Sensitivity to the two peptide bacteriocin plantaricin EF is dependent on CorC, a membrane‐bound, magnesium/cobalt efflux protein

Abstract Lactic acid bacteria produce a variety of antimicrobial peptides known as bacteriocins. Most bacteriocins are understood to kill sensitive bacteria through receptor‐mediated disruptions. Here, we report on the identification of the Lactobacillus plantarum plantaricin EF (PlnEF) receptor. Spontaneous PlnEF‐resistant mutants of the PlnEF‐indicator strain L. plantarum NCIMB 700965 (LP965) were isolated and confirmed to maintain cellular ATP levels in the presence of PlnEF. Genome comparisons resulted in the identification of a single mutated gene annotated as the membrane‐bound, magnesium/cobalt efflux protein CorC. All isolates contained a valine (V) at position 334 instead of a glycine (G) in a cysteine‐β‐synthase domain at the C‐terminal region of CorC. In silico template‐based modeling of this domain indicated that the mutation resides in a loop between two β‐strands. The relationship between PlnEF, CorC, and metal homeostasis was supported by the finding that PlnEF‐resistance was lost when PlnEF was applied together with high concentrations of Mg2+, Co2+, Zn2+, or Cu2+. Lastly, PlnEF sensitivity was increased upon heterologous expression of LP965 corC but not the G334V CorC mutant in the PlnEF‐resistant strain Lactobacillus casei BL23. These results show that PlnEF kills sensitive bacteria by targeting CorC.


| INTRODUC TI ON
Lactic acid bacteria (LAB) produce a diverse array of bacteriocins.
Bacteriocins are ribosomally synthesized peptides with bactericidal activity and are frequently most active against species that are highly related to the producer strains (Chikindas, Weeks, Drider, Chistyakov, & Dicks, 2018). Bacteriocins, and LAB bacteriocins in particular, have received considerable interest for their potential use in food preservation and pathogen inhibition (Cotter, Hill, & Ross, 2005). The most well-known among these bacteriocins is nisin, a class I bacteriocin produced by Lactococcus lactis which has been approved as a food additive in both Europe and the US since the 1980s (Gharsallaoui, Oulahal, Joly, & Degraeve, 2016). More recently, numerous LAB bacteriocin biosynthetic genes were found in human gastrointestinal tract and vaginal microbiomes, extending the potential functions of bacteriocins beyond foods to human health (Walsh et al., 2015;Zheng, Gänzle, Lin, Ruan, & Sun, 2015). However, the specific host cell targets for many of these peptides have yet to be identified, thereby limiting the expansion of bacteriocin use and accurate predictions of antimicrobial activity.
Lactobacillus plantarum is a well-characterized LAB species and most strains produce several bacteriocins called plantaricins (Nissen-Meyer, Oppegård, Rogne, Haugen, & Kristiansen, 2010). Plantaricin EF (PlnEF) was one of the first plantaricins to be isolated and characterized for its inhibitory spectrum against LAB (Anderssen, Diep, Nes, Eijsink, & Nissen-Meyer, 1998). It is a member of the IIb class of bacteriocins, or bacteriocins that require two different peptides in equal quantities for full activity (Nissen-Meyer et al., 2010). PlnEF consists of the 33 residue PlnE peptide and the 34 residue PlnF peptide. PlnE has a pair of amphiphilic α-helices at both the N-and C-terminal ends (Fimland, Rogne, Fimland, Nissen-Meyer, & Kristiansen, 2008). PlnF has a single, central α-helix that is polar at the N-terminal end and amphiphilic at the C-terminus (Fimland et al., 2008). PlnE contains two GXXXG motifs characteristic of class IIb bacteriocins, while PlnF contains a GXXXG-like motif (SXXXG). Nuclear magnetic resonance analysis of PlnEF suggests these regions provide an interaction point between the peptides (Fimland et al., 2008).
Assessment of the molecular orientation that these peptides adopt in artificial, cellular membrane-mimicking micelles revealed the most likely conformation is an anti-parallel coupling with the N-terminus of PlnE pointed into the cellular membrane and the N-terminus of PlnF closer to the extracellular environment (Kyriakou, Ekblad, Kristiansen, & Kaznessis, 2016).
In this study, we employed a forward genetics approach using genome comparisons to identify the PlnEF receptor in the sensitive strain L. plantarum NCIMB 700965 (LP965). This strain has been used as an indicator for plantaricin biosynthesis (Kjos, Snipen, Salehian, Nes, & Diep, 2010;Moll et al., 1999) and lacks a functional PlnEFimmunity protein Plantaricin I. Isolation and characterization of PlnEF-resistant mutants of LP965 led to the identification of CorC, a putative magnesium/cobalt efflux protein, as a target of PlnEF.

