Structure-Activity Relationships of the Enterococcal Cytolysin

Abstract

Enterococcal cytolysin is a hemolytic virulence factor linked to human disease and increased patient mortality. Produced by pathogenic strains of Enterococcus faecalis, cytolysin is made up of two small, post-translationally modified peptides called CylL_L” and CylL_S”. They exhibit a unique toxicity profile where lytic activity is observed for both mammalian cells and gram-positive bacteria that is dependent on the presence of both peptides. In this study, we performed alanine substitution of all residues in CylL_L” and CylL_S” and determined the effect on both activities. We identified key residues involved in overall activity and residues that dictate cell type specificity. All (methyl)lanthionines as well as a Gly-rich hinge region were critical for both activities. In addition, we investigated the binding of the two subunits to bacterial cells suggesting that the large subunit CylL_L” has stronger affinity for the membrane or a target molecule therein. Genome mining identified other potential two-component lanthipeptides and provides insights into potential evolutionary origins.

Graphical Abstract

Introduction

The enterococcal cytolysin (Figure 1) is a two-component lanthipeptide virulence factor produced by opportunistic pathogens.¹ Cytolysin is made up of two subunits, CylL_L” and CylL_S”, which work together in a 1:1 ratio to lyse both mammalian and bacterial cells.² In addition to serving as toxin, cytolysin has quorum sensing activity.^3–5 Enterococci are a major source of nosocomial infections and are the causative agent for a number of disease states, including endophthalmitis, endocarditis, and non-alcoholic hepatitis.^{1, 6–7} Recently, cytolysin was directly linked to hepatic cell death and severity of liver disease in human populations.⁶ The chemical and three-dimensional structures of the polycyclic CylL_L” and CylL_S” peptides have been reported (Fig. 1),^8–9 with CylL_L” unexpectedly showing alpha helical secondary structure in solution. The stable helical structure for a short peptide containing three cyclic thioethers is reminiscent of synthetic stapling methods to enforce helicity.^10–11 In contrast to the knowledge about cytolysin structure, the molecular basis of its unique dual activity remains unknown.^{1, 12}

Figure 1.

(A) Schematic structure of CylL_L” with rings A-C and hinge region highlighted in dashed ellipses. See Figure 2 for full chemical structure. (B) Schematic structure of CylL_S” with rings A and B highlighted in dashed ellipses. (C) Lowest energy NMR structure of CylL_L” in the compact conformation (PDB: 6VGT). (D) NMR structure of CylL_L” in an extended conformation (PDB: 6VGT). (E) Lowest energy NMR structure of CylL_S” (PDB: 6VE9). (F) Schematic structure of cerecidin A7.

CylL_L” and CylL_S” belong to a group of natural products called ribosomally synthesized and post-translationally modified peptides (RiPPs).¹³ Typically, RiPPs derive from linear precursor peptides that contain an N-terminal leader peptide and a C-terminal core peptide (Figure 2). The leader peptide serves as a recognition motif for the post-translational modification enzymes, which catalyze modifications in the core peptide. CylL_L” and CylL_S” contain a total of 16 modified residues that are installed by a single class II lanthipeptide synthetase, CylM.^{8, 14} Select serine and threonine residues in the core peptide are first phosphorylated, followed by phosphate elimination to form dehydroalanine (Dha) and dehydrobutyrine (Dhb), respectively (Figure 2A). A subset of the dehydroamino acids then undergo cyclization via Michael-type addition by cysteine thiols, yielding lanthionine or methyllanthionine (Figure 2).

Figure 2.

(A) Schematic representation of class II lanthipeptide biosynthesis. (B) Structures of CylL_L” and CylL_S” with non-canonical LL-(methyl)lanthionine macrocycles highlighted in orange and canonical DL-lanthionine macrocycles highlighted in blue.

Cytolysin was the first lanthipeptide for which it was shown that the stereochemistry of the Michael-type addition resulted in the noncanonical LL configuration of three of the five thioether crosslinks (Figure 2B),^{8, 15} as all previously stereochemically characterized lanthipeptides had the DL-stereochemistry. Chemical synthesis of analogs in which the LL stereochemistry in cytolysin was changed to the DL stereochemistry displayed a 10-fold decrease in minimal inhibitory concentration (MIC) against bacteria, but no change in hemolytic activity,² suggesting that structure-activity relationships might be different for the two activities, a notion previously suggested in the literature.¹⁶

