Competition between VanUG Repressor and VanRG Activator Leads to Rheostatic Control of vanG Vancomycin Resistance Operon Expression

Enterococcus faecalis BM4518 is resistant to vancomycin by synthesis of peptidoglycan precursors ending in D-alanyl-D-serine. In the chromosomal vanG locus, transcription of the resistance genes from the PYG resistance promoter is inducible and, upstream from these genes, there is an unusual three-component regulatory system encoded by the vanURSG operon from the PUG regulatory promoter. In contrast to the other van operons in enterococci, the vanG operon possesses the additional vanUG gene which encodes a transcriptional regulator whose role remains unknown. We show by DNase I footprinting, RT-qPCR, and reporter proteins activities that VanUG, but not VanRG, binds to PUG and negatively autoregulates the vanURSG operon and that it also represses PYG where it overlaps with VanRG for binding. In clinical isolate BM4518, the transcription level of the resistance genes was dependent on vancomycin concentration whereas, in a ΔvanUG mutant, resistance was expressed at a maximum level even at low concentrations of the inducer. The binding competition between VanUG and VanRG on the PYG resistance promoter allowed rheostatic activation of the resistance operon depending likely on the level of VanRG phosphorylation by the VanSG sensor. In addition, there was cross-talk between VanSG and VanR'G, a VanRG homolog, encoded elsewhere in the chromosome indicating a sophisticated and subtle regulation of vancomycin resistance expression by a complex two-component system.


Introduction
Vancomycin-resistant enterococci are a major cause of nosocomial infections and an important public health problem because the treatment options for the infections they cause are very limited [1]. Vancomycin, which can be the only antibiotic effective against multiresistant clinical isolates, acts by binding to the C-terminal D-alanyl-D-alanine (D-Ala-D-Ala) residues of peptidoglycan precursors blocking the extracellular steps in peptidoglycan synthesis [2]. Resistance in Enterococcus is mediated by nine types of operons that produce modified peptidoglycan precursors ending in D-Ala-D-Lac (vanA, -B, -D, and-M) or D-Ala-D-Ser (vanC, -E, -G, -L, and-N) to which vancomycin bind with a low affinity and from the elimination of the high affinity precursors ending in D-Ala-D-Ala [3][4][5][6].
Expression of the vancomycin resistance operons is regulated by VanS/VanR-type twocomponent signal transduction systems composed of a membrane-bound histidine kinase (VanS-type) and a cytoplasmic response regulator (VanR-type) that acts as a transcriptional activator [3]. The sensors modulate the levels of phosphorylation of the regulators. In the presence of vancomycin, VanS acts primarily as a kinase that autophosphorylates and transfers its phosphate to VanR. Phosphorylated VanR binds to the promoters upstream from the vanRS regulatory and resistance operons leading to increased transcription of the regulatory and resistance genes [7][8][9]. The phosphatase activity of VanS-type sensors is required for negative regulation of the resistance genes in the absence of vancomycin preventing accumulation of VanRtype regulators phosphorylated by acetylphosphate or by kinases encoded by the host chromosome [7,10].
VanG-type Enterococcusfaecalis clinical isolates from Australia and Canada are distinct from other Van-type enterococci. The chromosomal vanG cluster (Fig 1) confers resistance to vancomycin (MICs, 16 μg/ml) by inducible synthesis of precursors ending in D-Ala-D-Ser [11]. It contains the vanY G ,W G ,G,XY G ,T G resistance genes, the last three strictly required for resistance encode, respectively, a VanG ligase to synthesize D-Ala-D-Ser, a VanXY G D,Dcarboxypeptidase to hydrolyse D-Ala-D-Ala, and a VanT G membrane bound serine racemase to produce D-Ser (Fig 1). As opposed to the other van gene clusters, the vanG regulatory operon contains three genes, vanU G , vanR G , and vanS G , encoding a "three component" regulatory system (Fig 1). Additional gene vanU G encodes a transcriptional regulator belonging to the Xre protein family and of unknown function. The vanURS G genes are co-transcribed, even in the absence of vancomycin, from the P UG regulatory promoter, whereas transcription of the resistance genes is inducible and initiated from the P YG resistance promoter [11].
Cryptic vanG-like operons are common in Clostridium difficile, a major human pathogen which is a target for vancomycin, and a vanU G gene encoding a protein identical to VanU G was found in a clinical isolate (GenBank N°AVLW01000050). A VanU G -like protein (Gen-Bank N°YP002939420), 79% identical with VanU G , was detected in an Eubacterium associated with a two-component system controlling an ABC-type transporter and a protein (GenBank N°YP007781704) with 76% identity was reported in Ruminococcus bromii associated with a CheY related regulator and a partial vanG operon. These regulators have not been studied.
We report the role of VanU G in the transcription of the vanG operon in E.faecalis. We show that VanU G binds to the P UG regulatory and P YG resistance promoters and negatively regulates the vanURS G regulatory and resistance operons. In contrast, VanR G binds only to P YG . It thus appears that, upon induction by vancomycin, the VanS G sensor phosphorylates VanR G which competes and displaces VanU G from P YG leading to transcription of the resistance operon in a dose dependent manner. Thus, rheostatic regulation of resistance gene expression results from binding of a repressor and an activator encoded in a single operon to the same promoter.

