Inhibition of the Agrobacterium tumefaciens TraR quorum-sensing regulator. Interactions with the TraM anti-activator.

The Agrobacterium tumefaciens quorum-sensing transcriptional regulator TraR and its inducing ligand 3-oxo-octanoyl-l-homoserine lactone control conjugal transfer of the tumor-inducing plasmid, the primary virulence factor responsible for crown gall disease of plants. This regulatory system enables A. tumefaciens to express its conjugal transfer regulon preferentially at high population densities. TraR activity is antagonized by a second tumor-inducing plasmid-encoded protein designated TraM. TraM and TraR are thought to form an anti-activation complex that prevents TraR from recognizing its target DNA-binding sites. The formation and inhibitory function of the TraM-TraR anti-activation complex was analyzed using several different assays for protein-protein interaction, including surface plasmon resonance. The TraR-TraM complex forms readily in solution and is extremely stable (K(D) of 1-4 x 10(-9) m). Directed mutational analysis of TraM identified a number of amino acids that play important roles in the inhibition of TraR, clustering in two regions of the protein. Interestingly, several mutants were identified that proficiently bound TraR but were unable to inhibit its activity. This observation suggests a mechanistic separation between the initial assembly of the complex and conversion of TraR to an inactive form.

Conjugation of the Agrobacterium tumefaciens tumor-inducing (Ti) 1 plasmid is regulated by crown gall tumor-released compounds called opines and by the acylated homoserine lactone (acyl-HSL) 3-oxo-octanoyl-L-homoserine lactone (3-oxo-C8-HSL; see Refs. 1 and 2). Acyl-HSLs are diffusible pheromones produced by a variety of Gram-negative bacteria. At low population densities, acyl-HSLs are rapidly dissipated by diffusion into the surrounding environment. The contribution of each signal-producing cell is additive, and if population density increases, the relative acyl-HSL concentration is elevated, eventually reaching an inducing concentration and stimulating a programmed set of adaptive responses. The release and subsequent detection of acyl-HSLs therefore enable bacteria to monitor their own population density, a process often referred to as quorum sensing (3,4). Synthesis of acyl-HSLs in most cases is mediated by proteins that resemble the LuxI protein of Vibrio fischeri, whereas most of the receptors of these signals are acyl-HSL-dependent transcriptional regulators and resemble the V. fischeri LuxR protein.
A simplified model of acyl-HSL quorum sensing in A. tumefaciens is presented in Fig. 1. The TraI protein, encoded on the Ti plasmid, directs synthesis of 3-oxo-C8-HSL (5)(6)(7)(8)(9). Ti plasmid conjugal transfer (tra) genes are activated by TraR in response to the pheromone (6,10). Complexes of TraR with 3-oxo-C8-HSL bind to promoter elements, called tra boxes, upstream of at least five different tra operons on the Ti plasmid (11)(12)(13). By comparison to LuxR itself, TraR can be divided into two functional domains (4). A region in the amino-terminal half of the protein has conserved residues known to be required for 3-oxohexanoyl-homoserine lactone binding by LuxR (14). The carboxyl-terminal region of the protein contains a helix-turn-helix motif (position 193-211) that has been implicated in DNA binding for both TraR and LuxR (15,16). There is genetic and biochemical evidence that interaction with 3-oxo-C8-HSL results in the formation of stable TraR dimers that can bind tra boxes and activate transcription (17,18). One acyl-HSL molecule binds per monomer of TraR in an extremely stable complex that is partially dissociable with detergent treatment (19).
In contrast to most other LuxR-type proteins, the activity of TraR is directly influenced by several other regulatory proteins (Fig. 1). The TrlR protein (previously designated TraS) is a truncated homologue of TraR that contains sequences necessary for dimerization but lacks the helix-turn-helix DNA-binding motif (20,21). TrlR inhibits TraR through formation of inactive heterodimers (22). The activity of TraR is also under the influence of the Ti plasmid-encoded TraM regulator, a small protein thus far only identified in A. tumefaciens and other members of the family Rhizobiaceae (23)(24)(25). TraM inhibits the activity of the TraR protein, and this inhibition is absolutely required for the normal operation of the A. tumefaciens quorum sensor.
Mutants in traM are hyperconjugal and, in contrast to wild type, will transfer the Ti plasmid to recipient cells at high efficiency even at low densities of conjugal donors (6,23,26). In effect, TraM acts to set the threshold level of acyl-HSL-TraR complex required to activate the expression of Ti plasmid tra genes. Therefore, TraM plays a crucial role in determining what constitutes a bacterial quorum and when Ti plasmid conjugal transfer is initiated. Recent studies (27,28) of the TraM protein encoded by the nopaline-type Ti plasmid suggested that it inhibits activation of tra gene expression through formation of a putative anti-activation complex with TraR. Yeast two-hybrid studies and far-Western analysis of TraM binding to immobilized TraR suggest that TraM associates with a region in the carboxyl-terminal half of the transcription factor (27,28). Mutational analysis of the nopaline-type Ti plasmid (pTiC58) traM coding sequence identified two amino acid residues in the carboxyl terminus of the protein, where alterations resulted in reduced inhibitory function (27).
