All the strains (see below) used in minimal inhibitory concentration (MIC) assays were grown in BHI medium (Oxoid) at 30°C without shaking. LsbB, EntK1, EntEJ97, and BHT-B bacteriocins were synthesized by Pepmic Co., Ltd, China with 98–99% purity. Synthesized peptides were solubilized to the concentrations of 10.0–0.1 mg/ml in 0.1% (vol/vol) trifluoroacetic acid and stored at -20°C until use. Bacteriocin garvicin ML was purified to 95% as described by (Borrero et al., 2011). Bacteriocin activity was determined by microtiter plate assay as previously described (Holo et al., 1991). The MIC was defined as the minimal bacteriocin concentration that inhibited the growth of the indicator strain by at least 50% (50% of the turbidity of the control culture without bacteriocin) in 200 μl culture.

Circular dichroism (CD) spectra were recorded using a Jasco J-810 spectropolarimeter (Jasco International Co.) calibrated with D-camphor-10-sulfonate (Icatayama Chemical). All measurements were done using a quartz cuvette (Starna) with 0.1 cm path length at 25°C. Samples were scanned five times with a scanning rate of 50 nm/min with a bandwidth of 0.5 nm and response time of 1 s over the wavelength range 190–250 nm. Spectra were recorded at different (30 and 50%) trifluoroethanol (TFE) (Aldridge) concentrations at 25°C. The approximate α-helical content of the protein was estimated from its molar ellipticity at 222 nm (Scholtz et al., 1991).

The experiments were run on a sample containing 1.0 mM EntK1, 50% D3-TFE (99.5% D) (Aldrich), Milli-Q water and 0.2 mM of 4,4-dimethyl-4-silapentane-1-sulfonic acid (DSS) (Larodan).

2D NOESY (Jeener et al., 1979), 2D TOCSY (Braunschweiler and Ernst, 1983), 2D DQCOSY, ¹⁵N-HSQC (Davis et al., 1992), and¹³C-HSQC (Hurd, 1991) were recorded. The data was acquired on a 600 MHz Bruker Avance II spectrometer with four channels and a 5 mm TCI cryoprobe (Bruker Biospin). NOESY spectra with mixing times between 120 and 300 ms were obtained for both samples. TOCSY mixing times of 15–80 ms was used. The experiments were run at 298.15 K. Spectra were processed using the Topspin program (Bruker Biospin). DSS was used as a chemical shift standard, and ¹⁵N and ¹³C data were referenced using frequency ratios as described in (Wishart et al., 1995).

For visualization, assignment and integration of the spectra the computer program CARA was used (Keller, 2004). The spectra were assigned using standard methods (Wüthrich, 1986).

Dihedral angle restraints were obtained from the chemical shift values using the program TALOS-N (Shen and Bax, 2013). Nuclear Overhauser effect (NOE) distance restraints were calculated from the peak volumes in the NOESY spectra with NOESY mixing time 200 ms.

All structure calculations were made using the structure calculation program CYANA 2.1 (Guntert et al., 1997; Herrmann et al., 2002). The applied restraints are presented in Table 1. The annealing macro in CYANA calculated 100 structures. The 20 structures with the lowest energy were kept and analyzed further. The root-mean-square deviation (RMSD) was calculated and the structures were visualized using MolMol (Koradi et al., 1996).

EntEJ97 and EntK1-resistant mutants were obtained by spot-on-lawn assay using 20 μl of bacteriocin with concentration 1.0–0.1 mg/ml. After overnight incubation at 30°C, resistant colonies of E. faecalis LMG3358 (resistant to EntEJ97) and E. faecium LMG2787 (resistant to EntEJ97 and EntK1) appeared within the inhibition zones. Resistant isolates were picked and re-streaked on plates for pure cultures before frozen cultures containing 15% glycerol were made and stored at -80°C. The level of resistance against EntEJ97 and EntK1 was determined by a microtitre plate assay (Holo et al., 1991). DNA from the bacteriocin resistant mutants was extracted from overnight cultures with a GenElute^TM Bacterial Genomic DNA Kit (Sigma–Aldrich) and PCR of the rseP gene was performed using primers Ent, EF (F, M, R) (Table 2). For sequencing of E. faecium LMG2787 mutants with transposons inside rseP, additional primers (EF_R2, T1_F, T2_F, and T3_F) were created (Table 2). PCR products were purified with NucleoSpin Extract II (Macherey-Nagel, Düren, Germany) and sent to GATC Biotech, Germany, for sequencing.