Results
VanU G but not VanR G binds to the P UG regulatory promoter Primer extension of the region upstream from vanU G indicated that, irrespective of induction, the transcriptional start site for vanURS G was located 22 bp upstream from the translation initiation codon of vanU G [11]. The P UG promoter consists of -35 and -10 regions corresponding to δ70 recognition sequences separated by 17 bp (Fig 2A). To determine if VanU G and VanR G bind to the P UG regulatory promoter region and to identify putative specific binding sites, DNaseI footprinting experiments were carried out. A radiolabeled PCR probe corresponding to positions -247 to +110 relative to the transcription initiation site of P UG was incubated with increasing amounts of purified VanU G , VanR G , and VanR G phosphorylated (VanR G -P) by acetyl phosphate. The P UG region protected by VanU G depended on the protein concentration, extending from -70 to -20 (positions relative to the transcription initiation site) overlapping the -35 sequence at a low concentration (Fig 2B, lane 6) and from -70 to +10 at higher concentrations ( Fig 2B, lanes 7 and 8). The region (-70 to -20) contained two adjacent imperfect palindromic sequences likely corresponding to the binding motifs of VanU G (Fig 2A). As opposed to the wild-type fragment, two DNA fragments containing double mutations in the imperfect dyad symmetry operator of P UG were not retarded by VanU G , indicating a key role in VanU G binding (S1 Fig). The appearance of several DNase I hypersensitive sites (Fig 2B) corresponding to bending of the DNA duplex suggested binding of two VanU G monomers or dimers. This is consistent with the presence of two inverted repeats in the P UG region (Fig 2A) and with the two-step gel retardation (S1 Fig). In contrast to VanU G , VanR G and VanR G -P did not bind to the P UG promoter.

VanU G acts as a repressor of the P UG regulatory promoter
The vanG operon is part of a large genetic element and is transferable from E. faecalis BM4518 to E. faecalis JH2-2 from chromosome to chromosome [11]. Since clinical isolate BM4518 is not transformable, we studied the VanURS G system in transconjugant BM4522 (JH2-2::vanG) (S1 Table). To determine the role of VanU G on P UG , the vanU G , vanR G , and vanS G genes of BM4522 were inactivated individually by in-frame deletions leading to BM4720(ΔvanU G ), BM4721(ΔvanR G ), and BM4722(ΔvanS G ). Transcription of the regulatory genes was quantified by RT-qPCR. In BM4522, low level transcription occured at similar levels without and with various concentrations of vancomycin indicating that the P UG promoter was not inducible by vancomycin (Fig 2C). In the absence of vanU G , vanR G and vanS G were transcribed in the absence or presence of vancomycin at higher level ( 5-fold) from P UG indicating that VanU G acted as a repressor on this promoter region (Fig 2D). In the absence of vanR G or vanS G , transcription of the regulatory genes remained unchanged even in the presence of vancomycin.
To confirm regulation of P UG by VanU G , the vanURS G genes were cloned into vancomycin susceptible Escherichia coli NR698 [12] under the control of promoter P spank upstream from Regions protected from DNase I cleavage by VanU G are delineated by a bracket. The binding motif is composed of two 14-bp imperfect inverted repeats indicated in orange and purple and by arrows; the complementary bases are underlined. (B) DNase I footprinting analysis of the binding of VanU G to P UG . A 357-bp DNA fragment was amplified from the P UG promoter region using a labeled reverse primer (VanG126) to radiolabel the template strand. Increasing amounts of VanU G , indicated above each lane, were incubated with the DNA probe. The bracket indicates the region protected from DNase I cleavage by VanU G and the co-ordinates of protection relative to the transcriptional start site are indicated on the left. M is the A+G Maxam and Gilbert sequencing reaction lane of the probe used as a size marker and the nucleotide positions are indicated at the right. Transcription of the regulatory genes by RT-qPCR in transconjugant BM4522 (C) and deletant derivatives relative to the same genes of BM4522 (D). The strains are indicated at the bottom. Results are presented in arbitrary units normalized to the rpoB transcripts in the same strain and in BM4522 under similar conditions. Each strain, not induced or induced by vancomycin, was tested in triplicate in two independent experiments. The bars represent the means and the error bars the standard deviations; nd, not detectable. NI, not induced. Vm, vancomycin. VanG-Type Vancomycin Resistance Regulation P UG fused to a chloramphenicol acetyltransferase (CAT) reporter gene, the two promoters being separated by a transcription terminator (term) ( Table 1). Subsequently, each of the three genes was inactivated. E. coli RNA polymerase bound to the P UG promoter (S2A Fig) which was active in the new host, in the presence or in the absence of vancomycin (Table 1). CAT was produced at a maximum level in the absence of vanU G by plasmids pAT952(P spank termP UG cat), pAT966(P spank vanR G termP UG cat), and pAT969(P spank vanRS G termP UG cat) ( Table 1). In contrast, in the presence of VanU G , CAT production was decreased to similar basal levels by plasmids pAT965(P spank vanU G termP UG cat), pAT967(P spank vanUR G termP UG cat), and pAT968 (P spank vanURS G termP UG cat) ( Table 1). These results confirmed that VanU G acts as a strong repressor on the P UG promoter.
The VanR G S G two-component system is functional Transcription of the resistance genes is under the control of VanURS G and, as discussed above, VanU G negatively autoregulates vanURS G transcription from the P UG regulatory promoter. To determine if VanR G and VanS G acted as a two-component system and to study the putative interaction of VanU G with these proteins, VanU G , VanR G , and the cytoplasmic histidine kinase domain of VanS G were purified as C-terminal His-tag proteins (S1 Table). VanS G autophosphorylated in the presence of [γ-32 P]-ATP ( Fig 3A). When incubated with purified VanU G or VanR G , phosphorylated VanS G transferred its phosphate group to VanR G (Fig 3B) but not to VanU G (Fig 3E). Phosphorylation of VanR G was fast and efficient, occurring in less than a minute. To test the phosphatase activity of VanS G , hydrolysis of VanR G -P over time was analysed in the absence or in the presence of VanS G . Purified [ 32 P]-VanR G was stable in vitro for at least 30min and then dephosphorylated slowly ( Fig 3C); addition of purified VanS G increased dephosphorylation only slightly (Fig 3D-3G). These results indicate that VanRS G was functional and had characteristics similar to those of other VanRS-type two-component systems [7,9] and that VanU G did not affect phosphorylation nor dephosphorylation of VanR G and VanS G (Fig 3E and 3F). VanU G and VanR G bind to overlapping sites of the P YG resistance promoter To study the putative binding of VanU G and VanR G to the P YG region and to identify specific binding sites, DNaseI footprinting experiments were carried out. The inducible P YG promoter is composed of -35 (AAAACA) and -10 (TACAAT) regions separated by 16 bp which have similarity with δ70 recognition sequences, although the -35 sequence is not conserved consistent with the fact that the promoter is positively regulated (Fig 4B). Analysis of the P YG region revealed three 12-bp directly repeated VanR G binding motifs and a deduced consensus sequence (T/C)CGTANGAAA(T/A)T was analogous to that in the P R and P H vanA operon promoters [13]. In the P UG region, similar sequences were not found (Fig 2A) which could explain lack of VanR G binding. The radiolabeled probe corresponding to positions -163 to +69 relative to the transcription initiation point of the P YG promoter and containing the three conserved sequences was incubated with increasing amounts of purified VanU G , VanR G , and VanR G -P (Fig 4). The three proteins protected in a concentration-dependent manner an overlapping DNA region that included the three direct repeats. The P YG region protected by VanU G was much larger than that by VanR G and VanR G -P extending from -110 to -3 and overlapped the -35 sequence at 0.2 and 1μM ( Fig 4A, lanes 17 and 18). The P YG region protected by VanR G and VanR G -P extended from -100 to -56 at low concentration ( Fig 4A, bracket I, lanes 3 and 8) and from -100 to -43 at higher concentrations ( Fig 4A, bracket II, lanes 4 and 5, and 9 and 10). There were three binding motifs a, b, and c with different affinities for VanR G and VanR G -P in the P YG promoter region (Fig 4). Only a slight difference in affinity in favor of VanR G -P at 0.2μM was noted for the "a" site ( Fig

