The autoregulatory role of EsaR, a quorum-sensing regulator in Pantoea stewartii ssp. stewartii: evidence for a repressor function


  • Timothy D. Minogue,

    1. Department of Plant Science University of Connecticut, 302B AG Biotech, Storrs, CT 06269, USA.
    Search for more papers by this author
    • *For correspondence. E-mail; Tel. (+1) 860 486 4408; Fax (+1) 860 486 0534. Present addresses:

  • Markus Wehland-von Trebra,

    1. Institut für Kristallographie, Freie Universität Berlin, D-14195 Berlin, Germany.
    Search for more papers by this author
    • Institut für Klinische Pharmakologie und Toxikologie, Berlin, Germany.

  • Frank Bernhard,

    1. Institut für Kristallographie, Freie Universität Berlin, D-14195 Berlin, Germany.
    Search for more papers by this author
    • Universität Frankfurt, Biozentrum, Institut für Biophysikalische Chemie, Marie-Curie-Str. 9, D-69439 Frankfurt/Main, Germany.

  • Susanne B. von Bodman

    1. Department of Plant Science University of Connecticut, 302B AG Biotech, Storrs, CT 06269, USA.
    2. Department of Molecular and Cell Biology, University of Connecticut, Storrs, CT 06269, USA.
    Search for more papers by this author


Capsular polysaccharide synthesis and virulence in the plant pathogenic bacterium Pantoea stewartii ssp. stewartii requires the quorum-sensing regulatory proteins, EsaR and EsaI, and the diffusible inducer N-(3-oxo-hexanoyl)-L-homoserine lactone. Prior mu-tational studies suggested that EsaR might function as a repressor of quorum sensing in the control of capsular polysaccharide synthesis. Further, a lux box-like palindromic sequence coinciding with the putative –10 element of the esaR promoter suggested a possible negative autoregulatory role for EsaR. This report presents genetic evidence that EsaR represses the esaR gene under inducer-limiting conditions, and that addition of inducer promotes rapid, dose-dependent derepression. DNA mobility-shift assays and analyses by surface plasmon resonance refractometry show that EsaR binds target DNAs in a ligand-free state, and that inducer alters the binding characteristics of EsaR. Physical measurements indicate that the EsaR protein binds N-(3-oxo-hexanoyl)-L-homoserine lactone, in a 1:1 protein:ligand ratio, and that inducer binding enhances the thermal stability of the EsaR protein. These combined genetic and biochemical data establish that EsaR regulates its own expression by signal-independent repression and signal-dependent derepression. Additionally, we provide evidence that EsaR does not govern the expression of the linked esaI gene, thus EsaR has no role in controlling coinducer synthesis.


Bacteria express selected gene systems in a population-dependent manner by sensing self-produced, membrane-diffusible signals in a strategy called quorum sensing (QS) (Fuqua et al., 1996). The key elements of QS regulation in many Gram-negative bacteria are homologue proteins of LuxI, a N-acyl-homoserinelactone (AHL) signal synthase, and LuxR, an AHL-dependent response regulator. These two proteins control the expression of bioluminescence in the marine bacterium, Vibrio fischeri (Fuqua et al., 1994; Williams et al., 2000; Fuqua et al., 2001; Miller and Bassler, 2001; Withers et al., 2001). Alternative QS mechanisms, mediated by unrelated control factors, exist in other Gram-negative bacteria, most notably Vibrio harveyi (Bassler et al., 1994), and in several Gram-positive organisms (Dunny and Leonard, 1997; Kleerebezem and Quadri, 2001). In general, QS governs the control of diverse phenotypes, each benefiting a bacterium in a specialized habitat (Whiteley et al., 1999; Pierson, 2000; Whitehead et al., 2001).

Pantoea stewartii ssp. stewartii (P. stewartii) is the causative agent of Stewart’s wilt disease in sweetcorn and leaf blight in maize. Disease symptoms develop when the bacterium produces large amounts of a capsular polysaccharide (CPS), which blocks the corn xylem vessels and induces necrotic lesions (Coplin et al., 1992). CPS synthesis is a QS-controlled phenotype governed by the LuxI and LuxR homologue proteins, EsaI and EsaR (von Bodman and Farrand, 1995). Disruption of the signal synthase gene, esaI, leads to parallel loss of AHL, CPS production, and virulence. In contrast, mutations in the esaR gene give a hypermucoid phenotype irrespective of AHL (von Bodman et al., 1998). The simplest explanation for these observations is that EsaR functions as a repressor of CPS synthesis and that derepression requires inducing levels of AHL. The functions required for CPS synthesis are encoded by an extensive cps gene system (Dolph et al., 1988). This gene system is closely related to the wza gene cluster encoding the synthesis of the group I capsules, colanic acid in Escherichia coli (Reeves et al., 1996), and amylovoran in Erwinia amylovora (Bernhard et al., 1993). It is well established that these gene systems require transcriptional activation by the environmental response regulatory complex, RcsA/B (Gottesman and Stout, 1991; Kelm et al., 1997; Wehland et al., 1999). The difference in P. stewartii is that the level of RcsA/B-mediated control of stewartan synthesis is secondary to the dominant control by EsaR.