Our findings analyzing TraM from the octopine-type Ti plasmid pTiR10 are consistent with the formation of a TraR-TraM anti-activation complex. We expand on this previous work, analyze TraM inhibition at additional TraR-binding sites, and demonstrate TraR-TraM complex formation in solution. Moreover, kinetic analysis of the interaction between these two proteins using SPR reveals that the anti-activation complex assembles slowly but once formed is highly stable. Additionally, site-directed mutagenesis of traM has identified a substantial number of important residues required for productive inhibition of TraR, clustered in two regions of the protein. Phenotypic analysis of several mutant TraM proteins in vivo and in vitro suggests a mechanistic separation between binding to TraR and inhibition of its DNA binding activity.
Construction of an Affinity-tagged traM Derivative-The H 6 TraM derivative was obtained by PCR amplification of the traM coding sequence using synthetic oligonucleotides 5Ј-GGCGCTAGCGAACTG-GAAGATGCAAACGTGAC (which contains an NheI site (underlined)) and 5Ј-GCATCTCATTCCTTCGC. The 1027-bp amplification product, digested with NheI and BamHI (BamHI cleaves the fragment at an internal site downstream of the traM coding sequence to generate a 592-bp fragment), was ligated with NheI-BamHI-digested pRSETA (Invitrogen Corp, Carlsbad, CA) to create pMB107. In pMB107 the aminoterminal end of the traM-coding sequence is fused to the RGS(H) 6 affinity tag of pRSETA, under the control of the T7 promoter (P T7 ).
Site-directed Mutagenesis-Oligonucleotide-directed mutagenesis was used to construct mutations in the A. tumefaciens traM gene on pMB107. Mutations were introduced using the Quick-change Site-directed Mutagenesis Kit (Stratagene, La Jolla, CA), and mutated plasmids were checked with automated DNA sequencing by the Indiana Molecular Biology Institute. Following sequence confirmation the mutated traM alleles were PCR-amplified with Pfu Polymerase (Stratagene) using the oligonucleotides 5Ј-GGCGGTACCCTGATAAACAGGAAA-CAGCTATGGAACTGGAAGATGCAAACGTGACG (KpnI cleavage site, Shine-Delgarno sequence and start codon are underlined) and the pR-SETA reverse primer 5Ј-TAGTTATTGCTCAGCGG. The PCR product was digested with KpnI and BamHI (which cleaves within the pRSETA multiple cloning site internal to the 3Ј primer), followed by ligation to the BHR expression vector pBBR1-MCS5, also cleaved with KpnI and BamHI (32). These derivatives expressed the traM mutant alleles from the E. coli P lac promoter on pBBR1-MCS5. Translation of the lacZ␣ peptide was terminated at two tandem stop codons (in bold type) just downstream from the KpnI site, and translation of the native traM gene was initiated at its own ATG codon further downstream. The complete traM-coding sequence was determined for each BHR construct prior to phenotypic analysis.
In Vivo Activity of traM Alleles-The BHR plasmids described above were introduced into A. tumefaciens NTL4 (a strain cured of the Ti plasmid), harboring a plasmid carrying a P traI -lacZ fusion (pCF372) and plasmid expressing traR either strongly as a P tetR -traR fusion (pCF218) or moderately as a P BAD -traR fusion (pCF436). The plasmids expressing traR also carry the E. coli lacI Q repressor gene, which allows for controlled expression of the P lac -traM transcriptional fusions (33). Expression of P traI -lacZ in cultures grown in ATGN minimal media with a variety of supplements was monitored by ␤-galactosidase assays (33). For the traM mutant alleles, the ratio of traI-lacZ expression (Miller units of ␤-galactosidase activity) in the presence of the P lac -traM inducer IPTG relative to traI-lacZ expression in the absence of IPTG was calculated. This activity ratio was subtracted from 1 to obtain inhibitory activity. Percent inhibitory activity was determined as the ratio of inhibitory activities of mutant traM alleles to wild type traM.
