The Bovine Antimicrobial Peptide Lactoferricin Interacts with Polysialic Acid without Loss of Its Antimicrobial Activity against Escherichia coli

Simple Summary Bovine milk contains a high concentration of the protein lactoferrin. It is an important antimicrobial biomolecule, which is also present in other bodily fluids like blood and semen. However, not only the intact protein but also its cleavage products have antimicrobial activity. Perhaps, the best-known cleavage product of lactoferrin is the peptide lactoferricin that has significant antimicrobial capacity against a broad range of pathogens such as enterohemorrhagic Escherichia coli (EHEC). Interestingly, lactoferricin can interact with the sugar polymer polysialic acid, which is also present in milk, blood, and semen. In the present study, we tested if the binding to polysialic acid influences the biological activity of bovine lactoferricin. Remarkably, neither different amounts of polysialic acid nor different chain lengths of this sugar polymer influenced the antimicrobial activity of lactoferricin. The ability of polysialic acid to bind and not inactivate lactoferricin may allow the development of novel endogenous and biodegradable polysialylated surfaces and/or hydrogels, which can be loaded with the antimicrobial peptide lactoferricin for biomedical applications in veterinary and human medicine. Abstract The lactoferrin-derived peptide lactoferricin (LFcin) belongs to the family of antimicrobial peptides, and its bovine form has already been successfully applied to counteract enterohemorrhagic Escherichia coli (EHEC) infection. Recently, it was described that LFcin interacts with the sugar polymer polysialic acid (polySia) and that the binding of lactoferrin to polySia is mediated by LFcin, included in the N-terminal domain of lactoferrin. For this reason, the impact of polySia on the antimicrobial activity of bovine LFcin was investigated. Initially, the interaction of LFcin was characterized in more detail by native agarose gel electrophoresis, demonstrating that a chain length of 10 sialic acid residues was necessary to bind LFcin, whereas approximately twice-as-long chains were needed to detect binding of lactoferrin. Remarkably, the binding of polySia showed, independently of the chain length, no impact on the antimicrobial effects of LFcin. Thus, LFcin binds polySia without loss of its protective activity as an antimicrobial peptide.

To control the separated chain lengths, 100 ng of each fraction was mildly labeled with DMB in 80 µL DMB reaction buffer (overnight at 11 • C), and 20 ng were analyzed with a DNAPac™ PA100 column (4 mm × 250 mm, 13 µm; ThermoFisher Scientific) [52]. The used gradient was the following: The flow rate was 1 mL/min. PolySia digested with active endoneuraminidase (EndoN, 0.1 µg) was analyzed under the same conditions.

Electrophorese on Native Agarose Gel
For the interaction analysis, lactoferrin and LFcin were separated by native gel electrophoresis in the absence and presence of Neu5Ac and different chain length of polySia, as previously described in detail [12,53]. To this end, the samples were loaded on an agarose gel (2%, w/v) (lactoferrin and LFcin, 10 µg/lane; Neu5Ac and sialic acid polymers, 5 µg/lane) and separated for 5 h at 80 V (running buffer: 25 mM Tris/HCl, 19.2 mM glycine, pH 8.5). Afterwards, the proteins were fixed in 45% methanol/7.5% acetic acid (v/v) overnight and colored with Coomassie blue (Roti-Blue, Carl-Roth), and the gel was de-stained with 25% methanol.

ELISA
ELISA plates were coated with 20 µg/mL of lactoferrin (coating buffer, 15 mM Na 2 CO 3 , 35 mM NaHCO 3 , pH 9.6) for 2 h, followed by two washing steps. Afterwards, Neu5Ac and polySia fractions (160 µg/mL in PBS) were added, and incubation was carried out for 2 h. In a further step, murine anti-LFcin antibody (0.5 µg/mL) was applied to bind unoccupied LFcin binding sides. For a detailed description, please see [12].

