The antimicrobial metabolite nisin Z reduces intestinal tumorigenesis and modulates the cecal microbiome in Apc^Min/+ mice

ABSTRACT

Nisin Z, an antimicrobial metabolite produced by Lactococcus lactis, has been safely used as a food preservative for many years. Nisin Z also showed promising activity against various cancer types in vitro and significantly reduced tumor size in an ectopic head and neck cancer model. Here, we investigate the activity of nisin Z for colorectal cancer treatment and observed an in vitro reduction in cellular proliferation and a moderate enhancement in cell death. We next analyzed the effect of oral nisin Z administration in the Apc^Min/+ intestinal adenoma mouse model. We measured tumor burden along the gastrointestinal tract and observed a decrease in tumor burden in the middle region of the small intestine but not in the lower region or colon. Since tumor progression in the Apc^Min/+ model is exacerbated by an inflammatory environment, we next determined whether nisin Z impacts this in a direct or indirect manner. We show that nisin Z directly reduces NF-κB activation in a dose-dependent manner in a cell-based assay, and we observe a corresponding reduction in NF-κB protein expression in the middle region of the small intestine. In addition, nisin Z impacted the cecal microbiome composition as well as microbiota-associated plasma metabolites, leading to changes that are in part associated with beneficial outcomes. Interestingly, the Apc^Min/+ genotype differentially impacted the nisin Z-mediated differences in cecal microbiome composition and plasma metabolites compared to wild-type animals. In summary, our data suggest that the reduction in small intestinal tumor burden could be due to nisin Z’s contribution to a reduced pro-inflammatory environment. Future studies will reveal whether nisin’s localized effect is due to degradation of the peptidic compound in more distal regions of the gastrointestinal tract and focus on development of delivery systems to increase efficacy.

IMPORTANCE

With the increased incidence of colorectal cancer, especially among younger individuals, it is critical to study approaches that help with the prevention and treatment of this debilitating disease. Our study indicates that nisin Z, a bacterially produced peptide antibiotic, decreases the growth of colorectal cancer cells and moderately increases cell death in vitro. Oral administration of nisin Z in an intestinal adenoma mouse model revealed a reduction of tumor burden in the middle region of the small intestine. This decreased tumor burden might in part be attributed to a direct anti-inflammatory effect, as well as an indirect effect on the gut microbiota and their metabolites due to nisin Z’s antibacterial activity. Overall, we demonstrate a potential activity for nisin Z in the prevention or amelioration of inflammation-associated colorectal cancer, underscoring the significance of investigating the properties of bacterial natural products in human health.

INTRODUCTION

Colorectal cancer (CRC) is the third most common cancer type worldwide, typically affecting individuals over the age of 50. CRC can have genetic origins, causing an increased predilection for development of polyps in the colonic or rectal epithelium (1, 2). In addition, CRC progression can be influenced by a combination of factors, including lifestyle choices and genetic conditions (3, 4). The gut microbiota can play a crucial role in regulating CRC through various microbiome-dependent metabolites. For example, indoles resulting from gut microbial tryptophan metabolism play an important role in maintaining intestinal barrier homeostasis and attenuating CRC progression (5).

A common genetic mutation in the Apc gene can lead to familial adenomatous polyposis coli (Apc), causing people to sporadically develop colorectal adenomas (5). These characteristics are modeled by the C57BL/6 Apc^Min/+ mouse model, which predisposes mice to develop “multiple intestinal neoplasia” (Min), resulting in tumor formation across their small intestine and colon (6). The Apc^Min/+ model is readily amenable to testing medicinal interventions as well as lifestyle choices aimed at the prevention of CRC development. For example, high-fat diet consumption in the Apc^Min/+ model leads to an increase in intestinal tumor burden due to low-grade inflammation and altered epithelial permeability (7). In addition, this model has been extensively used to study the effects of dietary and lifestyle interventions, including micronutrients (8–11), caloric restriction (12), and exercise (13). The impact of dietary changes on CRC development in Apc^Min/+ mice is paired with changes in the gut microbiome and metabolome (4, 14, 15).

Nisin is a post-translationally modified peptide produced by Lactococcus lactis that has antibacterial properties against gram-positive bacteria. Due to its antibiotic activity, nisin Z is commonly used as a food preservative in dairy products and has received the “Generally Recognized as Safe” status by the US Food and Drug Administration (FDA). More recently, nisin and its variants have gained attention because of their potential use in medical applications, including gastrointestinal (GI) infections and treatment of certain cancer types (16–20). Most notably, nisin Z has potent activity in reducing head and neck cancer cell proliferation as well as angiogenesis (18). This effect has in part been attributed to increased calcium influx and subsequent cell cycle arrest mediated by the pro-apoptotic cation transport regulator CHAC1 (21). CHAC1 upregulation has been associated with various cancer types, including CRC (22, 23). Orally administered nisin Z significantly reduces tumor proliferation in a preclinical model (18), and is currently in a phase I clinical trial for squamous and head and neck carcinoma (24).

In this paper, we will determine the impact of nisin Z on cellular proliferation, inflammatory pathways, and the gut microbiome, as it affects CRC. We found that nisin Z decreases CRC cellular proliferation while slightly enhancing cell death. We also show a dose-dependent effect on NF-κB activation, as well as shifts in the gut microbiome and microbial metabolites due to nisin Z’s antibacterial properties.

