Bioluminescence based biosensors for quantitative detection of enterococcal peptide–pheromone activity reveal inter-strain telesensing in vivo during polymicrobial systemic infection

Enterococcus faecalis is a significant threat in the nosocomial setting due to the emergence of isolates that are multi-antibiotic resistant, refractory to the available therapies and equipped with a variety of pathogenicity determinants. This bacterium uses quorum-sensing systems to regulate its physiological processes, including the expression of virulence traits, to adapt and proliferate within a host. Here, we describe the construction and application of two bioluminescence-based reporter systems for the direct detection of the quorum-sensing regulated expression of (i) the gelatinase biosynthesis-activating pheromone (GBAP) and (ii) the cytolysin small subunit (CylLS) in natural samples. The two E. faecalis reporters conditionally expressed bioluminescence in the presence of GBAP and CylLS both in the supernatants of liquid cultures and in an agar-overlay assay in as little as three hours, with a high level of sensitivity. Biosensors employed to investigate the interaction between the fsr and cyl systems revealed that fsr impeded CylLS activity by 75%. Furthermore, we identified a clinical E. faecalis isolate that acted as a biological cheater, producing cytolysin only upon sensing CylLS-producers in its environment. This isolate enhanced its virulence during polymicrobial systemic infection of Galleria mellonella.

ion of the cyl locus. Mature cytolysin consists of two peptides-CylL S and CylL L -and acts as a cytolytic toxin through forming a complex in eukaryotic and prokaryotic cellular membranes that leads to membrane rupture. Once synthesised, the two precursors are post-translationally modified by the product of the gene cylM and are secreted into the extracellular environment by the CylB transporter. Outside the cell, they eventually undergo a proteolytic activation through the action of CylA. In the absence of target cells, the larger subunit, CylL L , forms a stable inactive complex with the small subunit CylL S and inhibits its cytolytic activity. However, in the presence of a target, CylL L binds with a higher affinity to the cellular lipid membrane than to CylL S . As a consequence, the locally accumulating free mature CylL S will reach a certain threshold concentration that leads to the de-repressed binding of the regulatory proteins CylR1 and CylR2 on P Lys and the autoinduction of the cytolysin operon. An additional gene, cylI, provides self-protection from the bactericidal effect of the toxin. Through this regulatory system, E. faecalis can therefore finely tune the expression of cyl genes according to the presence of target cells.
The Fsr system is a major virulence regulator in E. faecalis and comprises four genes that are responsive to the extracellular accumulation of the gelatinase biosynthesis-activating pheromone (GBAP) 15,17 . The P fsrB promoter initiates the transcription of an operon comprising three genes: fsrB, fsrD and fsrC. The fsrD gene encodes the precursor of GBAP, which is processed by the product of fsrB following its extracellular release. The local accumulation of the GBAP peptide is sensed by the histidine kinase FsrC, which is on the surface of E. faecalis, which then activates the response regulator FsrA by phosphorylation. FsrA acts as a transcription factor that up regulates expression through the P fsrB promoter as well as through a promoter that controls the coordinate expression of two virulence factors, the gelatinase GelE and the serine protease SprE.
In a recent study, our group constructed and assessed the use of two bioluminescence-based systems for the in vivo non-invasive monitoring of E. faecalis cytolysin-and gelatinase-promoter activity in the murine intestine and during the systemic infection of Galleria mellonella larvae and mice 18 . By determining the bioluminescence emission at different time points during the progression of an E. faecalis infection, we showed that both the gelatinase and cytolysin promoters were subjected to temporal regulation and that the expression of these traits was controlled in response to sensing diverse environmental conditions.
In this study, we describe the use of two bioluminescence-based reporter systems as biosensors for the direct detection and quantification of GBAP and CylL S in biological samples. The two biosensors are based on the P gelE and cylR2R1P cyl promoters that drive the GBAP-or CylL S -induced expression of the luxABCDE operon specifically in the presence of true pheromone producers. Our results showed that the biosensors are suitable for the rapid, sensitive and real-time detection of positive isolates directly in natural samples and demonstrated for the first time that enterococcal telesensing both in vitro and in vivo during polymicrobial infection in G. mellonella larvae is possible.