Purification of H 6 TraM Proteins and TraR-E. coli BL21/DE3 harboring pMB107 and several of its mutant derivatives were used to overproduce H 6 TraM fusion proteins. For purification, 250 ml of cells were incubated at 37°C with vigorous shaking in LB medium supplemented with Ap. When cultures reached an absorbance at 600 nm (A 600 ) of 0.6, IPTG was added to a final concentration of 0.5 mM to induce expression of the P T7 -H 6 traM gene. After 3 h of additional incubation, cells were collected by centrifugation (11,760 ϫ g, 10 min, 4°C) and suspended in Extraction Buffer (300 mM NaCl and 50 mM NaH 2 PO 4 , pH 7.0). Cells were disrupted by passage through a French pressure cell (Aminco, Urbana, IL) at 18,000 pounds/square inch. Unlysed cells and  6 TraM was eluted from the resin using a linear gradient of 5-500 mM imidazole in Extraction Buffer and was obtained as a large single peak at ϳ200 mM imidazole. Active TraR was purified from cultures of E. coli BL21/DE3 (pJZ358) cultured in the presence of 3-oxo-C8-HSL as described previously (19). Protein concentrations were determined by the technique of Gill and von Hippel (34) and confirmed using the Micro-BCA assay (Pierce). Gel Mobility Shift Assays of TraR DNA Binding Activity and TraM Inhibition-Plasmids pJZ304 and pCF361 were cleaved with the appropriate restriction endonuclease (see text) and end-labeled using [␣-32 P]dCTP and the Klenow fragment of DNA polymerase I by standard techniques. Gel mobility shift assays were performed essentially as described in Zhu and Winans (19). Briefly, a range of TraR concentrations was added to the labeled DNA (2,500 cpm) in reaction buffer (12 mM HEPES-NaOH, 4 mM Tris-Cl, 60 mM potassium glutamate, 1 mM EDTA, 1 mM dithiothreitol, 12% glycerol) and allowed to incubate for 30 min at room temperature. The reactions were loaded directly on 8% polyacrylamide (80:1 acrylamide/bisacrylamide) gels and electrophoresed. The gels were dried and analyzed using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). Unless otherwise indicated, TraR and H 6 TraM were allowed to preincubate for 20 min at 25°C, followed by addition of the labeled DNA to the binding reaction, and a subsequent 30-min incubation.
Radiolabeling of TraR and TraM-Radiolabeled preparations of TraR and H 6 TraM derivatives were obtained by a standard in vivo technique (30). LB-grown overnight cultures of E. coli BL21/DE3 harboring plasmids containing traR or traM genes expressed from the T7 promoter were diluted 100-fold into ATGN broth supplemented with an 18-amino acid mixture (lacking methionine and cysteine), thiamine (5 g/ml), and Ap (1 mg/ml). Organically synthesized 3-oxo-C 8 -HSL (1 M) was also added to cultures expressing TraR. Cultures expressing H 6 TraM derivatives were cultivated at 37°C, whereas those expressing TraR were cultured at 28°C. When cultures reached an A 600 of 0.5, IPTG was added (0.5 mM). After incubation for 30 min, rifampicin was added to inhibit the E. coli RNA polymerase. After 30 min, 50 Ci of [ 35 S]Met (5000 Ci/mmol) was added to the cultures, and incubation was continued for 3 h. Cells were harvested by centrifugation (11,760 ϫ g, 10 min, 4°C) and lysed using a French pressure cell as described above.
Gel Filtration Chromatography of TraR-TraM Complexes-To analyze complexes between the two regulatory proteins in vitro, E. coli lysates containing radiolabeled H 6 TraM, radiolabeled TraR, or both were prepared in TTEDG buffer (50 mM Tris-Cl, pH 7.9, 0.1% Tween 20, 0.5 mM EDTA, 1 mM dithiothreitol, 5% glycerol, 200 mM NaCl) as described above. The lysates were mixed in equal proportions and incubated for 15 h at room temperature to allow complete complex formation. Following incubation, these reactions were applied to a pre-equilibrated 15-ml Superdex 75 gel filtration column in a mobile phase of TTEDG at a flow rate of 0.2 ml/min. Fractions (0.2 ml) were analyzed by SDS-PAGE. Radioactivity was quantitated using a Phos-phorImager. RGS His monoclonal antibodies (Qiagen Inc.) were used to confirm the identity of the band representing presumptive H 6 TraM.
Assays of TraM Interactions with Immobilized TraR-We prepared [ 35 S]Met-labeled wild type and mutant H 6 TraM proteins from E. coli BL21/DE3 as described above. Whole cell lysates containing the ra-diolabeled H 6 TraM were generated using B-PER nonionic lysis reagent (Pierce) following the supplier's recommendations. Several dilutions of each lysate were fractionated on SDS-PAGE gels, and the dried gels were analyzed using a PhosphorImager. The amount of labeled H 6 TraM in each lysate was determined, and they were normalized for the concentration of radiolabeled H 6 TraM.