To express the E. faecalis rseP gene heterologously in S. pneumoniae, the gene was placed by homologous recombination in the genome of strain SPH131, behind the ComS-inducible P_comX promoter (ComRS system) (Berg et al., 2011). The P_comX-rseP construct was created by overlap extension PCR (Higuchi et al., 1988). First, the E. faecalis LMG3358 rseP was amplified using the primer pair ds171/ds172 with genomic E. faecalis DNA as template. The P_comX promoter and its ∼1000 bp upstream and downstream regions were amplified using the primer pairs khb31/khb36 and khb33/khb34, respectively. Genomic DNA derived from strain SPH131 served as template. The P_comX with its upstream region was fused to the 5′ end of the E. faecalis rseP gene using the primers khb31 and ds172. The P_comX downstream fragment was fused to the 3′ end of the rseP gene using primer pair ds171 and khb34. Finally, these two fragments were fused using primer khb31 and khb34 giving rise to P_comX-rseP. The Janus cassette (Sung et al., 2001) in strain SPH131 was replaced with the P_comX-rseP fragment by natural transformation, giving rise to strain ds218 (Table 3 and Supplementary Figure 1).

Streptococcus pneumoniae has a gene homologous to the lactococcal rseP. To avoid the potential background noise of the S. pneumoniae rseP, this gene was removed from the genome in strain ds218 using the Janus cassette (Sung et al., 2001). An rseP deletion cassette was constructed by amplifying Janus with the primers kan484F and RpsL41.R (Johnsborg et al., 2008) where genomic DNA from strain RH426 (Johnsborg et al., 2008) served as template. The Janus cassette was then fused to the ∼1000 bp upstream [primers th009 and th010 with genomic RH1 (Johnsborg et al., 2008) DNA as template] and downstream region (primers th011 and th012 with genomic RH1 DNA as template) of rseP using primers th009 and th012. The resulting fragment was used to transform strain ds218 resulting in the replacement of S. pneumoniae rseP with Janus generating strain ds219. Janus was then removed from strain ds219 by transforming with a fragment consisting of the rseP flanking regions only. This fragment was constructed by amplifying the rseP ∼1000 bp upstream region using primers th009 and ds87, while the ∼1000 bp downstream region was amplified with the primers ds88 and th012. The upstream and downstream fragments where then fused using the primers pair th009/th012. The resulting fragment was used to replace the Janus in strain ds219 resulting in strain ds220 (Table 3 and Supplementary Figure 1).

To create a strain expressing point mutated rseP genes we replaced the E. faecalis 3358 rseP in strain ds220 with Janus cassette giving rise to strain ds221. This Janus cassette was amplified with the primer pair khb31/khb34 and genomic DNA from strain SPH131 served as template. Selected residues in RseP were substituted with alanines by a PCR approach, using primer pairs listed in Table 2. The resulting DNA pair fragments were subsequently fused using the primers khb31 and khb34. The resulting fragment was used to replace Janus cassette in strain ds221 giving rise to strains ms1-5 (Table 3). Ectopic expression of the rseP gene in strain ds220 and ms1-5 was induced with the synthetic ComS peptide (NH2-LPYFAGCL-COOH) (Genosphere Biotechnologies) as described by Berg et al. (2011).

Circular dichroism spectra of EntK1 showed that it was unstructured in water but became structured in TFE solution (Supplementary Figure 2). Maximum structuring was obtained in 50% TFE (55% α-helical content). This concentration was consequently used in the NMR experiment.