Results and Discussion
Development of E. faecalis biosensors for detection of CylL S and GBAP pheromones. In a previous study, we utilised E. faecalis variants that expressed the luxABCDE cassette under the control of the cytolysin or gelatinase promoter to monitor the expression of the two factors during in vitro growth and during the infection of the mouse model. The results showed that the fsr and cyl QS-reporter systems were both modulated by environmental cues 18 .
In the current study, we constructed and employed lux-based biosensors for the detection and quantification of the cytolysin subunit CylL S and the gelatinase biosynthesis-activating pheromone (GBAP) of E. faecalis 19,20 .
The cytolysin biosensor was constructed by introducing the CylL S -responsive regulatory genes and the cyl promoter onto a lux-containing vector. The resulting plasmid, pSL101cylR2R1P cyl , was introduced into E. faecalis JH2-2 18 , a plasmid-free derivative of E. faecalis JH2, which lacks the cyl operon and is therefore unable to produce or sense the toxin 21 . For simplicity, in the following text, we will refer to the cytolysin biosensor as JH2-2 CBS.
To develop a functional GBAP biosensor system, we investigated the expression profile of both fsrB and gelE promoter-driven lux expression in E. faecalis MMH594. The growth of SL11 and SL13, which are, respectively P fsrB and P gelE reporter strains, at 37uC in GM17 medium was compared by monitoring the optical density at 620 nm. No significant difference in the growth rate or the final cell density was observed. This result indicated that the presence of the biosensor system did not hamper normal cell growth ( Figure 1). Both promoters led to the emission of bioluminescence throughout growth, with similar expression patterns; in both strains, the signal was low during the early exponential phase and increased from the beginning of the mid-logarithmic phase, before waning during the stationary phase. Nevertheless, under the condition of equal amounts of pheromone, P fsrB showed a lower level of specific activity, with only a 10-fold increase, whereas P gelE -driven luxABCDE expression increased 230-fold during growth. These results are consistent with the outcomes of previous studies showing that the phosphorylated response regulator FsrA had a lower binding affinity for the fsrB promoter than for the gelE promoter 15,22 . Due to the higher level of expression driven by the P gelE promoter under the conditions tested, pREG696 luxP gelE was therefore selected for introduction into E. faecalis V583fsrB* for use as the GBAP reporter (aka V583fsrB* GBS).
Biosensor proof of concept: Screening CylL S and GBAP production by genome-sequenced E. faecalis isolates. To test the ability of the JH2-2 CBS and V583fsrB* GBS to sense the presence of CylL S and GBAP producers in the environment, we employed nine genomesequenced E. faecalis strains of clinical and commensal origin and developed a screening method that utilised GM17-agar plates. The strains had been previously tested for their cytolytic phenotype and GBAP-production ability (S. Leanti La Rosa, L. G. Snipen, B. E. Murray, R. Willems, M. S. Gilmore, D. B. Diep, I. F. Nes, and D. A. Brede, submitted). The panel included the cytolysin-positive strains DS5 and 398, the GBAP-positive isolates E1Sol, V583 and V583DgelE, the GBAP-and cytolysin-producer MMH594, the GBAP-and cytolysin-negative strains T2 and CH188, and V583fsrB*, which harbours a mutation in the fsrB gene and is thus unable to synthesise the GBAP pheromone. Cells of the abovementioned strains were cultured on two GM17-agar plates that were individually overlaid with the biosensors. Induction of visible light emission by the CylL S-and GBAP-producers but not by the non-producers occurred by 3 hours following the application of the appropriate biosensor ( Figure 2). These bioluminescently tagged E. faecalis strains may therefore offer a simple, cost-effective and rapid method for determining the presence of cytolysin or GBAP producers in food, water, faecal and clinical samples. Furthermore, the systems were highly specific in sensing CylL S and GBAP and therefore would effectively prevent the false positive assumption of virulence traits based on only the detection of genes that might not necessarily lead to the corresponding phenotype.
