Lipid-free NisI: interaction with nisin and contribution to nisin immunity via secretion


*Corresponding author. Tel.: +358-9-191-593-39; fax: +358-9-191-593-22, E-mail address:


Nisin-producing Lactococcus lactis cells protect their own cytoplasmic membrane by specific immunity proteins, NisF/E/G and NisI, a transporter complex and a lipoprotein, respectively. A portion of NisI is secreted to the medium in a lipid-free form (LF-NisI). Here, kinetics of the interaction between nisin and LF-NisI was examined by surface plasmon resonance analysis. The affinity constant KD for the interaction was calculated to be in the micromolar range. Contribution of the secreted LF-NisI to nisin immunity was studied by replacing the lipoprotein specific nisI signal sequence with a secretion signal of non-lipoprotein origin. Secretion of LF-NisI in NisF/E/G-expressing L. lactis strain NZ9840 increased significantly its nisin tolerance suggesting that the lipid-free form of NisI could have a supportive role in nisin immunity.


Nisin is a post-translationally modified bacteriocin produced by Lactococcus lactis strains [1]. It is classified as a lantibiotic due to the modified residues including dehydroalanine, dehydrobutyrine, lanthionine and methyllanthionine [2]. Nisin interacts with the cell wall precursor Lipid II on the cytoplasmic membrane leading to the formation of nisin:Lipid II pore and dissipation of the proton motive force [3,4]. At high concentrations, nisin is able to form pores in the absence of Lipid II [5,6]. With Lipid II the pore is more stable and the membrane potential required for pore formation is reduced [7].

The membrane of L. lactis is sensitive to nisin. For membrane protection, nisin producers synthesize a complex immunity system consisting of four polypeptides: the membrane-bound lipoprotein NisI, and an ABC translocator including two integral membrane proteins, NisE and NisG, and cytoplasmic NisF [8–13]. Both NisI and NisF/E/G play important roles in nisin immunity. Disruption of the nisI gene causes approximately 80% reduction in nisin immunity [12,14]. However, even strong expression of the nisI gene in nisin-sensitive L. lactis without any nis genes increased the immunity level only by 25% compared to nisin producers [15], indicating that both immunity factors are crucial for full nisin immunity.

Mechanisms by which NisI protects the cell have been proposed, such as inhibition of pore formation activity [16], or assisting the NisF/E/G nisin translocator [9]. Experimental evidence for physical interaction between nisin and NisI has been presented by Stein et al. [13]. Nisin and NisI formed an insoluble and unstable complex proposing a function of NisI as a nisin intercepting molecule lowering the nisin concentration at the cell membrane. No data about the affinity level of the NisI:nisin interaction have yet been published.

Part of the expressed NisI has recently been shown to escape the lipid-modification and to be secreted in a lipid-free form (LF-NisI) to the growth medium both in a heterologous host Bacillus subtilis[13], and in a nisin producing L. lactis[17]. The purpose of LF-NisI secretion has not yet been studied. In this study, we showed the strength of the association of LF-NisI to nisin, and the dissociation rate of the LF-NisI:nisin complex by surface plasmon resonance analysis. Contribution of LF-NisI to nisin immunity was studied by changing the lipoprotein signal peptide of NisI to the signal peptide of the secreted non-lipoprotein Usp45 [18] resulting in efficiently secreted NisI with no lipid moieties. L. lactis cells secreting LF-NisI were shown to tolerate nisin better than the host cells.

2Materials and methods

2.1Bacterial strains, plasmids, and growth conditions

The bacterial strains and plasmids used are presented in Table 1. Escherichia coli strains were grown at 37 °C in Luria-Bertani medium. For plasmid selections, ampicillin or erythromycin (erm) was added to 100 or 250 μg/ml, respectively. L. lactis strains were grown in M17 medium (Difco) supplemented with 0.5% (w/v) glucose (M17G) at 30 °C. For plasmid selections, erm was used at a final concentration of 5 μg/ml (erm5). Analyzing the nisin immunity levels, L. lactis strains were first grown overnight in M17G + erm5. To induce expression of the nisFEG genes, 5 IU nisin/ml was added to the cultures of NZ9840 derivatives. Then, 0.3 μl from the overnight cultures was inoculated into 300 μl of M17G + erm5 containing different concentration of nisin (Sigma) in Bioscreen microtiter plates. The plates were grown in the Bioscreen C (Labsystems) at 30 °C for 36 h. Every hour the plates were shaken moderately for 10 s as well as optical density was measured with a wideband filter (420–580 nm).