RESULTS

Nisin Z reduces Caco-2 cell proliferation and enhances cell death

Given the promising effects of nisin Z in attenuating head and neck cancer (18), we investigated its potential activity on CRC. We first looked for the effect of nisin Z on Caco-2 (human colon adenocarcinoma) cell proliferation. Using an automated microplate live-cell imaging system, we monitored cellular confluency for 72 h at nisin Z concentrations up to 300 µM (1 mg/mL) (Fig. 1A and B). At concentrations of 30 or 150 µM, nisin Z treatment did not significantly affect the increase in Caco-2 confluency compared to the vehicle control, which reached ~80% confluency at 72 h. However, at a nisin Z concentration of 300 µM, we observed a significant reduction in the cellular confluency increase after 48 h. To investigate whether this reduced confluency increase is driven by a decreased proliferation rate or increased cell death, we next performed a LIVE/DEAD stain to test for potential cytotoxicity after 24 and 72 h nisin Z treatment (Fig. 1C). We observed a significant increase in the percentage of dead cells after exposure to 300 µM nisin Z at 72 h. In summary, these results indicate that nisin Z treatment dose-dependently affects Caco-2 cell growth, which could be the result of an antiproliferative activity combined with a modest cytotoxic activity.

Fig 1.

Nisin Z reduces Caco-2 cellular proliferation while increasing cell death. (A) The vehicle-only control group, as well as cells treated with 30 or 150 µM nisin Z, steadily proliferated and reached ~80% confluency at 72 h. Cells treated with 300 µM (1 mg/mL) nisin Z show reduced proliferation after 48 h of treatment. (B) Nisin Z significantly decreases proliferation 48 and 72 h post-treatment at 300 µM. The control group (as well as the 30 and 150 µM nisin Z treatments) had a ~3.2-fold increase in confluency at 72 h, whereas the increase for 300 µM nisin Z was only ~1.8-fold. P-values shown were calculated using two-way ANOVA (n = 6 repeats; *P < 0.05, ****P < 0.0001). (C) After 72 h, ~15% of cells in the vehicle-only control group are dead. Exposure to nisin Z for 72 h slightly increases the proportion of dead cells to 23%. Statistical analysis was performed with one-way ANOVA (n = 6; ***P < 0.001). Individual points represent individual experiments, bars represent group means, and error bars represent SD.

Oral nisin Z reduces the number and size of small intestinal adenomas in Apc^Min/+ mice

We next tested the effects of nisin Z on GI adenoma formation in the C57BL/6J Apc^Min/+ animal model. We selected this model because Apc^Min/+ mice are genetically predisposed to develop adenomas across their GI tract. This adenoma development can be modulated by therapeutic or environmental interventions, including modifications in the diet and microbiota (13, 25). To model an inflammatory environment that induces adenoma formation in our Apc^Min/+ model, similar to dietary-driven inflammation-induced intestinal adenoma formation in humans, the animals were fed a high-fat diet post-weaning (7, 15, 26). As Apc^Min/+ tumor formation is characterized by sex-dependent differences, we used all female animals in this study (27, 28). Nisin Z was provided to the experimental group in their drinking water for 12 weeks post-weaning at a concentration of 1 mg/mL (300 µM), which is within the typical range for oral antibiotics administered in animal models and corresponds to the concentration for which we observed bioactivity in our Caco-2 cell-based assay (Fig. 1). The control mice received regular drinking water. After 12 weeks on treatment, we dissected out the small and large intestines and prepared them for gross morphological analysis using our 3D-printable MIST device (29). The nisin Z group had significantly fewer tumors in the middle segment of their small intestine compared to the control animals (Fig. 2A). In addition, nisin Z treatment also resulted in an ~50% reduction in tumor area in this region (Fig. 2B). We did not observe significant differences in tumor number or area in other small intestinal regions or the colon (Fig. 2A and B).

Fig 2.

Oral administration of nisin Z in the drinking water of Apc^Min/+ mice reduces tumor numbers and sizes in the middle region of the small intestine (SI) after 12 weeks. (A) Tumor counts of the upper, middle, and lower SI, as well as the colon of Apc^Min/+ mice on regular drinking water or drinking water (control) containing 1 mg/mL nisin Z. (B) Tumor burden is represented by the sum of the tumor areas per mouse for each intestinal segment. Statistical analysis was performed using a Mann-Whitney test (n = 6–10 animals; **P < 0.01, and ***P < 0.001). Individual points represent individual animals, bars represent group means, and error bars represent SD.

Since our cell-based assays suggested that the effect of nisin Z might be mostly due to reduced proliferation, we next compared cyclin D1 expression in the middle small intestine and the colon. Cyclin D1 is a crucial cell cycle regulatory protein that permits the transition of G1 to S phase (30). We observed a significant decrease in cyclin D1 expression in the middle region of the small intestine of nisin Z-treated Apc^Min/+ mice (Fig. 3A). Unlike the colon, where nisin Z had no effect on cyclin D1 expression (Fig. 3B). Our results suggest that nisin Z could be associated with the observed lower tumor burden in the middle section of the small intestine by decreasing cyclin D1 expression and thereby reducing tumor cell proliferation.