To assess the feasibility of the biosensor systems to directly detect cytolysin-and GBAP-positive strains in natural samples growing on enterococci-selective media, mixed cultures of clinical E. faecalis isolates were plated on bile esculin agar (BEA) and overlaid with JH2-2 CBS or V583fsrB* GBS biosensors. In both cases, the biosensor overlay allowed the identification of bioluminescence-inducing colonies in as little as three hours, confirming that the screening method was effective in detecting a specific producer within a sample www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 8339 | DOI: 10.1038/srep08339 and is suitable for real-time monitoring of pathogens without the need for pure cultures ( Figure 3).
The JH2-2 CBS and V583fsrB* GBS detected high levels of CylLsand GBAP-pheromone activities in cell-free culture supernatants.
To corroborate the applicability of the CylL S and GBAP biosensors, we employed them for the direct detection and quantification of the two pheromones in liquid substrates. Whereas GBAP has been successfully isolated from the supernatants of E. faecalis cultures 17 , previous studies reported that cytolysin was clearly produced by E. faecalis growing on blood agar or when erythrocytes were added to broth cultures but a low level of activity was observed in liquid cultures 23,24 . It was therefore of interest to determine whether the CylL S and GBAP biosensors were able to specifically sense the local accumulation of mature pheromones in culture supernatants.
To investigate GBAP detection, serial dilutions of supernatants from overnight and exponential-phase cultures of E. faecalis V583, V583DgelE and V583fsrB* were mixed with an equal amount of a 50-fold dilution of the GBAP biosensor in its exponential-phase. Bioluminescence was detected in the supernatants of the E. faecalis V583 and V583DgelE cultures but not in those of the V583fsrB* cultures, confirming the specificity of the construct, which reacted only to the gelatinase biosynthesis-inducing pheromone ( Figure 4A). A highly significant correlation was found between the level of light emission in photons/s over a three-fold order of serial dilution of the samples of the supernatants of exponential-(R 2 . 0.90) and stationary-phase cultures (R 2 . 0.96), showing that this biosensor is an accurate and sensitive mean for quantifying the concentration of the GBAP pheromone. The measured GBAP activity was 320 GIU in exponential phase supernatants and 5120 GIU in overnight  central row: V583fsrB*, CH188, and 398; bottom row: MHH594, T2, and V583DgelE. The E. faecalis isolates were cultured on GM17 agar plates, and the biosensor was overlaid after an overnight incubation. Plates were kept at 37uC for 3 hours and imaged with a Xenogen IVIS Lumina II Imaging System (Calipers Corp., CA). A 10-fold higher light emission was induced by CylL S than by GBAP.
www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 8339 | DOI: 10.1038/srep08339 culture supernatant (data not shown). GBAP was detected at similar levels in the supernatants of both V583 and V583DgelE stationaryand exponential-growth phase cultures (P . 0.05, according to the Mann-Whitney test). This finding suggested that the gelatinase of these isolates did not inactivate or affect the biological activity of GBAP. The high levels of GBAP activity observed in the overnight cultures strongly suggested that the down-regulation of the fsr circuit during the stationary phase was not due to the inactivation of GBAP activity.
