR primers for the spaI''nisI hybrid were NIS154 (forward, homology to promoter P45) 5'-GAATTCCGTTAGGGGCTTGAACAAG-3', NIS226 (reverse I, homology to spaI; nisI elongation) 5'-AAATCATCATTTATCTCAACGGCAAATGCTTCAGTAAGATCTTTGGCATCAGAATCTTC-3', and NIS227 (reverse II, homology to nisI-part of NIS226 primer; further nisI elongation) 5'-AGCTGGGCCCTTAGTTTCCTACCTTCGTTGCAAGCTTAAAATCATCATTTATCTCAACG-3'. The nisI elongations added to the primers are underlined. The restriction site for ApaI added to the NIS227 primer is in bold. The spaI''nisI PCR product was restricted with MluI and ApaI, and cloned into MluIApaI-restricted pLEB580, thus replacing nisI in pLEB580, and resulting in the spaI''nisI expression plasmid pLEB627.

Nisin immunity of C-terminally truncated NisI fragments
In order to localize a site for nisin interaction in NisI (226 aa), a C-terminal deletion series of NisI was constructed. The deletion sites were chosen with the help of secondary structure prediction software on the NPS@ web server. The prediction of NisI secondary structure, and the sites chosen for truncations, are presented in Fig. 1(a). Fragments of the nisI gene lacking 15, 63, 87 and 222 nt from the 3' end were amplified by PCR from L. lactis N8 chromosomal DNA. The ΔnisI PCR products were cloned into the lactococcal expression vector pLEB565, downstream from the constitutive promoter P45. The constructed ΔnisI plasmids were transferred into L. lactis strains MG1614 (no nisin genes) and NZ9840 (ΔnisA, ΔnisI). The cloned nisI deletion fragments encoded C-terminally truncated NisI proteins lacking 5, 21, 29 and 74 aa.

(21K): Fig. 1. (a) Secondary structure predictions of NisI (226 aa) and SpaI (143 aa) according to the protein secondary structure programs on the NPS@ web server. The sites chosen for the C-terminal truncations are shown with arrows. (b) Sequences of the C-terminal 21 aa of NisI and SpaI.

To examine the nisin immunity levels, the ΔnisI transformants were grown in Bioscreen C, in medium containing different concentrations of nisin. The growth was measured as optical density, and compared with the strains carrying the cloning vector pLEB565 and the strains expressing native NisI. The growth curves showed that the shortest deletion (5 aa) decreased the immunity level of native NisI considerably, and that the 21 aa deletion, containing a putative hydrophobic 9 aa α-helix (Fig. 1a), decreased the immunity level even more, compared with the native NisI (Fig. 2a). Longer deletions (29 and 74 aa) did not further decrease the nisin immunity. Thus, the last 5 aa were shown to be important for the immunity function of NisI.

(29K): Fig. 2. Nisin immunity of L. lactis strains expressing native and C-terminally truncated NisI proteins. (a) L. lactis host MG1614. (b) L. lactis host NZ9840. Nisin concentrations are given in µg ml1. , NisI; , NisIΔ5; •, NisIΔ21; , NisIΔ29; , NisIΔ74; , vector pLEB565. Values are means (±SD) of three measurements.

According to expectations, nisin immunities of the NisFEG-producing NZ9840 NisI variants were higher than MG1614 transformants (Fig. 2b). In addition, as with native NisI, the truncated NisI increased the nisin immunity more in NZ9840 than in MG1614: approximately 56 µg ml¹ in NZ9840, compared with 0.350.5 µg ml¹ in MG1614 (Table 2). The nisin immunity levels given in Table 2 were determined from several Bioscreen growth experiments (Fig. 2; data not shown). In conclusion, the C terminus of NisI is essential for nisin immunity function.

Table 2. Nisin immunities of L. lactis strains expressing native NisI, C-terminally truncated NisI proteins, and the SpaI''NisI fusion