The controlled expression of the QS regulatory elements themselves is generally an integral feature of QS regulation (Shadel and Baldwin, 1992; Fuqua et al., 1994; Pesci et al., 1997; Whitehead et al., 2001). Thus, the strategy by which EsaR governs its own expression, and the expression of the linked esaI gene, may be indicative of its role as a repressor of CPS synthesis. Preliminary observations suggested that EsaR regulates its own expression by repression (von Bodman and Farrand, 1995), although the role of the AHL signal in this mode of regulation was unclear. The current study focuses on a well conserved lux box-like sequence, the esaR box associated with the esaR gene, to explore whether EsaR protein is genetically and biochemically programmed for a role as a repressor of QS.


Autorepression of the esaR promoter by EsaR

LuxR, TraR and related QS transcription factors regulate their own expression by activation (Shadel and Baldwin, 1992; Hwang et al., 1994; Seed et al., 1995; Fuqua et al., 1996; Fuqua and Winans, 1996; Fuqua et al., 2001). The linked esaI/esaR gene system of P. stewartii is convergently organized and features 3′ terminal ends that overlap by 21 basepairs (bp) (von Bodman and Farrand, 1995). The promoter of the esaR gene, not the esaI gene, bears a defined esaR box. This palindrome coincides with the putative –10 promoter sequence, which suggests that binding of EsaR at this site may block transcription and provide a mechanism for EsaR-mediated autorepression. To test this prediction, we developed an in vivo assay based on the coexpression of plasmids pTDM6 and pTDM7 in the E. coli Top10 host strain, TM67 (Table 1). In this assay, plasmid pTDM6 (Fig. 1A) contributes the esaR coding sequence expressed from the E. coli PBAD promoter, which is controlled by the AraC regulator, also encoded on the plasmid, as a function of L-arabinose induction (Guzman et al., 1995). Plasmid pTDM7 carries an esaR::lacZY reporter gene fusion designed to mea-sure the in vivo activity of the esaR promoter (Fig. 1A). Growth of strain TM67 in glucose-supplemented medium lacking L-arabinose yielded fully induced levels of β-galactosidase, whereas growth in the presence of 0.02%L-arabinose gave nearly 10-fold lower levels of β-galactosidase (Fig. 1B). Intermediary levels of reporter activity were inversely proportional to the amount of L-arabinose provided. These data confirm that EsaR acts as a repressor, and that this repression is AHL-independent.

Table 1. Strains and plasmids.
 Relevant genotypeReference or source
 TB1 E. coli, JM83 hsdR (rk-mk+), ΔaraNew England Biolabs
 Top10 E. coli, lacZΔM15 ΔlacX74 araD139 Δ (ara-leu) 7697Invitrogen
 DH5α E. coli, lacZΔM15 Δ (lacZYA-argF)U169 gyrA96Invitrogen
 TM6 E. coli Top10 (pTDM6)This study
 TM67 E. coli Top10 (pTDM6, pTDM7)This study
 TDM619 E. coli Top10 (pTDM6, pTDM19)This study
 DC283 P. stewartii, SS104, NalR Dolph et al. (1988)
 ESN51 P. stewartii, esaI::Tn5 von Bodman and Farrand (1995)
 ESΔIRP. stewartii, ΔesaI-esaR von Bodman and Farrand (1995)
 pBluescript-II KS+Cloning vector, ColE1 ori, ApRStratagene
 pBAD22Arabinose inducible expression vector, ApR Guzman et al. (1995)
 pBBR1MCS-3Broad range expression vector, TcR Kovach et al. (1995)
 pSVB5-18 esaI and esaR cloned pBluescript-II KS+, ApR von Bodman and Farrand (1995)
 pLKC480Source of lacZY-KmR cassette Tiedeman and Smith, 1988)
 pLKC481Source of lacZY-KmR cassette Tiedeman and Smith (1988)
 pSVB60 esaR re-cloned from pSVB5-18 into pBBR1MCS-3 as a
PstI–SalI fragment
This study
 pTDM6PCR-amplified esaR coding sequence cloned into pBAD22as
a NcoI–HindIII fragment
This study
 pTDM7 esaR::lacZY-KmR gene fusion in pSVB60This study
 pTDM18PCR-amplified esaI promoter and partial coding sequence
cloned into pBBR1MCS-3 as a XbaI–XmaI fragment
This study
 pTDM19 esaI::lacZY-KmR gene fusion cloned into the XmaI site of pTDM18This study
Figure 1.

Dose-dependent repression and derepression of an esaR translational fusion. The E. coli strain TM67 harbouring (A) plasmids pTDM6 and pTDM7 was grown (B) to different optical densities (600 nm) in the presence of 0% (◆), 0.0002% (•), or 0.02% (▴) L-arabinose; or (C) to an optical density (600 nm) of 0.6 in presence of 0.02%L-arabinose with 0 (▪), 100 pM (□) 100 nM (bsl00023), 100 μM (bsl00020) N-(3-oxo-hexanoyl)-L-homoserine lactone (OHHL), or 0%L-arabinose and 100 μM OHHL (bsl00024.

D. Strain TM67 was grown with 0% (◆) or 0.02%L-arabinose (▴) and harvested at different times after L-arabinose induction. Parallel cultures were treated with 100 μM OHHL at 1.5 h (□) or 2 h (○). The cells were harvested and assayed for β-galactosidase expressed as units (Sambrook et al., 1989) (B) or activity (Miller, 1972) (C and D).