VanU G allows rheostatic expression of the resistance genes
To study the consequences of the binding of VanU G and VanR G to overlapping regions of P YG on the expression of the resistance genes, the VanT G serine racemase was used as a reporter ( Fig 5). In clinical isolate BM4518 and transconjugant BM4522, synthesis of the serine racemase was dependent on the concentration of vancomycin ( Fig 5). In contrast, in BM4720(ΔvanU G ), the resistance operon was expressed at its maximum even at low concentrations of vancomycin. These results suggested that VanU G acts as a repressor of P YG and that, in its absence, there is no fine-tuning of resistance expression from this promoter. Thus, modulation of transcription by vancomycin was due to the phosphorylation level of VanR G mediated by VanS G provided that VanU G was present. Surprisingly, as in the wild-type strain, induction was dependent on the concentration of the inducer in BM4721(ΔvanR G ) ( Fig 5). This could be accounted for by the presence of a VanR homolog in the host. In fact, we found, in both E.faecalis BM4518 and transconjugant BM4522 which were entirely sequenced (GenBank N°PRJNA245745), a gene specifying a VanR' G protein with 65% identity with VanR G (S3A Fig). In BM4722(ΔvanS G ) there was no synthesis of VanT G in the presence of vancomycin indicating that VanR G and VanR' G are not phosphorylated in the absence of VanS G . Double mutant BM4723(ΔvanR G , ΔvanR' G ) derived from E. faecalis BM4721(ΔvanR G ) was susceptible to vancomycin (MIC, 1μg/ ml) and VanT G production was no longer inducible by vancomycin, indicating cross-talk between VanS G and VanR' G ( Fig 5). To avoid interference by this regulator, transcription from the P YG promoter was studied in E.coli NR698 since E. coli RNA polymerase was able to bind to this promoter (S2B Fig). The vanURS G , vanRS G , and vanUS G genes were cloned under the control of P spank upstream from the P YG transcriptionally fused to a cat gene generating pAT970 (P spank vanURS G termP YG cat), pAT971 (P spank vanRS G termP YG cat), and pAT972 (P spank vanUS G termP YG cat). In the absence of VanU G , induction by vancomycin led to similar levels of CAT synthesis in the strain harboring pAT971 (P spank vanRS G termP YG cat) whatever the concentration of the inducer, whereas with pAT970 (P spank vanURS G termP YG cat) CAT production depended on the vancomycin concentration (Table 2). These results confirmed that, in the presence of vancomycin, VanU G is required for rheostatic gene transcription from P YG and that VanR G phosphorylation is essential for expression of the resistance genes since, in the absence of this regulator in pAT972 (P spank vanUS G termP YG cat), the level of CAT activity was low, both without (74U±9) and with (104 U ± 13) vancomycin (0.30 μg/ml). In the absence of vancomycin, CAT activity was lower in E. coli producing vanU G encoded by pAT970 (P spank vanURS G-termP YG cat) than in its counterpart harboring pAT971 (P spank vanRS G termP YG cat). This confirms that VanU G acts as a repressor on the P YG resistance promoter ( Table 2). VanU G and VanR G compete for binding to the P YG resistance promoter Since VanU G and VanR G bound at overlapping sites of P YG , to assess a possible effect of VanR G on the binding of VanU G , we performed DNaseI footprinting assays on the labeled P YG probe with purified VanR G and VanU G (Fig 6). Low and medium concentrations (64 nM and 128 nM) of VanU G which allow binding to P YG were tested with increasing concentrations of VanR G . Upon addition of VanR G , the binding profile of VanU G faded while that of VanR G appeared and increased in a dose dependent manner (Fig 6A). In the reverse experiment two approriate concentrations of VanR G were challenged by increasing concentrations of VanU G and the binding of VanR G decreased also in the presence of VanU G (S4 Fig). In summary, VanU G alone did not allow transcription of the resistance genes ( Fig 6B). It thus appears that at a low concentration of vancomycin there was competition between VanU G and VanR G , the latter being partially phosphorylated, transcription of vanY G W G GXY G T G was low. In contrast, at high concentrations of vancomycin, VanR G was efficiently phosphorylated and able to displace VanU G leading to maximal transcription of the resistance genes from the P YG promoter.  The presence of vanU G reduces the fitness cost associated with expression of VanG-type resistance To study the role of VanU G in this sophisticated resistance mechanism, the fitness cost of BM4720(ΔvanU G ) compared with that of BM4522 in monocultures in the absence and in the presence of vancomycin (1 μg/ml) was analysed by determination of the growth rates ( Table 3).
The results showed that the growth rates of both strains were indistinguishable in the absence of vancomycin indicating that non induced VanG-type resistance is not costly for the host. In contrast, in the presence of vancomycin, the relative growth rate of BM4720(ΔvanU G ) (0.74) was significantly reduced when compared with that of BM4522 (0.93) indicating that increased expression of resistance was significantly more costly in the absence of vanU G .