Fig 3.

Nisin Z reduces cyclin D1 expression in the middle region of the small intestine of Apc^Min/+ mice. (A) Oral nisin Z (at 1 mg/mL) reduces cyclin D1 expression in the middle region of the small intestine (SI) after 12 weeks. (B) Cyclin D1 expression in the colon is not significantly affected by oral nisin Z treatment. Statistical analysis was performed using an unpaired t-test (n = 4 animals; *P < 0.05). mRNA expression levels were calculated based on the ΔΔ-CT method and, genes of interest were normalized to the housekeeping gene β-actin. Individual points represent individual animals, bars represent group means, and error bars represent SD.

Nisin Z attenuates the NF-κB mediated inflammatory response in vitro and in vivo

Since tumor development in the Apc^Min/+ model can be affected by several factors like diet, gut microbiota, as well as an inflammatory environment, we decided to examine these different factors further. As an antibiotic, nisin Z could alter the composition of the gut microbiota, which can result in an indirect effect on the inflammatory environment in the GI tract. Alternatively, nisin Z could exert a direct, microbiota-independent anti-inflammatory effect. In Apc^Min/+ mice, inflammation is associated with tumor progression through pathways such as the nuclear factor kappa B (NF-κB) (31–33). We tested the effect of nisin Z on NF-κB signaling using the human monocytic THP-1 Blue reporter assay. Addition of the Toll-like receptor four ligand lipopolysaccharide (LPS) results in an activation of the NF-κB pathway, which is quantified via the activity of a secreted embryonic alkaline phosphatase. Pretreatment of the cells with a range of different nisin Z concentrations (1.5–150 µM) for 4 h prior to the LPS challenge resulted in a dose-dependent reduction in NF-κB activation (Fig. 4A). This indicates that nisin Z could have the capability of directly reducing the response to a proinflammatory stimulus, even in the absence of the gut microbiota or their metabolites. Since we observed a modest cytotoxicity effect of nisin Z on Caco-2 cells, we next investigated whether the observed decrease in NF-κB activation at higher concentrations is due to modulation of the signaling pathway rather than increased cell death. We performed a live/dead stain and found no differences in THP1 viability across the concentration range tested (Fig. 4B). Taken together, these results show that nisin Z can have a modest, direct anti-inflammatory effect, which could contribute to reducing the inflammatory environment in our animal model.

Fig 4.

Nisin Z attenuates an LPS-induced proinflammatory response in a dose-dependent manner. (A) The NF-κB-mediated proinflammatory response is monitored using the THP-1 Blue monocyte reporter assay. Induction of NF-κB is coupled to an increase in the secreted embryonic alkaline phosphatase (SEAP), for which the activity is quantified using a colorimetric assay. A proinflammatory response is induced with 50 ng/mL LPS and addition of different nisin Z concentrations (1.5, 15, and 150 µM). (B) THP-1 Blue cells maintain their viability across the range of nisin Z concentrations in the NF-κB assay. Statistical analysis was performed using one-way ANOVA with a post-hoc Dunnett’s test (n = 6–8 repeats; ****P < 0.0001). Individual points represent individual experiments, bars represent group means and error bars represent SD.

We next investigated the anti-inflammatory effect of nisin Z (300 µM) on C57BL/6 Apc^Min/+ and corresponding C57BL/6 wild-type (WT) (Apc^+/+) mice by monitoring the expression of NF-κB. Protein extracts from the middle segments of the small intestine were analyzed by Western blot using an antibody for detecting total NF-κB (p65/RelA). The corresponding bands are quantified using densitometry analysis, and the NF-κB signal is calculated as the ratio of NF-κB to the level of housekeeping protein β-actin. Our quantification showed a significant nisin Z-dependent reduction in NF-κB levels in both Apc^Min/+ (Fig. 5A) and WT mice (Fig. 5B). Since NF-κB expression is associated with inflammation-driven cancer (34), the decrease in NF-κB signal suggests a reduction in the inflammatory environment in the GI tract, potentially contributing to the suppression of tumor growth in the mid-segment of the small intestine.

Fig 5.

Orally administered nisin Z reduces NF-κB expression in the middle segment of the small intestine of Apc^Min/+ and WT C57BL/6 mice. NF-κB levels were quantified by Western blot analysis and standardized to the expression of the β-actin housekeeping protein for control and nisin Z-treated mice. Nisin Z treatment reduces the level of total NF-κB signal in (A) Apc^Min/+ and (B) WT mice. NF-κB signal is defined as the expression ratio of NF-κB to β-actin. Statistical analysis was performed using an unpaired t-test (n = 4 animals; **P < 0.01, *P < 0.05). Individual points represent individual animals, bars represent group mean, and the error bar represents SD.