Derepression is AHL ligand-dependent

Proof for EsaR mediating QS through a repression mechanism would require evidence for inducing levels of the biologically relevant AHL signal, N-(3-oxo-hexanoyl)-L-homoserine lactone (OHHL), to promote derepression. Accordingly, we grew strain TM67 under stringent repressive conditions in presence of 0.02%L-arabinose. Parallel cultures were supplemented with levels of synthetic OHHL ranging from 0 μM to 100 μM. Derepression of the reporter function correlated directly to the quantity of OHHL supplied (Fig. 1C). More importantly, derepression occurred almost immediately upon addition of the OHHL (Fig. 1D). We conclude from these data that EsaR repression of its own gene is abolished by AHL.

EsaR does not control the expression of the esaI gene

We reported previously that P. stewartii strains mutated in the esaR locus synthesise wild-type levels of AHL (von Bodman and Farrand, 1995) suggesting that EsaR does not control the expression of the linked esaI gene. We confirmed this observation by measuring levels of β-galactosidase expressed from an esaI::lacZY translational gene fusion carried on plasmid pTDM19. When coexpressed with plasmid pTDM6 in E. coli strain TM619, this gene fusion produced identical levels of β-galactosidase regardless of whether or not the strain was induced with L-arabinose and exogenous AHL (data not shown). This therefore establishes that EsaR does not govern the esaI-encoded signal synthesis; in fact, esaI appears to be expressed constitutively.

Purification of native EsaR protein

DNA binding studies to establish the role of EsaR as a repressor in vitro require purified EsaR protein. Initial attempts to express EsaR as a His6-tagged fusion protein were hampered by protein solubility problems. We therefore opted to express EsaR in its native form in E. coli strain TM6, which harbours the recombinant plasmid pTDM6. This plasmid carries the esaR gene under the control of the L-arabinose inducible promoter, PBAD (Table 1 and Fig. 1A). EsaR protein was extracted from cultures grown in the presence of 0.02%L-arabinose and in the absence of the AHL ligand. This is an important point, because some LuxR-type proteins, most notably TraR, require AHL ligand to remain soluble in cellular lysates (Zhu and Winans, 1999; Qin et al., 2001). Growth of strain TM6 at 28°C allowed the purification of native EsaR protein that, when purified by column chromatography, was soluble to 0.4 mM. Active fractions were identified by DNA mobility-shift assays against a synthetic DNA target. These fractions were pooled, aliquoted, and stored at –80°C until needed for further biochemical analysis.

Analysis of EsaR binding in DNA mobility-shift assays

The in vivo data of Fig. 1 indicated that Apo-EsaR represses an esaR promoter gene fusion, presumably by binding to the esaR box located in this region. We therefore assayed purified protein against synthetic target DNAs in DNA mobility-shift assays to define the in vitro DNA binding characteristics of EsaR. Apo-EsaR formed a specific complex with a 28 bp DNA fragment bearing the native esaR box (Table 2). Moreover, the density of the shifted band, measured in digital pixels, was proportional to the amount of EsaR protein assayed (Fig. 2A, lanes 2–4). Addition of increasing levels of unlabelled probe DNA (20 mer esaR box) effectively inhibited complex formation (Fig. 2A, lanes 5–7) and produced a high intensity hybrid DNA species composed of labelled 28 mer and unlabelled 20 mer oligonucleotides that migrate between the 28 mer double-stranded and single-stranded DNA forms. Higher order complexes of EsaR protein were only faintly detected at protein concentrations at or above 200 nM.

Table 2. Oligonucleotides and primers used in this study.
Fragment/primerSequenceUsed for
  1. The underlined sequences indicate the esaR box.

Mobility-shift assay (Fig. 2)
Mobility-shift assay (Fig. 2)
Figure 2.

Interaction of purified EsaR protein with target DNA.

A. A synthetic, end-labelled 28 bp DNA fragment (1.6 μM) containing the EsaR recognition sequence (lane 1) was incubated with 10-fold increasing amounts (2 to 200 nM) of purified EsaR protein and resolved by native gel electrophoresis (lanes 2–4). Binding specificity of EsaR for the probe DNA was demonstrated with 10-fold increasing amounts of excess (5 to 500 μM) unlabelled competitor DNA bearing the 20 mer esaR box (lanes 5–7). Addition of OHHL at 10, 100 and 500 μM did not result in complex dissociation (lanes 8–10).

B. A binding constant (KD) of 3 × 10−8 M was calculated based on the pixel intensity of free and complexed probe DNA recorded with a Bio-Rad phosphorimager and evaluated with IMAGERONE software.

C. DNA mobility analysis of purified MBP-EsaR (M) hybrid protein treated with Xa factor protease at increasing incubation times resolves probe DNA complexed with the higher mass MBP-EsaR (M::M) homodimer, the MalE-EsaR::EsaR (M::E) heterodimer of intermediary mobility, and eventual complete conversion to the lower mass EsaR::EsaR (E::E) homodimer. Lane 1 resolves the M:M homodimer at 0 time incubation; lane 2 shows the E:E homodimer after 2 h of Xa factor treatment.