The NMR spectra were assigned using standard methods as described in “Materials and Methods.” Supplementary Figure 3 shows the assigned ¹⁵N HSQC spectrum of EntK1 in 50% TFE. Figure 1 shows connected strips from NOESY, the torsion angle restraints used in structure calculations and the Hα chemical shift indexes (CSIs). CSI (low Hα ppm values) (Matthews et al., 1992) indicates that there is an α-helix in EntK1 from residue 8 to 27 and TALOS-N torsion angle predictions (Shen and Bax, 2013) indicated backbone torsion angles (ϕ- and ψ-angle) consistent with α-helical regions from residue 8 to 25 (Figure 1).

A total of 561 (15.1 per residue) unique NOE connectivities were assigned. The total number of distance restrictions and their classes are shown in Table 1. The mid-range NOE connectivities that are especially important for α-helices, NN(i, i+2), αN(i, i+2), αN(i, i+3), αβ(i, i+3), and αN(i, i+4), are shown in Figure 1. In the presence of TFE, the observed connectivities indicated an α-helical region stretching from residue 8 to 26.

Structures of EntK1 based on the experimentally obtained constrains were calculated using CYANA. A superimposition of the structure ensemble of the 20 lowest energy structures of EntK1 and the cartoon depiction of the lowest energy structure of EntK1 are shown in Figures 2A,B, respectively. The NMR structure of EntK1 contains an α-helix from residue 8 to 24. The structure ensemble has been deposited to the Protein Data Bank with access code: 5L82 and the NMR data have been deposited to the Biomagnetic Resonance Data Bank with the access code 34006.

A sizable collection of Gram-positive bacteria from different species and genera was used as indicators to assess the inhibition spectra of the bacteriocins EntK1, EntEJ97, and LsbB (Table 4). EntQ, which is a member of the LsbB family bacteriocins, was not tested because synthetic EntQ peptides had poor activity, likely due to formation of disulphide bridges between and inside the molecules, since adding reducing agents (e.g., DTT) restored some of the activity of EntQ (data not shown). As seen in Table 4, the antimicrobial spectra of EntK1, EntEJ97, and LsbB differ greatly. LsbB was active only against L. lactis IL1403, one of the four strains of L. lactis tested and it had no activity against other genera and species. EntK1 was active mostly against E. faecium and E. hirae and some lactococcal strains. EntEJ97 showed the broadest activity spectrum, inhibiting E. faecium, E. faecalis, L. lactis, Staphylococcus aureus, Lactococcus garvieae, and Listeria monocytogenes. It is noteworthy that EntK1 appeared to have significantly lower MIC values (i.e., being more potent) than EntEJ97 against E. faecium, but was inactive against E. faecalis (Table 4). Similarly, all three bacteriocins were active against the L. lactis IL1403, but LsbB was about 340 and 70 times more potent than EntK1 and EntEJ97, respectively (Table 4).

The strain E. faecium LMG2787 was used to generate mutants resistant to EntK1 and EntEJ97. Since E. faecalis is resistant to EntK1, E. faecalis LMG3358 was used to generate mutants resistant only to EntEJ97. The two enterococcal strains were exposed to various concentrations of EntEJ97 and EntK1 at 30°C and relatively many resistant colonies appeared within the inhibition zones on agar plates (see below). For E. faecalis, 12 EntEJ97 resistant colonies were randomly selected and examined for resistance to EntEJ97 by microtiter assays. Similarly, six EntEJ97 and six EntK1mutants of E. faecium were selected for further analysis.

Two levels of E. faecalis resistance to EntEJ97 were found: seven colonies were highly resistant (HR) – at least 500 times more compared to the wild type (WT) strain. The second set of five colonies showed a lower resistance (LR) level, being about 16–32 times more resistant than the WT E. faecalis. For E. faecium only HR (at least 500 times against EntK1 and EntEJ97, compared to WT) type mutants were found. When all these mutants were challenged with the non-related bacteriocins BHT-B (Hyink et al., 2005) and garvicin ML (Borrero et al., 2011), they were found as sensitive as the WT strains (data not shown). These results clearly imply the involvement of specific resistance mechanism(s) toward EntEJ97/EntK1 amongst these mutants.