Discussion
Among the ubiquitous two-component regulators, VanR/VanS-type systems are one of the rare to control expression of genes mediating antibiotic resistance [3]. In the VanG-type strains, a membrane associated sensor kinase (VanS G ) which detects a signal associated with the presence of vancomycin in the environment and a cytoplasmic response regulator (VanR G ) that acts as a transcriptional activator are also present (Fig 1) and functional (Fig 3) but there is, in addition, a VanU G transcriptional regulator (Fig 1).
In the two main VanA-and VanB-type systems, the regulatory genes (vanRS) and the resistance genes are transcribed from independent and coordinately regulated promoters, but VanR is the only known direct regulator of the resistance genes [3,8,13]. In VanG-type strains, cotranscription of vanURS G is repressed from P UG by VanU G (Fig 2 and Table 1) and expression of the resistance genes from P YG is activated by VanR G and repressed by VanU G (Fig 5 and Table 2). Thus, VanU G regulates the resistance genes both directly, by binding to the P YG promoter region (Fig 4), and indirectly by repressing synthesis of VanR G S G (Fig 5). Like other members of the XRE protein family (S3B Fig) [14-16], VanU G binds to short repeated sequences which span the promoters (Fig 2A and 2B). Unlike the VanR and VanR B proteins which bind to their own promoters [8,13], VanR G does not regulate its own expression (Fig 2). No sequences similar to the VanR G consensus binding site are found in P UG (Figs 2 and 4).
VanR G , as VanR and VanR B , belongs to the OmpR-PhoB subclass of response regulators that have the peculiarity to bind to their target promoters in the unphosphorylated or phosphorylated form [8,13,17,18]. Phosphorylation of VanR and VanR B enhances the affinity of the proteins for their respective regulatory P R or P RB and resistance P H or P YB promoter regions allowing increased transcription of the regulatory and resistance genes [8,13]. In VanA-type strains, VanR and VanR-P bind to P R and P H regions which contain a single or two 12-bp conserved sites, respectively [13]. Comparison of the sequences of the P UG and P YG regions with the 12-bp consensus sequence spanned by VanR and VanR-P revealed three binding sites in the P YG region with a consensus sequence (Fig 4B) similar to that in VanA-type resistance [13]. As for the regulatory P R and resistance P H promoters, the positioning of these sites in P YG was upstream from the -35 motif. VanU G , VanR G , and VanR G -P protected overlapping regions, the two latter binding to P YG a and b sites with a higher affinity than to the c site (Fig 4). There are only two sites in the P H promoter but VanR generated a more extensive footprint (80 bp for P H ) than VanR G (42bp for P YG ) likely due to higher cooperativity of VanR. Although not essential for binding in vitro, phosphorylation of VanR G increased its affinity for the P YG resistance promoter (Fig 4). In the P UG promoter region no sequences similar to the consensus VanG-Type Vancomycin Resistance Regulation were found (Fig 2A) which could explain the absence of binding of VanR G and low-level transcription from the regulatory promoter. In many instances, regulation of gene transcription in E.coli occurs essentially through control of the phosphatase activity of the sensor [19,20]. In VanA-and VanB-type strains, the level of phosphorylation of VanR and VanR B is modulated by the kinase and phosphatase activities of the VanS and VanS B sensors [7,10,21]. Phosphatase activity is critical for response regulators, such as VanR and VanR B , whose phosphorylated form is highly stable, to ensure that the protein is not permanently activated. In VanG-type strains, in the absence of VanU G , induction by vancomycin led to maximal VanT G serine racemase (Fig 5) or CAT synthesis ( Table 2) even at low concentrations of the inducer. Since in the absence of VanU G there was no modulation of resistance genes transcription from the P YG promoter, this suggests that a low amount of VanR G -P is sufficient to induce the resistance operon. VanU G did not modulate VanR G and VanS G phosphorylation ( Fig 4F) and was not phosphorylated by VanS G (Fig 4E). Surprisingly, at least in vitro, the phosphatase activity of VanS G was not very efficient (Fig 4D) in comparison with those of VanS or VanS B [7,9]. Expression of VanG-type resistance was thus inducible by vancomycin due to the presence of VanU G as opposed to direct modulation of VanR activity by VanS in the other van operons. In the absence of vancomycin only VanU G bound to the P YG promoter; however when the concentration of vancomycin increased, VanR G being more efficiently phosphorylated by VanS G , displaced progressively VanU G allowing gradual transcription of the resistance genes (Fig 6) as it is likely the case with VanR' G , the VanR G homolog encoded elsewhere in the chromosome. In B. subtilis, when both repressors SinR and SlrR are bound to the degU promoter, they can be displaced by the response regulator DegU leading to activation of the degU gene [22]. Also in B. subtilis, CcpC activates aconitase gene citB expression whereas CodY binds to its promoter and represses citB transcription [23]; PutR which is an activator essential for transcription of the putBCP operon for proline utilization is displaced by the CodY repressor [24].
VanU G does not possess the characteristics of auxiliary regulators which can interact with histidine kinases, influencing signal perception and transduction. Nor does it interact with the response regulator to alter its phosphorylation status or its DNA binding ability, the recruitement of RNA polymerase on the promoter, or to sequester it through protein:protein interaction [25,26]. The results presented here show that competition between the VanU G repressor and the VanR G activator for binding to the P YG promoter may be responsible for the complex regulation of the resistance genes (Fig 6). This is an unusual example of rheostatic regulation of gene transcription due to binding competition between two regulators encoded in the same operon. It also elucidates an unsuspected strategy by which enterococcal clinical isolates regulate transcription of acquired genes for vancomycin resistance.
In previous work, we showed in VanB-type resistance that, despite the complex dual biochemical mechanism of resistance to vancomycin, its biological cost in enterococci is negligible when non induced, whereas a significant fitness reduction is observed when resistance is expressed in the presence of the inducer, the antibiotic itself [27]. Thus resistance is expressed exclusively when needed for bacterial survival. In VanG-type strains, tight regulation of resistance expression involves VanU G which can thus be considered as a compensatory component, drastically reducing the biological cost associated with vancomycin resistance in the presence of antibiotic.