DNA mobility-shift assays using a wide range of protein concentrations against a constant amount of labelled probe DNA yielded a binding affinity of 3 × 10−8 M (≈KD) for EsaR and the esaR box DNA palindrome (Fig. 2B). A calculated Hill coefficient of 0.99 indicates that EsaR binds the esaR box with little or no cooperativity. Interestingly, the addition of the OHHL signal ligand did not induce DNA–protein complex dissociation even when used at a concentration of 500 μM, and regardless of whether the protein was exposed to the signal ligand pre or post presentation of the DNA probe (Fig. 2A). This in vitro observation is puzzling considering that EsaR is highly responsive to OHHL in vivo.

EsaR binds to the DNA as a dimer

To further evaluate the oligomeric state of EsaR, we developed an assay based on the differential mobility of the native EsaR protein and a larger mass hybrid MBP-EsaR protein. Specifically, we fused the malE gene to the 5′-end of the esaR coding sequence separated by a Xa factor protease recognition linker sequence (Riggs, 2000). Standard DNA binding assays, containing affinity-purified MBP-EsaR, a radiolabelled 28 bp esaR promoter DNA fragment, and a constant amount of Xa factor, were incubated for different times before gel electrophoresis. Figure 2C shows that at zero time incubation the larger mass MBP-EsaR homodimer was present exclusively. Prolonged periods of incubation gradually yielded the MBP-EsaR::EsaR heterodimer of intermediate mobility, followed by complete transformation to the lower migrating EsaR::EsaR homodimer. The formation of only one intermediate complex suggests that EsaR binds the DNA target as a dimer under the conditions assayed. These observations are consistent with our previous size-exclusion chromatographic results, which show that Apo-EsaR fractionates with a relative molecular mass of dimeric EsaR (data not shown).

Analysis of EsaR binding by surface plasmon resonance

Surface plasmon resonance (SPR) is a refractometry-based technique that allows measurement of biomole-cular interactions in real time as changes of mass concentrations on a sensor surface (Rich and Myszka, 2001). We used this method to gain additional insight into the in vitro molecular recognition dynamics of EsaR for its DNA target. A 70 bp biotinylated PCR product of the native esaR promoter was conjugated to a dextran streptavidin-coated (SA) sensor chip at a surface density of 120 resonance units (RUs). Injection of EsaR protein analyte over a range of nM concentrations showed a steady increase in RUs with 344 nM EsaR protein yielding a sensogram that reflects saturation binding (Fig. 3A). The RU values of the sensograms displayed were corrected for nonspecific interactions between EsaR protein and a similar length non-target DNA measured in a parallel reference cell. The data were analysed with the BIAEVALUATION 3.0 software, assuming a simple 1:1 Langmuir binding model to determine a KD value of 9 × 10−9 M (χ2 = 6.01). This binding constant is in good agreement with the 3 × 10−8 M value derived from DNA mobility-shift assays (Fig. 2B).

Figure 3.

Interaction of purified EsaR protein with a PCR-generated DNA fragment containing the esaR box evaluated by SPR refractometry.

A.The target DNA was immobilized on a SA strepdavidin-coated gold sensory chip at 120 RU (1:1 Rmax of 120). EsaR protein analyte was injected over a range of concentrations (22 nM to 1.4 μM, displayed are 86, 172 and 344 nM) at a flow rate of 25 μl min−1 with an association time of 180 s, and a dissociation time of 600 s (shown only 350 s).

B. SPR assay with 344 nM EsaR protein and 0, 172 nM and 1 μM concentrations of OHHL added to the buffering system during the association phase only. The RU values shown in both sensograms are difference RU values of the sample cell and parallel reference cell measuring the non-specific protein–DNA interactions.

Surface plasmon resonance analysis using the same sensor chip and a constant EsaR concentration of 344 nM, but varying amounts of AHL in the buffer system during the association phase of the experiment yielded proportionally lower RU values (Fig. 3B). The sensograms were similar to those obtained when measuring lower EsaR protein concentrations (Fig. 3A). These results suggest that signal ligand may prevent AHL-bound EsaR from interacting with the immobilized DNA target thereby reducing the effective concentration of binding competent EsaR protein.

Analysis of the EsaR/AHL interaction by fluorescence spectroscopy

EsaR contains three tryptophan residues; two located in the N-terminal region (W54 and W82) and one (W191) in the C-terminal helix–turn–helix domain. LuxR homologue proteins utilise regions in the N-terminus to dimerize and to bind signal ligand (Choi and Greenberg, 1992; Hanzelka and Greenberg, 1995; Qin et al., 2001; Zhu and Winans, 2001). Such intermolecular interactions generally invoke structural changes, which, in turn, alter the intrinsic fluorescence intensity of the protein. We took advantage of the intrinsic fluorescence of EsaR to assess the effect of ligand binding on the structural integrity of the protein (Fig. 4A). Measurements taken at wavelengths between 300 nm and 400 nm produced a fluorescence maximum around 340 nm. This fluorescence decreased in intensity with increasing concentrations of the OHHL ligand. Maximal quenching of approximately 50% re-quired 370 nM ligand, and higher concentrations did not further quench fluorescence. These assays employed 370 nM EsaR, therefore suggesting a 1:1 molar ratio of protein–ligand interaction. These results agree with similar fluorescence studies reported previously for CarR (Welch et al., 2000).

Figure 4.

Biophysical analysis of EsaR–N-acyl-homoserinelactone (AHL) interactions. The EsaR fluorescence was determined at a concentration of 370 nM.

