Bioassay for Nisin in Milk, Processed Cheese, Salad Dressings, Canned Tomatoes, and Liquid Egg Products

Abstract

A sensitive nisin quantification bioassay was constructed, based on Lactococcus lactis chromosomally encoding the nisin regulatory proteins NisK and NisR and a plasmid with a green fluorescent protein (GFP) variant gfp_uv gene under the control of the nisin-inducible nisA promoter. This strain, LAC275, was capable of transducing the signal from extracellular nisin into measurable GFPuv fluorescence through the NisRK signal transduction system. The LAC275 cells detected nisin concentrations of 10 pg/ml in culture supernatant, 0.2 ng/ml in milk, 3.6 ng/g in processed cheese, 1 ng/g in salad dressings and crushed, canned tomatoes, and 2 ng/g in liquid egg. This method was up to 1,000 times more sensitive than a previously described GFP-based nisin bioassay. This new assay made it possible to detect significantly smaller amounts of nisin than the presently most sensitive published nisin bioassay based on nisin-induced bioluminescence. The major advantage of this sensitivity was that foods could be extensively diluted prior to the assay, avoiding potential inhibitory and interfering substances present in most food products.

Bacteriocins, ribosomally synthesized antimicrobial peptides, are produced by various bacterial species, including members of the lactic acid bacteria group (23). The bacteriocins produced by lactic acid bacteria inhibit the growth of other gram-positive bacteria, which include food-borne pathogens and food spoilage bacteria, such as Bacillus cereus, Clostridium botulinum, Clostridium perfringens, Listeria monocytogenes, and Staphylococcus aureus (28). One of these bacteriocins, nisin, is commercially available and used as a food preservative (E234) in over 50 countries around the world (7), including the United States, countries in the European Union, and Asia. Nisin is a 34-amino-acid peptide produced by Lactococcus lactis (28), and three different natural nisin variants have been discovered: nisin A (17), nisin Z (16, 22), and nisin Q (32). It is suitable for many types of foods from liquid to solid foods, chilled to warm-storage foods, and canned to packaged foods (28). It is mainly used in dairy products, such as processed cheese, cheese spreads, and puddings, but it is also used to preserve salad dressings, vegetables, and even beer (7). However, regulations concerning the levels of nisin allowed in foods differ around the world (1). The activity of nisin decreases during food processing and storage, due to the temperature, pH, and components of the food (6). Therefore, the ability to quantify nisin is essential for monitoring nisin quantities added into foods, as well as its stability throughout the product's shelf life.

Various detection methods (10) based on growth inhibition, such as horizontal agar diffusion (29), immunochemistry (2, 3, 5, 15, 26, 27), and nisin-induced reporter gene expression (25, 31), have been developed to detect and quantify nisin (Table 1). The widely used agar diffusion assay is not able to distinguish nisin from other interfering substances present in the food, causing false-positive results (29). Although immunochemical methods are more sensitive than the agar diffusion assay, they are not totally reliable, due to cross-reactions with compounds structurally related to nisin, which can be present in the testing material (14).

TABLE 1.

Different types of nisin detection and quantification methods and their detection limits in various matrices

Method	Detection limit(s)	Reference
ELISAa for NisA	0.5 ng/ml in pure solution; 250 ng/ml in processed cheese	15
CD-ELISAb for NisA (monoclonal antibodies)	10 ng/ml in pure solution	27
CD-ELISAb for NisA (polyclonal antibodies)	5 ng/ml in pure solution	26
Dot blot immunoassay for NisZ	375 ng/ml in pure solution; 155 ng/ml in milk and whey	3
ELISA for NisZ	0.75 ng/ml in pure solution; 3.5 ng/ml in MRSc medium; 1.7 ng/ml in milk	2
Competitive enzyme immunoassay for NisZ	78 ng/ml in pure solution; 87 ng/ml MRSc medium; 106 ng/ml in milk; 90.5 ng/ml in whey	5
Agar diffusion	12.5 ng/ml in pure solution	29
Nisin bioluminescence assay	0.0125 ng/ml in pure solution; 0.0125 ng/ml in M17GSd medium; 1 ng/ml in milk	31
Green fluorescent protein-based nisin bioassay	2.5 ng/ml in M17Ge medium; 45 ng/ml in milk; 900 ng/g in processed cheese; 1,000 ng/g in salad dressing	25

^aELISA, enzyme-linked immunoabsorbent assay.