Lysates of E. coli BL21/DE3 (pJZ358) were fractionated by SDS-PAGE and electrotransferred to nitrocellulose membranes. The membrane was washed extensively in Tris-buffered saline plus 0.01% Tween 20 (TTBS) and blocked with TTBS ϩ 5% skim milk (Blotto). The blocked membranes were incubated with separate BPER lysates containing different [ 35 S]Met-H 6 TraM proteins on a rocking platform for 17 h at room temperature. Following incubation, the membranes were washed five times in TTBS, air-dried, and analyzed using a PhosphorImager.
Surface Plasmon Resonance (SPR) Assays-The BIACORE 3000 system (BIA-CORE Inc., Uppsala, Sweden) was used to perform SPR analysis of TraR-TraM interactions. TraR was immobilized via primary amine cross-linking to the CMD fibers of a CM5 research grade chip. Carboxyl groups on the CMD fibers were first activated by injection of a mixture of N-hydroxysuccinimide and N-ethyl-NЈ-(dimethylaminopropyl)-carbodiimide. TraR (50 nM) purified as a complex with 3-oxo-C8-HSL was injected in maleate buffer (5 mM maleic acid, pH 7.0) into the flow cell at 2 l/min to allow covalent bonding to the activated CMD fibers. Unreacted CMD fibers were titrated using acidified ethanolamine (pH 8.5), and the response units (RU) of TraR immobilized on the chip surface were determined. H 6 TraM was bound to a CM5 chip by conjugating the protein via a thiol linkage to its single dispensable Cys at position 64 (see "Results") using maleimide coupling chemistry. To do this, a CM5 chip was activated using the N-hydroxysuccinimide/Nethyl-NЈ-(dimethylaminopropyl)-carbodiimide mixture described above, followed by ethylenediamine treatment. The heterobifunctional crosslinker Sulfo-MBS (Pierce) was injected and conjugated to the CMD fibers on the sensor chip surface. H 6 TraM (650 nM) was injected in maleate buffer and reacted with the immobilized cross-linker via its thiol side chain. Remaining active maleimide residues were titrated using 100 mM L-cysteine. Immobilized TraR and immobilized H 6 TraM surfaces were used to analyze TraR-TraM interactions. The analyte, TraR or H 6 TraM, was injected in a buffer containing HEPES-buffered saline (10 mM HEPES, pH 7.4, 150 mM NaCl) with 3 mM EDTA and 0.005% polysorbate 20 (HBS-EP). Binding was monitored as an increase in RU. The dissociation of the complex was monitored as a decrease in RU following termination of analyte injection and continued mobile phase flow. The surface was regenerated with a short pulse of 6 M guanidine hydrochloride (GdnHCl) for 1 min at a flow rate of 10 l/min as described by Schuster et al. (35). HBS-EP buffer was applied at 25 l/min for 5 min to remove GdnHCl. Regeneration was checked by repeated injections of analyte. Determination of kinetic constants was performed at low ligand surface densities over a range of protein concentration in a randomized series with injections performed in triplicate for each concentration. Reported values are based on averaged data for each analyte concentration in triplicate.

TraM Inhibits TraR-DNA Complex Formation in Vitro-To
test the anti-activator function of TraM, gel mobility shift assays were performed using a DNA fragment containing tra box I, a TraR-binding site located between divergent pTiR10 tra operons (Fig. 1 TraI is a LuxI-type acyl-HSL synthase that catalyzes production of 3-oxo-C8-HSL (filled dots). TraR is a LuxR-type transcriptional regulator, and TraM is an anti-activator of TraR. TrlR is a truncated, inhibitory version of TraR. Classes of activated genes are as follows: (a) Ti plasmid conjugal transfer (tra) genes; (b) the traM gene; and (c) traR and several genes directly upstream of traR.
( Fig. 2, panel A). Preincubation of TraR with high concentrations of purified H 6 TraM completely abolished complex formation. No additional complexes were observed, indicating that H 6 TraM does not itself bind to the tra box. Heat denaturation of H 6 TraM prior to these assays destroys its inhibitory activity (Fig. 2, panel B).
The inhibition of DNA binding activity is dose-dependent with respect to H 6 TraM, with a 32-fold molar excess of H 6 TraM (320 nM) over TraR providing full inhibition in 20 min coincubation period assays (Fig. 2, panel A). A one-to-one molar ratio of H 6 TraM to TraR provides virtually complete inhibition after extended incubation periods (Ͼ15 h), whereas reactions with excess TraR cannot be completely inhibited (data not shown). Addition of H 6 TraM to preformed TraR-DNA complexes gave similar results, indicating that TraM can disrupt TraR-DNA complexes. Inhibition was completely unaffected by addition of 3-oxo-C8-HSL up to 30 M (250-fold molar excess over TraM, data not shown).