| Selection of PlnEF-resistant LP965 mutants
Plantaricin E (PlnE) and plantaricin F (PlnF) were synthesized and combined in equal molar ratios (PlnEF) prior to measuring for their inhibitory activity against LP965. The MIC 50 of PlnEF against LP965 was 12.5 nM, and LP965 did not grow in the presence of PlnEF at concentrations above 50 nM when measured over a 24 hr period (data not shown).
A total of five putative PlnEF-resistant isolates were randomly selected and confirmed for PlnEF resistance using soft agar inhibition (data not shown) and growth assays ( Figure 1). All five isolates (designated EF.A to EF.E) exhibited a fourfold increase in PlnEF MIC 50 values (MIC 50 >50 nM) compared to wild-type LP965 (WT). Growth rates and maximum optical density (as measured by OD 600 ) of all isolates were increased relative to the WT strain in the presence of 25 nM PlnEF ( Figure 1). Conversely, the growth rates of the PlnEFresistant mutants did not significantly deviate from WT LP965 when incubated in MRS lacking PlnEF (ANCOVA of growth rates

| PlnEF-resistant isolates contain a mutation in CorC, a membrane-bound, magnesium/cobalt efflux protein
A high-confidence reference genome for LP965 was constructed using long (PacBio) and short (Illumina) read sequence data. The DNA sequence of the PlnEF-resistant strain EF.A was also obtained (PacBio). Alignments between the LP965 and EF.A genomes revealed three chromosomal point mutations and five gaps (a result of putative indels). To rule out sequencing errors, the eight genomic regions with sequence variations were amplified by PCR and subjected to DNA sequencing (Table A1)

| EF.A CorC G334V mutation resides in a cysteine-β-synthase domain
The CorC protein is predicted to be a membrane-bound protein containing four transmembrane domains ( Figure A3). Protein annotation predicts two cysteine-β-synthase domains (CBS, also known as Bateman domains (Baykov, Tuominen, & Lahti, 2011)) and two transporter-associated domains (PFAM03471, Figure A3). The CorC G334V mutation in strains EF.A-EF.E is localized in the second of the two CBS domains.
To elucidate a potential molecular basis for CorC interactions with PlnEF, structural modeling was performed for the WT LP965 CorC protein sequence region from Y216 to G351, encompassing the C-terminal CBS. This was accomplished using the resolved crystal structure of the CorC protein produced by Oenococcus oeni (PDB ID:3OCO), which shares 41% amino acid identity to CorC in LP965 ( Figure A4). Our Y216 to G351 model of LP965 CorC indicated that this region might be in- F I G U R E 2 Effects of plantaricin EF (PlnEF) on cellular concentrations of ATP. Cells of LP965 and the PlnEF-resistant isolate EF.A were initially energized with 10 mM glucose (arrow at 1 min). At 10 min (second arrow), 25 nM of PlnEF, 25 μM nisin, or water (NT) was added to each culture. The avg ± SD of n = 3 replicates is shown