^bCD-ELISA, competitive direct enzyme-linked immunoabsorbent assay.

^cde Man, Rogosa, and Sharpe broth.

^dM17 broth containing 0.5% (wt/vol) glucose and sucrose.

^eM17 broth containing 0.5% (wt/vol) glucose.

Wahlström and Saris (31) developed a bioassay based on nisin-induced expression of the bioluminescence genes originating from Xenorhabdus luminescens. They constructed a Lactococcus lactis strain that contained a plasmid with the luxAB gene fused under the control of the nisF promoter and regulated by the NisR and NisK proteins. The genes nisR and nisK are part of the 11-gene cluster that encodes the proteins required in nisin biosynthesis, regulation, and self immunity (21). The genes are arranged into two nisin-inducible operons, nisA/Z/QBTCIPRK and nisFEG (21). The gene nisK encodes an integral membrane-bound sensor kinase, which is activated by extracellular nisin, causing the autophosphorylation of a specific histidyl moiety located at the cytoplasmic side of the membrane (13, 20). The phosphoryl group is then transmitted to the response regulator protein NisR (11, 20). The phosphorylated NisR binds to two promoters, nisA/Z/Q and nisF (20). In the presence of nisin, the NisRK two-component signaling system induces expression of the luxAB gene, resulting in measurable bioluminescence. However, a drawback of this assay is that the growth stage of the indicator cells affects luciferase activity. Therefore, it is difficult to process multiple samples at the same time (31). To avoid this drawback, Reunanen and Saris (25) developed a green fluorescent protein (GFP)-based nisin microplate bioassay by constructing a L. lactis strain, LAC240, which contained a plasmid with nisR and nisK and the reporter gfp gene under the control of the nisF promoter. The sensitivity of this method was less than that of the luciferase assay, with values of 45 ng/ml and 1 ng/ml in milk, respectively (25, 31), but multiple samples could be analyzed simultaneously without considering the energetic state of the cells.

Due to the lack of a nisin quantification method that is capable of processing multiple samples and at the same time having high sensitivity, we constructed in this study a new indicator strain that can be used in the nisin-induced GFP microplate bioassay. With this new strain, the detection limit of nisin was much lower than with the previous indicator strain and that of any existing nisin quantification method presently reported.

MATERIALS AND METHODS

Construction of the indicator strain.

The gfp_uv gene was amplified from the plasmid pGFPuv (Clonetech Laboratories Inc., Palo Alto, CA) by PCR. The primers (Oligomer, Helsinki, Finland) were 5′AGA AAT CAT GAG TAA AGG AGA AGA AC3′ (G0576) and 5′AGC TGC ATG TGT CAG AGG TTT TCA3′ (G0575). The primer G0576 introduced a BspHI site (underlined), which gives compatible ends in respect to NcoI to the 5′ terminus of the gfp_uv gene. The PCR product was introduced to the pCR4-TOPO by the TOPO TA cloning kit (Invitrogen Life Technologies, California), from which the gfp_uv gene was excised with BspHI and EcoRI. The plasmid pNZ8048 (20), which contains the nisA promoter (9), was restricted with NcoI and EcoRI enzymes. The gfp_uv gene was ligated to the linearized pNZ8048, and the resulting plasmid (pLEB651) was electroporated (18) into the non-nisin-producing Lactococcus lactis strain NZ9000 (20), which contains the regulatory genes nisR and nisK in its chromosome. The electroporated cells were plated at 30°C on M17 (Oxoid, Ltd., Hampshire, England) containing 0.5% (wt/vol) glucose and 0.5 M sucrose (M17GS) with chloramphenicol (10 μg/ml). The resulting indicator strain was designated LAC275.

GFP_uv nisin bioassay.