TraM Inhibition of TraR DNA Binding Activity Is Promoterindependent-The region upstream of traI contains two sequences that resemble tra box I ( Fig. 1 and Fig. 3, panel A). The promoter proximal element (tra box II) is required for transcriptional activation of traI, and, as with tra box I, is centered at Ϫ42.5 nucleotides upstream of the transcription start site (11). Another element (tra box III) is located at position Ϫ167 relative to traI but is not required for traI expression (11,36). A plasmid carrying both tra box II and III (pCF361) was cleaved to separate the two elements from each other, generating fragments II and III (Fig. 3, panel A). Gel mobility shift assays with TraR using this DNA preparation resulted in two distinct complexes (Fig. 3, panels B and C). To determine which fragment was present in each complex, the two fragments were gel-purified and used separately in gel mobility shift assays. As expected, the faster migrating complex contained fragment III, whereas the slower migrating complex contained fragment II (Fig. 3, panel C). To test whether TraM could inhibit binding of TraR to these sites, we preincubated H 6 TraM with TraR, and we tested TraR for DNA binding activity. H 6 TraM inhibited the binding of each site with the same potency as observed with tra box I (Fig. 3, panel D).
Gel Filtration Chromatography of the TraR-TraM Complex-Luo et al. (28) demonstrated that TraM will bind to TraR that has been strongly expressed in E. coli extracts and immobilized on nitrocellulose membranes. To examine the presumptive complex between TraM and TraR in solution, we used gel filtration chromatography on mixtures of E. coli extracts containing TraR and H 6 TraM (both radiolabeled). TraR purified from cells grown in the presence of 3-oxo-C8-HSL elutes with an approximate molecular size of 52 kDa (Fig. 4, panels A and D), whereas TraR produced in the absence of the pheromone elutes as 26-kDa monomer (17). H 6 TraM in E. coli extracts elutes as a species having a molecular mass of 36 kDa, which is slightly less than 3 times its monomer molecular mass of 12.8 kDa, suggesting that it migrates as a trimer or perhaps a rapidly eluting dimer (Fig. 4, panel B). Co-incubation of these two extracts resulted in co-elution of all the labeled TraR and a portion of the H 6 TraM at an approximate molecular size of 60 kDa (Fig. 4, panel C). The mass of this complex suggests that it may consist of two TraR monomers and one TraM monomer, or of one TraR monomer and several TraM monomers. When higher protein concentrations were used, both proteins eluted as higher molecular mass complexes (data not shown). Simi- larly, co-expression of TraR and H 6 TraM in E. coli from a single plasmid with dual P T7 promoters (pCF444) revealed a broad peak of labeled protein that eluted at high molecular mass (data not shown).
SPR Analysis of TraR-TraM Interactions-SPR is a technique that allows the measurement of molecular interactions in real time (see Ref. 37 for a review). TraR purified as a complex with 3-oxo-C8-HSL was conjugated to a CM5 chip (see "Experimental Procedures"). For this experiment, a relatively low TraR surface density (576 RU, 1:1 R max of 282) was used in kinetic analyses. Injection of purified H 6 TraM at nanomolar concentrations over the TraR chip resulted in a steady increase in RU (Fig. 5, panel A). In contrast to the specific TraR interaction, H 6 TraM exhibits very little nonspecific interaction with the identical surface prepared without protein or with equivalent amounts of ␤-lactoglobulin as a noninteracting protein (data not shown). The RU values obtained with immobilized TraR were corrected for this limited background signal. Following injection, bound H 6 TraM dissociated from the TraR surface extremely slowly (Fig. 5, panel A). In parallel experiments, sensor chips containing TraR at higher density gave qualitatively similar results (data not shown).
In order to reuse a chip for SPR analysis, bound analyte must be stripped from the surface (a process called regeneration) after each injection, usually with high salt or mild detergents. However, these treatments did not effectively remove H 6 TraM from the TraR chip. The only reagent that reproducibly did so was 6 M GdnHCl applied as a brief pulse (10 l at 10 l/min flow rate), followed by continued HBS-EP buffer flow (Ͼ5 min at 25 l/min) to stabilize TraR prior to the next injection. Chips treated in this way bound H 6 TraM with the same kinetics as chips not treated with denaturant (data not shown). Although it is not certain that 3-oxo-C8-HSL bound to TraR on the CM5 chip is retained after exposure to the denaturant, in solution 3-oxo-C8-HSL is not removed from TraR by GdnHCl treatment (17). Kinetic analysis of H 6 TraM interaction was performed over a range of concentrations from 9 to 300 nM (Fig. 5, panel  A). The results obtained were fit to a Langmuir binding model using BIA-Evaluation 3.0 software ( 2 value of 7). The association and dissociation rates calculated from these data provide a reasonable estimate of the interaction (Table II). Binding of H 6 TraM to the TraR surface proceeds at 6.98 ϫ 10 4 binding events M Ϫ1 s Ϫ1 , whereas dissociation of bound H 6 TraM occurs at 6.5 ϫ 10 Ϫ5 dissociations s Ϫ1 . This suggests that for this assay format, the initial binding event is relatively slow, but once formed the H 6 TraM-TraR complex is highly stable.