Similarly, the JH2-2 CBS facilitated the specific detection of active CylL S in cell-free supernatants. JH2-2 CBS sensed up to 640 CIU in the supernatants of cells in the exponential phase of growth and 5120 CIU in overnight culture supernatant (data not shown). The exhibited levels of bioluminescence showed a linear dose-response dependency with the CylL S concentration (R 2 . 0.97). This trend allowed the quantitative measurement of CylL S activity during bac-terial growth. CylL S was found to accumulate in large amounts during the exponential growth phase of the CylL S -producers 398 and MMH594 but was found to be totally absent in cultures of the non-producer T2 ( Figure 4B). To the best of our knowledge, the presence of a high level of biologically active CylL S pheromone in the absence of target cells has not been reported hitherto. Surprisingly, the content of CylL S increased during the growth of in vitro monocultures, despite the fact that CylL S and CylL L are known to form non-haemolytic oligomers 25 . This result implied that cyl signalling was highly responsive even when the level of toxin activity was low.
Impairment of cytolysin production is mediated by the metalloprotease GelE. Interestingly, we observed a 2-fold difference in the levels of CylL S production in 398 and MMH594 exponential-phase cultures, and a 4-fold higher level of CylL S in the supernatants of late . Interestingly, the effect of the co-presence of these two virulence traits was much less than the sum of their main individual effects, described as saturation or antagonistic effects. It was previously reported that GelE contribute to the virulence of E. faecalis by triggering the proteolytic degradation of a broad range of host substrates [26][27][28] , and gelE expression was found to have a profound effect on the secretome of E. faecalis 29,30 . GelE activity was found to be required for a variety of processes, such as the regulation of the display of the surface adhesin Ace 31 and the activation of the primary autolysin AtlA and its contribution to biofilm formation 32 . Additionally, the production of gelatinase was reported to impair other cellular activities, such as conjugation due to the degradation of sex pheromone-related peptides 33 . Taken together with the results of utilising the CylL Sbiosensor detection system, we considered the possibility that gelatinase might also have an impact on cytolysin. To test this hypothesis, we employed the JH2-2 CBS biosensor to detect CylL S accumulation in the supernatants of exponential-and stationaryphase cultures of the gelatinase-overproducing E. faecalis strain 398::pCG ( Figure 5). Under both of the tested conditions, our data showed that introducing gelatinase into the Cyl producer 398 leads to a 4-to 8-fold reduction of CylL S activity (p , 0.05, according to the results of Mann-Whitney test).
We previously observed that cytolysin is expressed mostly during the late exponential phase and that its expression gradually subsides during the stationary phase 18 . This trend is consistent with a scenario in which GelE is able to degrade part of the CylL S that was produced during the active growth phase and that GelE continued to exert its proteolytic activity long after the production of CylL S subsided. The fact that Cyl production was reduced by 75% was consistent with our previous observations regarding the pathogenicity of E. faecalis in C. elegans . It thus appears probable that this phenomenon was highly effective in vivo.
Interstrain CylL S -mediated communication demonstrated the occurrence of an in vivo cheating behaviour. Quorum sensing is a specific type of telesensing that allows cell-cell communication via small diffusible molecules that are released to explore the surrounding environment 34 . In its simplest form, quorum sensing enables the bacteria to control the production of molecules that are released into the extracellular environment and become available not only for sibling producer cells but also for any other cells present. However, it has been reported that individuals that do not respond to quorum-sensing signals act as cheaters, not incurring the metabolic cost of producing those released molecules while benefiting from those secreted by cooperators 35 . The cheater gains a fitness advantage over the quorum-sensing positive strains. The production of extracellular metabolites, such as virulence factors, may lead to the recognition of the bacteria and ultimately destruction by the host immune system 36 . Therefore, quorum sensing can be used to trigger concerted gene activation in the microbial community to effectively respond to the prevailing conditions, which could be activating processes for evading a host or escaping the host's defences.  Studies have shown that the cyl locus is subject to genetic instability both in vitro and in vivo 37,38 . A number of investigators have reported phenotypic instabilities with no obvious explanations at the genetic level, indicating that silent or non-functional cyl genes occur in the genomes of enterococcal isolates of different origins [39][40][41] . In a previous report, we identified an apparent incongruence between the cyl genotype and phenotype in the clinical isolate T2 42  Here, we performed a detailed comparative sequence analysis of the cyl locus of a subset of genome-sequenced E. faecalis strains, including the strain MMH594, and we detected the presence of an IS6770 element that was integrated into the 39-end of the cylA gene in T2 (Fig. 6A). The insertion causes a premature stop and the production of a truncated, and presumably non-functional, CylA (Fig. 6B).