During the biosynthesis of cytolysin, the majority of the leader peptide is cleaved and the polycyclic peptide exported out of the cell by CylB, a peptidase domain-containing ABC transporter.^{14, 17} The remaining six residues of the leader peptide are removed by an extracellular serine protease, CylA,^18–19 to produce the mature CylL_L” and CylL_S” peptides. The biosynthesis of cytolysin has been successfully reconstituted in Escherichia coli as a heterologous host.⁸ In the heterologous expression system, the linear CylL_L or CylL_S peptides were co-expressed with CylM, and the CylM-modified peptides were isolated and proteolyzed in vitro with recombinant CylA, yielding the mature products. Prior to the development of the heterologous expression platform, cytolysin was isolated from the native producer organism through co-culturing with mammalian erythrocytes.³

Cytolysin-like two-component peptides are also produced by other bacteria such as the lactic acid bacterium Carnobacterium maltaromaticum C2.²⁰ C. maltaromaticum has been used to inhibit growth of Listeria monocytogenes in fish and meat products;²¹ unlike cytolysin, carnolysin does not have hemolytic activity.²⁰ Recent large scale bioinformatic analysis of predicted lanthipeptides encoded in the bacterial and archaeal genomes revealed that CylL_S-like precursor peptides are the third most abundant amongst all lanthipeptides.²² Unexpectedly, CylL_L-like precursor peptides are of significantly lower abundance, implying that a large number of CylL_S-like peptides function in a stand-alone manner. Indeed, investigation of the cerecidins, CylL_S-like peptides produced by Bacillus sp. that do not encode a CylL_L-like subunit (Figure 1), demonstrated antimicrobial activity at mid to low micromolar range without the need of a partner peptide.²³

Cytolysin is highly potent with MICs in the nanomolar range against many gram-positive bacteria. This degree of potency is often thought to require a molecular receptor rather than non-specific membrane lysis of antimicrobial peptides, which typically occurs at micromolar concentrations.²⁴ Other examples of lanthipeptides with specific molecular receptors are nisin,²⁵ mersacidin,²⁶ nukacin ISK-1,²⁷ and a group of two-component lantibiotics that all bind to lipid II,^28–29 and the duramycin group of compounds that recognize phosphatidylethanolamine.^30–31 In an effort to understand the structure-activity relationships of cytolysin, we performed in this study individual alanine substitution of every residue in both subunits. Each His₆-tagged mutant precursor peptide (CylL_L and CylL_S) was post-translationally modified by CylM in E. coli, purified by metal affinity chromatography, and its leader peptide removed by CylA. The peptides were characterized by mass spectrometry and evaluated for lytic activity against rabbit erythrocytes and antibacterial activity against Lactococcus lactis sp. cremoris. This strategy identified key motifs and residues that are required for bioactivity and cell-type specificity. Additionally, we performed sequential binding studies to determine which subunit binds first to the bacterial cell membrane and hence may interact with a putative specific receptor. Finally, we show the distribution of these peptides in the currently sequenced bacterial genomes.

Results and Discussion

Variant peptide library preparation

Single alanine substitution variants were generated using recombinant expression in E. coli BL21 (DE3) cells.⁸ The gene encoding each mutant precursor peptide was purchased as a double-stranded DNA fragment (Table S1) and assembled into the first multiple cloning site of pRSFDuet that encoded the cytolysin lanthipeptide synthetase, CylM, in the second multiple cloning site. The sequence of each assembled plasmid was confirmed by Sanger sequencing and the plasmids were used to transform the E. coli expression host. After co-expression, the CylM-modified His₆-tagged precursor peptide was isolated from cells and purified by immobilized metal ion affinity chromatography (IMAC). The purified peptide was next treated with recombinant CylA protease^18–19 to liberate the mature modified core peptide. The post-translationally modified core peptide was further purified using semi-preparative reversed-phase high-performance liquid chromatography (RP-HPLC) and analyzed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS, Table S2). CylM demonstrated remarkable substrate tolerance as all Ser/Thr residues that are dehydrated in the wt peptides were also dehydrated in the variants (Figures S1–S47). All variants, except for CylL_L”-G23A, were isolated in sufficient quantities for bioactivity determination. CylL_L”-G23A could not be isolated as co-expression of CylL_L-G23A with CylM resulted in no detectable product by MALDI-TOF MS after IMAC (see Methods). All peptides that had significantly reduced biological activities (see below) were also further analyzed to ensure full cyclization by reaction with the Cys-selective alkylation agent N-ethylmaleimide (Figures S49–S73).³²

