The membrane topology of immunity proteins for the two‐peptide bacteriocins carnobacteriocin XY, lactococcin G, and lactococcin MN shows structural diversity

Abstract The two‐peptide bacteriocins produced by Gram‐positive bacteria require two different peptides, present in equimolar amounts, to elicit optimal antimicrobial activity. Producer organisms are protected from their bacteriocin by a dedicated immunity protein. The immunity proteins for two‐peptide bacteriocins contain putative transmembrane domains (TMDs) and might therefore be associated with the membrane. The immunity protein CbnZ for the two‐peptide bacteriocin carnobacteriocin XY (CbnXY) was identified by heterologously expressing the cbnZ gene in sensitive host strains. Using protein topology prediction methods and the dual pho‐lac reporter system, we mapped out the membrane topology of CbnZ, along with those of the immunity proteins LagC and LciM for the two‐peptide bacteriocins lactococcin G and lactococcin MN, respectively. Our results reveal wide structural variety between these immunity proteins that can contain as little as one TMD or as many as four TMDs.

discovery of bacteriocins and their structures, it has become apparent that this scheme is too restrictive. For this reason, an updated classification system has recently been proposed, in which class I broadly contains any post-translationally modified bacteriocin, class II still consists of unmodified peptides, and class III includes large, heat labile bacteriocins such as bacteriolysins and tailocins (Acedo, Chiorean, Vederas, & van Belkum, 2018;Alvarez-Sieiro et al., 2016).
The linear two-peptide bacteriocins are class II bacteriocins and require the combined action of two different peptides to elicit full antimicrobial activity. They are expressed as precursor peptides, with an N-terminal leader sequence, and the core peptides contain characteristic GXXXG or GXXXG-like motifs (in which a glycine has been replaced by alanine or serine (Nissen-Meyer, Oppegård, Rogne, Haugen, & Kristiansen, 2010. Genetic evidence for a two-peptide bacteriocin was first reported in 1991 (van Belkum, Hayema, Jeeninga, Kok, & Venema, 1991), and it was later identified as lactococcin MN (LcnMN) (van Belkum, 1994). A year later, lactococcin G (LcnG) was the first two-peptide bacteriocin to be isolated (Nissen-Meyer, Holo, Håvarstein, Sletten, & Nes, 1992). Since then, numerous other two-peptide bacteriocins have been reported. The gene clusters for these bacteriocins consist of at least five genes, which encode for the precursor bacteriocins, a dedicated immunity protein, a transport protein, and an accessory protein that may be involved in bacteriocin secretion. In many cases, additional genes that encode a three-component regulatory system are also located on, or near, the operon with the structural genes (Nissen-Meyer et al., 2011). The gene clusters encoding for LcnG, LcnMN, and the recently discovered two-peptide bacteriocin carnobacteriocin XY (CbnXY) (Acedo et al., 2017;Quadri et al., 1997;Tulini et al., 2014) are illustrated in Figure 1.
Mode-of-action studies have revealed that many of the two-peptide bacteriocins kill target cells by creating pores in the cell membrane, resulting in leakage of ions and small molecules, and loss of the proton motive force . Since most of these bacteriocins display narrow spectra of activity, it is likely that they require a membrane-associated receptor. Recently, the receptors for several different two-peptide bacteriocins have been discovered (Ekblad, Nissen-Meyer, & Kristensen, 2017;Heeney, Yarov-Yarovoy, & Marco, 2019;Kjos et al., 2014;Oppegård, Kjos, Veening, Nissen-Meyer, & Kristensen, 2016). The three-dimensional structures of LcnG , plantaricin EF , plantaricin JK (Rogne, Haugen, Fimland, Nissen-Meyer, & Kristiansen, 2009), CbnXY (Acedo et al., 2017), and plantaricin S (Ekblad & Kristiansen, 2019) have been solved. In all cases, the peptides were unstructured in aqueous environments, but assumed helical conformations when exposed to membrane-mimicking conditions. The immunity proteins for several two-peptide bacteriocins are known, but to date there have been no structural studies exploring these proteins and their mechanism of immunity remains a mystery. Some immunity proteins for this class of bacteriocins show homology to the Abi family of proteins, which are putative membrane-bound metalloproteases characterized by three conserved motifs (EXXXR, FXXXH, and an invariant histidine) (Kjos, Snipen, Salehian, Nes, & Diep, 2010), and it has been suggested that these immunity proteins function by proteolytically degrading their cognate bacteriocins (Kjos et al., 2010;Lages, Mustopa, Sukmarini, & Suharsono, 2015). In other cases, such as with LagC (the immunity protein for LcnG), it is likely that the immunity protein interacts directly both with the bacteriocin and its cellular receptor (Oppegård, Emanuelsen, Thorbek, Fimland, & Nissen-Meyer, 2010). Mutational analysis of the immunity protein for mutacin IV has identified several key residues in the C-terminus that are essential for immunity (Hossain & Biswas, 2012). In all cases, the immunity proteins for this class of bacteriocins are predicted to contain transmembrane domains (TMDs) (Nissen-Meyer et al., 2011), but aside from that, there appears to be very little similarity between these proteins.
Elucidating the structural features of the immunity proteins for the two-peptide bacteriocins is crucial to understanding how these proteins impart protection to producer organisms. Here, we have identified the immunity protein for CbnXY and analyzed its membrane topology. As a comparison, we have also determined the orientation F I G U R E 1 Schematic representation of the gene clusters for lactococcin G (accession number: FJ938036.1) (Nes et al., 1995) and carnobacteriocin XY (accession number: L47121.1) (Quadri et al., 1997), and of partial gene cluster for lactococcin MN (accession number: MN231270) (van Belkum et al., 1991). Open reading frames (ORFs) are indicated by arrows in the proposed direction of transcription. For ORFs with deduced functions, the color of the arrow indicates function, as listed in the legend LcnMN, respectively, using the dual pho-lac reporter system.