We then tested the hypothesis that E. faecalis strains exhibiting a cytolysin-negative phenotype but harbouring the elements of the cyl locus could act as social cheaters that cease production of quorumcontrolled cytolysin and benefit from or take advantage of the release of CylL S by cooperators. A variant of E. faecalis T2 tagged with pSL101cylR2R1P cyl (named T2 CBS) was employed; the assessment of cytolysin production mediated by sensing the local accumulation of mature CylL S subunits in the environment was performed by growing E. faecalis T2 CBS in proximity to the cytolysin-positive strain MMH594 on blood-agar plates (Figure 7). A diffuse and bright haemolytic zone was observed in the area in which the two strains were in proximity to each other (Fig. 7A) indicating that E. faecalis T2 CBS was able to detect and use the CylL S released by the MMH594 cells to trigger its own cytolysin production. Imaging analysis showed the 'flare'-like induction of bioluminescence throughout the streak of T2 CBS ( Figure 7C). Haemolysis or the 'flare' effect was not detected when JH2-2, a strain lacking all of the genes necessary for cytolysin production, was streaked near T2 CBS cells (data not shown). In this case, bioluminescence arose only from the JH2-2 CBS cells in close proximity to MMH594 cells. No increase in the level of haemolysis or light emissions was observed when MMH594 or T2 CBS cells were cultivated in proximity to E. faecalis CH188, which lacks a complete cyl operon 43 (Figures 7B and 7D). In addition, no haemolysis was observed when T2 CBS was cultivated alone or in proximity to its parental strain on blood-agar plates (data not shown). To test whether the cytolysin-positive phenotype was stably imposed on the T2 cells, cells that had been induced once were re-streaked on blood agar. However, these cells consistently reverted to a cytolysinnegative phenotype (data not shown). Based on these results, we hypothesised that E. faecalis T2 acted as a social cheater, adopting the strategy of reducing its metabolic burden by avoiding the production and release of the toxin and benefiting from a growth advantage over cytolysin-producing cells. However, this strain retained its immunity to the bactericidal effect and its ability to exploit the release of CylL S into the environment by a primary producer to efficiently spread during a polymicrobial infection.
In vivo cytolysin telesensing during infections of G. mellonella larvae exacerbated the virulence. A previous study indicated that appropriate regulation of both the gelatinase and the cytolysin promoter occurred during the mono-infection of G. mellonella larvae with lux-tagged E. faecalis MMH594 variants and that these traits were conditionally induced in response to the insect haemocoel environment 18 . To investigate whether the interstrain communication observed in T2 CBS (as described above) could affect pathogenicity in vivo, we performed a polymicrobial infection of G. mellonella larvae with E. faecalis T2 CBS in combination with the cytolysin-producer 398. Injecting 2 3 10 6 T2 CBS cells resulted in an LT 50 of approximately 13 hours ( Figure 8A) and low levels of bioluminescence ( Figure 8B). Injecting an equal amount of inoculum of T2 CBS in combination with 398 at a dose of approximately 2 3 10 5 cells into the haemocoel of the insects resulted in an increased level of virulence, with an LT 50 of 11 hours (P 5 0.015, chi-square test with 1 degree of freedom, T2 CBS 1 398 versus T2 CBS). Moreover, the 100-fold increase in bioluminescence observed at 6 hours post infection was indicative of interstrain CylL S communication. To explore the possibility that the increased pathogenicity could be due to the presence of 398 or was a result of induced cytolysin production by T2, a similar experiment was performed using the JH2-2 CBS instead of the T2 CBS, because the JH2-2 strains lacks the cyl operon. In this experiment, the JH2-2 CBS demonstrated CylL S interstrain communication, with a level of bioluminescence induction similar to that observed when T2 was used as the biosensor ( Figure 8B), but there was no increase in the level of virulence ( Figure 8A). Collectively, these results suggested that in vivo telesensing was a credible scenario during polymicrobial enterococcal infections that potentially could contribute to increasing the severity of such infections. These findings might have broad implications because a number of apparently cyl-negative phenotypic isolates contain intact cyl loci or a deletion in the cylA proteinase gene 40,41 . It is conceivable that such strains would both be immune to the cytolysin toxin and demonstrate cytolytic activity when sharing a habitat with a CylA-expressing strain. Alternatively, clinical isolates such as T2, which bears defective cyl loci, might have evolved during infection via a mechanism similar to that by which V583 descended from V586 44 .