Oral nisin Z treatment differentially affects the cecal microbiome composition and microbiome-dependent plasma metabolites in WT and Apc^Min/+ mice

Since nisin Z is primarily known for its antibiotic properties, we next assessed its potential indirect anti-inflammatory impact via the gut microbiota and their metabolites. We used 16S rRNA sequencing to determine the cecal microbiota composition of Apc^Min/+ mice on nisin Z versus control drinking water. To account for potential differences due to host genotype, we compared these data to the cecal communities of C57BL/6J WT female littermates, which were randomized into similar control and nisin Z drinking water groups. To account for potential cage effects, we included cecal samples from animals that were housed in two distinct cages for each group. As can be expected from an antibiotic intervention, nisin Z reduced the alpha diversity in WT mice, as estimated by the Chao1 diversity index (Fig. 6A). Interestingly, the Apc^Min/+ alpha-diversity was less impacted by nisin Z treatment, though a similar reduced trend was observed. Canonical correspondence analysis highlights the difference in beta-diversity between the control WT and Apc^Min/+ communities, which seem to converge in a separate cluster upon nisin Z treatment (Fig. 6B). The total diversity plot shows general trends in response to nisin Z treatment that are observable in both the WT and Apc^Min/+ backgrounds (Fig. 6C). Most notably, nisin Z reduces the relative abundance of Faecalibaculum and Lactobacillus at the expense of relative increases in Helicobacter, Akkermansia, Odoribacter, Lachnospiraceae (A2), Bilophila, and Alistipes. These differences are also reflected in the total diversity plot (Fig. 6D) and clustering heat map (Fig. 6E), which additionally revealed a decimation of Romboutsia upon nisin Z treatment and underscores the contributions of Lachnospiraceae, Faecalibaculum, and Lactobacillus to community differences in Apc^Min/+ mice. Our data show that nisin Z causes shifts in the microbial communities, as would be expected from an antibiotic, and in addition highlights potential differences in community and nisin Z response between WT and Apc^Min/+ littermates.

Fig 6.

Orally administered nisin Z differentially affects the cecal microbiome composition in WT and Apc^Min/+ mice. (A) Alpha-diversity (as estimated by Chao1 diversity index) is significantly more impacted by nisin Z treatment in WT animals. Pairwise analysis was done using Wilcoxon rank sum t-test. (B) Canonical correspondence analysis shows convergence of the beta-diversity of WT and Apc^Min/+ mice treated with nisin, in contrast to the rather distinct control microbial communities for these mouse genotypes. We used PERMANOVA analysis (R² = 0.36, P = 0.014). (C) Stacked bar chart of relative abundances (left y-axis) of the top genera for each of the four groups. (D) Differential abundance analysis highlighting the genera that are significantly altered by nisin Z treatment or mouse genotype. Pairwise test used was “metagenomeSeq” with Benjamini-Hochberg correction. (E) Heatmap of the differentially abundant taxa among the different groups. Statistical test used was ANOVA with Benjamini-Hochberg correction. The color gradient represents the relative abundance z-score (n = 4–5 animals; *P < 0.05).

To inquire whether the observed differences in gut microbiome composition could have anti-inflammatory implications, we analyzed peripheral plasma from WT and Apc^Min/+ mice using a targeted liquid chromatography-tandem mass spectrometry (LC-MS/MS) panel for microbiome-associated metabolites. In line with the microbiome composition data, we observed host genotype-dependent effects on the systemic concentration of certain metabolites in response to nisin Z treatment (Fig. 7). In Apc^Min/+ mice, the plasma concentration of the microbiome-influenced metabolite serotonin was increased with nisin Z treatment, whereas the amounts of phenyl sulfate and 4-hydroxyhippuric acid were lower compared to the control animals. In WT animals, nisin Z was found to impact the plasma concentration of hippuric acid. Interestingly, nisin Z treatment resulted in genotypic differences in the plasma concentrations of 4-hydroxyphenylpyruvic acid, indoxyl sulfate, and phenylpyruvic acid.

Fig 7.

Oral nisin Z treatment differentially affects gut microbiome-dependent plasma metabolites in WT and Apc^Min/+ mice. Peripheral plasma concentrations of microbiome-dependent metabolites were measured by LC-MS/MS, including phenyl sulfate, hippuric acid, serotonin, indoxyl sulfate, and 4-hydroxyhippuric acid. Statistical analysis was performed using the Kruskal-Wallis nonparametric test (n = 7–10 animals; *P < 0.05, **P < 0.01, and ***P < 0.001. Individual points represent individual animals, bars represent group means, and error bars represent SD.

DISCUSSION

Nisin Z is a post-translationally modified ribosomal peptide antibiotic produced by certain strains of L. lactis subsp. lactis and is commonly used in food preservation (35). More recently, the compound has gained interest for its potential anti-tumor activity in head and neck cancer (18). Additional in vitro studies have revealed similar anti-proliferative effects on CRC cell lines (19, 20, 36, 37), though no in vivo efficacy has yet been demonstrated for this cancer type. Our in vitro observations that nisin Z reduces Caco-2 proliferation and has a modest cytotoxic effect are in line with these prior studies. We also explored a potential direct anti-inflammatory effect of nisin Z via modulation of the NF-κB pathway, and this might be linked to the observed reduction in cellular proliferation (38). NF-κB signaling has been linked to G1-S phase transition as well as inflammatory pathways, and we found that nisin Z can exhibit modest anti-inflammatory activity through this pathway. Overall, a reduction in inflammation is associated with reduced tumor formation in the Apc^Min/+ model (38, 39) as well as with a better overall CRC prognosis (33).