The population of TraR proteins immobilized on the CM5 chip described above is by definition heterogeneous, crosslinked at a fraction of the potential primary amines exposed on the surface of TraR. To test the reversibility of the SPR format and generate a more homogeneous surface, we immobilized H 6 TraM by its single dispensable cysteine (Cys-64, see below) using maleimide cross-linking chemistry. This presumably generates a homogeneous surface with each H 6 TraM liganded identically. Kinetic analysis with a low density (140 RU) H 6 TraM surface and TraR as the analyte results in binding that is ϳ2-fold more rapid than when TraR is the ligand (1.2 ϫ 10 5 events M Ϫ1 s Ϫ1 , Table II and Fig. 5, panel B). However, the complex also dissociates ϳ2.5-fold more rapidly (4.9 ϫ 10 Ϫ4 events s Ϫ1 ). Overall, the liganded TraR bound to the H 6 TraM analyte is 4-fold more stable than the liganded H 6 TraM-TraR analyte complex (estimated K D values of 9.3 ϫ 10 Ϫ10 and 4.1 ϫ 10 Ϫ9 , respectively, Table II).
Site-directed Mutagenesis of Residues Conserved between TraM Homologues-To identify TraM residues likely to be important for function, we aligned the protein sequences of octopine-type and nopaline-type TraM proteins as well as a functional TraM orthologue found on the plasmid pNGR234a from Rhizobium sp. NGR234 (24). 2 This alignment revealed 18 absolutely conserved residues (Fig. 6, panel A). Site-specific mutagenesis was used to generate derivatives of pMB107 containing single mutations in each conserved residue.
All conserved residues except alanine itself were converted to alanine, whereas conserved alanines were converted to glycine. A single nonconserved cysteine at position 64 was also mutated to a serine. Each of these traM alleles was also introduced without a His 6 tag into a BHR plasmid so that they are expressed from the lac promoter (P lac ) of the vector. These plasmids were introduced into strain NTL4(pCF372)(pCF218) and NTL4(pCF372)(pCF436). Both strains lack the Ti plasmid and harbor a P traI -lacZ fusion on pCF372. The first strain strongly expresses traR (cloned in pCF218), whereas the second strain expresses low levels of TraR (cloned in pCF436, see Ref. 38). In both of these strains, wild type traM caused significant inhibition of 3-oxo-C8-HSL-dependent traI expression. Inhibition required the addition of IPTG to induce the P lac -traM fusion (Table III). In the strain expressing high levels of TraR, traI expression was reduced 100-fold by TraM. In the strain that weakly expressed TraR, traI expression was completely inhibited by TraM (Table III). Full inhibition required addition of IPTG to induce TraM expression, although some inhibition was observed even in the absence of IPTG (Table III) In the strain strongly expressing traR, 8 of the 19 mutations showed 90% or greater loss of activity (Table IV) (Table IV). Western blot analysis of the A. tumefaciens lysates expressing traM with polyclonal antibodies raised against purified H 6 TraM suggested that all of the mutant proteins were relatively abundant, except for the TraMH40A derivative (Fig. 6, panel B). All of these lysates, including that of the wild type, contained the fulllength TraM protein and a single, slightly smaller degradation product.
Direct Binding Assay Reveals Differences between TraM Mutant Proteins-To determine whether reduced binding efficiency for TraR could explain the defects observed for the traM mutants in our genetic analysis, we performed an assay that directly measures binding of TraM to immobilized TraR. All of the TraM proteins that exhibited reduced activity in vivo were overexpressed as His 6 -tagged proteins in E. coli (see "Experimental Procedures"). Extracts from BL21/DE3(pJZ358) (which overexpresses TraR) were fractionated by SDS-PAGE and electrotransferred to nitrocellulose membranes. Extracts containing radiolabeled mutant or wild type H 6 TraM were 2 X. He and C. Fuqua, manuscript in preparation.  6. Mutation of conserved TraM residues and analysis of stability. Panel A, positions of the TraM mutations generated in the current study. Panel B, Western blot of lysates from A. tumefaciens cells harboring pCF372 (P traI -lacZ), pCF218 (P tetR -traR), and each of the pBBR1-MCS5 derivatives expressing traM alleles from P lac . Lysates prepared from cultures grown in the presence of 0.5 mM IPTG, electrophoresed on SDS-PAGE, transferred to nitrocellulose, and probed with anti-TraM antiserum. Antibody binding was detected using horseradish peroxidase-conjugated goat anti-rabbit antibodies and SuperSignal PicoWest chemiluminescence substrate, followed by exposure to x-ray film.
