The number of bacteria cells living in the average human body is roughly equal to the number of somatic cells and almost each bacterial species can produce at least one kind of bacteriocin [27]. These molecules appear to be critical factors influencing the proper microbiome composition of the skin, mucous membranes, as well as gastrointestinal and urogenital tracts [11]. The current interest in bacteriocins is mainly related to their potential use as antibacterial compounds and antibiotic substitutes. However, the impact of bacteriocins on human or animal physiology, especially on the host's immune system, is poorly documented.
Our previous studies showed that BacSp222 bacteriocin and its succinylated forms exhibit significant pro-inflammatory properties manifesting in NF-ĸB activation in murine immune cell lines and human neutrophils, resulting in increased production of selected cytokines and NO. Therefore, in the present study, we aimed to elucidate the mechanism of bacteriocin recognition by immune system cells. First, we observed that BacSp222 undergoes internalization into P388.D1 cells without damaging membrane integrity. So far, the direct interaction of other bacteriocins with eucaryotic cells, remarkably immune cells, has been poorly documented. For example, Martínez-García and colleagues showed that FITC-labeled bacteriocin AS-48 is endocytosed into Trypanosoma brucei depending on clathrin and temperature. However, the observed endocytosis was closely related to the formation of autophagic vacuoles and, consequently, to the toxic effect of the tested bacteriocin on the protozoan [28]. In contrast, Dreyer et al. showed that nisin, plantaricin 423, and bacST4SA labeled with NHS fluorescein, used in non-toxic concentrations, could bind to the cell membrane or penetrate Caco-2 epithelial cells and human umbilical vascular endothelial cells (HUVEC). However, these studies did not specify the exact cellular location of these bacteriocins [29].
Next, we aimed to identify receptors involved in BacSp222 recognition by host cells and mediating intracellular signals leading to stimulation of inflammatory response. Bacteriocin BacSp222 is a cationic, 50-amino acid peptide secreted into the environment by the opportunistic strain of Staphylococcus pseudintermedius 222 [14]. Therefore, the most likely receptor responsible for recognizing BacSp222 seemed to be a receptor belonging to the pattern recognition receptors (PRRs). PRRs are specialized receptors that are a significant part of the innate immune system. Their primary function is to recognize fragments of conserved molecular structures typical of microorganisms (so-called pathogen-associated molecular patterns, PAMPs) and initiate the immune response [30]. Since our previous observations indicated that bacteriocin BacSp222 with chemically cleaved formylated methionine exerted lower pro-inflammatory activity than its natural forms, BacSp222, or suc-K20-BacSp222 [15], we first focused on receptors belonging to FPRs (specifically FPR1 or FPR2) and known to mediate intracellular signalling after binding peptides containing formylated methionine at the N-terminus [31]. However, as was stated in the Results section, our separate bioluminescent assays on cAMPZen FroZen human recombinant CHO-K1 cells overexpressing FPR2 suggested that BacSp222 was not able to stimulate such receptors [20]. To confirm such results, in the present study, we observed that specific FPR antagonists Boc-2 (antagonist for FPR1, and at concentrations above 10 µM also antagonist for FPR2 [32]) and WRW4 (antagonist for FPR2 [33]) did not affect bacteriocin-induced TNF production by RAW 264.7 and P388.D1 cells. Therefore, based on these observations, we excluded FPRs as receptors involved in BacSp222 bacteriocin recognition. At the same time, we observed that FPR1 and FPR2 agonists such as fMLP and WKYMVM, unlike bacteriocin BacSp222, did not stimulate TNF expression. This observation is in line with the results of Monsel et al. and Eierman et al., who have shown that human neutrophils and monocytes stimulated with fMLP did not produce TNF [34, 35].
