In silico characterization of DdE and DdF proteins
The EntDD14 cluster is composed of 10 genes: ddABCDEFGHIJ. Genes ddAB encode the two peptides of EntDD14, while ddHIJ encode the ABC transporter. The other remaining 5 genes, ddCDEFG, were allocated to proteins of unknown functions. Of note, genes ddEF, that encode two proteins of 141 (DdE) and 458 (DdF) amino acid residues, respectively displayed homology with proteins carrying PH domains. The BLAST tool analysis revealed the presence of the conserved protein domain family YdbT (Genbank accession COG3428), within the primary structure of the DdF protein. Remarkably, this domain is based on the uncharacterized membrane protein YdbT, that contains PHb2 (bacterial pleckstrin homology) domain, that was first described in Bacillus subtilis. In the genome of B. subtilis ssp. subtilis 168 strain, the protein YdbT, of 493 amino acid residues is associated to heterologous antibiotic resistance (Genbank accession NP_388341.1), although no experimental evidence supported the allocation of such a function. Upstream of YdbT, we detected another PH domain-containing the protein named YdbS, of 159 amino acids, which has the conserved protein domain family YdbS (Genbank accession COG3402). The reported data from the Protein Data Bank (PDB) revealed that YdbS protein has two transmembrane (TM) domains and one PHb2 domain, while the YdbT protein has six TM domains and three PHb2 domains (Fig. 1a).
Of note, genes encoding homologous proteins of YdbT and YdbS were detected in the genomes of bacteria such as Staphylococcus aureus, Corynebacterium glutamicum or L. innocua (Table 1). In all these species, the two genes encoding YdbS- and YdbT-like proteins are contained in a same locus exhibiting a common transcriptional orientation, as is the case of ddE and ddF genes. Further, DdE and DdF proteins share sequence homology with B. subtilis YdbS and YdbT, respectively, though they are only 17% and 18% identical (Table 1, Fig. 1a).
According to BLAST analyses in the E. faecium L50 strain, which produces the leaderless two-peptide enterocin L50, counterparts of the DdE and DdF proteins exist, and displayed 79% and 74% homology to those found in the EntDD14 operon. Moreover, counterparts of these proteins were found in other enterococci strains which have homologous EntDD14 clusters as reported in a previous study [19].
To gain more insights on DdE and DdF, we analyzed in silico their secondary structure and TM domains prediction. This analysis supports the membrane localization of these proteins. Indeed, DdE contains two TM domains, whereas DdF contains six TM domains (Fig. 1b), and they seem to have similar structural organization such as YdbS and YdbT proteins of B. subtilis (Fig. 1a). In addition, a structural model was predicted for DdE and DdF using the I-Tasser program that utilises the resolved 3D structure of proteins deposited in the Protein Data Bank. The C-score is the confidence value for I-Tasser modelling which is between -5 and 2 and where values > -1.5 are considered as a correct global topology model [26, 27].
This topological prediction assigns the DdE and DdF protein models scores of -4.23 and -3.08, respectively (Fig. 1c), meaning a relatively low-confidence model. However, several proteins sharing DdE or DdF related structures are involved in translocation and transport across the membrane or belong to ABC transporters, as shown in Table 2. These in silico analyses strongly predict that the DdE and DdF proteins are located in the cell membrane and would have a role in EntDD14 transport.
PH domain-containing proteins DdE and DdF are essential for EntDD14 transport
To confirm our in silico analyses, we deleted genes encoding DdE or DdF and analysed the resulting phenotype of the mutant strains. Deletion of each gene was performed by homologous recombination, using the thermosensitive vector pLT06 [23]. Of note, E. faecalis 14 ΔddE and ΔddF mutant strains were obtained and their genetic backgrounds were confirmed by PCR and sequence analyses. Antibacterial assessment of cell-free supernatant (CFS) from ΔddE or ΔddF mutants was performed by the well-known agar diffusion test (ADT), against Listeria innocua ATCC33090 as bacterial target. Importantly, no inhibitory activity was detected, arguing the absence of EntDD14 in the CFS of the mutant strains (Fig. 2a). Therefore, each independently deleted gene entailed the total loss of antimicrobial activity. To confirm this hypothesis, MALDI-TOF/MS analysis was applied on CFS gathered from each mutant strain, as well as from the wild-type (WT). As expected, EntDD14 was not detected in the CFS from mutant strains (Fig. 2b), conversely that from the WT exhibited a typical peak of 5.2 kDa, corresponding to that of EntDD14, as previously reported by Caly et al. [18].
These independent ways of investigation enabled us to claim that abolition of DdE or DdF activity controls export from the cytoplasm of EntDD14. To strengthen this statement, trans-complementation assays were conducted upon cloning the ddF gene into the Gram-positive replicative plasmid pAT18 [28]. The E. faecalis ∆ddF-complemented strain was generated in the presence of erythromycin. However, the presence of the antibiotic is not compatible with the antimicrobial assay. A study of plasmid stability showed that after 10 and 30 generations without selection pressure, the number of bacteria still harbouring the pAT18:ddF recombinant-vector was 95% and 89%, respectively (data not shown). Thus, we performed all the assays with the complemented strain without erythromycin selection. Following this, the ∆ddF-complemented strain was able to secrete again EntDD14 into the CFS, as confirmed by the ADT (Fig. 2a), and MALDI-TOF/MS analyses (Fig. 2b).
These genetic experimental data showing that E. faecalis 14 lacking DdE or DdF protein is clearly unable to transport or translocate out of the cytoplasm EntDD14 bacteriocin reinforcing the predictions of the in silico analysis that allocated them a key role in the transport machinery. This surprising result suggests a new pathway in the mode of transport involving PH domain-containing proteins and likely also in the mode of action of the leaderless two-peptide EntDD14.