Bacterial Growth Assay
Bacterial growth was analyzed with the Bacteria Counting Colorimetric Assay Kit (BioVision, Milpitas, CA, USA), which was also used to characterize the impact of histones on E. coli [49]. During all the following steps, E. coli was cultured at 37 • C under shaking. To generate a preculture, LB medium (1% NaCl [w/v], 1% peptone [w/v], 0.5% yeast extract [w/v]) was inoculated with a frozen E. coli stock and incubated overnight. With this preculture, a main culture was inoculated and grown until an OD (600 nm) of 0.29−0.32 was reached. For the bacterial growth experiments, 50 µL of LB medium was added to a 96-well plate. In addition, LB medium containing LFcin (200 µg/mL) and/or polySia (400, 200, or 100 µg/mL), Neu5Ac (400 µg/mL), fractionated polySia (400 µg/mL), or enzymatically cleaved polySia (400 µg/mL) was applied. For the enzymatic digestion of the polymers, polySia (6 mg/mL) was treated with endo N (6.7 µg/mL, 3 h, 37 • C). To the 50 µL of differently modified LB media, 40 µL of bacteria solution (~2.4 × 10 8 bacteria/mL) and 10 µL of WST/ECS solution (reagents of the Bacteria Counting Colorimetric Assay Kit) were added. Thus, the final concentration of LFcin is 100 µg/mL. The bacterial growth was measured for 150 min in 30 min intervals, at a wavelength of 450 nm.

A Lower DP of PolySia Is Sufficient to Mediate the Binding to LFcin in Comparison to Lactoferrin
Antimicrobial peptides act together as a functional complex to attack the bacterial membrane [54]. If a switch of several LFcin molecules from one polySia chain to the bacterial membrane is possible, it is conceivable that such an accumulation of several LFcin molecules on a polySia chain supports the cooperation of the peptides in the formation of damaging complexes. The LFcin peptides would be directly located in a functional neighborhood.
To calculate the loading capacity of a polySia chain, it is important to determine the precise number of linked sialic acid residues which are needed to initiate the interaction. In polySia, the degree of polymerization necessary for the interaction with human lactoferrin or bovine LFcin was previously narrowed down to fractions consisting of polymers with a DP between 15 and 24 sialic acid residues [11,12]. Fractions with shorter chains, consisting of 2 up to 14 linked Neu5Ac residues, showed no reliable interaction with lactoferrin and LFcin in these studies. For the present experiments, groups containing only two main chain lengths (DP 2-3, 4-5, 6-7, 8-9, 10-11, 12-13, 14-15, 16-17, 18-19, and 20-21) were collected. In the collected fractions, the chain lengths were tested by anion-exchange chromatography after fluorescent labeling with DMB ( Figure 1a). It was not necessary to collect monoSia, since Neu5Ac is commercially available.
To analyze the ability of polySia to interact with lactoferrin, native agarose gels were loaded with LFcin and/or sialic acid chains of different lengths. In addition, lactoferrin was applied by using an identical experimental setup. Thereby, an interaction could be visualized by a migration shift of the peptide/protein to the positive pole. This migration shift might be caused by the polyanionic charge of polySia during the attachment to its binding partners. In the case of lactoferrin, a first weak interaction may start with a DP of 18-19 ( Figure 1b). The fraction DP 20-21 significantly influences the migration of lactoferrin. In contrast, already shorter chains influence the migration of LFcin. Here, a detectable interaction started with DP 10-11 ( Figure 1c).
Since lactoferrin is much bulkier than LFcin, it is likely that an interaction with short polySia chains may only slightly influence the migration in the gel. Thus, an interaction with short chains might be overlooked. In order to confirm the obtained results, lactoferrin was coated on ELISA plates. Thereafter, polySia fractions according to their DP were added. Subsequently, an antibody against the LFcin-containing domain of lactoferrin was added. The antibody could easily bind the unoccupied N-terminal domain, whereas polySia inhibited the binding [12]. In line with the results from the native agarose gel, polySia with DP 20-21 and unfractionated polySia were able to inhibit the antibody binding ( Figure 1d). Thus, in comparison with LFcin, twice-as-long polymers were needed to mediate an interaction.
Animals 2020, 10, 1 5 of 11 unoccupied N-terminal domain, whereas polySia inhibited the binding [12]. In line with the results from the native agarose gel, polySia with DP 20-21 and unfractionated polySia were able to inhibit the antibody binding ( Figure 1d). Thus, in comparison with LFcin, twice-as-long polymers were needed to mediate an interaction.  The differences between LFcin and lactoferrin regarding the required chain length for polySia binding might be the results of conformational changes of LFcin after proteolytic release. It is known that bovine LFcin can change its conformation from an α-helical structure to a twisted β-sheet. This transformation might be also the reason for a better binding of LFcin to bacterial membranes and the stronger antibacterial effect of LFcin in comparison to lactoferrin [5,6,55]. In addition, in lactoferrin, large molecular surface areas of LFcin are hidden by other domains of the protein. Furthermore, it needs to be mentioned that molecular dynamic simulation suggested a second binding domain in lactoferrin for the terminal end of a polySia chain [11]. The interaction with two different protein domains is probably needed to stabilize the interaction in the case of lactoferrin.