Another family of PRR receptors activated during the first stages of microbial infection is the TLR family. These receptors are expressed on all innate immune cells [36]. Their ligands are various molecules that differ in hydrophobicity, size, and structure, including different proteins, lipids, and nucleic acids [37]. TLRs molecules comprise extracellular leucine-rich repeats (LRR), transmembrane domains, and intracellular toll–interleukin 1 receptor (TIR) domains. The extracellular domain is responsible for ligand recognition, while the intracellular domain is involved in signal activation in the cell [38]. TLRs expressed on the cell surface primarily recognize proteins, peptidoglycans, and lipopeptides, while intracellular TLRs recognize predominantly nucleic acids [39]. In our research, we used genetically modified HEK-Blue cells that overexpressed the most crucial human cell surface TLRs: TLR2, TLR4, and TLR5. These cells were not devoid of any endogenous receptor. We observed that TLR2 was the only receptor activated after exposure to all tested forms of BacSp222, including its deformylated form, -fM-BacSp222. Although TLR2 is a membrane receptor, its endocytosis is necessary for the cell's response to LTA [40]. Moreover, activation of the NF-κB-dependent signaling pathway by human THP-1 monocytes or human TLR2-overexpressing HEK-Blue reporter cells stimulated with LTA or Pam3CSK4 requires receptor-dependent endocytosis, and inhibition of this process affects the level of TNF released by the cells [41]. We believe that the internalization of BacSp222 by P388.D1 cells, shown in this study, is most likely just a consequence of endocytosis of the TLR2-BacSp222 complex.
The central role of TLR2 is to recognize infection with Gram-positive bacteria. As was previously stated in the Results section, ligands for this receptor are mainly LTA, a major constituent of the Gram-positive bacteria cell wall. TLR2 also recognizes lipopeptides, atypical LPS, as well as selected lipoarabinomannans and zymosans [24]. Such a variety of recognized ligands is related to the fact that TLR2 can also form heterodimers, e.g., with TLR1, TLR6, and, in humans, TLR10 [42]. Murine cells do not express TLR10 because the murine TLR10 gene homolog is a pseudogene [43]. Our observations indicated that BacSp222 is recognized by both human and murine cells. Therefore, in the further stage of our research, we focused exclusively on the interaction between bacteriocin and TLR2/TLR1 as well as TLR2/TLR6 heterodimers. We used HEK-Blue cells with deleted endogenous TLR1 and TLR6, which expressing exogenous human heterodimers TLR2/TLR1 or TLR2/TLR6. Bacteriocin and its modified forms activated only TLR2/TLR6 heterodimer, indicating that TLR6 is indispensable for signal transmission. Therefore, our results clearly demonstrated that the studied bacteriocin is a novel ligand for the TLR2/TLR6 heterodimer. According to the published data, known ligands for this heterodimer are LTA [25, 26], diacylated lipopeptides (e.g. mycoplasmal macrophage activating lipopeptide-2 [44]) and zymosan [45]. At the same time, for the TLR2/TLR1 heterodimer, several protein ligands have been described, such as the B pentamer heat-labile enterotoxin, produced by Escherichia coli [46], or the BspA protein, produced by Tannerella forsythia [47].
Of course, the chemical nature of BacSp222 is fundamentally different from mentioned known typical TLR2/TLR6 ligands - namely LTA. Therefore, it was crucial in our work to prove that such a compound did not contaminate the studied bacteriocin preparations. LTA is a complex polymer formed of repeating units of polyhydroxy alkanes, glycerol, and ribitol, joined via phosphodiester linkages and anchored by a lipid moiety to the membrane of Gram-positive bacteria. LTA molecules have been grouped into different types, assuming the chemical nature of substituents decorating polyglycerol-phosphate subunits, the length of the whole polymer, and the nature of the glycolipid anchor in the cellular membrane [48, 49]. Such a chemical diversity makes the quantitative analyses of LTAs very difficult. Therefore, we decided on indirect determinations of total phosphorus in bacteriocin preparations, taking advantage of the fact that all types of LTA contain organic phosphodiester bonds, absent, in turn, in molecules of the peptides studied. As described in detail in the Materials and methods section, we elaborated a sensitive microplate phosphorus analysis technique, which involves converting the total organic phosphorus into its inorganic form, subsequent reaction with a mixture of molybdate-antimony reagent and then a spectrophotometric measurement. The results of such analyses confirmed that all peptides used in our study lack phosphorus and, thus, also possible LTA contamination. Therefore, observed TLR2/TLR6 activation is a straightforward consequence of bacteriocin recognition and binding. However, further studies are necessary to elucidate what specific part of the bacteriocin molecule is responsible for interacting with the TLR2/TLR6 heterodimer.