Residues that control bioactivity and specificity of cytolysin

CylL_L” and CylL_S” exhibit synergistic lytic activity against mammalian and bacterial cells. Therefore, all purified mutant peptides were tested in triplicate with the wild-type counterpart subunit. The hemolytic activity was assessed by first determining the dose-response curve of the wild-type (wt) peptides (Figure S74). The Hill slope of the dose-response curve for hemolytic activity was 1.5 ± 0.3, perhaps suggesting weak positive cooperativity, but it is difficult to distill mechanistic information given the complexity of the molecular events that may underlie cell lysis. For comparison, the hemolytic activity of all peptide variants were tested at the EC₈₀ of 0.28 μM for wt cytolysin. The MIC of wt cytolysin against L. lactis sp. cremoris as indicator strain was determined in GM17 liquid growth media (32 nM, Table 1), and all subsequent antibacterial activity assays of the variants were conducted using peptide concentrations ranging from 2 to 256 nM. CylL_L” alone did not inhibit bacterial growth at concentrations up to 10 μM. However, CylL_S” alone inhibited bacterial growth with an MIC of 5 μM.

Table 1.

Bioactivity profiles of alanine-substituted cytolysin peptides when combined with the wt partner peptide. n.a. not accessible. Peptides in bold font are mutants of ring forming residues. Residues 6, 8, 9, 11–13, 16, and 17 are Ala in wt CylL_L” and therefore were not mutated. Similarly, residues 4, 14, and 18 are Ala in CylL_S”. Green fields are arbitrarily set as within one dilution of the wt MIC and as 80–125% of wt for hemolytic activity. Red fields are for MIC values >2 dilutions of the MIC of wt cytolysin and for hemolytic activity <40% of wt. Yellow fields are used for intermediary values. For the hemolytic experiments, all peptide concentrations were 0.28 μM.

Peptide	MIC (nM)a	Relative Hemolysis (%)b	Peptide	MIC (nM)a	Relative Hemolysis (%)b
CylLL”	32	100 ± 4	CylLS”	32	100 ± 4
CylL L ”-T1A	> 256	12 ± 1	CylL S ”-T1A	> 256	25 ± 11
CylLL”-T2A	> 256	12 ± 0	CylLS”-T2A	128	95 ± 8
CylLL”-P3A	128	104 ± 3	CylLS”-P3A	32	120 ± 0
CylLL”-V4A	> 256	55 ± 1	CylL S ”-C5A	>256	17 ± 1
CylL L ”-C5A	> 256	13 ± 0	CylLS”-F6A	64	45 ± 14
CylLL”-V7A	32	108 ± 5	CylLS”-T7A	32	114 ± 3
CylLL”-T10A	128	123 ± 6	CylLS”-I8A	128	14 ± 1
CylL L ”-S14A	> 256	35 ± 9	CylLS”-G9A	256	74 ± 7
CylLL”-S15A	256	82 ± 10	CylLS”-L10A	64	78 ± 6
CylL L ”-C18A	> 256	12 ± 1	CylLS”-G11A	32	102 ± 7
CylLL”-G19A	16	109 ± 5	CylLS”-V12A	16	60 ± 9
CylLL”-W20A	64	34 ± 1	CylLS”-G13A	32	62 ± 14
CylLL”-V21A	256	96 ± 4	CylLS”-L15A	64	94 ± 13
CylLL”-G22A	256	112 ± 8	CylLS”-F16A	256	13 ± 2
CylLL”-G23A	n.a.	n.a.	CylL S ”-S17A	>256	22 ± 11
CylLL”-G24A	> 256	12 ± 0	CylLS”-K19A	128	111 ± 3
CylLL”-I25A	32	48 ± 4	CylLS”-F20A	64	92 ± 6
CylLL”-F26A	32	117 ± 12	CylL S ”-C21A	>256	12 ± 0
CylLL”-T27A	16	124 ± 1
CylLL”-G28A	256	106 ± 1
CylLL”-V29A	64	60 ± 13
CylLL”-T30A	> 256	45 ± 8
CylLL”-V31A	16	82 ± 9
CylLL”-V32A	8	107 ± 2
CylLL”-V33A	32	96 ± 7
CylL L ”-S34A	> 256	12 ± 1
CylLL”-L35A	128	60 ± 8
CylLL”-K36A	16	63 ± 17
CylLL”-H37A	64	70 ± 14
CylL L ”-C38A	128	13 ± 0

^aMinimum inhibitory concentration against L. lactis sp. cremoris.

^bRelative hemolytic activity against rabbit red blood cells compared to wt peptides. Values shown are mean ± standard deviation of the mean for experiments performed in triplicate.