Here, we report the first in vivo investigation of the potential anti-intestinal adenoma activities of nisin Z using the Apc^Min/+ animal model. We observed a decrease in the number and size of tumors in the middle section of the small intestine. This was associated with a decrease in both cyclin D1 expression as well as NF-κB levels, indicating that a sufficient concentration of bioactive nisin Z might be present in this GI region. We did not observe a nisin Z-dependent effect on the tumor burden or cyclin D1 expression in the colon. We speculate that nisin Z might be partially degraded along the GI tract due to its proteinaceous nature, thereby not retaining sufficient concentrations to exert anti-tumor activity in the lower small intestine and colon. The apparent lack of effect in the upper small intestinal segment could potentially be due to differences in transit time and pH. Despite the potential partial degradation of nisin Z in the upper GI tract, it still had a significant impact on the cecal microbiome composition, as well as microbiome-dependent plasma metabolites. It remains to be determined whether these effects are due to a modulation of the small intestinal microbiota with downstream impact on the cecum or whether sufficient amounts of nisin Z make it to the lower GI tract to exert antibiotic activity. Nisin was found to exert similar effects on the microbiota in a porcine model, suggesting survival through the GI tract (40). While an earlier study of the GI transit for the related compound lacticin 3147 did not detect any intact compound across the GI tract by mass spectrometry, potentially due to a lower administered oral dose (41). The nisin Z dose used in our in vivo experiments (300 µM) was chosen because this corresponds to ~1 mg/mL in the drinking water, which is within the typically used range in our lab for oral antibiotics. This dosage is not expected to be toxic to the animals since a 90-day study in rats showed no toxic effects for doses up to 3,000 mg/kg/day (42). On average, a 30 g adult mouse consumes 5 mL of water daily, corresponding to a daily dose of 5 mg of nisin Z, or ~167 mg/kg. This dosage is in the same range as another study into the compound’s anti-tumor effect, where experimental groups were administered 400 or 800 mg/kg for the treatment of head and neck cancer (18). The nisin Z concentration range used in our human monocytic THP-1 Blue reporter experiments (1.5–150 µM) was chosen below the in vivo oral dose to account for degradation in the GI tract and assess cellular responses under controlled conditions that better mimic physiological concentrations.

We observed genotype-specific changes in the gut microbiota of WT and Apc^Min/+ mice upon treatment with nisin Z. As expected from its antimicrobial activity spectrum, nisin Z causes an overall decrease in the relative abundance of gram-positive bacteria. Several of the differentially abundant genera in our data have previously been implicated in CRC. Nisin Z treatment increased the abundance of Bacteroides in our Apc^Min/+ mice. This is beneficial, as a decrease in Bacteroides abundance has been associated with worse CRC outcomes, though this is highly strain-dependent (43). Additionally, we observed an increase in Akkermansia with nisin Z treatment, which has previously been linked to a reduction in intestinal inflammation through the NF-κB pathway, leading to a decrease in tumor formation (44). An increase in Odoribacter might protect the host against CRC through enhanced cellular apoptosis (45, 46). However, this is strain-dependent, as certain strains of Odoribacter are associated with CRC (47). We also observed some unexpected changes in the microbiota, which could have a negative impact on the GI inflammatory environment. In contrast to previously published work, nisin did not inhibit the growth of Helicobacter in our animals (48). Certain Helicobacter species (such as Helicobacter pylori) have been associated with the development of gastric and CRC (49), yet this is unlikely to be the case in our animals, which were raised in specific pathogen-free conditions. Furthermore, nisin Z treatment increased the abundance of Bilophila in Apc^Min/+ mice. Bilophila spp. are prominent producers of hydrogen sulfide, which can enhance inflammation and cellular proliferation, potentially worsening CRC (50). Overall, the gut microbiome analysis by us, as well as others, suggests that nisin Z has anti-inflammatory properties and impacts microbial composition and diversity (40, 51).

We performed a functional interrogation of the microbiota by measuring microbiome-associated metabolites in the peripheral plasma. Apc^Min/+ mice receiving nisin Z treatment showed increased serotonin levels. The majority of serotonin is produced in the gut from tryptophan, and this process is influenced by the gut microbiota (52). Lactobacillus spp. are among these serotonin regulators, and while their abundance was previously found to be directly correlated to serotonin levels (53), Lactobacillus was actually reduced in our nisin Z-treated animals. Additionally, we observed a decrease in the 4-hydroxyhippuric acid plasma levels upon nisin Z treatment of Apc^Min/+ mice. It has previously been reported that 4-hydroxyhippuric acid promotes CRC cell growth in vitro and enhances tumorigenesis in Apc^Min/+ mice (54). In addition, patients with advanced CRC exhibit higher levels of 4-hydroxyhippuric acid in their tumor tissues (54). We speculate that the reduction in 4-hydroxyhippuric acid upon nisin Z treatment could contribute to the inhibition of tumor formation in our model and plan to test this in future experiments. While the 4-hydroxyhippuric acid levels in our WT animals did not change, we observed a similar decrease in hippuric acid upon nisin Z treatment instead. This observation is counterintuitive since hippuric acid is generally associated with health-associated parameters, including restriction of colonic damage and reduction in pro-inflammatory cytokines (55). The reduction in hippuric acid levels in the plasma of nisin Z-treated WT mice is associated with decreases in Eubacterium, Lachnospiraceae, and Romboutsia (Fig. 6D and E). Taken together, these observations indicate that nisin Z-associated gut microbial changes could affect the differential production of hippuric acid and 4-hydroxyhippuric acid. We also observed a decreasing trend in the plasma concentrations of the uremic toxins, indoxyl sulfate, and p-cresol sulfate, in nisin Z treated Apc^Min/+ mice. Bacteroides spp. are prominent producers of indoxyl sulfate via the metabolism of tryptophan to indoles (56). Despite a relative increase in Bacteroides spp. abundance, we observed a decrease in indoxyl sulfate plasma concentration, seemingly contradicting prior observations. In addition, Coriobacteriaceae spp. and Clostridium spp. have been reported to utilize tyrosine to produce p-cresol sulfate (57). Prior studies have revealed an association between both uremic toxins and an increase in pro-inflammatory cytokines, exacerbating CRC (58–61). This is in line with our observed direct relationship between Coriobacteriaceae spp. abundance and p-cresol sulfate concentration. Taken together, the reduced levels of uremic toxins could contribute to an overall decrease in luminal inflammation in nisin Z-treated Apc^Min/+ mice.