PolySia Has No Impact on the Antimicrobial Activity of LFcin
To prove the ability of polySia to influence the antibacterial properties of LFcin, both molecules were tested for their ability to inhibit the growth of bacteria. In bacterial growth assays, LFcin and polySia were separately tested, in addition to a combination of LFcin and different concentrations of polySia. As expected, bacterial growth was inhibited by LFcin (Figure 2a,b). In contrast, polySia had no statistically significant impact on bacterial growth (Figure 2b). Surprisingly, when LFcin was applied together with polySia, its antimicrobial activity was not influenced by polySia. The growth curves in the presence or absence of polySia were very similar, and no statistical differences were observed after 150 min.
The differences between LFcin and lactoferrin regarding the required chain length for polySia binding might be the results of conformational changes of LFcin after proteolytic release. It is known that bovine LFcin can change its conformation from an α-helical structure to a twisted β-sheet. This transformation might be also the reason for a better binding of LFcin to bacterial membranes and the stronger antibacterial effect of LFcin in comparison to lactoferrin [5,6,55]. In addition, in lactoferrin, large molecular surface areas of LFcin are hidden by other domains of the protein. Furthermore, it needs to be mentioned that molecular dynamic simulation suggested a second binding domain in lactoferrin for the terminal end of a polySia chain [11]. The interaction with two different protein domains is probably needed to stabilize the interaction in the case of lactoferrin.

PolySia Has no Impact on the Antimicrobial Activity of LFcin
To prove the ability of polySia to influence the antibacterial properties of LFcin, both molecules were tested for their ability to inhibit the growth of bacteria. In bacterial growth assays, LFcin and polySia were separately tested, in addition to a combination of LFcin and different concentrations of polySia. As expected, bacterial growth was inhibited by LFcin (Figure 2a,b). In contrast, polySia had no statistically significant impact on bacterial growth (Figure 2b). Surprisingly, when LFcin was applied together with polySia, its antimicrobial activity was not influenced by polySia. The growth curves in the presence or absence of polySia were very similar, and no statistical differences were observed after 150 min.  In order to test weather short or long chains of polySia may have an impact on LFcin's antimicrobial effect, endoN was used to cut the polySia mixture [56]. EndoN works very well because it degrades polySia to mainly short oligomers with a DP < 8 [57]. The digestion was controlled by anion-exchange chromatography after DMB labeling. The obtained chromatograms demonstrated that the degradation was sufficient (Figure 3a). However, also the resulting fragments of polySia exhibited no ability to influence the effects of LFcin (Figure 3b). As described above, a DP higher than 9 was needed to visualize an impact on the migration of LFcin (Figure 1c). Thus, the obtained degradation products (DP < 8) might be too short. In order to test weather short or long chains of polySia may have an impact on LFcin's antimicrobial effect, endoN was used to cut the polySia mixture [56]. EndoN works very well because it degrades polySia to mainly short oligomers with a DP < 8 [57]. The digestion was controlled by anion-exchange chromatography after DMB labeling. The obtained chromatograms demonstrated that the degradation was sufficient (Figure 3a). However, also the resulting fragments of polySia exhibited no ability to influence the effects of LFcin (Figure 3b). As described above, a DP higher than 9 was needed to visualize an impact on the migration of LFcin (Figure 1c). Thus, the obtained degradation products (DP < 8) might be too short. For this reason, five groups of polySia with chain lengths between 8 and 48 were collected, and the DP was analyzed by HPLC (Figure 4a). On the basis of the results of native gel electrophoresis, the chains have the theoretical capacity to bind one to four LFcin molecules. These fractions were used in combination with LFcin for the bacterial growth assays. The results in Figure 4b clearly depict the capacity of LFcin to inhibit bacterial growth despite the presence of polySia fractions of rising DPs. Thus, polySia can bind LFcin, but it has no impact on its antimicrobial activity. Comparable results were also obtained for histones H1 and H2A, as previously described [49].