Substitution of the ring-forming residues in both peptides with Ala (CylL_L”-T1A, C5A, S14A, C18A, S34A, C38A and CylL_S”-T1A, C5A, S17A, C18A) revealed that the (methyl)lanthionine rings are essential for both antibacterial and hemolytic activities (Table 1). CylL_L”-C38A appeared to retain modest antibacterial activity, but mass spectral analysis demonstrated the addition of glutathione (Figure S1), likely to Dha34 that can no longer form the lanthionine-containing C-ring, with the effect of such addition unknown. However, replacement of Ser34 with Ala, which also prevented formation of the C-ring, resulted in a variant (CylL_L”-S34A) that did not show any bioactivity. Hence, all five rings in the two peptides appear essential for antibacterial and cytolytic activities.

Two additional non-ring forming residues, CylL_L”-Thr2 and CylL_L”-Gly24, also appeared to be critical for both biological activities of CylL_L” (Table 1). Sequence alignment of all ten unique CylL_L-like core peptides encoded in the genomes revealed that the amino acids at positions 2 and 24 are always Thr and Gly, respectively (Figure 3A). Thr2 is likely important for activity because after dehydration to Dhb, this residue is important for driving the stereochemistry of the A-ring.^{8, 15} Interestingly, CylL_L-G24 is fully conserved within the “hinge region” of CylL_L” that spans Gly22-Gly23-Gly24. The solution NMR structures of CylL_L” revealed two unique conformations of the peptide, a compact conformation and an extended conformation, that can interconvert through the flexibility of the Gly₃ hinge region (Figure 1C and D).⁹ The extended conformation spans a length of 45–49 Å, about the size of a bacterial lipid bilayer. Abolishment of both bioactivities in CylL_L”-G24A suggests that flexibility of the hinge region may be crucial for cytotoxic activity. The hinge region also appeared important for the biosynthesis because CylL_L”-G23A was the only variant out of 48 prepared in this study that could not be obtained. We infer that the corresponding precursor peptide is a poor substrate for CylM and/or that the unmodified peptide is degraded in the cell by proteases. Indeed, attempts to isolate the His₆-CylL_L-G23A precursor peptide for in vitro modification by CylM were unsuccessful due to observed truncation of the peptide during expression (see Methods). Two other residues in the large subunit, Val4 and Thr30, are important for antibacterial activity but less so for hemolytic activity. Position 4 is always Val/Ile in the alignment of CylL_L” peptides and position 30 is always Ser/Thr.

Figure 3.

(A) Sequence alignment of all unique CylL_L-like core peptide sequences in the currently sequenced genomes. (B) All unique CylL_S-like core peptides encoded near genes coding for CylL_L-like peptides. (C) CylL_S core peptide alignment for all unique cerecidin-like core peptides that do not have a CylL_L peptide encoded in the same genome. Fully conserved residues are colored in green and moderately conserved residues are colored in lime. Also shown is the designed cerecidin-like variant of CylL_S, CerL_S.

In CylL_S”, two non-ring forming residues, Ile8 and Phe16, appeared important for both antibacterial and hemolytic activity as replacement with Ala had clear detrimental effects (Table 1). Alignment of all unique CylL_S-like core peptides in the genomes that are associated with a CylL_L partner peptide shows that position 8 is typically a bulky, aliphatic amino acid, whereas the amino acid identity at position 16 is more divergent (Figure 3B).

In addition to residues that abolish or negatively affect both activities when mutated, we observed that mutation of certain residues resulted in differential activity, where one biological activity was mostly retained while the other was reduced or eliminated. Such differential effects were most obvious for CylL_L”-P3A, CylL_L”-T10A, CylL_L”-S15A, CylL_L”-V21A, CylL_L”-G22A, and CylL_L”-G28A in the large subunit and CylL_S”-T2A and CylL_S”-K19A in the small subunit, which all retained near full hemolytic activity whereas the antibacterial MIC was increased at least 4-fold. More subtle effects were observed for CylL_L”-W20A, CylL_L”-I25A, CylL_L”-V29A, CylL_S”-F6A, CylL_S”-L10A, CylL_S”-V12A, and CylL_S”-G13A for which hemolytic activity was reduced whereas antibacterial activity was less strongly affected. CylL_S”-G9A suffered an 8-fold increase in antibacterial MIC, and a minor reduction in hemolytic activity.