| PlnEF increases the effects of metal stress
Because CorC is annotated as a putative magnesium/cobalt efflux protein, we sought to characterize whether the CorC G334V mutation altered LP965 sensitivity to different divalent metal cations.
However, there were no significant differences in the growth rates or final optical densities between LP965 and the CorC mutant when incubated in MRS in the presence of growth-inhibiting concentrations of MgSO 4 , CoSO 4 , C 6 H 8 FeNO 7 , CuSO 4 , ZnSO 4 , or MnSO 4 ( Figure A5).
Next, we investigated whether the CorC mutation altered L. plantarum growth in the presence of both PlnEF and high con-

| Heterologous expression of wild-type LP965 corC increases L. casei BL23 sensitivity to PlnEF
Numerous attempts to truncate or delete LP965 corC were unsuccessful using methods commonly applied for genetic modification of L. plantarum (Aukrust, Brurberg, & Nes, 1995). Efforts to delete Top-down and (c) side-views of (monomer) LP965 CorC protein region from Y216 to G351 shown in ribbon representation and colored by a rainbow scheme from N-terminus region (blue) to C-terminus region (red). All sidechains are shown in stick representation, except for G334, which is shown in space-filling representation. Template-based modeling was accomplished using PDB: 3OCO or truncate corC in L. plantarum NCIMB 8,826 (LP8826) were also unsuccessful. Because LP8826 is amenable to genetic manipulation (Yin et al., 2017), we concluded that CorC is likely an essential protein for L. plantarum.
Therefore, to confirm that the G334V CorC mutation is required for PlnEF resistance, we introduced corC from WT LP965 and the EF.A mutant into L. casei strain BL23. Strain BL23 contains a homolog to CorC (59% amino acid identity), but is at least twice as resistant to PlnEF as LP965 (MIC 50 = 25 nM). Expression of wild-type LP965 corC in L. casei BL23 increased the sensitivity of that organism to PlnEF from 25 nM to 12.5 nM ( Figure 5). Conversely, no increase in PlnEF sensitivity was found when the EF.A CorC G334V mutant was ex- These results confirm that wild-type LP965 corC increases sensitivity to PlnEF and that the single G334V amino acid substitution is sufficient to confer resistance.