Conclusion
Interestingly, it was shown that polySia and lactoferrin can interact in several bodily fluids, that polySia supports lactoferrin to inhibit the release of NET, and that polySia influences the binding of lactoferrin to already exposed NET fibers. Since polySia also interacts with the lactoferrin-derived peptide LFcin, an effect of polySia on the biological function of the antimicrobial peptide was also conceivable. Since the binding of lactoferrin is initiated by its LFcin-containing domain, a modulation of LFcin binding to NET or to polysialylated immune cells, such as dendritic cells, is likely [58].
Recently, a therapeutic effect of bovine LFcin after oral intake was reported for the prevention of enterohemorrhagic E. coli (EHEC) infection in a mouse model [59], demonstrating, like numerous other studies, the potential of this antimicrobial peptide in veterinary and human medicine (reviewed in [34]). However, neither different amounts of polySia nor different polySia DPs can influence this ability of LFcin. Particularly, the ability to bind and not inactivate LFcin might be a big For this reason, five groups of polySia with chain lengths between 8 and 48 were collected, and the DP was analyzed by HPLC (Figure 4a). On the basis of the results of native gel electrophoresis, the chains have the theoretical capacity to bind one to four LFcin molecules. These fractions were used in combination with LFcin for the bacterial growth assays. The results in Figure 4b clearly depict the capacity of LFcin to inhibit bacterial growth despite the presence of polySia fractions of rising DPs. Thus, polySia can bind LFcin, but it has no impact on its antimicrobial activity. Comparable results were also obtained for histones H1 and H2A, as previously described [49].
advantage, if polysialylated surfaces can be loaded with LFcin. One additional point to note is that lactoferrin, LFcin, and polySia are endogenous biomolecules and biodegradable. For this reason, polySia and LFcin might be a powerful combination to develop novel therapeutic strategies, such as polysialylated surfaces and/or hydrogels which can be equipped with the detachable antimicrobial peptide LFcin.

Conclusions
Interestingly, it was shown that polySia and lactoferrin can interact in several bodily fluids, that polySia supports lactoferrin to inhibit the release of NET, and that polySia influences the binding of lactoferrin to already exposed NET fibers. Since polySia also interacts with the lactoferrin-derived peptide LFcin, an effect of polySia on the biological function of the antimicrobial peptide was also conceivable. Since the binding of lactoferrin is initiated by its LFcin-containing domain, a modulation of LFcin binding to NET or to polysialylated immune cells, such as dendritic cells, is likely [58].
Recently, a therapeutic effect of bovine LFcin after oral intake was reported for the prevention of enterohemorrhagic E. coli (EHEC) infection in a mouse model [59], demonstrating, like numerous other studies, the potential of this antimicrobial peptide in veterinary and human medicine (reviewed in [34]). However, neither different amounts of polySia nor different polySia DPs can influence this ability of LFcin. Particularly, the ability to bind and not inactivate LFcin might be a big advantage, if polysialylated surfaces can be loaded with LFcin. One additional point to note is that lactoferrin, LFcin, and polySia are endogenous biomolecules and biodegradable. For this reason, polySia and LFcin might be a powerful combination to develop novel therapeutic strategies, such as polysialylated surfaces and/or hydrogels which can be equipped with the detachable antimicrobial peptide LFcin.
Author Contributions: All authors conceived and designed the experiments. A.K., K.Z., and C.E.G., and S.P.G. performed the experiments and/or analysis. A.K. and S.P.G. wrote the paper. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by a grant from the Deutsche Forschungsgemeinschaft (GA 1755/1-2).