As the lactococcal RseP has previously been found to serve as receptor for LsbB, we suspected that the RseP homologous in Enterococcus might serve the same function for the LsbB sequence related EntEJ97/EntK1 bacteriocins. The DNA regions containing rseP in all the EntEJ97/EntK1-resistant mutants (high and low) were therefore obtained by PCR and sequenced. Different types of mutations were found. For E. faecalis, all seven HR mutants contained either one or three consecutive 8-bp repeats of the sequence CAAAAAAT within the rseP gene, while the WT rseP has two such repeats. On the other hand, all E. faecium HR mutants carried a transposon within rseP (Figure 3). In all cases, these mutations caused frameshift in rseP and premature termination, indicating that a functional ResP is necessary for the sensitivity toward EntEJ97/EntK1. Surprisingly, no mutations were found in rseP from all five EntEJ97 LR mutants of E. faecalis, indicating that additional genes can affect the sensitivity of the bacterium to these bacteriocins. This latter aspect will be further mentioned in the “Discussion” section.

In order to confirm that the expression of rseP is sufficient to confer sensitivity to EntEJ97 and EntK1, the rseP gene from E. faecalis LMG3358 was heterologously expressed in the distantly related and non-sensitive host S. pneumoniae strain SPH131 as described in (Berg et al., 2011). To avoid potential background noise the endogenous rseP of S. pneumoniae SPH131 was removed by homologous recombination, giving rise to the clone S. pneumoniae ds220 with the endogenous rseP deleted but with the enterococcal rseP expressed under the control of the inducible promoter P_comX (Berg et al., 2011). While non-induced cells were not sensitive to EntEJ97 and EntK1, they became sensitive to the bacteriocins upon induction of rseP (Table 5). The results indicate that the enterococcal rseP is indeed directly involved in the sensitivity to EntEJ97 and EntK1. Further, it should be noted that the induced clone (expressing the E. faecalis rseP) was most sensitive to EntEJ97 (MIC of 18 nM), intermediate-sensitive to EntK1 (MIC of 700 nM), and poorly sensitive to LsbB (MIC > 7000 nM). This observation is line with the results presented in Table 4, where the E. faecalis strains showed a corresponding differentiation in sensitivity to the three bacteriocins.

Members of the RseP protein family have a conserved proteolytic motif (HExxH, where x is any amino acid) located within the first transmembrane helix (Koide et al., 2007). To analyze whether this active site is important for bacteriocin sensitivity, we changed the invariant residues in this motif (i.e., H18, E19, and H22), one by one, and three altogether in the enterococcal RseP to alanine and then assessed for bacteriocin sensitivity. In addition, the non-conserved residue Y24 located nearby the active site was replaced with alanine to serve as a control. The result showed that all clones expressing the RseP protein in which conserved residues were changed to alanine became less sensitive to EntEJ97 and EntK1, especially when all three conserved residues were replaced with alanine (30 times less sensitive to EntEJ97 than the WT) (Table 5). In the control clone (Y24 > A) no changes in the bacteriocin sensitivity were observed. This result demonstrates that the active site of RseP is essential for bacteriocin activity.

RseP protein is known to play a key role in bacterial stress response in Escherichia coli, Bacillus subtilis, and other bacteria (Alba et al., 2002; Varahan et al., 2013; Kim, 2015). We therefore examined how elevated growth temperature as a stress factor, influences the development of EntEJ97 and EntK1 resistance in E. faecium and E. faecalis. As expected resistant mutants only appeared at 30°C but not at 45°C (Figure 4). Cell counts of WT cultures grown for 8 h at 30 and 45°C in BHI broth showed that the cell number at 30°C was only about 1.2 times higher than that at 45°C (data not shown), implying that the lack of resistant mutants at the elevated temperature was not due to poor growth of WT cells but rather due to lack of development of resistant mutants. This is further supported by the fact that EntEJ97/EntK1-resistant mutants with a non-functional rseP gene (described above) obtained at 30°C could not grow at 45°C, while WT cells grew well (data not shown).