In future studies, we plan to compare the efficacy of direct nisin Z treatment with administration of probiotic L. lactis subsp. lactis. While our results indicate a potential direct effect of nisin Z on proliferation and inflammatory response in cell-based assays (Fig. 1 and 4), such a side-by-side comparison will help identify whether the indirect effects of other microbiota-associated metabolites are more significant. Administration of the natural producer might circumvent our issue with potential nisin Z degradation prior to reaching the colon. On the other hand, the probiotic strain might not survive or exert bioactivity in the lower GI tract, and hence, we are working on the development of hydrogel-based encapsulation devices that will safeguard their encapsulated L. lactis subsp. lactis or nisin Z cargo while passing through the stomach and small intestine (62). We hypothesize that this encapsulation will enhance nisin Z’s effectiveness at reducing the colonic inflammatory environment and tumor burden. We will examine nisin Z’s molecular integrity using LC-MS. In addition, we are intrigued by the genotypic differences in microbiota composition and associated metabolites between the WT and Apc^Min/+ animals. While both are on a C57BL/6 background, we can tease apart the relative contributions of Apc mutation and the consequences of a colonic tumor environment using the azoxymethane/dextran sulfate sodium (AOM/DSS) model. This model achieves colonic tumor formation as a result of treatment with the AOM carcinogen in combination with DSS-based intestinal barrier disruption using WT C57BL/6 animals, and we expect it will allow a more direct comparison within genetically identical animals.

While our study was the first to demonstrate an in vivo effect for nisin Z on intestinal tumor formation, it has several limitations. Our in vitro and in vivo results suggest an anti-inflammatory activity via potential interaction with the TLR-NF-κB pathway, though at this point, it remains to be speculated what the molecular target of nisin Z might be in CRC. Teasing apart the mechanisms will be complex, since in addition to a potential direct anti-inflammatory effect of nisin Z, its antibiotic activity could lead to indirect anti-inflammatory effects through modulation of the gut microbiota and their associated metabolites. Future studies will focus on the identification of the cellular target, as well as testing which microbial metabolites contribute to the inhibition of colonic tumor formation.

Conclusion

We showed that nisin Z can have an anti-proliferative effect in an Apc^Min/+ intestinal tumor mouse model via the downregulation of cyclin D expression and a reduction in NF-κB. We speculate that the localized effect in the small intestine could be due to degradation further along the GI tract as well as differences in physiological parameters earlier in the GI tract. We did not uncover a unifying mechanism that can explain our observations but identified direct anti-inflammatory, as well as indirect effects on the microbiota and their metabolites. We hypothesize that these individual factors, in combination with the continuous administration of nisin Z, contribute to a net reduction in the inflammatory environment, which in turn leads to the reduced small intestinal tumor burden.

MATERIALS AND METHODS

Cell culture

THP-1 Blue NF-κB Cells (InvivoGen) were cultured in customized medium and sub-cultured following the manufacturer’s protocol. The custom medium consists of Roswell Park Memorial Institute Medium (RPMI) 1640, 20 mM HEPES solution (Gibco), 1% GlutaMAX (Gibco), 1% penicillin/streptomycin, and 100 µg/mL Normocin (InvivoGen) and supplemented with 10% fetal bovine serum (FBS). The selection medium additionally contains 10 µg/mL blasticidin (InvivoGen). THP-1 cells were sub-cultured in selection medium every 4 days and harvested by centrifugation at 400 × g for 4 min. Supernatant was removed, and cells were resuspended in fresh selection medium. Cell culture was typically maintained between 7 × 10⁵ and 1 × 10⁶ cells/ml. CRC (Caco-2) cells (ATCC, USA) were cultured according to ATCC’s recommendations. Cells were sub-cultured weekly in complete medium, consisting of Eagle’s Minimum Essential Medium + 20% FBS and fed one to two times per week. Sub-culturing and assays were performed at 80% confluency (1 × 10⁵ cells/cm²).

Nisin Z solution preparation

“Ultrapure Nisin Z” (Handary S.A., Belgium) solutions were prepared fresh on the day of the experiments by vortexing for 30 seconds and sonicating for 15 min to ensure homogeneity. Next, solutions were passed through a 0.22 µm filter (MilliporeSigma, USA). For Caco-2 experiments, nisin Z was dissolved in complete media, and for THP1-Blue experiments in custom media without blasticidin, and for animal studies, nisin Z was dissolved in drinking water.