Bacterial strains, plasmids, and growth conditions
The origin and properties of the strains and plasmids are described in S1 Table. Escherichia coli TOP10 (Invitrogen, Groningen, The Netherlands) and NR698 (susceptible to vancomycin) [12] were used as a host for recombinant plasmids. E. coli BL21λDE3 [28], in which the T7 RNA polymerase gene is under the control of the inducible lacUV5 promoter carries the pREP4 plasmid allowing co-expression of the GroESL chaperonin to optimize recombinant protein solubility [29]. E. coli TG1 RepA [30] was used as a host for constructions in the pAT944(pGhost9Ocat) vector (S1 Table). Kanamycin (50μg/mL) was used as a selective agent for cloning PCR products in the pCR-Blunt vector (Invitrogen). Ampicillin was used to select pUC1813 [31]. pDR111 (gift from David Rudner, Harvard University), which harbors the P spank promoter between two fragments of the B.subtilis amyE gene, is a derivative of the P spac-hy plasmid pJQ43 containing an additional lacO binding site to achieve a better repression in the absence of the IPTG inducer. P spank is a lacI repressible IPTG inducible-promoter for gene overexpression. Spectinomycin (60μg/mL) and chloramphenicol (10μg/mL) were added to the medium to prevent loss of plasmids derived from pDR111(P spank ) and pAT944(pGhost9Ocat), respectively. Enterococcus faecalis JH2-2 is a derivative of strain JH2 that is resistant to fusidic acid and rifampin [32]. In all experiments, strains were grown in brain heart infusion (BHI) at 37°C with shaking at 110 rpm.

Promoter DNA labeling
Labeled P UG (357 bp) and P YG (233 bp) fragments were generated by PCR with BM4518 total DNA as a template and primer pairs VanG12-VanG126 and VanSG6-YG10 (S2 Table), respectively, using a combination of an unlabeled primer with an end-labeled primer (625nM) with T4 polynucleotide kinase (0.075 U/μl) (New England Biolabs) and [γ 32 P]-ATP (3000 Ci/mmol) (Perkin Elmer). The PCR reactions were carried out in a 50-μl volume and the products purified as described [8].

Gel shift assay
Purified labeled PCR products corresponding to wild-type and mutated P UG promoter region fragments were recovered from a 6% polyacrylamide gel and used as a probe for the gel shift assay after addition of 100 μl of ammonium acetate (0.5 M) diluted in Tris buffer (10 mM, pH8.5) overnight at 37°C. The P UG and mutated P UG probes (10,000cpm each) were incubated with various concentrations of purified VanU G regulator at 30°C for 20min in 20 μl of 50mM Tris-HCl (pH7.8) containing 20 mM MgCl2 and 0.1 mM dithiothreitol (DTT). After addition of the DNA dye solution (40% glycerol, 0.025% bromophenol blue and 0.025 xylene cyanol), the mixture was loaded on a 7.5% polyacrylamide gel in the absence of protein denaturants. The gels were dried and analysed by autoradiography.

DNase I footprinting
Complexes with the labeled promoter regions (5nM) were formed for 30 min at 30°C in 15 μl of buffer C (20 mM Hepes pH 8.0, 5 mM MgCl2, 50 mM potassium glutamate, 5 mM DTT, and 500μg/ml bovine serum albumin) using RNA polymerase of E. coli at 50 nM or VanU G , VanR G , or VanR G -P at increasing concentrations. For DNase I experiments, 1.5 μl of DNase I solution (1 μg ml -1 in 10 mM Tris-HCl, 10 mM MgCl2, 10 mM CaCl 2 , 125 mM KCl) were added and incubated at 30°C for 10s when the labeled promoter regions were alone, or for 20 s when when RNA polymerase or VanU G , VanR G or VanR G -P were present in the mixture. The reaction was stopped and all the samples were extracted, precipitated, washed, resuspended, and loaded on a sequencing gel as described [8]. Protected bands were identified by comparing the migration with that of the same fragment treated for the A+G sequencing reaction [33]. The gels were analysed by autoradiography.

Quantitative real-time RT-qPCR
Enterococci grown in 100 ml of brain heart infusion in 250-ml bottles, with and without vancomycin, at 37°C with shaking at 110 rpm to OD 600 = 0.8 were harvested. RNA was prepared using the Fast RNA ProBlue kit (MBP Biomedicals) according to the manufacturer's protocol, treated with DNase (Turbo DNA-free, Invitrogen), and checked for the absence of contaminant DNA in a standard PCR, using the same primers as for the RT-PCR. RNA concentrations were determined by measuring absorbance with a NanoDrop2000 (ThermoScientific). cDNA synthesis and RT-qPCR were performed with a Light Cycler RNA amplification kit SYBR greenI (Roche Diagnostic GmbH) in a total reaction volume of 19μl with 0.5 μM gene-specific primers (VanG129-VanG102 for vanU G , VanRG2-VanRG10 for vanR G , VanSG2-VanSG10 for vanS G , and rpoB5-rpoB12 for rpoB) (S2 Table) according to the manufacturer's instructions. Amplification and detection of specific products were performed using the LightCycler sequence detection system (Roche) with the following cycle profile: 1cycle at 55°C for 20 min for the reverse transcription step, followed by 1 cycle at 95°C for 30 s, 45 cycles at 95°C for 5 s, 52°C for 15 s, and 72°C for 15 s. The level of every gene transcript was normalized relative to rpoB transcript levels.