Using bioluminescence imaging, we were able to follow the realtime dynamics of the CylL S levels in intact insects during a mortality assay ( Figure 8B). No light was detectable at any time point in the larvae infected with 398, which confirmed the absence of any back- www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 8339 | DOI: 10.1038/srep08339 ground bioluminescence. A basal level of light emission was observed in the insects that had been injected with T2 CBS; however, the level of bioluminescence reached a peak at 5 h post infection and diminished progressively. Despite this basal signal, a 716-fold increase in the level of bioluminescence was observed in the larvae that had been co-infected with both isolates. The signal caused by infection with the mixed culture of T2 CBS and 398 was 60-fold higher than the basal signal; the signal progressively increased and peaked at 8 h postinfection, concomitant with when the insects began to die, and per-sisted at this level throughout the death of the wax moths. Similarly, co-infection with JH2-2 CBS and 398 gave rise to a 69-fold higher bioluminescence emission than the basal signal from larvae injected with JH2-2 CBS ( Figure 8B).
Taken together, the evident change in the levels of virulence and P cyl -driven luxABCDE expression that was observed during G. mellonella co-infection supported the hypothesis that a small population of CylL S producers can induce cytolysin synthesis by responsive cheaters during a polymicrobial infection.  www.nature.com/scientificreports

Conclusions
In this report, we describe the construction of two biosensors for the simple and rapid detection of CylL S and the gelatinase biosynthesisactivating pheromone in E. faecalis. We showed that these reporters can be used for the real-time identification of pheromone producers both on agar plates and from culture supernatants and illustrated the applicability of the constructs to natural samples, alleviating the need for pure cultures. The biosensors were used to monitor the production of the pheromones GBAP and CylL S during growth and they detected quantitative between-strain differences in CylL S activity.
Pursuing this observation, we showed that gelatinase had an antagonistic activity toward cytolysin production, probably through its proteolytic degradation of the cytolysin-toxin subunits. To the best of our knowledge, this is the first example of an antagonistic interaction between two virulence traits that has been demonstrated in E. faecalis.
We also used the biosensors to elucidate a novel type of social cheating mechanism, which enabled conditional toxin production in a certain strain of E. faecalis. The production of cytolysin by the cheater strain was dependent upon its recognition of the accumulation of CylL S produced by another cyl-expressing strain in its surrounding environment. The relevance of this mechanism was corroborated during the polymicrobial infection of the G. mellonella model system. Therefore, these reporter systems represent a powerful tool for studying E. faecalis pathogenicity, which will allow the population dynamics of cytolysin and GBAP producers to be monitored and will expand the current knowledge of the expression and functional activity of E. faecalis genes in microbe-host interactions.

Methods
Bacterial strains, plasmids and growth conditions. The bacterial strains used in this study are listed in Table 1. Unless otherwise indicated, the E. faecalis strains were routinely cultivated at 37uC, without agitation, in M17 broth (Oxoid LTD, UK) supplemented with 0.4% w/v glucose (GM17). The E. coli strains were grown at 37uC, with shaking, in Luria-Bertani (LB, Oxoid LTD, UK) broth. The following antibiotic concentrations were used for the enterococci: spectinomycin, 500 mg/mL and chloramphenicol, 20 mg/mL. The following antibiotic concentrations were used for the E. coli: spectinomycin, 200 mg/mL and chloramphenicol, 10 mg/mL.