Table 1.  Bacterial strains and plasmids used in this study
Strain or plasmidRelevant propertiesReference/Source
 Escherichia coli TG1Transformation host[19]
 E. coli ECO395Production of LF-NisI-GST fusion for purification of lipid-free NisI[11]
 Lactococcus lactis N8Nisin Z producer[20]
 L. lactis MG1614Plasmid-free derivative of NCDO712, Niss[21]
 L. lactis NZ9840Derivative of nisin A producer strain NZ9700 [10]; deletions in nisA, nisI, and nisP. Nisin-inducible nisFEG.Oscar Kuipers, The Netherlands
 L. lactis LAC230MG1614 carrying plasmid pLEB579; no nisin immunity factorsThis work
 L. lactis LAC231MG1614 carrying plasmid pLEB580; wild-type NisI[15]
 L. lactis LAC273MG1614 carrying plasmid pLEB643; secreting lipid-free NisIThis work
 L. lactis LAC276NZ9840 carrying plasmid pLEB579; NisF/E/GThis work
 L. lactis LAC277NZ9840 carrying plasmid pLEB580; NisF/E/G + wild-type NisIThis work
 L. lactis LAC278NZ9840 carrying plasmid pLEB643; NisF/E/G + secreting lipid-free NisIThis work
 pLEB5792.9 kb, Ermr, no NisI[22]
 pLEB5804.2 kb, Ermr, wild-type NisI[15]
 pLEB628P45 +ssusp45 cloned into pCR4-TopoJustus Reunanen, Finland
 pLEB6434.2 kb, Ermr, lipid-free NisIThis work

2.2DNA techniques

Standard procedures were used for agarose gel electrophoresis and PCR amplifications of DNA fragments [19]. L. lactis and E. coli were transformed by electroporation as described previously [23,24]. Restriction enzymes, DNA-ligase, T4-polynucleotide kinase, and Shrimp alkaline phosphatase were used as recommended by the manufacturers.

2.3Construction of LF-NisI secretion plasmid

The nisI gene without its signal sequence was amplified by PCR with primers NIS234 (forward) 5′-AGATCTCGAGTATCAAACAAGTCAAAAAAAGGTG-3′ and NIS235 (reverse) 5′-AGATCTCGAGGGATCCCTAGTTTCCTACCTTCGTTGCAAG-3′. Restriction sites for XhoI and BamHI added to the primers are underlined. The PCR product was translationally fused as an XhoI-fragment to the signal sequence of the lactococcal gene usp45 previously cloned in plasmid pLEB628. The ssusp45-nisI fusion was then cloned as a BamHI-EcoRI-fragment to the plasmid pLEB580, under the control of the constitutive promoter P45 [25,26]. The wild-type nisI gene in pLEB580 was thus replaced with the ssusp45-nisI fusion. The correct construction was verified by PCR and restriction analysis.

2.4Protein purification

Nisin Z was purified as described previously [27] and stored at 4 °C in water adjusted with HCl to pH 2.5. LF-NisI was produced as a glutathione S-transferase fusion protein in E. coli and purified using glutathionine-linked affinity chromatography [11]. After elution of LF-NisI from the thrombin treated glutathionine sepharose column, the elutant in 50 mM Tris pH 7.5, 0.2 M NaCl was run through a Superdex 75 (HR10/30) gel filtration column (Pharmacia) followed by HPLC-MonoQ HR5/5 (Pharmacia) purification using a Varian 5020 liquid chromatograph. The purified LF-NisI was desalted by filtration using a Filtron Omega with a 10 kDa cutoff. LF-NisI was stored in 20 mM Tris-HCl pH 7.5, 1 mM DTT.