| Bacterial strains and culture conditions
The bacterial strains used in this study are listed in Table 1.
Carnobacteria cultures were grown in brain heart infusion (BHI) media

| Construction of plasmids to identify the immunity protein for CbnXY
The plasmids and primers used in this study are listed in Tables 1   and 2

| Bacteriocin activity and immunity assays
Spot-on-lawn testing was performed to assess the sensitivity of the transformants, the producer organism (LV17B), and the sensitive strains (LV13 and A9b-) to purified CbnXY, as previously described (Acedo et al., 2017). Briefly, a 500 μM solution of CbnXY (total bacteriocin concentration, equimolar) was serially diluted twofold and 10 μl of each dilution was spotted onto a BHI agar plate that had been overlaid with soft agar containing the organism to be tested.
Plates were incubated for ~18 hr at room temperature and inspected for zones of inhibition.

| In silico prediction of membrane topology of CbnZ, LagC, and LciM
Membrane topology models of CbnZ, LagC, and LciM were con- prote in/). Using these topology models (Figure 2), proper deletion mutants were designed for each immunity protein to use with the dual pho-lac reporter system.

| CbnZ functions as the immunity protein for CbnXY
In order to identify the immunity protein for CbnXY, we first examined the operon containing the cbnXY structural genes to identify which gene might encode for the immunity protein. This gene cluster (Figure 1) contains several additional genes responsible for the production of the bacteriocin CbnB2, including genes that encode for regulatory and transport proteins (Kleerebezem, Kuipers, de Vos, Stiles, & Quadri, 2001;Quadri et al., 1997). Two additional open reading frames of unknown function, orf-3 and orf-7, also reside in the gene cluster. Since orf-3 was located immediately downstream of cbnXY, it was a likely candidate to encode the immunity protein. However, the gene product is a short TA B L E 2 Oligonucleotides used in this study protein (42 amino acids) with just one putative TMD. On the other hand, orf-7 was located further away from cbnXY and on the opposite strand, but it encoded for a longer protein (109 amino acids) with three putative TMDs. As such, we decided to express both of these genes in two different CbnXY-sensitive strains by cloning orf-3 and orf-7 downstream of the strong constitutive promoter of the expression vector pMG36c, giving plasmids pMG36c-orf3 and pMG36c-orf7, respectively, and use these plasmids to transform strains LV13 and A9b-.
Spot-on-lawn testing revealed that transformants harboring either pMG36c or pMG36c-orf7 showed no increase in resistance to CbnXY when compared to the sensitive strains, indicating that orf-7 does not seem to provide protection against the bacteriocin.
However, transformants with pMG36c-orf3 displayed a 250-fold increase in resistance to CbnXY, thus revealing that orf-3 encodes an immunity protein, which we named CbnZ. Interestingly, these transformants were still susceptible to CbnXY at the highest concentrations tested, whereas the producer organism (LV17B) was completely immune. Similar observations have been reported for the heterologous expression of other immunity proteins, and it has been suggested that this difference in immunity may be due to poor levels of transcription of the immunity protein genes when under control of a different promoter in the heterologous host (Flynn et al., 2002;Quadri et al., 1995).