Several residues in CylL_L” (Val7, Gly19, Phe26, Thr27, Val31, Val32, and Val33) and CylL_S” (Pro3, Thr7, Gly11, Leu15, and Phe20) do not appear important for either activity. Overall, antibacterial activity appears to be more sensitive to amino acid perturbations than hemolytic activity (Table 1), although it must be noted that the two assays measure different aspects of activity. In general, substitution of the highly conserved amino acids resulted in loss of biological activity.

Finally, mutation of several highly or completely conserved amino acids had only minor effects on activity (Table 1). The most surprising of these was mutation of the completely conserved Lys36 in CylL_L”, which was anticipated to be important as it is the only amino acid with a positively charged side chain in the large subunit. Despite complete conservation and similarity in position to Lys19 in CylL_S”, which is important for antibacterial activity (see below), Lys36 in CylL_L” was not critical for antibacterial nor hemolytic activity. Clear effects were also observed for some residues that are not conserved. CylL_L-Trp20 and CylL_S-Lys19 lack high sequence conservation (Figure 3A,B), but were previously suggested to possibly interact with the target membrane.⁹ We observed that CylL_L”-W20A and CylL_S”-K19A lose hemolytic or antibacterial activity, respectively. Despite low sequence conservation, these residues appear to be important for the specificity of cytoxicity for different cell types.

Structure activity relationship studies of cerecidin A7 (CerA7) were previously reported.²³ In general, the analogous mutations in CylL_S” resulted in similar effects on antibacterial activities as those in CerA7 with a few notable exceptions. CylL_S”-T2A exhibited reduced activity, whereas CerA7-T2A appears to be completely inactive. CylL_S”-F16A displayed reduced antibacterial activity whereas in CerA7 the corresponding position is alanine in the wt peptide. Lastly, CerA7-T13A exhibited 4-fold higher antimicrobial activity than wt CerA7, but CylL_S” possesses leucine at position 15 and substitution by Ala did not affect either activity.

In summary, through single alanine substitution, we identified residues that are important for the biological activity and cell-type specificity of enterococcal cytolysin. These results with cytolysin variants suggest that the molecular targets in bacteria and mammalian cells may be related molecules that have small differences in structure or abundance in different cell types.

CylL_L” associates with the bacterial cell membrane

Previous studies on two-component lanthipeptide systems demonstrated that one subunit is responsible for target engagement, while the other subunit facilitates pore formation.^{28–29, 33–35} Thus, we examined the order of binding for cytolysin as previously described for other two-component systems.^33–34 The sequential binding experiment was conducted side-by-side with the two-component antibacterial haloduracin. Haloduracin is composed of two subunits, Halα and Halβ (Figure 4A),³⁶ with Halα binding to the cell wall precursor lipid II, followed by binding of Halβ to the initial complex to induce pore formation.^{29, 34} When CylL_L” was added to Lactococcus lactis sp. cremoris cells followed by washing and subsequent addition of CylL_S”, with the concentration of both peptides at the MIC (32 nM), no strong growth inhibition phenotype was observed (Figure 4B). Reversing the order of the experiment did not change the outcome. A similar experiment with Halα and Halβ (64 nM) showed a growth delay only when Halα was added first (Figure 4C), consistent with previous studies. Both sets of experiments were then repeated at a concentration corresponding to eight times the MIC. For both Halα and CylL_L” cell growth was completely inhibited up to 16 h if they were added first followed by washing the cells and addition of the second component. Importantly, without adding the second subunit, no growth inhibition was seen even at 8x MIC.

Figure 4.

Schematic structures of (A) haloduracin α and (B) haloduracin β. (C) Sequential binding assay for cytolysin against Lactococcus lactis sp. cremoris with CylL_L” added first at 1X MIC (32 nM, black square), CylL_S” added first at 1X MIC (32 nM, red circle), CylL_L” added first at 8X MIC (256 nM, purple diamond), and CylL_S” added first at 8X MIC (256 nM, gold triangle); in all cases the cells were washed and the partner peptide was added at the same concentration. Alternatively, CylL_L” was added without CylL_S” (256 nM, cyan triangle), and CylL_S” was added without CylL_L” (256 nM, brown hexagon). (D) Sequential binding assay for haloduracin with Halα added first at 1X MIC (64 nM, black square), Halβ added first at 1X MIC (64 nM, red circle), Halα added first at 8X MIC (512 nM, purple diamond), and Halβ added first at 8X MIC (512 nM, gold triangle); then the cells were washed and the partner peptide was added at the same concentration. Alternatively, Halα was added without Halβ (512 nM, cyan triangle), and Halβ was added without Halα (512 nM, brown hexagon).