2.5SDS-Polyacrylamide gel electrophoresis and Western blot analysis

After electrophoresis in 15% SDS-polyacrylamide gels, proteins were transferred to an Immobilin-P PVDF-filter (Millipore) using a Bio-Rad electroblot. The filter was probed for 1 h with a 1:600 dilution of the KH1422 rabbit antiserum [11], which recognizes the NisI protein, and developed using the Western Breeze chemiluminescent detection kit (Invitrogen).

2.6Surface plasmon resonance analysis

The interaction studies between LF-NisI and nisin were performed using the BIACORE 2000 (Pharmacia Biosensor). Surface plasmon resonance is widely used for studying kinetics of biomolecular interactions [28]. Nisin, LF-NisI and bovine serum albumin (BSA) were covalently bound to the carboxymethylated dextran surface of the CM5 sensor chip by amine coupling using the Amine Coupling Kit (Pharmacia Biosensor). One of the four flow cells was left blank as a control. LF-NisI (50 μg/ml) and BSA (100 μg/ml) were injected on these four flow cells and the response was detected. The response values are expressed in resonance units (RU). One RU represents roughly to a change in protein concentration of 1 pg mm−2 on the sensor surface. For kinetic calculations, nisin was immobilized on surfaces of three flow cells without chemical coupling in concentrations of 15, 30, and 60 μM in 40 mM acetic acid, pH 4.9, 0.1% Tween 20. The three flow cells were stabilized in parallel by extensive wash (25 min at 20 μl/min) with running buffer (20 mM Hepes, 150 mM NaCl, 2.5 mM CaCl2, 0.005% Surfactant P20, pH 7.4). Uncoated flow cell was used as a control. LF-NisI was then injected to the four flow cells in concentrations of 1.1, 2.2, 4.4, and 8.9 μM in the running buffer for 3 min at 20 μl/min. Pre-defined model “1:1 binding with drifting baseline” in BIAevaluation software (Pharmacia Biosensor) was used for kinetic evaluation and curve fitting. The closeness of the fitted model to the experimental data was estimated by the statistical value χ2. According to the BIAevaluation software handbook, χ2 values below ten are considered acceptable.


3.1Surface plasmon resonance analysis of the LF-NisI:nisin interactions

As NisI had previously been shown to interact with nisin [13], we determined the strength of the interaction by surface plasmon resonance analysis. In order to study the interaction between nisin and LF-NisI, these polypeptides were purified and covalently bound to the matrix of a biosensor chip on separate flow cells with BSA as a control. LF-NisI and BSA were run through the flow cells in the BIACORE 2000 and the interaction of the molecules with the covalently bound polypeptides on the sensor chip was detected as changes in the surface plasmon resonance. Attempt to use nisin as the flowing molecule failed, due to high levels of unspecific binding to every flow cell covering the possible specific interactions. The results showed that LF-NisI interacted with nisin but not with BSA or itself (Fig. 1), and that BSA did not interact with nisin, LF-NisI or itself. The baseline for the flow cell with the covalently immobilized nisin drifted due to the continuous dissociation of noncovalently bound nisin. After extensive washing of the sensor surface the baseline of the flow cell no longer drifted. However, no binding of LF-NisI to immobilized nisin could be observed under these conditions (results not shown). This indicated that the previously observed response (Fig. 1) of the LF-NisI:nisin interaction required noncovalently bound nisin. Therefore, for kinetic analyses, nisin had to be immobilized noncovalently on the surface of the sensor chip. Conditions were found in which the noncovalent immobilization of nisin was relatively stable allowing binding of LF-NisI to be observed (Fig. 2). The on and off rates of LF-NisI:nisin complex were measured with LF-NisI concentrations from 1.1 to 8.9 μM. Measurements were hampered by stickiness of nisin to injection channels prior to flow cells in the sensor chip, as observed by unspecific binding to uncoated control flow cell during washing steps and LF-NisI injections. The association (ka) and dissociation (kd) rate constants were calculated from curves in which the analyte concentration was 2.2 μM (Fig. 2; Table 2). The obtained ka and kd propose the equilibrium dissociation constant KD in the micromolar range (0.6–2 μM), showing that the affinity between nisin and LF-NisI is not high. The χ2 values were in acceptable range (< 10) showing that the model fitted well to the experimental data (Table 2).