Construction of SpaI''NisI hybrid
To determine whether the C terminus of NisI is needed for the correct folding of the protein, or whether it is involved in the interaction with nisin, the 21 aa C-terminal fragment was produced in L. lactis. Since NisI is an extracellular membrane-anchored lipoprotein, intracellular expression of the 21 aa fragment would have no effect on nisin immunity. Therefore, the 21 aa fragment should be produced into its natural location, i.e. the extracellular side of the membrane. It was thought that correct localization would probably be achieved by fusing the 21 aa peptide to the C-terminally truncated subtilin immunity lipoprotein SpaI, since SpaI is secreted and attached to the membrane in a way that is similar to that of NisI. SpaI is also a suitable carrier, as it does not interact with nisin. The amino acid sequences of NisI and SpaI show 17 % homology only, which is considered to be insignificant. The sequences of the 21 C-terminal amino acids of NisI and SpaI are shown in Fig. 1(b). The C-terminal part of SpaI secondary structure resembles NisI structure, and it also contains a putative hydrophobic α-helix, which might facilitate the correct conformation of the protein (Fig. 1a). The last 63 nt in spaI were replaced by the last 63 nt from nisI, by using two sequential PCR reactions. The spaI''nisI hybrid gene, cloned into the lactococcal expression vector pLEB580, encoded a hybrid protein consisting of 122 aa of SpaI fused with 21 aa of NisI. As a control, the native spaI gene was cloned into the same vector. The resulting plasmids were electroporated into L. lactis host strains MG1614 and NZ9840, resulting in strains LAC266 (MG1614 SpaI), LAC267 (MG1614 SpaI''NisI), LAC303 (NZ9840 SpaI) and LAC304 (NZ9840 SpaI''NisI).

Nisin immunity conferred by the SpaI''NisI hybrid
To analyse the capacity of the SpaI''NisI hybrid protein to protect cells against nisin and subtilin, the SpaI''NisI-expressing L. lactis transformants were grown in Bioscreen C plates containing different concentrations of nisin or subtilin. The growth curves of the transformants are shown in Fig. 3. L. lactis strains producing SpaI were as sensitive to nisin as the host strain, confirming that SpaI did not have any protective ability against nisin. However, SpaI protected the transformants against subtilin, verifying the functionality of the SpaI protein in L. lactis. As expected, expression of native NisI in L. lactis strains increased their nisin immunity to a remarkable extent. Interestingly, the hybrid SpaI''NisI also protected the cells against nisin, albeit to a lesser extent than native NisI. Nevertheless, it is notable that the NisI C-terminal 21 aa fragment was capable of converting the subtilin immunity protein to the nisin immunity protein. Moreover, the SpaI''NisI fusion was shown to protect lactococci against subtilin. In the NisFEG-expressing strain NZ9840, no enhancement of NisFEG by the SpaI''NisI fusion was observed, since the SpaI''NisI fusion increased the nisin immunity level in NZ9840 by the same amount as that observed in the nisin-negative strain MG1614, i.e. an increase of 1 µg ml¹ in both strains (Table 2). In conclusion, the results made it evident that the C terminus of NisI is physically involved in nisin immunity function, doubtless by specific interaction with nisin.

(40K): Fig. 3. Nisin immunity and subtilin resistance of L. lactis strains expressing the SpaI''NisI fusion, and native NisI and SpaI. (a) L. lactis host MG1614. (b) L. lactis host NZ9840. Nisin concentrations are given in µg ml1. The subtilin used was from sterilized B. subtilis ATCC 6633 culture supernatant. , NisI; , SpaI; , SpaI''NisI fusion; vector pLEB579. Values are means (±SD) of three measurements.

In previous studies of the mechanism of nisin immunity, the nisin immunity protein NisI (226 aa) has been shown to specifically interact with nisin (Stein et al., 2003; Takala et al., 2004). In this study, we showed that it is the C terminus of NisI that is responsible for the nisin-specific interaction. The last 5 aa in NisI (residues 222226), containing one positively charged residue, were crucial for immunity function, since only 22 % (0.5 µg ml¹) of the nisin immunity capacity of native NisI (2.25 µg ml¹) remained after deletion of these residues, when expressed in L. lactis MG1614 (Table 2). Based on the secondary structure prediction, the C-terminal 5 aa are not part of any obvious structure (Fig. 1a). On the other hand, a 9 aa α-helix, including six hydrophobic and two negatively charged residues, was predicted in the position 206214 aa. Deletion of the C-terminal 21 aa (residues 206226), containing the putative 9 aa α-helix, decreased the immunity level further: only 14 % (0.35 µg ml¹) of the immunity capacity of native NisI was left when the deletion mutant was expressed in MG1614. The critical role in nisin immunity of the last 5 aa could be due to loss of stability of the C-terminal 9 aa α-helix. The 53 aa region preceding the C-terminal α-helix (residues 153205) consists mostly of random coil (Fig. 1a), and does not seem to be essential, since the immunity level of the Δ74 aa strain was virtually the same as the Δ21 aa variant. Similar C-terminal truncations have been constructed in the lantibiotic Pep5 immunity peptide PepI (Hoffmann et al., 2004). Also in PepI, the C terminus has been shown to be essential for immunity function, whereas N-terminal mutations affected the transport of the protein out of the cell. It has been proposed that the mode of action of PepI is that it binds to the target of the bacteriocin, thus shielding the target from the lantibiotic. However, since the sizes (69 vs 226 aa), the distributions of charged residues (equally distributed in NisI; positively charged C terminus in PepI) and the modes of action (NisI interacts with the bacteriocin; PepI interacts with the target of the bacteriocin) are different in PepI and NisI, it is probably a coincidence that the C termini of these immunity factors are essential for their immunity functions. Even though the exact mode of action for NisI has not yet been conclusively demonstrated, it is evident that the immunity function of NisI involves specific interaction with nisin (Stein et al., 2003; Takala et al., 2004). The mode of action of NisI may not include interaction with lipid II, the target of nisin, since externally added NisI had no significant affinity to cells, and it did not protect cells against nisin (Koponen et al., 2004); if NisI had affinity for lipid II, it would have interacted with it when added externally. Nevertheless, it cannot be excluded that membrane-bound NisI could hamper nisin from reaching lipid II. To compete with lipid II for nisin binding, NisI should bind nisin before it reaches lipid II, since the affinity of nisin for lipid II is much higher than for NisI (Breukink et al., 1999; Takala et al., 2004). Therefore, the dissociation of nisin from the nisinlipid II complex by NisI would be ineffective. Even though the function of NisFEG is to transport nisin from the membrane into the external environment (Stein et al., 2003), the ability of NisFEG to dissociate the nisinlipid II complex has not been shown.