A. EsaR titration with increasing amounts of OHHL at 0, 3.7, 7.4, 37, 74, 185, 470, 740 nM and 3.7 μM.

B. Interaction of EsaR with 3.7 μM OHHL, N-(3-oxo-dodecanoyl)-L-homoserine (ODHL), and L-homoserine lactone (HL).

The specificity of EsaR for its signal ligand may vary depending on the length and substitution of the acyl side chain. Comparative fluorescence data obtained from separate experiments containing equal amounts of either L-homoserine lactone (HL), N-(3-oxo-dodecanoyl)-L-homoserine Lactone (ODHL) and OHHL, gave different intrinsic fluorescence spectra (Fig. 4B). The signal ligands were added in 10-fold molar excess and the recorded reductions in EsaR fluorescence were 50%, 35% and 15% respectively. These data suggest that EsaR binds the physiologically relevant OHHL species most efficiently, shows reduced affinity for ODHL, and has minimal affinity for the HL moiety alone.

EsaR is more heat-stable in the presence of the signal ligand

We wished to evaluate whether the addition of the preferred ligand, OHHL, affects the stability of the EsaR protein and/or prevents the purified EsaR protein from precipitation at extremely high concentrations. We therefore monitored the effect of OHHL on the thermal stability of purified EsaR by circular dichroism (CD) spectroscopy. As shown in Fig. 5A, the thermal stability of Apo-EsaR was relatively low with denaturation beginning at 48°C and completed at 58°C. The addition of 10-fold excess OHHL increased the thermal stability of EsaR by about 10°C, with no denaturation seen below 53°C and complete denaturation requiring 68°C (Fig. 5B).

Figure 5.

Thermal stability of EsaR with and without AHL. Circular dichroism (CD) spectra of 370 nM EsaR were determined at increasing temperatures in the absence and presence of OHHL. The bottom curve in both diagrams represent the spectrum at 28°C, the top curves represent the spectrum at the highest temperature.

A. CD spectra without OHHL measured at temperatures ranging from 28°C to 63°C.

B. CD spectra with OHHL measured at temperatures ranging from 28°C to 73°C.


This study establishes several critical facts related to the role of EsaR as a repressor of QS in P. stewartii. First, it demonstrates that EsaR regulates its own expression by repression and AHL-mediated derepression; second, that purified EsaR dimerizes and binds DNA without AHL ligand; third, that AHL interacts specifically with EsaR protein, and induces structural changes that may neutralise its DNA binding affinity; and fourth, that EsaR has no role in the control of the linked signal synthase gene, esaI. These data correlate well with our earlier prediction that EsaR functions as a repressor of CPS synthesis in P. stewartii (von Bodman and Farrand, 1995; von Bodman et al., 1998).

A mechanism for QS by repression is unexpected because the majority of LuxR-type regulators are signal-responsive transcription activators (Fuqua et al., 1996, 2001; Whitehead et al., 2001). The LuxR class of proteins exhibits an overall sequence identity of 18–25%, with three critical regions of higher conservation (Whitehead et al., 2001). These include an N-terminal region for AHL signal binding, a more centrally located domain for protein oligomerization, and a helix–turn–helix C-terminal domain for DNA binding (Stevens and Greenberg, 1999; Fuqua et al., 2001; Whitehead et al., 2001). However, the QS regulatory proteins found in Erwinia-type bacterial species, including EsaR of P. stewartii (formerly E. stewartii) (von Bodman and Farrand, 1995), CarR (Welch et al., 2000) and ExpR of Erwinia carotovora ssp. carotovora (Ecc) (Andersson et al., 2000), have a higher degree of relatedness among them. This subfamily of QS regulators is also distinctive in its ability to dimerize and bind DNA in the absence of signal ligand. Based on these characteristics, it was proposed that the LuxR-type proteins fall into two basic groups (Andersson et al., 2000; Qin et al., 2001; Whitehead et al., 2001).

We propose here that the CarR, ExpR and EsaR, subfamily of QS regulators can be divided further based on their discrete functional differences. Specifically, CarR, the QS regulator of carbapenem synthesis in E. carotovora (Ecc) dimerizes and binds DNA in the absence of AHL, but forms higher order complexes in the presence of inducing amounts of AHL to promote activation of the car target gene system (Welch et al., 2000). In fact, mutational disruption of CarR results in a null-carbapenem phenotype (McGowan et al., 1995). CarR also controls its own expression by AHL-dependent activation (Whitehead et al., 2001). The ExpR regulatory proteins described for two different E. carotovora strains (Andersson et al., 2000) (Whitehead et al., 2001) and Erwinia chrysanthemi (Nasser et al., 1998) also bind target promoters in an AHL-free state, and were thought to control exoenzyme synthesis negatively. However, mutational inactivation of the expR gene has no, or only a minor, effect on exoenzyme expression. The implication is that in these strains QS regulation is redundant and that the repressive role of ExpR may be one of AHL sequestration rather than direct regulation (Andersson et al., 2000). The ExpR protein of E. chrysanthemi was also shown to bind to the expR promoter at a well conserved lux box palindrome (Reverchon et al., 1998; Andersson et al., 2000). However, mutational inactivation of ExpR had no effect on expR expression (Reverchon et al., 1998).