Figure 1.

Surface plasmon resonance analysis. LF-NisI was injected to four flow cells containing covalently bound nisin, BSA, or LF-NisI. One flow cell was left intact as a blank control. The concentrations of the immobilized polypeptides per mm2 of the flow cell surface at time point zero were 5.8 ng (nisin), 1.7 ng (LF-NisI), 1.8 ng (BSA), and 0.03 ng (blank). The baseline of the nisin-coated flow cell drifted indicating dissociation of noncovalently bound nisin. After a washing step, BSA was injected as a control. The response is expressed in resonance units (RU). One RU represents roughly to a change in protein concentration of 1 pg mm−2 on the sensor surface.

Figure 2.

Kinetic analysis of the LF-NisI:nisin interaction by surface plasmon resonance. Nisin was noncovalently bound to the sensor surface. After stabilization of the flow cells the nisin concentration per mm2 of the flow cell surface at time point zero were (a) 0.6 ng, (b) 1.2 ng, and (c) 1.9 ng. Then, LF-NisI (2.2 μM) was injected (3 min, 20 μl/min), and the association of LF-NisI to surface-bound nisin was detected. Dissociation of the nisin-bound LF-NisI was determined during 5 min after the LF-NisI injection. Uncoated flow cell was used as assay control and the response values thereof were subtracted from the values obtained from the flow cells containing immobilized nisin. The association and dissociation rates calculated from the curves are presented in Table 2. Broken line represents an extrapolation of drifting baselines. The response is expressed in resonance units (RU). One RU represents roughly to a change in protein concentration of 1 pg mm−2 on the sensor surface.

Table 2.  Kinetic constants of the LF-NisI:nisin interaction calculated from BIAcore sensorgrams in Fig. 2
Ligand, nisina (ng mm−2)Analyte, LF-NisIb (μM)ka (M−1 s−1)kd (s−1)KA (μM−1)KD (μM)χ2
  1. aNisin bound to sensor surface after stabilization of the flow cells.

  2. bLF-NisI concentration in the injected liquid. × 1036.2 × 10− × 1037.2 × 10− × 1047.4 × 10−

3.2Secretion of lipid-free NisI

In our previous study [17], we showed that NisI in a wild-type nisin producer strain did not merely exist as a lipoprotein, but was also secreted to the growth medium in a lipid-free form. To facilitate the study of a possible role of the secreted LF-NisI in nisin immunity, the NisI lipoprotein must be completely secretable, i.e., without signals targeting it to the lipid-modification. Therefore, the lipoprotein signal sequence and the site for lipid-modification (Cys1) of NisI were replaced by a signal sequence of a non-lipoprotein origin. The XhoI site used in construction of ssusp45-nisI fusion provided two extra amino acids, Leu and Glu, to the N-terminus of the LF-NisI core protein. The constructed plasmid, pLEB643 containing the secretion cassette for LF-NisI, was practically identical to pLEB580 expressing wild-type nisI, except that the signal sequence of nisI gene had been altered. Plasmids pLEB580 and pLEB643 were transformed into the nisin-sensitive L. lactis strain MG1614 (resulting in strains LAC231 and LAC273) and into the nisFEG expressing strain NZ9840 (resulting in strains LAC277 and LAC278). The localization of NisI in cultures of these transformants was determined by Western blotting (Fig. 3). Expression of wild-type NisI in LAC231 and in LAC277 resulted in a strong band for the cell fraction, and a faint band for the supernatant. This showed that the majority of the produced wild-type NisI existed as a lipoprotein anchored to the cell membrane, and only a small portion was secreted to the medium. In contrast, a strong LF-NisI band was seen in the supernatant fraction of LAC273 and LAC278 cultures, but only a dim band in the cell fraction. This demonstrated that expression of the nisI gene from pLEB643 in two different L. lactis strains caused efficient LF-NisI secretion. The level of NisI expression in all studied transformant strains was estimated to be approximately the same. Also, it was shown that the LF-NisI secretion in transformant strains was approximately at the same level as in the nisin producer strain N8 (Fig. 3).