used to probe the individual nitrocellulose membranes. Specific labeling of a band corresponding to TraR in the immobilized lysates was observed (Fig. 7). Identical lysates without TraR were not labeled, and probing with equivalent lysates lacking H 6 TraM also did not label TraR in the extracts (data not shown). A second, more weakly labeled protein significantly smaller than TraR was also observed and may represent a degradation product. Comparison of labeling between the mutants revealed that TraMH40A, R41A, L54A, G94A, and P97A exhibited significantly reduced binding to TraR (Fig. 7). Several mutant TraM proteins exhibited modest reductions of binding to TraR (L29A and Y72A) whereas other mutant proteins bound TraR at wild type levels (A36G, I37A, L43A, A81G, Q82A, and L93A).  a Plasmids refers to constructs supplying traR and traM. All assays were performed in A. tumefaciens NTL4 harboring pCF372 (traI-lacZ). b Miller units, see "Experimental Procedures." c IPTG, 0.5 mM final concentration. d TraR was expressed from pCF218 (P tetR -traR) or pCF436 (P BAD -traR). e Strains carried a pBBR1-MCS5 (vector control) or pGB200 (P Lac -traM). f Crude acyl-HSL semipurified from ethyl acetate extracts of cell-free culture fluids from an A. tumefaciens strain engineered to overproduce 3-oxo-C8-HSL. Added to induction medium at ϳ0.5 M final concentration. a Strongly expressed traR supplied from pCF218 P tetR -traR derivative, see Table I. b Weakly expressed traR supplied from pCF436 P BAD -traR derivative in the absence of L-arabinose, see Table I. c All traM alleles expressed as P lac -traM fusions cloned in pBBR1-MCS5, see Table I. d Miller units. e 100% inhibitory activity was determined for wild type traM as a comparison of ␤-galactosidase activity in cultures grown with and without IPTG (0.5 mM), of A. tumefaciens NTL4 harboring pCF372 (traI-lacZ), and either pCF218 or pCF436 and pGB200 (P lac -traM) in the presence of exogenous crude acyl-HSL (ϳ0.5 M). Percent inhibitory activity calculated for each traM mutant allele was compared with the inhibitory activity of wild type traM carried on pGB200.

Gel Mobility Shift Inhibition Assays with Purified TraM Mutants-Six
identically to the wild type proteins, and all proteins had equivalent solubility. These proteins were compared with the wild type for the ability to inhibit the binding of TraR to the tra box I element in vitro (Fig. 8). Several of the mutant proteins exhibited significant, yet reduced ability to inhibit TraR activity (A36G, I37A, and L93A). In contrast, the TraM H40A, R41A, and Q82A mutant proteins were severely debilitated in TraR inhibition. It is of note that the Q82A mutant protein, despite strong TraR interactions as assessed in the direct binding assay (Fig. 7), was unable to inhibit TraR activity at any concentration tested. No inhibition of TraR and no supershifted species (e.g. ternary complexes) were observed, even when higher concentrations of the Q82A mutant were tested (100 M, data not shown).
SPR Analysis of TraM Mutant Proteins-Kinetic analysis of purified mutant and H 6 TraM derivatives over a range of concentrations (9 -300 nM) was performed with a single TraR-CM5 sensor chip (Table II). All of the mutant proteins exhibited alterations in binding to TraR consistent with that observed in the direct binding assay (Fig. 7). Sensorgrams were generated with 150 nM of either wild type or mutant proteins (Fig. 9). The A36G mutant has binding interactions with TraR that are very similar, perhaps even more rapid than wild type TraM. The Q82A and I37A mutants are also comparable with wild type TraM in the rate of association but dissociate more rapidly. In contrast, TraM mutants H40A and R41A exhibited reduced rates of association and more rapid dissociation. The L93A mutant was dramatically reduced for association with TraR, but once the complex formed, it was as stable as that between TraR and the wild type protein. Interestingly, although most of the binding curves fit imperfectly to Langmuir binding kinetics (as with the wild type), the highly defective mutant H40A mutant was an extremely good fit ( Fig. 9 and Table II). DISCUSSION TraM is a potent inhibitor of the TraR transcriptional regulator (27,28). In this study, we use several different approaches (gel filtration, direct binding to immobilized TraR, and SPR analysis) to demonstrate that TraR and TraM form a highly stable complex. In addition, we show that the formation of the TraM-TraR complex is coincident with inhibition of TraR DNA binding activity and that this is independent of the DNA sequences bound by TraR. These observations suggest that physical association with TraM promotes a general inactivation of the DNA binding activity of TraR. TraM therefore joins a growing list of prokaryotic regulatory proteins that exert their effects on gene expression by interacting directly with a transcription factor. Examples are well documented in the lytic cycles of bacteriophages P22 (Ant-c2) and P1 (coi-c1), in the sporulation of Bacillus subtilis (SinI-SinR), in genetic competence (ComS-ComK), and in maltose utilization in E. coli (MalY-MalT and MalK-MalT) (39 -44). Additionally, several alternative factors in a variety of bacteria are regulated by anti-factors and anti-anti-factors that modulate and can even reverse RNAP holoenzyme formation by physically sequestering the factors (45)(46)(47)(48).