The nisin bioassay was performed essentially as described by Reunanen and Saris (25). Briefly, nisin was added to the 1:100 diluted LAC275 cells in M17 containing 0.5% (wt/vol) glucose, 0.1% Tween 80 (M17GTw), and 10 μg/ml chloramphenicol, so that concentrations of nisin varied from 0 to 70 pg/ml. After overnight incubation at 30°C and removal of 175 μl of medium from each well, GFP_uv fluorescence was measured with the Fluoroscan Ascent 374 fluorometer with Ascent software, version 2.4.2 (Labsystems, Helsinki, Finland). The excitation and emission filters were 373 nm and 538 nm, respectively. The fluorescence was measured as relative fluorescence units (RFU).

Detection of nisin from milk, cheese, salad dressings, canned tomatoes, and liquid egg.

The processed cheese (23% fat; Valio, Ltd., Helsinki, Finland), low-fat milk (1.5% fat; Valio, Ltd., Helsinki, Finland), Thousand Island dressing (22% fat; Saarioinen, Ltd., Huittinen, Finland), and French dressing (48% fat; Saarioinen, Ltd., Huittinen, Finland) were prepared for the GFPuv nisin bioassay as described by Reunanen and Saris (25). All dilutions of food samples were done in 0.1% Tween 80 dissolved in distilled H₂O acidified to pH 2.5 with HCl (0.1% Tween 80). The diluted cheese (25 mg/ml) and milk (1:4) samples were spiked with nisin so that the concentration of nisin in the final assay concentration ranged from 0 to 90 pg/ml and 0 to 300 pg/ml, respectively. The liquid egg (Scanegg, Ltd., Piispanristi, Finland) was diluted 1:100 with 0.1% Tween 80. Various amounts of nisin were added so that the nisin concentration in the assay of the diluted egg varied from 0 to 200 pg/ml. However, to test the robustness of the assay, three other more difficult food matrices, canned tomatoes and two types of salad dressings, were directly spiked with nisin. One gram of Thousand Island dressing or French dressing was directly spiked with 0 to 14 ng and 0 to 18 ng nisin, respectively, prior to adjusting the volume to 40 ml by 0.1% Tween 80. The canned, crushed tomatoes in tomato juice (Euro Shopper, Italy) were prepared for the nisin bioassay in the same manner as the salad dressings, and 0 to 15 ng of nisin was added to 1 g of crushed tomatoes. For each well on a microtiter plate, 50 μl of diluted food sample with or without nisin was combined with 175 μl of 1:100 LAC275 diluted in M17GTw culture medium. Otherwise, the GFP_uv nisin bioassay was performed as described above.

RESULTS

Construction of the L. lactis LAC275 nisin indicator strain.

The GFP microplate bioassay (25) was improved by constructing the plasmid pLEB651 and transforming it into the L. lactis NZ9000 strain. The plasmid contained the gfp_uv gene under the control of the strong P_nisA promoter, while the NZ9000 strain contained the nisin signal transduction genes nisR and nisK in its chromosome. These new L. lactis LAC275 cells were then tested if they could sense nisin in the environment and transduce this signal via the signal proteins NisK and NisR, resulting in expression of the green fluorescent protein.

The background fluorescence of medium and cells in the nisin bioassay.

The M17GTw culture medium alone (no LAC275 cells included) had a background fluorescence of 47.1 ± 2.8 RFU when the GFPuv nisin bioassay was used, while the fluorescence increased slightly to 52.6 ± 0.5 RFU when LAC275 cells, but no nisin, were present in the bioassay. Since the culture medium, cells and food matrices had an effect on the background fluorescence, the background fluorescence (fluorescence of samples not spiked with nisin) was subtracted from the fluorescence values of the samples containing different concentrations of nisin. All nisin fluorescence bioassays were repeated two to three times; each repetition contained four to six parallel wells for each nisin concentration. From the repeated experiments and replicate wells, the lowest nisin concentrations (minimum detection limits) were determined as relative fluorescence unit values that were repeatedly higher than the background fluorescence.

The nisin detection limit and linear dose-response area of L. lactis LAC275.