Figure 3.

Localization of NisI in L. lactis cultures by Western blotting analysis. The NisI polypeptide was recognized by KH1422, a polyclonal NisI-specific rabbit antiserum. Cell, 50 μl of centrifuged cells from overnight L. lactis cultures (N8 cells 10 μl). Sup, 10 μl of supernatants from the same cultures as cell fractions. N8, wild-type nisin-producer; MG1614, nisin-negative host strain; LAC231, MG1614 transformed with wild-type NisI plasmid pLEB580; LAC273, MG1614 transformed with lipid-free NisI plasmid pLEB643; LAC277, NZ9840 transformed with wild-type NisI plasmid pLEB580; LAC278, NZ9840 transformed with lipid-free NisI plasmid pLEB643.

3.3Contribution of secreted LF-NisI to nisin immunity

To investigate if the secreted LF-NisI is able to protect the cell against nisin, L. lactis strains MG1614 and NZ9840 containing plasmids pLEB579 (no NisI), pLEB580 (lipoprotein NisI), and pLEB643 (secreted LF-NisI) were grown with different concentrations of nisin. NZ9840 transformants were first grown overnight with 5 IU nisin/ml to induce the nisFEG expression for determination of possible co-operation between LF-NisI and NisF/E/G. The growth curves of the transformants in different nisin concentrations are shown in Fig. 4. Cells secreting LF-NisI tolerated higher nisin concentrations and grew faster than those without NisI. The protection against nisin by secretion of LF-NisI in MG1614 (LAC273) was, however, of minor importance. On the contrary, LF-NisI export in nisFEG expressing cells provided a significant increase in nisin immunity (Fig. 4(b)), compared to the host strain NZ9840. Even though the immunity level was clearly lower than in cells expressing wild-type nisI, not even the highest concentration, 2000 IU nisin/ml, could completely kill the cells producing NisF/E/G and secreting LF-NisI.

Figure 4.

Growth characteristics of L. lactis strains in different nisin concentrations. (a) L. lactis MG1614 (no nisin immunity genes). (b) L. lactis NZ9840 (nisin-inducible nisFEG expression: ▵, pLEB579, no NisI; ▪, wild-type NisI production from pLEB580; □, secretion of lipid-free NisI from pLEB643. Concentration of nisin in IU/ml is indicated in each figure.


In this study we have shown by surface plasmon resonance analyses that the secreted part of the nisin immunity protein NisI physically interacts, albeit weakly, with its cognate bacteriocin nisin. This interaction was not detectable when the flow cell of the sensor chip contained only covalently bound nisin. This could be explained by sterical hindrance. When bound covalently to the flow cell, nisin (3.3 kDa) is in very close vicinity of the sensor surface. However, nisin is known to adsorb to different surfaces [29]. Here, nisin had affinity to the carboxymethylated dextran surface, since it constantly adsorbed nisin without chemical coupling. After stabilization of the flow cells, the baseline drift caused by dissociation of the noncovalently adsorbed nisin from the surface was reduced enough allowing kinetic examination of LF-NisI:nisin interaction. Even though there were two simultaneous dissociation events, nisin from the surface and LF-NisI from nisin, the latter was clearly the dominating one. LF-NisI dissociated from nisin considerably faster than the baseline drifted (Fig. 2). Also, the drift was taken into account in curve fitting by using the pre-defined model “1:1 binding with drifting baseline” in the BIA evaluation software. The performance of surface plasmon resonance analysis in determination of affinity constants of biomolecular interactions ranges from milli- to picomolar area [28]. The obtained value for the equilibrium dissociation constant KD for LF-NisI:nisin interaction was in the micromolar range (0.6–2 μM), which indicates a rather weak affinity, but in a range to be expected for an interaction that is reasonable to dissociate easily. As the produced nisin is targeted for killing competing cells and for cell signaling [30], and not for forming stable complexes with an immunity protein, it is to be expected that the affinity of nisin to NisI is low. Immunity proteins of bacteriocin producers can function either by direct physical interaction with the bacteriocin or by binding to structures of the cell needed for bacteriocin activity [16]. NisI functions by specific binding to nisin, as shown here and by Stein et al. [13]. In their study, NisI and nisin formed a centrifugable complex, which was easily dissolved in aqueous environment releasing active nisin.