The E. faecalis CylL S -biosensor strains (CBS) were constructed by introducing the pSL101cylR2R1P cyl vector into E. faecalis strains JH2-2 and T2 18 . To develop a GBAP biosensor (GBS), the following two vectors were employed: pREG696 luxP fsrB 45 , which contained a 480-bp segment of the upstream region and the ATG codon of fsrB fused to the luxABCDE cassette of pPL2 lux and the axe-txe stability module, and pREG696 luxP gelE 18 .
The vectors were propagated in E. coli GeneHogs, and the plasmid DNA was isolated using the E.Z.N.A. Plasmid Mini Kit I (Omega Bio-tek, USA) prior to transferring it into the E. faecalis strains via electro-transformation 46 . The transformants were selected using GM17 plates containing spectinomycin.
Assessment of P fsrB and P gelE promoter activities during in vitro growth. The growth and the luminescence expression driven by the P fsrB and P gelE promoter of the SL11 and SL13 E. faecalis strains, respectively, were evaluated as described previously 47 . Briefly, E. faecalis overnight (ON) cultures were diluted 100-fold using GM17 medium, grown at 37uC until they reached an optical density at 620 nm of 0.2 and then diluted 100-fold again using fresh GM17 broth. A 300-ml aliquot of the culture was added to the wells of a 96-well plate (Nunc, Thermo Fisher Scientific, Denmark) and the plate was incubated at 37uC under static conditions in a Spectrostar Nano microplate reader (BMG Labtech). The absorbance at 620 nm was measured at 15-min intervals for 7 h. For the bioluminescence measurements, 300 ml of the same culture was added to the wells of a black 96-well plate (Nunc, Thermo Fisher Scientific, Denmark), which was incubated at 37uC under static conditions in the chamber of a Xenogen IVIS Lumina II Imaging System (Calipers Corp., CA). Luminescence was measured every 15 min for 7 h, with a binning factor of 16, F-stop of 1 and an exposure time of 1 minute.
To determine the specific promoter activities, the photons/second emissions of the different lux-tagged E. faecalis cultures were normalised according to their growth, as expressed as the optical density (OD) at 620 nm.
Validation of the specificity of the CylL S -and GBAP-producing colony-screening assays. Single colonies of genome-sequenced E. faecalis isolates that are known to produce CylL S and/or GBAP were cultured overnight in GM17 broth at 37uC. A 2.5ml aliquot of each culture was spotted onto two GM17 agar plates, which were incubated for 16 h at 37uC. The plates were subsequently overlaid with 10 mL of GM17 soft agar (0.8% w/v agar) that had been tempered to 50uC and were seeded with 200 ml of an overnight biosensor culture. After incubation at 37uC for 3 h, the plates were visualised using a Xenogen IVIS Lumina II Imaging System (Calipers Corp., CA), using a binning factor of 16, F-stop of 1 and an exposure time of 1 minute.
Identification of CylL S and GBAP producers on enterococci-selective plates. A mixed culture containing two known CylL S and GBAP producers was plated on two bile-aesculin agar (BEA, Oxoid LTD., UK) plates, which were incubated at 37uC overnight. The dilution factor of the culture was adjusted so that the plating resulted in the production of 20-30 colonies per plate. After performing the biosensor overlay as described above, the plates were maintained at 37uC for 3 hours and were visualised using a Xenogen IVIS Lumina II Imaging System (Calipers Corp., CA), using a binning factor of 16, F-stop of 1 and an exposure time of 3 minutes.