EsaR, in contrast to ExpR and CarR, has a well defined activity as a repressor of its own expression, and the hypermucoid phenotype of esaR null mutants suggests that it represses CPS synthesis as well. Specifically, EsaR negatively regulates an esaR::lacZ reporter gene expressed in vivo in E. coli in a dose-dependent manner (Fig. 1B), and EsaR responds rapidly to exogenously provided AHL for derepression (Fig. 1C). The in vitro binding properties of EsaR support the genetic data showing that Apo-EsaR exists primarily as a homodimer complex and exhibits specificity for the esaR box DNA target with an affinity characteristic of these proteins (Fig. 2A and C). In addition, higher order oligomeric complexes of EsaR, characteristic of CarR and ExpR, are detected only when assaying EsaR protein at or above 200 nM concentrations, with or without OHHL present (Fig. 2B). A significant inconsistency between the in vivo and in vitro data is the observation that addition of OHHL does not promote EsaR–DNA complex dissociation in the DNA mobility-shift assays (Fig. 2A). This is surprising for three reasons. First, the fluorescence-quenching experiments (Fig. 4) show that OHHL interacts specifically and stoichiometrically with purified EsaR protein. Second, the SPR data (Fig. 3B) show that when EsaR is exposed to increasing levels of OHHL, the concentration of binding competent EsaR decreases proportionally. Third, ExpR of E. chrysanthemi releases DNA in response to AHL, although the ligand concentrations used in the ExpR DNA mobility-shift assays (Reverchon et al., 1998) were substantially higher than the amounts used in the current study. The SPR data indicate that AHL reduces the binding affinity of Apo-EsaR. However, additional SPR experiments are needed to establish whether AHL promotes DNA–protein complex dissociation, or whether DNA bound EsaR protein has a structural conformation that is unfit to perceive the AHL signal. AHL-mediated derepression of the esaR reporter gene fusion in vivo is rapid regardless of the time lag between arabinose induction to express EsaR and the addition of OHHL (Fig. 1C). These data suggest that holoenzyme formation does not require nascent protein synthesis as is predicted for TraR protein of Agrobacterium tumefaciens (Zhu and Winans, 1999). In A. tumefaciens, AHL is thought to serve as a scaffold for proper TraR folding and protection from cellular proteolysis. Presumably, EsaR folds properly in the absence of AHL to assume a stable, DNA binding-proficient conformation. It is conceivable, that preformed Apo-EsaR undergoes an AHL-induced structural transformation that perturbs the binding domain, or causes the protein dimer to dissociate into AHL-bound monomers. In either case, AHL may render EsaR protein susceptible to proteolysis. Proteolytic inactivation of EsaR could account for the rapid derepression observed in vivo, while also explaining the inability of AHL to induce complex dissociation in the in vitro DNA mobility-shift assays. Experiments are in progress to test whether proteolysis is a factor in the mechanism of AHL-dependent derepression of EsaR.

A model by Qin and colleagues (Qin et al., 2001) proposes that Apo-TraR of A. tumefaciens partitions to the cytoplasmic membrane to isolate monomeric TraR from the pool of intracellular AHL under non-inducing conditions. It is proposed that TraR dissociates from the membrane and dimerizes when the membrane-permeable AHL concentration reaches inducing levels. In contrast, EsaR does not appear to be membrane-associated; in fact, EsaR is soluble in its native form without AHL ligand, unlike TraR. If, as predicted, membrane association is to prevent premature TraR dimerization, then EsaR protein may not share this requirement because it is naturally dimeric and seeks DNA targets under AHL-limiting conditions.

EsaR is highly responsive to AHL-induced derepression in the control of its own gene expression. More significantly, it responds gradually and proportionally to the amount of AHL provided, not suddenly at a given inducing concentration. If this reflects the normal dynamics of EsaR and AHL co–inducer interaction, then it is unclear how EsaR governs the expression of CPS synthesis in a strictly cell density-dependent manner when AHL production is constitutive. We suggest that AHL-induced derepression of the esaR gene may be important in this respect. Specifically, we propose that increased cellular levels of EsaR sequester the constitutively generated cellular pools of AHL to prevent premature derepression of CPS synthesis at lower cell densities. Accordingly, a cell population would become ‘quorate’ when the cellular levels of AHL exceed the maximum levels of EsaR expressed. This rationale would explain also the apparent paradox of AHL-dependent derepression of the esaR gene to generate more of the QS repressor when ap-proaching AHL-inducing conditions. EsaR-mediated autorepression may also serve as an intrinsic mechanism of QS modulation.

N-Acyl-homoserinelactone (AHL) signal synthesis in P. stewartii is independent of EsaR and appears to be constitutive. This is in contrast to the paradigm of QS regulation that requires LuxR-AHL activation of the cognate luxI signal synthase gene. It is difficult to predict when and why esaI regulation became EsaR-independent; but it may have been in response to EsaR assuming a role as a repressor, because repression of the esaI gene may be counterproductive in the overall scheme of QS. Conversely, EsaR may have assumed a repressor role in response to deregulated synthesis of AHL to preserve the QS control of CPS synthesis and virulence. We know from previous studies that the deregulated synthesis of CPS is disadvantageous for P. stewartii to successfully colonize the plant host (von Bodman and Farrand, 1995).