Autophosphorylation of VanS G
Autophosphorylation of VanS G (40 μg) was performed in a final volume of 100 μl of buffer A (final concentrations: 50 mM Tris-HCl, 50mM KCl and 1 mM MgCl2, pH7.5). The reaction was initiated by the addition of 5 μl of ATP (1mM final) containing 200 μCi of [γ-32P]ATP and incubated at room temperature for 1 h. ATP was removed using 500 μl Sephadex G-50 spin column equilibrated with buffer A. The reaction was quenched by the addition of 5 μl of β-mercaptoethanol-stop solution (Sigma), followed by electrophoresis on 12% NuPAGE Novex Bis-Tris gels (Invitrogen) in MOPS buffer (1X), and autoradiography.

Phosphorylation of VanU G and VanR G by VanS G
Phosphotransfer to purified VanU G and VanR G were carried out in buffer A. The reaction was initiated by the addition of 10 μl of the purified autophosphorylation reaction mixture of VanS G (40 μg) described above to a 15 μl reaction mixture containing VanU G or VanR G (55 μg each). After incubation for various periods of times at room temperature, the phosphotransfer reactions were quenched by the addition of stop solution (Sigma) followed by electrophoresis on 12% NuPAGE Novex Bis-Tris gels (Invitrogen) in MOPS buffer (1X) and autoradiography.

Hydrolysis of phospho-VanU G and phospho-VanR G by VanS G
The VanU G (220 μg) and VanR G (225 μg) response regulators were labelled with acetyl[ 32 P] phosphate for 1 h at room temperature as described above, and 52 μg of VanS G histidine kinase were added, and incubation was pursued for various periods of times. Aliquots (10 μl) were withdrawn at designated time points and the reactions were stopped, followed by electrophoresis on 15% SDS-polyacrylamide gels and autoradiography.