| D ISCUSS I ON
We identified CorC, a putative magnesium/cobalt exporter, as the receptor for the L. plantarum bacteriocin PlnEF. These findings are in agreement with previous studies on PlnEF showing that this bacteriocin causes cation efflux (Moll et al., 1999). We hypothesize that PlnEF anchors to CorC and either inserts into the lipid bilayer directly or, alternatively, PlnEF inserts through CorC to cause disrupted metal homeostasis. In either case, the activity of PlnEF is distinct from the L. plantarum bacteriocin Plantaricin JK which is known to induce anion efflux and for which the cell surface receptor was identified as a protein in the APC transporter family (Oppegård et al., 2016). The results are similarly consistent with prior work showing the bacterial receptor for PlnEF is different from lactococcin A and other class II pediocin-like bacteriocins known to bind to proteins in the mannose phosphotransferase system (Diep, Skaugen, Salehian, Holo, & Nes, 2007). Our findings also conform with the expectation that there is a specific protein receptor for PlnEF which causes targeted damage to the cell, as opposed to receptors for broad-spectrum bacteriocins like nisin which cause more pervasive effects on the cell membrane and the release of intracellular ATP (Breukink et al., 1999).
Because metal concentrations vary significantly throughout microenvironments, bacteria contain dedicated proteins for the import of essential metals and export of excess/toxic metals (Barwinska-Sendra & Waldron, 2017). The L. plantarum reference strain WCFS1 contains 42 annotated metal cation transporters (Kleerebezem et al., 2003). L. plantarum strains have long been known to have a relatively high requirement for manganese (MacLeod & Snell, 1947), and the best understood metal transport systems in this species are those with affinity for Mn 2+ (Groot et al., 2005). Presently, transport systems for magnesium and the trace metals Zn 2+ , Co 2+ , Cu 2+ and Fe 2+ have not been well characterized in L. plantarum. In other bacteria, magnesium import, in particular, has been extensively studied and linked to four classes of transporters (Groisman et al., 2013;Moomaw & Maguire, 2008;Shin et al., 2014). CorA, the most widely distributed of these transporters, has been shown to maintain both Mg 2+ and Co 2+ homeostasis in L. lactis (Mills et al., 2005). Some bacteria also carry systems for Mg 2+ export, although this function is not as well understood. Putative Mg 2+ efflux proteins have been identified as CorB, CorC, and CorD in Salmonella (Gibson, Bagga, Miller, & Maguire, 1991) and Shigella (Zhang, Ren, Zhu, Li, & Wang, 2010) and YhdP in Bacillus (Akanuma et al., 2014). These proteins might also export other metals such as Zn 2+ , as indicated for CorB and CorC in Pseudomonas stutzeri (Vaccaro et al., 2016).
CBS domains are frequently found in two to four tandem copies in both cytosolic-and membrane-associated proteins and are present in proteins from all domains of life (Baykov et al., 2011). In eukaryotes, the cyclin M (CNNM) family of proteins mediate Mg 2+ transport and share some structural similarities to bacterial CorC and CorB including the C-terminal CBS pairs (Hirata, Funato, Takano, & Miki, 2014). The CBS domains of magnesium transporters are reported to be important for magnesium and ATP binding (Armitano, Redder, Guimarães, & Linder, 2016;Baykov et al., 2011). Recently, a Staphylococcus aureus protein with a domain structure similar to Salmonella CorB (magnesium protection factor A (MpfA); SA00657) was found to be essential for S. aureus growth in magnesium concentrations as low as 10 mM (Armitano et al., 2016).  (Zhang et al., 2010). The magnesium import protein CorA has also been shown to change conformational states in response to magnesium binding (Matthies et al., 2016), thereby indicating a dynamic process.
PlnEF-resistant mutants were only incrementally (4X) more resistant to the bacteriocin than WT LP965. By comparison, lactococcin G-resistant mutants of L. lactis had MIC 50 values at least 1,000 to 10,000 times greater than the wild-type strain (Kjos et al., 2014). This difference could be due to the fact that those mutations were predicted to encode truncated forms of the Upp receptor protein (Kjos et al., 2014); whereas only a single amino acid substitution in CorC was found among the five LP965 PlnEF-resistant mutants examined here. Heterologous expression of the LP965 and EF.A CorC proteins in L. casei also confirmed the importance of the single G334V point mutation in conferring PlnEF resistance. Because we did not find other mutants and because we were unable to construct a CorC deletion mutant in either LP965 or LP8826, our results indicate that similar to S. aureus MpfA (Armitano et al., 2016), CorC is essential for maintaining magnesium metal homeostasis in L. plantarum.
Determining how bacteriocins exert antimicrobial activity is important for elucidating microbe-microbe interactions and application potential in food and intestinal ecosystems. Identification of bacteriocin cellular receptors will advance our understanding of the inhibitory spectra, functional significance, and resistance mechanisms associated with the myriad of bacteriocins currently known.
We recently demonstrated that the PlnEF system is important for L. plantarum mediated protection against diet induced obesity in mice (Heeney et al., 2018). The gut microbiota and colonic metabolomes were not altered with L. plantarum consumption. Instead, the L. plantarum plantaricin system was correlated with increased production of the tight junction protein ZO-1 in the intestinal epithelium. Application of the purified peptides on differentiated Caco-2

| Bacterial strains and growth conditions
Strains and plasmids used in this study are listed in Table 1

| Bacteriocin peptide synthesis
The full-length peptide sequences of plantaricins taken from the published genome of L. plantarum WCFS1 (Refseq: NC_004567.2) were downloaded from NCBI and trimmed to delete export-signal peptide sequences. Leaderless forms of plantaricin E (FNRGGYNFGKSVRHVVDAIGSVAGIRGILKSIR), and plantaricin A (PlnA) (KSSAYSLQMGATAIKQVKKLFKKWGW) were chemically synthesized by Genscript. Plantaricin F (VFHAYSARGVRNNYKSAVGPADWVISAVRGFIHG) was synthesized by Thermo-Fisher. Peptides were 98%-99% pure and diluted in ultra-pure, molecular grade water (Ambion) prior to being stored at −20°C until use.