In subsequent experiments, we verified whether TLR2 antagonists could inhibit the interaction of bacteriocin with the receptor. We used two different antagonists, sparstolonin B, and TL2-C29. The first molecule is a selective antagonist of TLR2 and TLR4, inhibiting the recruitment of MyD88 to the intracellular TIR subunit of TLR2 or TLR4 [22]. On the other hand, TL2-C29 is a non-competitive antagonist that specifically binds to the TIR domain of TLR2, affecting the interaction between MyD88 and the intracellular domain of TLR2 [23]. We observed that sparstolonin B decreased the level of TNF released by RAW 264.7 cells stimulated with both LTA, BacSp222 as well as suc-K20-BacSp222, and in the case of P388.D1 cells, only LTA-stimulated cells. Furthermore, TL2-C29 affected the TNF level released by RAW 264.7, and P388.D1 cells stimulated with LTA, BacSp222, suc-K20-BacSp222, and, in the case of P388.D1 cells, also -fM-BacSp222. In contrast, in the case of the human TLR2/6 heterodimer, only TL2-C29 significantly reduced the level of released SEAP into the culture medium. The lack of TLR2/TLR6 inhibition by sparstolonin B is most likely related to its low concentration in our experiments (6 µM for RAW 264.7 and P388.D1 cells, 20 µM for HEK-Blue cells). Liang et al. tested on RAW 264.7 cells the inhibitory effect of sparstolonin B in 10 µM or 100 µM concentrations [50], but in our study, these concentrations were toxic to RAW 264.7 cells (data not shown). Mistry et al. showed that TL2-C29 inhibits the signaling of human TLR2/TLR6 and TLR2/TLR1 heterodimers and murine TLR2/TLR1 heterodimer [23]. However, our studies on RAW 264.7 and P388.D1 cells showed that the TL2-C29 inhibited the interaction of bacteriocin with human TLR2/TLR6 and the murine TLR2/TLR6. Our results align with the results shared by the TL2-C29 manufacturer (InvivoGen web page, https://www.invivogen.com/tlr2-inh-c29, accessed on 02.02.2023) and indicate that TL2-C29 is also an antagonist for the murine TLR2/TLR6 and, in higher concentrations, TLR2/TLR1 dimer. Moreover and importantly, inhibition of BacSp222-stimulated TLR2/TLR6 signaling by sparstolonin B and TL2-C29 indicates that the adapter protein MyD88 is involved in the signaling cascade. Summarizing, our studies revealed the TLR2/TLR6/MyD88/NF-κB pathway as a potential mechanism explaining the inflammatory response activated by BacSp222 bacteriocin in human and murine cells.
The primary limitation of the other reports documenting the influence of different bacteriocins on the immune response of the host or inflammatory reactions is the usage of insufficiently pure peptide preparations, especially those not verified for endotoxins content. To date, only five peptide bacteriocins have been reliably verified for their activity toward eukaryotic cells: microcin J25 [51–53], nisin [54, 55], pyocin S5 [56], as well as avicin A and acidocin A [57]. All these reports concerned carefully purified peptides, and the studies were performed in vitro, using different cell lines, and/or in vivo, on mouse, rat, or Galleria mellonella animal models. The researchers verified mainly the influence of studied peptides on inflammatory reactions. And so, three bacteriocins, microcin J25, nisin, and pyocin S5, revealed evident anti-inflammatory properties. On the other hand, two pediocin-like bacteriocins, avicin A and acidocin A, demonstrated different pro-inflammatory activity, manifested by an increase in the production of pro-inflammatory cytokines and chemokines by human primary monocytes [57]. However, the mechanism of such a phenomenon is, to date, puzzling. Kindrachuk et al. studied signalling pathways activated by nisin and showed that this bacteriocin stimulated the phosphorylation of p38 mitogen-activated protein kinases (p38 MAPKs), Akt serine/threonine kinase (Akt), and cAMP-responsive element binding protein (CREB) in human primary monocytes [55]. However, these results could not be repeated with the human monocyte THP-1 cell line. Although such work is intriguing, it contains numerous technical shortcomings, making it difficult to draw clear conclusions concerning the mechanism of nisin activity toward studied cells. Moreover, in our previous study, we used nisin as a control bacteriocin for studying the proinflammatory properties of BacSp222. In all experiments, the concentration of BacSp222 and nisin was identical, but we observed immune cell activation only in response to BacSp222 [15]. We believe the present study is the first to identify a specific eukaryotic cell receptor responsible for recognizing the pro-inflammatory bacteriocin molecule.