| Immunity proteins for the two-peptide bacteriocins are structurally diverse
CbnZ is a small hydrophobic protein of just 42 amino acids, and while there are a few examples of short immunity proteins for this class of bacteriocins, such as BrcI (53 amino acids), EnkIaz (52 amino acids), and AbpIM (55 amino acids) (Flynn et al., 2002;Ishibashi et al., 2014;McCormick et al., 1998), most other immunity proteins for this class of bacteriocins are substantially longer and are predicted to have at least four or five TMDs (Nissen-Meyer et al., 2011). For example, the immunity proteins LagC (Nes, Håvarstein, & Holo, 1995) and LciM (van Belkum et al., 1991) consist of 110 and 154 amino acids, respectively, and are predicted to have at least four TMDs (Nissen-Meyer et al., 2011). We were therefore interested in further exploring the structural diversity among these various immunity proteins by investigating whether CbnZ, LagC, and LciM are indeed membrane-associated, and by mapping out of the orientation of the TMDs of these proteins across the membrane.
The dual pho-lac reporter system of the pKTop expression vector is a convenient method to characterize the membrane topology of proteins (Karimova & Ladant, 2017). This system employs two E. coli proteins, alkaline phosphatase (PhoA) and β-galactosidase (LacZ), which are only active in the periplasm and cytoplasm, respectively.
PhoA converts the substrate X-Pho into an insoluble blue compound, whereas LacZ converts Salmon-Gal into a red precipitate. Thus, when these proteins are fused to the C-terminus of a putative TMD, the orientation of the TMD can easily be determined. To use this system, we began by identifying the putative TMDs and loop regions for CbnZ, LagC, and LciM. Based on the results of membrane protein topology prediction methods (Figure 2), one TMD was predicted for CbnZ, and four TMDs were predicted for both LagC and LciM.
These models also predicted that the N-and C-termini for LagC were located on the outside of the cytoplasmic membrane, whereas the N-and C-termini for LciM were predicted to be located inside the cytoplasm. From these models, we identified key residues to fuse to the pho-lac system of the pKTop vector. In total, we designed and constructed ten different fusion proteins: four for LagC, one for CbnZ, and five for LciM. We used the HindIII and BamHI restriction sites of the pKTop vector, which necessarily introduced an additional eight non-native amino acids to the N-termini of each fusion protein.
We determined the activity of the reporter system by grow-  the eight non-native amino acids at the N-terminus were not responsible for the unexpected result of LagC-31 (data not shown).
The orientation of TMDs across a membrane is strongly influenced by the "positive-inside rule," and in general, topology is determined by the most positive loop (von Heijne, 1989(von Heijne, , 2006 The presence of aspartic acid residues at positions 22 and 27 in LagC will likely further force the TMD of the LagC-31 construct to reverse its orientation across the membrane. This would explain the unexpected blue color of transformants containing LagC-31. Based on our experimental data, in combination with our predicted topology models, we conclude that LagC is membrane-associated with four TMDs that are most likely oriented as shown in Figure 2.

| CON CLUS IONS
We have determined that CbnZ functions as the immunity protein for the two-peptide bacteriocin CbnXY. To date, this is the shortest known immunity protein for this class of bacteriocins. Using the pho-lac reporter system, we have also confirmed that CbnZ, along with the immunity proteins for LcnG and LcnMN, is membrane-associated. Our results reveal that the immunity proteins for these two-peptide bacteriocins display great structural variety: They range in length, number of TMDs, and orientation across the membrane. Our findings are an important first step toward uncovering the structural features of this intriguing group of proteins. Further investigations, such as labeling studies, may help uncover how these immunity proteins provide protection to their producer organisms.

ACK N OWLED G M ENTS
We are grateful to Gouzel Karimova at Institute Pasteur, France, for providing us with the pKTop plasmid. Funding for this work was provided by Discovery Grant RGPIN-2014-05457, from the Natural Sciences and Engineering Research Council of Canada (NSERC).
APB was the recipient of an NSERC undergraduate student research award (USRA) in 2018.

CO N FLI C T O F I NTE R E S T S
None declared.

AUTH O R CO NTR I B UTI O N S
APB, SRvdE, and LAMV performed the experiments. All authors were in involved in data analysis. LAMV and MJvB designed and supervised the study, and wrote the manuscript. All authors read and approved the final manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analyzed during this study are included in this published article.

E TH I C S S TATEM ENT
None required.