The C-terminally truncated NisI proteins were expressed in two different L. lactis strains: nisin-negative MG1614, and NZ9840 expressing NisFEG for partial nisin immunity. A possible co-operation between NisFEG and NisI has been suggested, based on observations that NisI increases the nisin immunity level of Lactococcus cells to a greater extent when NisFEG is present (Ra et al., 1999; Takala et al., 2004). Similar co-operation with the LanFEG complex has been observed with the lantibiotic nukacin ISK-1 immunity protein NukH (Aso et al., 2005). However, unlike NisI, NukH inactivates the bacteriocin after binding (Okuda et al., 2005). Therefore, this LanFEG co-operative similarity does not throw much light upon the exact function of NisI in the nisin immunity mechanism, since NisI does not inactivate nisin (Stein et al., 2003; Koponen et al., 2004). Here, similar to native NisI, the truncated NisI proteins increased the nisin immunity to a greater extent in NisFEG-expressing NZ9840 than in MG1614. This showed that by co-operating with NisFEG, the truncated NisI had higher capacity to protect cells against nisin. Thus, it can be concluded that the C terminus of NisI is involved in interaction with nisin, and is not involved in co-operation with NisFEG. An alternative explanation is that C-terminal deletions hampered the correct folding of the NisI protein, and in that way influenced its function. The next objective was to find out which of these two explanations is correct: does the C terminus of NisI bind nisin, or does it affect the correct conformation of the protein?

The capacity of the C-terminal 21 aa fragment to provide nisin immunity, i.e. without the rest of the NisI protein, was investigated. The cellular location and function of the subtilin immunity lipoprotein SpaI has been shown to be similar to NisI, except that SpaI is specific to subtilin, and does not interact with nisin (Stein et al., 2005). Thus, SpaI would offer a carrier for NisI-fragment targeting. For targeting the 21 aa fragment to its natural location on the cytoplasmic membrane, the fragment was fused to C-terminally truncated (Δ21 aa) SpaI. The expression of the SpaI''NisI hybrid in two L. lactis strains increased their nisin immunity significantly, demonstrating that the 21 aa NisI fragment was able to convert the subtilin immunity protein to the nisin immunity protein. As a result, it can be concluded that not only is the C terminus of NisI responsible for the correct folding of NisI protein, but it must also be involved in the specific interaction with nisin. Unlike native NisI and the C-terminally truncated NisI, the increase in nisin immunity level conferred by SpaI''NisI expression in the NisFEG strain NZ9840 was the same as in the nisin-negative strain MG1614. Since SpaI''NisI increased the nisin immunity level by 1 µg ml¹, with or without NisFEG, it is obvious that the fusion does not co-operate with NisFEG. Thus, as already concluded from the results with C-terminally truncated NisI, the NisI C terminus is not responsible for NisFEG co-operation. According to the results presented here, the C terminus of NisI provides specificity for the nisinNisI interaction.

This work was financed by the Academy of Finland, project number 1211494. The authors want to thank Ms Kaisu Nevalainen for her excellent technical assistance.