When Halβ was added first at 8x MIC (512 nM), followed by washing and addition of Halα, a more modest growth delay was observed; Halβ does not bind to lipid II on its own.²⁹ Similarly, when CylL_S” was added first at 8x MIC (256 nM) followed by washing of the cells and addition of CylL_L”, a modest growth delay was observed. Our data therefore suggest that CylL_L” binds to the membrane, followed by CylL_S” binding and subsequent pore formation. These data may indicate that CylL_L” binds to a specific target in the membrane but potentially with less affinity than Halα binding to lipid II, which has been estimated to have a K_D of ~14 nM.²⁹ Alternatively, the larger size of the mostly hydrophobic CylL_L” peptide may explain a higher affinity for the membrane. A previous study also reported higher affinity of CylL_L” towards artificial lipid membranes compared to CylL_S”.³

Based on sequence homology to the cerecidins and their reported bioactivity in the absence of a CylL_L peptide, we had initially predicted that CylL_S” would be the subunit that might have affinity to a target. This would be analogous to the putative lipid II binding motif in ring B of Halα being found in single component lantibiotics such as nukacin II²⁷ or mersacidin.³⁷ But our findings in this work suggest that it may be CylL_L” that binds a membrane bound receptor molecule and that may have increased the antibacterial activity and broadened the spectrum of activity. To further investigate the factors that differentiate the activities of the one-component cerecidins and the two-component cytolysin, we generated a variant of CylL_S” that is similar to cerecidin A7 by encoding a C-terminal Lys and deletion of Phe6 and Thr7. This variant was fully dehydrated when co-expressed with CylM and termed CerL_S” (Figure 3C, S48). We tested its activity by itself against L. lactis sp cremoris resulting in an MIC of 10 μM, similar to the MIC of 5 μM of CylL_S”. When CylL_L” was added to the assay, no antibacterial activity was observed in stark contrast to the strong potentiation observed upon adding CylL_L” to wt CylL_S”. Additionally, hemolytic activity was not observed for CerL_S” with or without the addition of CylL_L”. Thus, the synergistic activity of the cytolysin peptides is highly specific.

Cytolysin and cerecidins diverged from a common ancestor

In an effort to understand the evolutionary relationship between the highly potent two-component cytolysin that has activity against both bacterial and mammalian cells and the much less potent cerecidins that only display antibacterial activity, a phylogenetic tree was constructed using the maximum likelihood method with sequences for the respective class II lanthipeptide synthetases (Figure S75).^38–39 CinM, a distantly related lanthipeptide synthetase, was defined as the outgroup.⁴⁰ The resulting tree contains three distinct main clades. One clade (blue, Figure S75) is made up of the lanthipeptide synthetases that produce cytolysins as defined by two distinct substrate peptides of different length (CylL_S and CylL_L orthologs) that are encoded nearby, one that will generate a product with two (methyl)lanthionines and one with three (methyl)lanthionines. In some cases, extra copies of CylL_L and/or CylL_S-like peptides were found. One cluster encoded in Enterococcus caccae is an exception and encodes just a single CylL_S subunit. The organisms encoding these systems are mostly enterococci but also other Firmicutes. Another clade (purple) contains the cerecidins. Although their biosynthetic gene clusters often have multiple precursor peptides, they are highly similar in sequence and length and resemble CylL_S” (e.g. Figure 3C). A third clade consists of enzymes in Bacillus species that again have two precursor peptides of distinct length. This clade can be further subcategorized into two groups based on the identity of the core peptide sequences of the substrates (Figure S75B). To the best of our knowledge no characterized examples of these peptides have been reported. We suggest the name bibacillins I and II (“bi-” as in two precursors, “-bacillin” as all current members are in Bacillus species) to describe these groups of lanthipeptides. Whether they are two-component systems and whether they have activity against both bacteria and mammalian cells are interesting questions for future investigation. The tree shows that the biosynthetic enzymes involved in producing cytolysin and cerecidins evolved from a common ancestor, whereas they are more distantly related to the lanthipeptide synthetase involved in biosynthesis of the bibacillins (Figure S75A).

Conclusion

In this study we identified key residues required for the biological activity of enterococcal cytolysin, a molecule with direct links to human disease. Cytolysin can lyse both mammalian and bacterial cells with high potency, the only such example amongst currently known lanthipeptides. Some residues are required for both hemolytic and antibacterial activity, which include all ring forming residues, and a Gly in the hinge region of the large subunit. Other residues were identified that endow the peptides with specificity against one cell type over the other. Our data is consistent with a model in which CylL_L” associates with a presumed membrane target, followed by binding of the CylL_S” subunit and pore formation. Further studies will be required to identify this putative target.