Plasmid construction
The plasmids were constructed as follows.
Construction of pAT940, pAT941 and pAT942. pAT940(pET28ΩvanU G ) and pAT941 (pET28ΩvanR G ). A 225-bp BsaI-XhoI fragment corresponding to the vanU G coding sequence amplified with UG1 and UG2 (S2 Table) and a 705-bp BsaI-XhoI fragment corresponding to the vanR G coding sequence amplified by using oligonucleotides RG1 and RG2 (S2 Table) and BM4518 [11] total DNA as a template, were cloned in the NcoI and XhoI sites of modified pET28 [35] to generate plasmids pAT940(pET28ΩvanU G ) and pAT941(pET28ΩvanR G ). Oligodeoxynucleotide UG1 contained a BsaI restriction site designed to generate a cohesive end compatible with NcoI and 16 bases complementary to codons 1-6 of vanU G of BM4518 (S2 Table). Oligodeoxynucleotide UG2 contained a XhoI site replacing the TGA stop codon and 21 bases complementary to codons 69-75 of vanU G . Oligodeoxynucleotide RG1 contained a BsaI restriction site designed to generate a cohesive end compatible with NcoI and 16 bases complementary to codons 1-6 of vanR G of BM4518. Oligodeoxynucleotide RG2 contained a XhoI site replacing the TGA stop codon and 21 bases complementary to codons 229-235 of vanR G .
pAT942(pET28ΩvanS G ). A cytoplasmic portion of the vanS G gene of strain BM4518 was amplified using BM4518 total DNA as a template and primer pair SG1-SG3 (S2 Table). Oligodeoxynucleotide SG1 contained a BsaI restriction site designed to generate a cohesive end compatible with NcoI, and 16 bases complementary to codons 88-93 of vanS G . Oligodeoxynucleotide SG3 contained a XhoI site in place of the TAG stop codon and 21 bases complementary to codons 361-367 of vanS G . The 842-bp pCR product from vanS G was digested by BsaI and XhoI and cloned between the NcoI and XhoI restriction sites of plasmid pET28 to generate plasmid pAT942(pET28ΩvanS G ).
Construction of pAT944(pGhost9Ocat). The XbaI cassette containing the chloramphenicol acetyltransferase cat gene with its own promoter was amplified from DNA of plasmid pAT943(pUC1318OPcat) with primers pG9CAT NH2 and pG9CAT COOH (S2 Table) which contain a XbaI restriction site allowing the replacement of the XbaI fragment containing the erythromycin resistance gene in pGhost9 [36] to generate plasmid pAT944(pGhost9Ocat).
Construction of pAT945(pGhost9CmOΔvanU G ), pAT946(pGhost9CmOΔvanR G ), pAT947(pGhost9CmOΔvanS G, ), and pAT973(pGhost9CmOΔvanR' G ). The vanU G , vanR G , and vanS G genes of the vanG operon and vanR' G from BM4518 were inactivated by deletion using splicing-by-overlap extension PCR in two steps and cloned into the thermosensitive shuttle plasmid pAT944(pGhost9Ocat) using XhoI and PstI restriction sites to generate plasmids pAT945(pGhost9CmOΔvanU G ), pAT946(pGhost9CmOΔvanR G ), pAT947(pGhost9C-mOΔvanS G ), and pAT973(pGhost9CmOΔvanR' G ). The primers used for the construction of the deletant alleles and the extent of the deletions are reported in S2 Table. A SmaI restriction site was added in the primers to screen for integration in the corresponding chromosomal gene. Briefly, the remnants of the vanU G , vanR G , vanS G and vanR' G genes of BM4518 were first amplified from total DNA of BM4518 as a template using primers UG3-UG4 and UG5-UG6 for ΔvanU G , UG3-RG4 and RG5-RG7 for ΔvanR G , SG4-SG5 and SG6-SG7 for ΔvanS G , RG10-RG11 and RG12-RG13 for ΔvanR' G and, in a second step, the resulting PCR products were amplified with UG3 plus UG6, UG3 plus RG7, SG4 plus SG7, and RG10 plus RG13 respectively, to obtain ΔvanU G , ΔvanR G , ΔvanS G and ΔvanR' G .
Construction of pAT949 and derivatives. Plasmid pAT949(pDR111OP spank cat) was constructed by cloning the HindIII-SphI fragment of pAT948(pUC1813Ocat) carrying the cat cassette in pDR111(P spank ) digested with the same enzymes allowing a directional cloning of the cat reporter gene under the control of the inducible P spank promoter.
pAT950 (pDR111OP spank termcat). A 66-bp HindIII-SalI fragment corresponding to the transcription terminator of gene 32 from bacteriophage T4 [37] was amplified by PCR with oligodeoxynucleotides T4F-HindIII and T4R-SalI/NheI (S2 Table). Primer T4F-HindIII contained HindIII and NheI restriction sites. Primer T4R-SalI/NheI contained SalI and NheI restriction sites. The HindIII and SalI restriction sites allowed directional cloning of the transcription terminator (term) from bacteriophage T4 under the control of the inducible P spank promoter and upstream from the cat reporter gene of the pAT949(pDR111OP spank cat) shuttle vector.
pAT951(pDR111OP spank vanU G cat). The vanU G gene of BM4518 was amplified using primer pair UG NH2 and UG COOH (S2 Table) and total DNA of the corresponding strain as a template. Oligodeoxynucleotide UG NH2 contained BsaI and HindIII restriction sites, a RBS, and 6 bases complementary to vanU G including the ATG (translation initiation) codon. Oligodeoxynucleotide UG COOH harbored SalI and NheI restriction sites, the stop codon, and 15 bases complementary to the 3' end sequence of vanU G from BM4518. The BsaI and SalI restriction sites allowed directional cloning of a 249-bp fragment of vanU G downstream from the inducible P spank promoter and upstream from the cat gene of the pAT949(pDR111OP spank cat) shuttle vector to generate pAT951(pDR111OP spank vanU G cat).
pAT952(pDR111OP spank termP UG cat) and pAT953(pDR111OP spank vanU G P UG cat). The regulatory P UG (183 bp) promoter was amplified by PCR from BM4518 total DNA with oligodeoxynucleotides PUG1 and PUG2 (S2 Table). Primers PUG1 and PUG2 contained a NheI and a SalI restriction site, respectively, which allowed directional cloning of P UG upstream from the cat gene of pAT950(pDR111OP spank termcat) to generate pAT952(pDR111OP spank termP UG cat) or allowed directional cloning of P UG downstream from vanU G and upstream from the cat reporter gene of pAT951(pDR111OP spank vanU G cat) to generate pAT953 (pDR111OP spank vanU G P UG cat).
pAT954(pDR111OP spank vanR G P UG cat). A 754-bp HindIII-NheI fragment corresponding to the vanR G coding sequence with its RBS, initiation and stop codons was amplified by PCR from BM4518 with oligodeoxynucleotides RG NH2 and RG COOH (S2 Table). Primer RG NH2 contained a HindIII restriction site. Primer RG COOH comprised SalI and NheI restriction sites, the stop codon, and 14 bases complementary to the 3' end of vanR G from BM4518. The HindIII and NheI restriction sites allowed directional cloning of the vanR G gene under the control of the inducible P spank promoter and upstream from P UG and the cat gene of pAT952 (pDR111OP spank termP UG cat).
pAT956(pDR111OP spank vanUR G P UG cat), pAT958(pDR111OP spank vanRS G P UG cat), pAT960(pDR111OP spank vanURS G P UG cat) pAT961(pDR111OP spank vanRS G P YG cat)and pAT962(pDR111OP spank vanURS G P YG cat). The vanUR G , vanRS G , and vanURS G genes of BM4518 were amplified using primer pairs UG NH2 -RG COOH , RG NH2 -SG COOH , and UG NH2 -SG COOH (S2 Table), respectively, and BM4518 total DNA as a template. Oligodeoxynucleotides UG NH2 and RG NH2 harbored a HindIII restriction site and 21 bases complementary to the sequence upstream from vanU G or 17 bases complementary to the sequence upstream from vanR G . Primers RG COOH and SG COOH contained each SalI and NheI restriction sites, the stop codon and 14 or 13 bases complementary to the 3' end of respectively vanR G and vanS G of BM4518. The HindIII and SalI restriction sites allowed directional cloning of vanUR G , vanRS G , and vanURS G upstream from the cat reporter gene of shuttle vector pAT949(pDR111OP spank cat) carrying the inducible P spank promoter to generate pAT955 (pDR111OP spank vanUR G cat), pAT957(pDR111OP spank vanRS G cat), and pAT959 (pDR111OP spank vanURS G cat). The 183-bp NheI-SalI fragment carrying the P UG promoter obtained above by amplification was cloned in pAT955(pDR111OP spank vanUR G cat), pAT957 (pDR111OP spank vanRS G cat), and pAT959(pDR111OP spank vanURS G cat) digested with the same enzymes to generate pAT956(pDR111OP spank vanUR G P UG cat), pAT958 (pDR111OP spank vanRS G P UG cat), and pAT960(pDR111OP spank vanURS G P UG cat). The 177-bp NheI-SalI fragment carrying the P YG resistance promoter amplified by PCR from BM4518 DNA with primers PYG1 and PYG2 (S2 Table) was cloned in pAT957 (pDR111OP spank vanRS G cat), and pAT959(pDR111OP spank vanURS G cat) digested with the same enzymes to generate, respectively, pAT961(pDR111OP spank vanRS G P YG cat)and pAT962(pDR111OP spank vanURS G P YG cat).
pAT972(pDR111OP spank vanUS G termcat). The 1,144-bp fragment containing the vanS G gene of BM4518 was amplified using primer pair SG NH2 -SG COOH (S2 Table) and total DNA of the corresponding strain as a template. The NheI and SalI restriction sites allowed directional cloning of vanS G downstream from the vanU G gene and upstream from the cat gene of pAT951(pDR111OP spank vanU G cat) to generate pAT963(pDR111OP spank vanUS G cat).
The EcoRI fragment harboring the vanUS G ' genes from pAT963(pDR111OP spank vanUS G cat) was replaced by the EcoRI fragment carrying the vanRS G ' genes of pAT971(pDR111OP spank van-RS G termP YG cat) to generate pAT972(pDR111OP spank vanUS G termcat).
In Gram-positive bacteria, pGhost9 [36] which replicates at 28°C but is lost above 37°C, allowed construction of E.faecalis BM4522 derivatives by insertional inactivation. Plasmids pAT945(pGhost9CmOΔvanU G ), pAT946(pGhost9CmOΔvanR G ), and pAT947(pGhost9C-mOΔvanS G ) were electrotransformed into E. faecalis BM4522 [11] to generate, respectively, BM4720(ΔvanU G ), BM4721(ΔvanR G ), and BM4722(ΔvanS G ) (S1 Table). Plasmid pAT973 (pGhost9CmOΔvanR' G ) was electrotransformed into E. faecalis BM4721(ΔvanR G ) to generate the double mutant BM4723(ΔvanR G , ΔvanR' G ). Transformants were selected at the permissive temperature (28°C) on M17 plates containing 10g/ml of chloramphenicol and 0.5% glucose. A colony of each transformant was inoculated into 50 ml of M17 broth containing 0.5% glucose and incubated for 2h at 28°C. The culture was then shifted to a non-permissive temperature (42°C) for 2 h and integrants, following a first recombination event, were selected at 42°C on M17 agar containing chloramphenicol (10g/ml). Plasmid excision, by a second recombination event, was favored by subculturing at 28°C in the absence of chloramphenicol and plasmid loss was screened for by plating at 42°C on M17-glucose followed by replica plating on chloramphenicol. The integration locus was determined by PCR following digestion with SmaI and sequencing.