Loss of DdE or DdF protein leads to overproduction of the EntDD14 operon
To gain further insight into the EntDD14 mode of transport, a transcriptional analysis was carried out to evaluate the expression of genes involved in the production and transport of EntDD14, primarily those supposed to constitute the ABC transporter (ddHIJ). This gene expression experiment was conducted at 5 h (end of logarithmic phase) and 24 h (stationary phase) of bacterial growth of the WT strain and its isogenic derivatives ΔddE and ΔddF mutant stains, and the results are shown in Fig. 3.
Regarding these results, all the genes tested are constitutively expressed by the WT strain, both at the end of the exponential phase and in the stationary phase (24 h). This means that the WT strain is accustomed to tolerating the presence of EntDD14 at a level that does not interfere with its growth or development. In other words, the WT strain must have an intrinsic level of resistance or immunity to its own bacteriocin that may be due to the balance between the production and evacuation of the enterocin as reflected in the expression of the genes constituting its operon structure.
Indeed, when this expression balance is disrupted by turning off either of the ddE or ddF genes, the resulting mutants react differently from the WT and we observe more disturbance at the end of the exponential phase (Fig. 3a) than in the stationary phase (Fig. 3b) but the changes go in the same direction in the two cases.
At the end of the log phase, ddA and ddB genes were 4.6- and 3.5-folds overexpressed in the ΔddE mutant, and 3.6- and 3-folds in the ΔddF mutant (Fig. 3a). For both situations, overexpression of ddA and ddB genes suggests that they may be influenced by the ddE and ddF genes which could lead, in their corresponding mutants, to an overproduction of EntDD14. Among the genes involved in its extracellular export, mainly ddF of the ΔddE mutant is clearly overexpressed (2.7-folds) but at a lower level than for the structural ddA and ddB genes which suggests an additional deficit in the ability to evacuate the enterocin outside the cell. As for the other genes of the ABC transporter system (ddHIJ), they are overexpressed by a factor of about 2 and mainly in the ΔddE mutant.
This situation occurs also at the stationary phase but with lower overexpression levels and only for the ΔddE mutant since there is even a slight down expression of ddAB genes in ΔddF mutant (Fig. 3b) and this may be due to a much-reduced metabolic activity.
These data indicate overall that (i) ddAB genes, and (ii) those coding for ABC transporter are expressed in the mutant-strains deprived of DdE and DdF proteins, but the cells are unable to externalize EntDD14 outside of the cytoplasm; allowing EntDD14 to accumulate inside the cells. To investigate this point, total intracellular proteins extracted from ΔddE or ΔddF mutant strains were analysed by MALDI-TOF/MS, and compared to those extracted from the WT and the Δbac mutant strain, formerly obtained by knocking-out ddAB genes and characterized for its inability to produce EntDD14 [19]. In both ΔddE and ΔddF mutant-strains, a peak corresponding to EntDD14, with a molecular size of 5.2 kDa was detected (Figs. 4c, 4d, respectively). Of note, this peak was also detected in the WT but not in the Δbac mutant strain (Figs. 4a, 4b, respectively). The intensity of the peaks was 400 AU for WT, 1800 AU for ΔddE and 1,600 AU for ΔddF.
As expected, these data confirmed that EntDD14 is more accumulated in ΔddE and ΔddF mutant strains than the WT, arguing that these proteins have a key role in EntDD14 transport out of the cytoplasm.
EntDD14 accumulated inside the cytoplasm induces cell-toxicity
EntDD14 as above-stated accumulates inside the cell, when DdE or DdF is missing. To verify the probable deleterious effect of this accumulation, we compared the kinetic growth of the mutants and complemented strains to that of the WT strain (Fig. 5). These growth curves revealed discrepancies. The latency phase of the mutant strains is extended in the first hours of growth, but reached the same OD600nm than that of the WT strain at the entrance of the stationary phase and remains constant throughout the 24 h of the experiment (Fig. 5a). The growth rate (µ) of the mutant strain, ΔddE was 1.06 ± 0.06 and that of ΔddF was 1.08 ± 0.07, and does not differ from that of the WT strain, 1.16 ± 0.05. The complemented ΔddF-Comp strain shows the same behaviour as the WT strain, with a slightly lower growth rate, 1.01 ± 0.06, which can be ascribed to the presence of the pAT18:ddF plasmid. Overall, the mutant-strains have registered a loss in cell viability (Fig. 5b). The CFU counts indicate that all strains had reached approximately the same number of viable cells at the end of the exponential phase, ~3×109 CFU·mL-1. However, after 24 h of growth, ΔddE and ΔddF mutant strains have registered 1 log reduction in CFU/mL compared to the WT strain, which represents a 90% loss of cell viability. The loss of cell-viability is not necessarily correlated to loss of turbidity of the bacterial culture, as cell lysis seems not to occur.
To confirm this cell-viability feature in the ΔddE and ΔddF mutant strains, we performed a confocal microscopy analysis using the live/dead Bacterial Viability Kit. The ∆ddF and ∆ddE mutant strains showed similar numbers of bacterial cells, but very low live/dead ratio compared to the WT and the ∆ddF-complemented strains (Fig. 5c), revealing abundance of bacteria with compromised membranes and uncultivable. Therefore, the overall results support that, in addition to provoke cell lysis, the intracellular accumulation of EntDD14 is deleterious in the mutant cells lacking ddE or ddF genes.