The detection limit and linear dose-response area using the indicator strain LAC275 was first obtained for the M17G culture medium by spiking the medium with different amounts of nisin. Nisin concentrations of 10 to 70 pg/ml could be detected reliably (Fig. 1A). The L. lactis LAC275 strain was then tested with different food matrices. The food materials caused a film on top of the indicator cells in the microtiter plates, affecting the measurement of fluorescence. Therefore, the food samples were diluted with 0.1% Tween 80. In 25-mg/ml diluted processed cheese, nisin concentrations of 20 to 90 pg/ml (Fig. 1B) were detectable, while the linear dose-response for 1:4 diluted milk was from 50 to 300 pg/ml (Fig. 2A). The liquid egg was diluted 1:100 as growth of LAC275, which is sensitive to lysozyme present in eggs, was impaired when the egg was diluted less. The dose-response relationship in the liquid egg ranged from 20 to 180 pg/ml (Fig. 2B). All the nisin amounts listed above are given as final assay concentrations. Two types of salad dressings were used in the study: Thousand Island dressing as a mayonnaise-based salad dressing and French dressing as a vinegar- and oil-based salad dressing. The linear dose-response area for Thousand Island dressing was 1 to 14 ng/g (Fig. 3A), while it was a slightly broader for French dressing, from 1 to 18 ng/g (Fig. 3B). The detectable range for nisin in canned tomatoes was from 1 to 15 ng/g (Fig. 3C). The detection limits from all food products analyzed with the LAC275 strain in the nisin bioassay were much lower than those for the LAC240 indicator strain (Table 2).

FIG. 1.

Standard curves for nisin using L. lactis LAC275 in the GFP_uv nisin bioassay, given as final assay concentrations in M17G culture medium containing 0.1% Tween 80 and 10-μg/ml chloramphenicol (M17GTw) (A) and in 25 mg processed cheese diluted in 1 ml of 0.1% Tween 80 (B). The means and standard deviations of four replicate wells are shown. From the fluorescence values obtained for samples containing nisin, the background fluorescence values (samples not supplemented with nisin) of 50.3 ± 0.8 RFU for M17GTw and 47.9 ± 2.5 RFU for processed cheese were subtracted to obtain the relative fluorescence units. The R² values for the linear curves for the M17GTw culture medium and for the cheese were 0.9906 and 0.9686, respectively.

FIG. 2.

Standard curves for nisin in 1:4 diluted low-fat milk (four replicates) (A) and in 1:100 diluted liquid egg in 0.1% Tween 80 (five replicates) (B), given as final assay concentrations. The averages and error bars are shown. The background fluorescence values (49.7 ± 0.6 RFU for milk and 52.4 ± 0.4 RFU for liquid egg) were subtracted from the obtained RFUs. The corresponding R² values for the curves were 0.9856 (milk) and 0.9771 (liquid egg).

FIG. 3.

Standard curves for foods spiked with nisin: Thousand Island salad dressing (A), French salad dressing (B), and canned tomatoes (C). A total of 1 g of each food was diluted in 40 ml of 0.1% Tween 80. The means and standard deviations of five replicate wells are shown. The fluorescence values obtained from samples without added nisin (43.8 ± 4.5 RFU for Thousand Island dressing, 46.7 ± 1.2 RFU for French dressing, and 45.8 ± 1.0 RFU for canned tomatoes) were subtracted from the measured RFUs. The R² values for the Thousand Island dressing, French dressing, and canned tomatoes were 0.9905, 0.9927, and 0.9896, respectively.

TABLE 2.

The detection limits of GFP-based nisin bioassays using the L. lactis LAC275 strain developed in this study and L. lactis strain LAC240 previously developed (25) in different food matrices

Matrix	Detection limit for a:		Sensitivity comparison factorb
	LAC240	LAC275
M17Gc	2.5 ng/ml	10 pg/ml	250×
Low-fat milk	45 ng/ml	0.2 ng/ml	225×
Processed cheese	0.9 μg/g	3.6 ng/g	250×
Thousand Island salad dressing	1 μg/g	1 ng/g	1,000×
French salad dressing	1 μg/g	1 ng/g	1,000×
Liquid egg	Not tested	9 ng/g

^aThe nisin concentrations are indicated as the concentrations present in undiluted foods.

^bThe ratio of the detection limits of LAC275 to LAC240.

^cM17 broth containing 0.5% (wt/vol) glucose.