Finally, we wish to emphasize that we cannot rule out the possibility that EsaR may function under certain conditions also as a gene activator, nor can we predict at this stage whether the expression of the cps gene system is under the direct control of EsaR. We have to consider the formal possibility that the main role of EsaR is to sequester cellular AHL and that a second QS regulatory system may actually govern CPS synthesis. However, such a mechanism is difficult to reconcile with the fact that the esaI-esaR double mutant P. stewartii strain ESΔIR exhibits a hypermucoid phenotype in absence of cellular EsaR or AHL. Experiments to define the role of EsaR in the control of CPS and to search for additional QS systems in P. stewartii are in progress.

Experimental procedures

Bacterial strains, plasmids, oligonucleotides and DNA techniques

The Escherichia coli strains DH5α (Life Technologies) and Top10 (Invitrogen) were used as cloning hosts and were grown at 37°C on nutrient agar plates, Luria–Bertani (LB) broth, M9 minimal, or RM medium (1 × M9 salts, 1 mM MgCl2, 2% casamino acids, 0.4% glucose) in presence of 100 μg ml−1 of ampicillin, 10 μg ml−1 of tetracycline, and 10 μg ml−1 of kanamycin, where applicable. All relevant plasmids and strains are listed in Table 1. The stewartii strains were grown at 28°C in LB in presence of 30 μg ml−1 of nalidixic acid. DNA techniques were performed by standard methods as described (von Bodman and Farrand, 1995; von Bodman et al., 1998). DNA fragments were amplified using Ex Taq Polymerase (Takera/Panvera) and synthetic oligonucleotides (Table 2) ordered to specification from Sigma Genosys.

Plasmid constructions

We amplified the esaR gene using PCR primers, F_esaRNcoI and R_esaRHindIII (Table 2) to create a 5′-NcoI cloning site to overlap and include the ATG translation initiation codon, and a 3′HindIII site downstream of the stop codon. This fragment was inserted into the similarly digested ex-pression vector, pBAD22 (Guzman et al., 1995) resulting in plasmid pTDM6 (Fig. 1A). The esaR gene was re-cloned as a PstI–SalI fragment from plasmid pSVB5-18 (von Bodman and Farrand, 1995) into the broad-host-range vector, pBBR1MCS-3 (Kovach et al., 1995) to generate plasmid pSVB60 (Table 1). This plasmid was used to create a translational reporter gene fusion by inserting the lacZY-KmR cassette from plasmid pLKC481 (Tiedeman and Smith, 1988) cloned as an XmaI fragment into the internal MroI site to generate pTDM7 (Fig. 1A and Table 1). Polymerase chain reaction (PCR) amplification of the esaI promoter and partial coding sequence used primers F_esaIXbaI and R_esaIX-maI that provided 5′-XbaI and 3′-XmaI restriction sites (Table 2) for cloning into plasmid pBBR1MCS-3 to create pTDM18 (Table 1). Plasmid pTDM18 was used to create an in-frame reporter gene fusion by linking the lacZYKmR cassette released as a XmaI fragment from plasmid pLKC480 (Tiedeman and Smith, 1988) to the esaI coding sequence at the XmaI site to create pTDM19 (Table 1).

β-Galactosidase activity assay

Production of β-galactosidase was quantified as described (von Bodman and Farrand 1995). Cells were grown in RM minimal medium, diluted to an optical density (OD)600 of 0.05 and allowed to grow to an OD600 of 0.1 before inducing with L-arabinose. N-(3-oxo-hexanoyl)-L-homoserine lactone (OHHL) was supplied at the indicated concentrations either at the point of L-arabinose induction, or at defined times post L-arabinose induction. β-Galactosidase levels were expressed as β-galactosidase units (Sambrook et al., 1989) or β-galactosidase activity (Miller, 1972). Each experiment was performed in triplicate and was repeated at least two times.

Expression and purification of EsaR protein

Native EsaR protein was purified from E. coli strain TOP10 (pTM6). This strain was grown at 28°C in 1 l of LB containing 100 μg ml−1 of ampicillin. The cells were harvested when the culture reached an optical density OD600 of 0.6, resuspended in 10% glycerol, and stored as stock inocula at –80°C. Cells from 18 l volume fermentations, induced with 0.02%L-arabinose at an OD600 of 0.6 were harvested 4 h post induction. The cells were resuspended in 180 ml of buffer (50 mM Tris, pH 7.5, 10% glycerol) and broken by three passages through a French press (20 000 psi). Cell debris was removed by centrifugation (30 000 g for 30 min) and the soluble fraction was passed through a 0.2 μm cellulose filter (Millipore) before fractionation by heparin affinity perfusion chromatography (BioCAD, Poros 20HE). The column was equilibrated with 450 ml of TBP buffer (20 mM Tris bis-propane buffer, pH 7.5). The soluble lysate was applied to the column in four 5 ml-volume injections and the column was washed with 20 ml of TBP buffer after each injection. Bound protein was eluted with a two-step gradient of TBP buffer containing 400 mM and 800 mM NaCl respectively. Fractions containing EsaR were identified by standard 15% SDS-polyacrylamide gel electrophoresis (SDS–PAGE) using protein molecular weight markers (prestained, broad range, New England Biolabs) for size reference, and differential expression of EsaR protein in un-induced and arabinose-induced lysates. Fractions containing EsaR were pooled and concentrated 16-fold with a Centriprep YM-10 filter (Millipore). The filtrate was applied to a size exclusion column (S-100 Sephracyl, 100KD exclusion) and eluted with a 1 bed-volume of TNG buffer (20 mM Tris, pH 7.5, 500 mM NaCl,10% glycerol). Fractions containing EsaR were identified by SDS–PAGE. EsaR was stored in TNG at –80°C.