Detection of CylL S and GBAP in E. faecalis supernatants. CylL S and GBAP detection in the supernatants of cultures of genome-sequenced E. faecalis producers was performed using the biosensors JH2-2 CBS and V583fsrB* GBS, respectively. Single colonies of E. faecalis 398, MMH594, T2, V583, V583DgelE and V583fsrB* were inoculated into 5 mL of GM17 broth, and the cultures were incubated overnight at 37uC. The cultures were then diluted 15100 using fresh GM17 broth and grown at 37uC; aliquots of the cultures were taken when the OD 620 values were 0.1, 0.25, 0.5 and 1.0. After centrifugation at 10,000 3 g for 10 minutes to remove the cells, the culture supernatants were collected and stored at 4uC until use. The supernatants were subjected to serial 2-fold dilutions in a volume of 100 ml/well in a 96-well plate (Nunc, Thermo Fisher Scientific, Denmark), and 100 ml of a 503 diluted exponential phase culture or an ON culture of the appropriate biosensor was added to each well. After incubation at 37uC for 3 h, the levels of bioluminescence emission were measured using a Xenogen IVIS Lumina II Imaging System, using a binning factor of 16, F-stop of 1 and an exposure time of 1 minute.
One CylL S inducing unit (CIU) was defined as the reciprocal of the highest level of dilution of a sample that provided a 2-fold increase in the level of bioluminescence of a 0.2-mL aliquot of a JH2-2 CBS culture compared with the level of basal expression. Similarly, 1 GBAP-inducing unit (GIU) was defined as the reciprocal of the highest level of dilution of a sample that provided a 2-fold increase in the level of bioluminescence of a 0.2-mL aliquot of a 0.2 mL culture of V583fsrB* GBS compared with the level of basal expression.
In vitro detection of cytolysin-mediated interstrain communication. The cytolysin-positive E. faecalis strain MMH594 was inoculated onto a blood-agar plate (Brain-heart infusion agar supplemented with 5% (v/v) defibrinated horse blood, 1% (w/v) glucose and 0.03% (w/v) L-arginine (Sigma-Aldrich)) by creating a curved streak that reached the centre of the plate. Samples of E. faecalis CH188 or T2 CBS, which exhibited a negative haemolytic phenotype, were similarly streaked on the right side of the plate, 2-mm from and not touching the MMH594 cells. As a control, a plate was similarly seeded with CH188 and T2 CBS cells. The plates were incubated at 37uC for 16 h under anaerobic conditions. Imaging of the plates was performed as described above.
In vivo CylL S telesensing during polymicrobial systemic infection of G. mellonella using E. faecalis. G. mellonella larvae were infected with E. faecalis as previously described, with some minor modifications 47 . Briefly, exponential-phase cultures of E. faecalis 398 and T2 CBS grown in GM17 broth were washed three times using a sterile 0.9% saline solution and were brought to concentrations of 2 3 10 7 6 0.8 3 10 7 CFU/mL and 2 3 10 8 6 1.8 3 10 8 CFU/mL, respectively. For the purpose of coinfection, 398 and T2 CBS cells were mixed in a ratio of 1510 v/v. Larvae (weighing approximately 3 mg and approximately 3 cm in length) were injected through the left hindmost proleg with 10 ml of E. faecalis solution using a Hamilton 710SNR 100-mL syringe (Hamilton Company) fitted with a 30 G needle (BD Microlance 3). For each assay, 10 insects were used in triplicate and the experiment was independently repeated at least two times. The larval survival rate was determined at 20 hours after infection. For real-time visualisation of E. faecalis infection in G. mellonella, five individual insects were injected as described above and were placed in duplicate in a 4.0-cm Petri dish. The plates were incubated at 37uC in the chamber of a Xenogen IVIS Lumina II imaging system (Caliper Life Sciences, CA) and the bioluminescence emissions were recorded at 30-minute intervals for 20 hours. The LT 50 values (time at which the lethality of 50% of the insects was reached during a 20-hour period) were employed to compare the level of infectivity of the inoculum. This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder in order to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ www.nature.com/scientificreports