Caco-2 proliferation assay

Caco-2 proliferation over time was measured using an Incucyte Live-Cell Analysis System (Sartorius, USA). Caco-2 cells were seeded (4 × 10⁴ cells/well) and incubated overnight at 37°C + 5% CO₂, then treated with 30, 150, or 300 µM of nisin Z, and compared to a vehicle only control. Confluency (%) was measured at 37°C + 5% CO₂ every 4 h for 72 h, using standard cell-by-cell analysis, where the software’s algorithm identifies unlabeled, discrete cells in phase images.

NF-κB reporter assay

THP1-Blue cells (InvivoGen, USA) were seeded (1 × 10⁴ cells/well) into a 96-well plate and incubated overnight at 37°C + 5% CO₂. Cells were pretreated with customized RPMI vehicle, or various concentrations of nisin Z (1.5, 15, and 150 µM) in RPMI for 4 h. After incubation, 50 ng/mL LPS (Invitrogen, USA) was added, and cells were further incubated overnight. Alkaline phosphatase produced by the cells was quantified using a p-Nitrophenyl Phosphate substrate kit (Thermo Fisher, USA), according to the user guidelines. Absorbance was measured after 30 min at 405 nm in a BioTek Synergy HT Microplate Reader (Agilent, USA).

Cell culture live/dead assay

Caco-2 or THP-1 Blue cells were seeded at 1 × 10⁶ cells/well in a 6-well plate. Caco-2 cells were treated with 300 µM nisin Z for 24 or 72 h. Next, Caco-2 cells were dissociated from the wells using TrypLE Express (Gibco, USA). THP-1 Blue cells were treated with 1.5, 15, or 150 µM nisin Z. Cell staining with LIVE/DEAD Fixable Blue Dead Cell Stain Kit for UV excitation (Invitrogen, USA) was performed according to the user guide. Finally, the fixed cell suspension was analyzed by flow cytometry using a FACSymphony A5 SE (BD Biosciences, USA), and the results were analyzed using FlowJo_v10.10.0 (BD Biosciences, USA).

Animal studies

C57BL/6J WT and C57BL/6J Apc^Min/+ breeders were purchased from The Jackson Laboratory, USA, and housed in the Biological Resources Unit of the Cleveland Clinic Lerner Research Institute. At 8 weeks, C57BL/6 WT females and C57BL/6 Apc^Min/+ males were cohoused for breeding. The genotype of the pups was determined using allele-specific PCR. For experimental groups, three to four mice were cohoused per cage, in at least two cages per experimental group. Experimental female WT and Apc^Min/+ mice were placed on a high-fat diet with 60 kcal% lard-derived fat (Diet# D12492, Research Diets, Inc.) at 5 weeks old. Animals were randomly assigned to groups provided with nisin Z (300 µM) or regular drinking water. Animal body weight and food intake were monitored weekly and after 12 weeks on treatment, animals were sacrificed using a ketamine (120 mg/kg)-xylazine (16 mg/kg) cocktail injected intraperitoneally. Peripheral plasma was collected and prepared for LC-MS/MS analysis to detect gut microbiome-related metabolites. The GI tract of the Apc^Min/+ mice was dissected, and the small intestine was sectioned into three equally sized regions: upper, middle, and lower small intestine. The small intestinal segments and colon were longitudinally sliced and spread onto a wax surface using our 3D-printed Mouse Intestinal Slicing Tool (29). Intestinal adenomas were enumerated by two separate individuals, and their areas were measured using ImageJ (NIH, USA). Sections of the small intestine were used for Western blot analysis, small intestine and colon sections were used for gene expression analysis, and cecal contents were extracted for 16S rRNA sequencing of microbiome composition.

Small intestine Western blot analysis

Sections of the small intestine’s mid-segment (15 mg) were extracted using lysis buffer containing Pierce RIPA buffer (Thermo Fisher, USA) and 1% Halt Protease Inhibitor Cocktail (Thermo Fisher, USA). The samples were then homogenized at 30 Hz for 10 min and centrifuged at 14,000 × g for 10 min (4°C). The extracted protein was quantified using the Pierce Bradford Protein Assay Kit (Thermo Fisher, USA) and diluted in Laemmli buffer with 2-mercaptoethanol. Proteins were separated using a 12% Mini-PROTEAN TGX Gel (Bio-Rad, USA) in Tris/Gly/SDS buffer at 100V for 90 min, and sizes were identified using Prestained Protein Marker (PL00001, Proteintech, USA), transferred onto a 0.2 µm polyvinylidene difluoride membrane (MilliporeSigma, USA), using wet electroblotting at 25V, overnight (4°C). The membranes were blocked using 5% bovine serum albumin (BSA) (wt/vol) in Tris-buffered saline with Tween 20. Membranes were then incubated overnight at 4°C with primary NF-κB p65 (D14E12 XP, Cell Signaling Technology, USA), followed by secondary anti-rabbit IgG, HRP-linked antibody for 1 h (Cell Signaling Technology, USA) at 1:1,000 dilution in 5% BSA. The housekeeping β-actin (Proteintech, USA) was diluted at 1:10,000 in 5% BSA and incubated overnight (4°C), followed by HRP Goat anti-mouse IgG (Poly4053, BioLegend, USA) at 3:10,000 in 5% BSA for 1 h. Images were captured using an Amersham Imager 600 (GE, USA) and quantified by densitometry using ImageJ (NIH, USA).