As part of NisI in the wild-type nisin producer is secreted to the medium, and since NisI binds nisin, we suggested previously [17] that constant secretion of LF-NisI followed by binding nisin may contribute to NisI-mediated nisin immunity. To test this hypothesis an expression plasmid directing synthesis of secreted LF-NisI was constructed. The leader peptide of NisI is a strongly conserved lipoprotein signal peptide with a functional site for lipid-modifications [11,31]. This signal peptide was replaced with the Usp45 signal, which does not target NisI for lipid-modification but only for secretion [18]. In addition to the signal peptide, also the amino acid content in the N-terminus of the exported mature protein is important for efficient secretion [32]. The XhoI site used for fusing the Usp45 signal to NisI was chosen because it provided two particular amino acids, Leu and Glu, to the N-terminus. These residues, especially negatively charged Glu in N-terminus of exported proteins, have been shown to significantly improve secretion [33]. Expression of the ssusp45-nisI fusion resulted in efficient secretion of LF-NisI in L. lactis, which tolerated nisin better than the host cells, indicating that the secreted LF-NisI could have a supportive role in nisin immunity in nisin producing cells. The lipid-free NisI-mediated immunity was however lower than the immunity level of lipoprotein NisI, affirming that membrane-bound NisI is needed for full nisin immunity.

The NisF/E/G complex exports membrane bound nisin back to the culture supernatant [12,13,16]. The contribution of LF-NisI to nisin immunity was considerably higher in the cells containing the NisF/E/G transporter than in those with no nisin immunity factors. This phenomenon suggests that LF-NisI assisted the NisF/E/G complex in exporting nisin from cell membrane to environment, or the NisF/E/G transporter supported the protective function of excreted LF-NisI possibly by providing a high local concentration of nisin close to the cytoplasmic membrane from which it could be intercepted by secreted LF-NisI. In our previous study [17] we showed that isolated and purified LF-NisI added to growth medium did not protect cells against nisin. Consequently, the secreted LF-NisI-mediated nisin resistance observed in this work had to take place on the way from membrane to growth medium.

It is unlikely that NisI with or without lipid moieties would block the interaction between nisin and Lipid II, since LF-NisI:nisin interaction is too weak to prevent the 10 to 50-fold higher affinity of nisin to Lipid II [34]. Nevertheless, secretion of LF-NisI may partly prevent nisin from approaching the cell membrane and Lipid II by intercepting nisin in the cell wall, as proposed for the function of lipoprotein NisI by Stein et al. [13]. Further, LF-NisI would act as a carrier transferring nisin away from the cell surface and releasing active nisin to the environment. The nisin, intercepted by LF-NisI, could originate from the external environment or from the membrane, if first translocated by NisF/E/G. The latter would explain the higher protection level provided by LF-NisI in NisF/E/G expressing cells. Membrane protection provided by secreted LF-NisI, instability of the NisI:nisin complex in the aqueous environment [13], and weakness of the physical interaction between nisin and LF-NisI shown here support the idea that secretion of LF-NisI may in this way be an essential part of the nisin immunity mechanism. This proposes a novel mechanism of immunity for a cell to protect its cytoplasmic membrane against the lethal action of a membrane depolarisating amphiphilic peptide.


This work was funded by the Finnish Academy and Valio Ltd. Hilkka Lankinen is acknowledged for her assistance with the BIACORE 2000. The authors thank Kari Kylä-Nikkilä and Justus Reunanen for cloning of ssusp45, and Kaisu Nevalainen for her technical assistance.