| Bacteriocin activity assays
The antimicrobial activity of PlnEF against target cells was tested as previously described (Kjos et al., 2014) using a spectrophotometer to monitor optical density (Synergy 2, Biotek instruments). The minimum inhibitory concentration (MIC 50 ) was defined as the peptide concentration (the sum of both peptides [in a 1:1 ratio]) that inhibited growth by 50% after 6 hr incubation.

| Quantification of corC expression levels
RNA was extracted from three separate exponential-phase cultures of WT LP965, EF.A, or L. casei BL23 strains grown in MRS as previously described (Tachon, Lee, & Marco, 2014).

| Soft-agar growth inhibition assay
Single colonies of naïve and PlnEF-resistant LP965 strains were used as indicator strains in soft agar assays as previously described (Kjos et al., 2014) with strain L. plantarum NCIMB 8826 used as the producer. A resistant LP965 phenotype was detected by a reduced zone of inhibition compared to the naïve strain after overnight incubation at 37°C.

| Isolation of genomic DNA and whole genome sequencing
DNA was isolated from WT and EF.A stationary phase cultures grown in MRS by phenol-chloroform extractions, followed by ethanol precipitation (Sambrook & Russell, 2006

| CorC mutagenesis
LP965 was not amenable to genetic manipulation with techniques described to transform this strain previously (Aukrust et al., 1995) or by adjusting parameters (glycine concentrations, recovery times, voltage) described therein. Repeated attempts to delete the corC open reading frame (ORF) in LP8826 (WCFS1 lp_2671) using methods previously established for clean deletions in this strain (Yin et al., 2017) or variations of those methods (glycine concentrations, recovery times, voltage, incubation temperature) were equally unsuccessful. Primers used for these attempts are listed in Table A3.

| Heterologous expression of L. plantarum corC in L. casei BL23
The corC genes from LP965 and EF.A were amplified with primers A and D using genomic DNA as a template (Table A3). The product was digested with EcoR1-HF and SacI-HF (New England Biolabs) and ligated into the multiple cloning site of pJIM2246 (Renault, Corthier, Goupil, Delorme, & Ehrlich, 1996) resulting in pJIMDH1 (WT LP965 CorC) and pJIMDH2 (EF.A CorC). E. coli DH5α was transformed and selected for as previously described (Yin et al., 2017). Plasmids were isolated from E. coli and amplified by PCR using corC primers (Table   A1) and DNA sequencing for verification. L. casei BL23 electrocompetent cells were prepared and transformed as previously described (Welker, Hughes, Steele, & Broadbent, 2015

| Statistics
Growth assays were conducted with technical replicates in triplicate and data are representative of at least two independent experiments. Data are presented as average values ± SD. Analysis of covariance (ANCOVA) was used to determine whether the growth rates of cultures in logarithmic phase were statistically different from one-another. Student's t test with Welch's correction was used to determine significant differences between final optical densities of cultures and levels of corC expression.

ACK N OWLED G EM ENTS
This manuscript is based upon work supported by the National Science

CO N FLI C T O F I NTE R E S T S
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

AUTH O R S CO NTR I B UTI O N
DDH wrote the manuscript and designed and conducted experiments. VYY produced 3D models and edited. MLM designed experiments and edited the manuscript.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
LP965 WT sequence data was deposited into the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/) with accession numbers CP023490 to CP023495. The EF.A PlnEF-resistant strain was deposited in NCBI with accession numbers CP026505 to