Gel retardation assays

Synthetic DNA fragments used in these assays are listed in Table 2. Double-stranded DNA was obtained by mixing complementary oligodeoxyribonucleotides heated to 95°C and slowly cooled to room temperature. These DNAs were labelled at their 3′-ends with [α-32P]-dATP, specific activity 3000 Ci mmol−1 (Perkin Elmer) in presence of Klenow DNA polymerase (Amersham). DNA binding reactions used varying concentrations of purified EsaR incubated at 28°C for 30 min with 1.6 μM concentration of 32P-labelled DNA fragments in 20 μl reactions using a buffer consisting of 20 mM Hepes, pH 7.6, 1 mM EDTA, 10 mM (NH4)2SO4, 1 mM DTT, 0.2% Tween-20, 30 mM KCl, 50 mg ml−1 of uncut λ-DNA and 150 μg ml−1 of BSA. After adding 5 μl of loading buffer, each reaction mixture was fractionated by electrophoresis at 4°C on a native 6% polyacrylamide gel in 1× TBE buffer (Tris-Borate-EDTA, electrophoresis grade, Fisher Biotech). Gels were dried using a vacuum gel drier. An imaging screen-K (Bio-Rad) was exposed overnight and phosphorescence signal was detected with the Molecular Imager FX system (Bio-Rad). The bands were analysed with QUANTITYONE quantification software (Bio-Rad).

Surface plasmon resonance technique

A BIAcore X instrument (BIAcore) was used to perform SPR measurements (Rich and Myszka, 2001). Activation of the SA streptavidin-coated sensor chip was performed following the manufacturer’s recommended procedure. Non-specific and specific biotinylated DNA were coupled to a streptavidin matrix of two separate chips to yield approximately 120 resonance units (RU). The resulting sensor chips were analysed in parallel flowcells. All experiments were carried out at a flow rate of 25 μl min−1. EsaR analyte was presented in a running buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 10 mM DTT, 0.1 mM EDTA). Bovine serum albumin (BSA) (200 ng μl−1) and λ-DNA (8 ng μl−1) were added to each protein sample before injection. Protein mixtures containing various concentrations (ranging from 22 nM to 1.4 μM) of purified EsaR protein were injected allowing an association time of 180 s and a dissociation time of 350 s. The reference flow cell featuring a random DNA target of the same size as the probe DNA was used to subtract unspecific DNA–protein interactions. Regeneration of the chip surface was achieved by removing all bound proteins with a pulse of 5 ml of 0.05% SDS in running buffer. Data analysis was performed using the BIAEVALUATION 3.0 program to determine the binding properties of the protein assuming 1:1 Langmuir kinetics.

MBP-EsaR hybrid protein

The EsaR protein was expressed from plasmid pMBP-EsaR that features the esaR coding sequence cloned into the pMAL-c2x plasmid (New England Biolabs) to generate a C-terminal fusion to the maltose-binding protein for expression in E. coli host strain TB1. The hybrid protein was purified by amylose resin (New England Biolabs) affinity chromatography. Bound proteins were eluted using four volumes of buffer (20 mM Tris, 1 mM EDTA, 200 mM NaCl) with a linear gradient of maltose from 0 mM to 0.6 mM. Proteolytic cleavage of the hybrid protein was achieved with Xa factor (New England Biolabs), using supplier-recommended conditions and varying times of incubation. DNA mobility-shift assays were performed as described above.

Fluorescence spectroscopy

Fluorescence measurements were performed with a Shimadzu RF-5001 PC spectrofluorimeter equipped with a Hellma Cuv-O-Stir magnetic cuvette stirrer and a Lauda RM6 thermostat. All experiments were carried out in 50 mM Tris pH 7.5, 100 mM NaCl, 0.1 mM EDTA. Purified EsaR protein was used for all experiments. Before measuring, the cuvettes containing 1.5 ml of the samples were equilibrated to 28°C and carefully stirred to ensure homogeneity. Spectra between 290 nm and 480 nm were recorded with an excitation wavelength of 284 nm and a scan speed of 0.8 nm per second. To avoid errors caused by sample dilution, the autoinducer titrations were performed by adding 1 μl of suitable stock solutions.

Circular dichroism measurements

Measurements were obtained with a Jasco J-600 spectropolarimeter equipped with a Lauda RC6 thermostat. All experiments were carried out in 20 mM Hepes pH 7.0, 500 mM NaCl, 5% glycerol. Spectra from 200 nm to 260 nm were recorded using a cuvette of 2 mm path length at a scanning speed of 20 nm per minute. For each temperature, the cuvette was allowed to equilibrate for 10 min before recording the spectrum.


We thank Professor Wolfram Sänger for inviting and supporting T.D.M. to visit his laboratory and to conduct part of this work at the Institut für Kristallographie, Freie Universität Berlin. We are grateful to Werner Schröder for supplying us with oligonucleotides, and J. Schneider-Mergener and C. Landgraf for use of the BIAcore instrument. We also thank John Clark, Clay Fuqua and Maria Koutsoudis for helpful discussions and critical evaluation of this manuscript. This work was supported in part by the USDA Agricultural Experiment Station grant CONSOO712 to S.B.vonB.