Methods

Cloning of single alanine mutant library

Plasmid assembly and propagation were conducted in chemically component NEBTurbo cells (New England Biolabs). Expression of recombinant peptides and proteins was performed after transformation of chemically competent BL21 Star (DE3) cells (Thermo Fisher Scientific) with the appropriate plasmid. Single stranded DNA primers were purchased from Integrated DNA Technologies and double stranded DNA gene fragments were purchased from Twist Biosciences. Individual linear DNA fragments were amplified from the single alanine substitution library with compatible homology arms for plasmid assembly using Phusion DNA polymerase. Either pRSFDuet-CylM or pET28b was linearized by PCR using Phusion DNA polymerase. The PCR products were purified by QIAprep spin column (Qiagen) and the concentration of DNA was quantified by Nanodrop absorbance assay (Thermo Fisher). The linear DNA fragments were mixed at 1:10 molar ratio backbone to insert and assembled using NEB HiFi Master Mix at 50 °C for 1 h. The assembly reaction was directly used to transform chemically competent NEBTurbo cells and plated on LB agar + kanamycin (50 μg/mL). All plasmids were confirmed by Sanger sequencing.

Peptide expression and purification

Terrific Broth (TB; 1.5 to 3.0 L) supplemented with 50 μg/mL kanamycin was inoculated with 1:100 dilution of overnight E. coli BL21 Star (DE3) cells containing the plasmid of interest. The cultures were grown at 37 °C with shaking to an OD₆₀₀ of 0.8. The peptide expressions were induced by addition of 0.3 mM IPTG, followed by incubation at 18 °C for an additional 16 h. The cells were harvested by centrifugation (6,000xg for 12 min), and the pellets were stored at −70 °C. Frozen cell pellets were thawed and resuspended in 50 – 100 mL of LanA Buffer B1 (6.0 M guanidinium hydrochloride, 0.5 mM imidazole, 0.5 M NaCl, 20 mM NaH₂PO₄, pH 7.5) and lysed by homogenization (Avestin). The lysate was clarified by centrifugation (25,000×g for 45 min) and applied onto a pre-equilibrated gravity column containing 1 mL of His60 SuperFlow resin (Takara Bio). The resin was washed with 15 mL of LanA Buffer B2 (4.0 M guanidinium hydrochloride, 30 mM imidazole, 0.3 M NaCl, 20 mM NaH₂PO₄, pH 7.5), followed by 15 mL of LanA co-expression Wash Buffer (30 mM imidazole, 0.3 M NaCl, 20 mM NaH₂PO₄, pH 7.5). The peptide was eluted with 15 mL of LanA co-expression Elution Buffer (0.5 M imidazole, 0.3 M NaCl, 20 mM NaH₂PO₄, pH 7.5) and stored at −20 °C. Because co-expression with CylM did not produce any product, CylL_L-G23A was also expressed as the linear precursor peptide and purified as described above to attempt in vitro modification with purified CylM. The elution fraction was desalted using Peptide Cleanup C18 Pipette Tips (Agilent) and analyzed by MALDI-TOF MS. The major peak corresponded to a peptide ending at Thr30, either because of incomplete translation or proteolytic degradation. Thus, it did not prove possible to access CylL_L”-G23A.

Peptide cleavage and HPLC purification

Recombinant CylA protease was expressed and purified as previously described.¹⁹ The peptide solution was thawed and treated with 100 μL of CylA (0.1 mg/mL) overnight at ambient temperature. The digested peptide solution was acidified with addition of 2% trifluoroacetic acid (TFA), centrifuged at 4500xg for 10 min, filtered through a 0.45 μm PVDF syringe filter and subjected to HPLC purification. Semi-preparative HPLC was performed on an Agilent 1260 Infinity II preparative liquid chromatography system connected to a 1260 Infinity II VWD and a 1260 Infinity II fraction collector. The peptides were purified on a Phenomenex Jupiter Proteo column (4 μm, 90 Å, 10 × 250 mm) using 0.1% TFA in H₂O (solvent A) and 0.1% TFA in CH₃CN (solvent) running at 4 mL/min. A gradient method was utilized with the following steps: 2% B for 10 min, ramp to 76% B over 25 min, ramp to 100% B over 5 min, hold at 100% B for 5 min, and ramp to 2% over 5 min.