Enzyme assays
For preparation of extracts, 8 ml of an overnight culture were added to 100 ml of broth in the absence or in the presence of vancomycin and strains were grown until OD 600 = 0.8 in 250 ml bottles with shaking at 110 rpm. The cells were harvested by centrifugation, washed in 0.1M phosphate buffer pH 7.0, resuspended in the same buffer, lysed by sonication, followed by centrifugation at 10,000 g during 45 min. The resuspended pellet for VanT G racemase [11] and supernatant for CAT activity, were assayed as described [38].

Genome sequencing, assemblies and annotation
Total DNA from BM4518 and BM4522 strains was purified and sequencing library preparation was carried out using the Nextera DNA Sample Preparation kit (Illumina, San Diego, CA), according to manufacturer's specifications. Quality and quantity of each sample library was measured on an Agilent Technologies 2100 Bioanalyzer (Santa Clara, CA). Libraries were normalized to 2nM, multiplexed and subjected to 250-bp paired end sequencing (Illumina MiSeq). On average, 5 million high-quality paired-end reads were collected for each strain, representing >220-fold coverage of the~2.9 Mb genomes. Reads were assembled de novo utilizing CLC Genomics Workbench (CLC bio, Cambridge, MA). Functional annotations were performed using a custom pipeline as described previously [39].

Determination of growth rates
Growth rates were determined in microplates coupled to a spectrophotometer iEMS reader (Labsystems). Strains were grown overnight at 37°C without or with 1 μg/ml of vancomycin. The cultures were diluted at OD 0.15 into 10 ml of broth without or with vancomycin (1μg/ml) and grown at 37°C with shaking until the beginning of the stationary phase. The cultures were diluted 1/1,000 to inoculate 10 5 bacteria into 200 μl of broth in a 96-well microplate that was incubated overnight at 37°C with shaking. Absorbance was measured at 600 nm every 3 min. Each culture was replicated three times in the same microplate. Growth rates performed in three independent experiments were determined at the beginning of the exponential phase and the relative growth rates were calculated as the ratio of the growth rate of the strain induced by vancomycin versus that of the non induced strain.  Table) leading to the WT and corresponding mutated (mutant 1) promoter region, respectively. A DNA fragment (293 bp) was obtained with labeled VanG12 plus PUG4 and labeled VanG12 plus mutated PUG6 primers (S2Table) leading to the WT and corresponding mutated (mutant 2) promoter region, respectively. Numbering relative to the transcription start site is indicated above the sequences. Only bases differing from the WT sequence are shown in the mutated fragments. (B) Gel shift analysis. The labeled fragments corresponding to the WT and mutated (mutant 1 and mutant 2) promoter regions were incubated in the absence or in the presence of decreasing concentrations of purified VanU G protein indicated above the lanes.  Table) to radiolabel the template strand and the DNA probe was incubated without and with δ70 RNA polymerase at 50 nM. (B) A 233-bp DNA fragment was amplified from the P YG promoter region using a labeled reverse primer (YG10) (S2 Table) to radiolabel the template strand and the DNA probe was analysed similarly. The brackets indicate the regions protected from DNase I cleavage by δ70 RNA polymerase, and the co-ordinates of protection relative to the transcriptional start site are indicated on the right. M is the A+G Maxam  A 233-bp DNA fragment was amplified from the P YG region using a labeled reverse primer (YG10) (S2 Table) to radiolabel the template strand. Increasing amounts of VanU G and two fixed amounts of VanR G indicated at the top were incubated with the DNA probe. The bracket indicates the region protected from DNase I cleavage by VanR G and/or VanU G and the co-ordinates of protection relative to the transcriptional start site are indicated on the left. M is the A+G Maxam and Gilbert sequencing reaction lane of the probes used as a size marker and the nucleotide positions are indicated at the right. (TIF) S1