DISCUSSION

The L. lactis strain LAC275 constructed in this study contained the gfp_uv reporter gene encoding fluorescing GFPuv under the control of the nisin-inducible nisA promoter and the regulatory genes nisR and nisK in its chromosome. This combination allowed sensitive nisin quantification from various foods ranging from milk to difficult matrices such as liquid egg. The GFPuv nisin bioassay described in this study was up to 1,000 times more sensitive (Table 2) than the GFP nisin bioassay with the LAC240 strain developed by Reunanen and Saris (25). The detection limit of the bioassay was increased, due to the specific construction of strain LAC275. The GFPuv fluoresced 18 times more brightly than the wild-type GFP when expressed in Escherichia coli (4). Due to the strong fluorescence, even small amounts of nisin-induced GFPuv could be detected and therefore partly explain why the sensitivity of the assay described in this study was high. In addition, the nisA promoter in LAC275 used in this assay is a stronger promoter than the nisF promoter used in LAC240, resulting in a higher level of expression of the gfp_uv gene (8, 12). The expression levels of nisRK regulatory genes are important in the final expression of the reporter gene under the control of the nisA promoter (19, 30). The nisRK genes were integrated into the chromosome of the indicator strain LAC275 rather than into a plasmid, as in the case with LAC240, since Pavan et al. (24) have shown with Lactobacillus plantarum that chromosomal location of the nisRK genes produced a more genetically stable recombinant strain than plasmid-encoded NisRK, resulting in better reproducibility and dose-dependent nisin induction.

Not only was the nisin detection limit improved by the new LAC275 strain, the time to perform the bioassay was also shortened. This was due to the fact that the GFPuv protein, compared to the red-shifted P11 GFP variant used in LAC240 (25), did not require a 30-min −20°C incubation and approximately 20-min thawing of the cells for protein maturation before fluorescence measurement. Therefore, the nisin-induced fluorescence of LAC275 cells could be measured directly without the freezing step.

The specificity of the GFP nisin bioassay is good as the NisRK pathway and produces approximately the same responses to both variants nisin A and nisin Z, and no response with subtilin, the bacteriocin structurally most similar to nisin (31). Therefore, as nisin Q (32) is structurally similar to the other variants of nisin, the GFPuv-based nisin bioassay described in this paper might possibly be used to detect and quantify nisin Q as well.

Food may contain interfering materials which can cause problems when nisin is quantified. One major advantage of the low detection limit of nisin in the LAC275-based nisin bioassay is that the foods for nisin quantification can be extensively diluted, thereby avoiding the adverse effects of potential interfering materials. The processed cheese, liquid egg, and the salad dressings caused problems without dilutions in the fluorescence measurements; therefore, these foods had to be diluted prior to the nisin bioassay. Due to the sensitivity of the LAC275-based nisin bioassay, food samples could be diluted hundreds of times without the assay's losing the ability to detect nisin present in the food. For example, in the United States, a maximum level of 250 mg nisin per kilogram of food is allowed to be used in pasteurized cheese and processed cheese spreads, while only 15 mg of nisin per kilogram is allowed in sauces and nonstandard salad dressings (1). Pasteurized, chilled soups can contain up to 5-mg/kg nisin and liquid egg products can contain 15-mg/kg nisin with written permission from the U.S. Department of Agriculture (1). The LAC275 indicator strain was tested using the GFPuv nisin bioassay on processed cheese, salad dressings, and liquid egg. The maximum amount of nisin allowed in the United States in processed cheese may contain 10⁴ times more nisin and the salad dressing and the liquid egg both may contain approximately 10³ times more nisin than the detection limit of LAC275 (Table 2). Therefore, these foods can be diluted extensively for nisin quantification; at the same time, interfering materials will be diluted as well.

The method described in this paper was able to detect 0.2 ng of nisin per ml in milk, while presently the most sensitive nisin quantification, the luciferase assay (31), could detect only 1 ng of nisin per ml in milk. Clearly, the nisin fluorescence bioassay with the new indicator strain LAC275 has the lowest detection limits presently described. Furthermore, no sensitive nisin quantification method has previously been described for salad dressings, canned tomatoes, and liquid egg. Therefore, using this strain for quantification and detection of nisin from these and other foods should be a preferable choice.