GI tissue gene expression analysis

Small intestine and colon tissue sections (50–100 mg) were lysed for 5 min at 30 Hz using Tissuelyser II (Qiagen, Germany). RNA was isolated using TRIzol (Invitrogen, USA) and treated with the DNase Removal Kit (Invitrogen, USA) to remove DNA contaminants. The concentration of RNA was determined by nanodrop and normalized to 750 ng/µL. cDNA was synthesized using the qScript Supermix (QuantaBio, USA) according to the manufacturer’s instructions and diluted for use with the SYBR Green Master Mix (Applied Biosystems, USA) and quantified by RT-PCR (StepOnePlus RT-PCR system, Applied Biosystems, USA). mRNA expression levels were calculated based on the ΔΔ-CT method, and genes of interest were normalized to the housekeeping gene β-actin. Primers utilized: cyclin D1, 5′-AGACCTTTGTGGCCCTCTG-3′ – F and 5′-GGCAGTCCGGGTCACACT-3′ – R. β-actin, 5′-TTCCTTCTTGGGTATGGAATCCT-3′ – F and 5′-TTTACGGATGTCAACGTCACAC-3′.

16S rRNA amplicon sequencing

DNA from cecal contents was extracted using DNeasy PowerSoil Pro Kit (Qiagen, Germany). Purified DNA isolate is sequenced utilizing MiSeq Reagent Micro Kit v2 (Illumina, USA). 16S rRNA gene amplicon sequencing and bioinformatics analysis were performed using methods explained earlier (63–65). Briefly, raw 16S rRNA amplicon sequence and metadata were demultiplexed using split_libraries_fastq.py script implemented in QIIME2 (66). Demultiplexed fastq file was split into sample-specific fastq files using split_sequence_file_on_sample_ids.py script from QIIME2. Individual fastq files without nonbiological nucleotides were processed using Divisive Amplicon Denoising Algorithm pipeline (v1.16) (67). Taxonomic status was assigned to each ASV using the SILVA database (2.8.0, available at DOI 10.5281/zenodo.14169025). The output of the dada2 pipeline (feature table of amplicon sequence variants [an ASV table]) was processed for alpha and beta diversity analysis using phyloseq (3.21) (68) and microbiomeSeq (69) packages in R. We analyzed variance (ANOVA) among sample categories while measuring the α-diversity measures using plot_anova_diversity function in microbiomeSeq package. Permutational multivariate analysis of variance (PERMANOVA) with 999 permutations was performed on all coordinates obtained during CCA with the ordination function of the vegan package (2.3–5). Pairwise correlation was performed between the microbiome (genera) and metabolomics (metabolites) data using the microbiomeSeq package.

Statistical analysis

Differential abundance analysis was performed using the random-forest algorithm, implemented in the DAtest package (2.8.0) (70). Briefly, differentially abundant methods were compared with false discovery rate (FDR), area under the (receiver operator) curve, empirical power (power), and false positive rate. Based on the DAtest’s benchmarking, we selected metagenomeSeq and ANOVA as the methods of choice to perform differential abundance analysis. We assessed the statistical significance (P  <  0.05) throughout, and whenever necessary, we adjusted P-values for multiple comparisons according to the Benjamini-Hochberg method to control FDR (71). Linear regression (parametric test) and Wilcoxon (nonparametric) test were performed on genera and ASV abundances against metadata variables using their base functions in R (version 4.1.2; R Core Team, 2021).

Gut microbiome-related metabolite LC-MS/MS analysis

Peripheral plasma samples were prepared for analysis by adding an internal standard mix in a 4:1 vol/vol ratio, as detailed in a previously published method (72). Samples were vortexed and centrifuged at 20,000 × g for 10 min (at 4°C), and the supernatant was transferred to glass sample vials. LC-MS/MS analysis was performed on a LCMS-8050 Triple Quad (Shimadzu Scientific Instruments, USA) using an XSelect C18 column (2.5 µM, 2.1 mm × 150 mm) (Waters, Ireland). Column temperature was kept at 40°C. Solvents A and B were 0.1% acetic acid in water and 0.1% acetic acid in acetonitrile, respectively. Samples were run according to the following gradient: 0.0–2.0 (0% B), 2.0–5.0 (40% B), 5.0–7.5 (85% B), 7.5–8.0 (100% B), 8.0–10.0 (100% B), and 10.0–13.0 min (0% B) and with a flow rate of 0.4 mL/min. Data were analyzed using Lab Solution software (Shimadzu Scientific Instruments, USA).

DATA AVAILABILITY

All data are presented in the paper or available on request. Raw sequence files from the 16S rRNA gene sequencing were deposited in Zenodo’s Sequence Read Archive (DOI: 10.6084/m9.figshare.28990487.v1).

ETHICS APPROVAL

All animal studies and procedures were previously approved by the Institutional Animal Care and Use Committee at the Lerner Research Institute of the Cleveland Clinic.