Mass spectrometry and NEM alkylation assay

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) experiments were performed on a Bruker UltrafleXtreme MALDI-TOF/TOF MS. HPLC purified samples were spotted onto a MALDI target plate using 1 μL of 25 mg/mL super DHB (Sigma) in 80% CH₃CN/H₂O + 0.1% TFA. For N-ethylmaleimide (NEM) alkylation, peptides (1 μM) were reduced in citrate buffer (100 mM, pH 3.5) containing TCEP (0.1 mM) at 30 °C for 15 min. The solution containing the reduced peptides was adjusted to pH ~ 8 and NEM (0.1 mM) was added to initiate the alkylation reaction, which was allowed to proceed at 30 °C for 30 min. The samples were desalted using Peptide Cleanup C18 Pipette Tips (Agilent) prior to MALDI-TOF MS analysis.

Antibacterial assay for MIC determination

Minimum inhibitory concentration was determined in 96-well microtiter plates. Overnight cultures of L. lactis sp. cremoris were diluted to 10⁵ CFU/mL in GM17 media (M17 + 0.5 % glucose). In each well, 50 μL of cell suspensions were added followed by each peptide. The peptides were serially diluted to reach a final concentration of 256, 128, 64, 32, 16, 8, 4, and 2 nM. The cells were incubated at 30 °C for 18 h and OD₆₀₀ was measured on a Synergy H4 Hybrid Multi-Mode Microplate Reader (BioTek).

Hemolysis assay

Defibrinated rabbit erythrocytes (Hemostat) were washed in cold PBS by centrifugation at 1000×g for 3 min, decanting, and resuspension in fresh PBS until the supernatant remained colorless. Washed erythrocytes were diluted to a final concentration of 5% v/v. In a 96-well microtiter plate, 90 μL of diluted cells were added and pre-incubated at 37 °C for 30 min. To determine the EC₈₀, the wild type cytolysin peptides were serially diluted to reach a final concentration of 2, 1, 0.5, 0.25, 0.125, 0.0625, 0.0312, and 0.0156 μM in a final volume of 150 μL. As positive hemolytic control, Triton X-100 was added to a final concentration of 0.03% v/v in a final volume of 150 μL. For negative hemolytic control PBS was added to a final volume of 150 μL. Both cytolysin peptides were added at 1:1 molar ratio and incubated at 37 °C for 1 h. The cells were centrifuged at 1000×g for 10 min and 20 μL of the supernatant was removed. The supernatant was diluted in 180 μL of PBS and absorbance was measured at 415 nm. The % hemolysis was calculated with the following equation: $\%\mathit{\text{ Hemolysis}}=\left(\frac{Ab{s}_{\mathit{\text{sample}}}}{Ab{s}_{\mathit{\text{Triton}} X-100}}\right)*100$. The data for the wt cytolysin was fit to a simple dose-response equation in OriginPro 2020: $y=A_1+\frac{\left(A_2-A_1\right)}{1+{10}^{(\text{log}{x}_0-x)p}}$ where p = the Hill slope and y = % hemolysis, providing p=1.5 ± 0.3. When the Hill equation was used $E=\mathit{\text{Emax}}*\frac{1}{1+{\left(\frac{E{C}_{50}}{x}\right)}^n}$, which assumes binding of n ligands to a receptor and a measured response E, a Hill coefficient of 1.3 ± 0.1 was obtained. Given the potentially complex relationship between the measured entity (release of hemoglobin from red blood cells) and a molecular binding event, it is not clear whether the cooperativity suggests an oligomeric complex involving multiple cytolysin molecules. All peptide variants were then tested as described above at a single concentration of 0.28 μM.

Sequential binding assay

Sequential binding experiments were performed as previously described.³⁴ Briefly, an overnight culture of L. lactis sp. cremoris was diluted 1:20 in fresh GM17 media (~10⁷ CFU/mL). In a 96-well microtiter plate, 75 μL of cells were added to 25 μL of serially diluted peptides. The cells were incubated with peptides at ambient temperature for 20 min. The cells were centrifuged at 4500×g for 10 min. The supernatant was removed, and the cells were washed two additional times with fresh media. After washing, the cells were resuspended in 75 μL of fresh GM17 media and transferred to a new 96-well microtiter plate containing 25 μL of the corresponding subunit at equal molar concentration. The OD₆₀₀ was measured in 30 min intervals for 18 h at 30 °C.

Figure 5.

Summary of structure-activity relationship studies of CylL_L” and CylL_S” where red represents reduction/loss of both biological activities, green represents loss/reduction of only antibacterial activity, and yellow represents loss/reduction of only hemolytic activity. * Not assessed (see text).