Gel filtration analysis of complexes assembled by mixing whole cell lysates with 35 S-labeled proteins suggests that the TraR-TraM complex formed under these conditions is slightly larger than the 52-kDa TraR dimer (Fig. 4, panel C). The most likely conformation of this complex is a dimer of TraR plus a monomer of TraM or a trimer of TraM plus a TraR monomer. Repeated attempts to determine the exact stoichiometry of TraR and TraM in the complex were confounded by differential labeling efficiencies for the two proteins. Additionally, we observed that the apparent molecular mass of this complex was highly variable depending upon protein concentrations, further hindering our efforts at determining the ratio of TraR to TraM in the complex (data not shown). The physiological relevance of such higher order complexes is not clear.
The K D of the complex that forms between TraR and TraM, as estimated from SPR experiments using either TraR or H 6 TraM as ligand, ranges from 1 ϫ 10 Ϫ9 M to 4 ϫ 10 Ϫ9 M (Table  II). Our findings suggest that this complex does not form rapidly (0.7-1 ϫ 10 5 M Ϫ1 s Ϫ1 ) but dissociates slowly (0.65-4.9 ϫ 10 Ϫ4 s Ϫ1 ). This in vitro finding is consistent with the direct binding experiments with immobilized TraR, in which bound H 6 TraM continued to increase for up to 17 h of incubation ( Fig.  7 and data not shown). The relatively slow rate of association is also reflected in the large molar excess of TraM over TraR required to achieve full inhibition in our gel shift inhibition experiments (Figs. 2 and 3) as well as those reported previously (28). In these experiments, TraR and TraM were only briefly co-incubated (20 min to 1 h). The observation that nearly complete inhibition is achieved with longer incubation times (Ͼ15 h) at a 1:1 TraM to TraR molar ratio, but is incomplete below 1:1, suggests that the TraR-TraM interaction is noncatalytic and relatively slow in vitro. In contrast, Luo et al. (28) observed that induction of a P BAD -traM fusion by addition of L-arabinose to the culture resulted in relatively rapid inhibition (1-2 h) of TraR-controlled target gene expression. Inhibition in both experimental formats is clearly fostered at high TraM:TraR ratios, although the association of TraM may be less efficient in vitro.
Stable complexes between TraM and TraR were detected using SPR over a range of protein concentration, with TraR as the ligand and TraM as the analyte or vice versa (Fig. 5). The calculated dissociation constant was about 4-fold smaller when TraR was the ligand than when TraM was the ligand (Table II and Fig. 5). This could be due to differences in the chemical cross-linking employed to bind the two proteins to the chip. Mutagenesis of Cys-64, the residue used to immobilize H 6 TraM as the ligand, does not affect the inhibitory activity of traM in vivo (Table IV). However, cross-linking to the CMD fiber at Cys-64 may modify the relative TraR-TraM interactions and result in the observed differences. Such effects would be averaged when TraR was the ligand as there are many different potential primary amines where cross-linking would occur (Fig.  5). In both formats, the kinetics of complex formation were best fit using Langmuir one-to-one binding. However, although the fit using the Langmuir model is reasonable, it is not ideal (see 2 Table II), suggesting that the association of H 6 TraM and TraR may be more complex than simple one-toone binding or that additional interactions increase the overall complexity of the reaction. The association curves for both ligand configurations are slow to reach a steady-state plateau, even over long injection times (Figs. 5 and 9). This may reflect the stability of the TraR-TraM interaction or that TraM and TraR from the analyte phase continue to associate after formation of an initial one-to-one complex. Effective regeneration of the stable TraR-TraM complex from the CM5 surface required treatment with 6 M GdnHCl. This is similar to the tenacious interactions between proteins of the bacterial chemotaxis machinery, which also required GdnHCl for effective regeneration in SPR analysis (35). The in vitro inhibition of TraR activity by TraM is unaffected by 3-oxo-C8-HSL (this study and Ref. 28). Considering these observations, we propose that in vivo, dissociation of TraR and TraM from the complex is not physiologically relevant and that this is a one-way, inhibitory process. This mechanism is similar to the proposed dead-end inhibitory complex formed between the SinI inhibitor and the SinR repressor of sporulation genes during initiation of endospore development in B. subtilis and contrasts reversible regulatory complexes such as those typified by the MalK and MalY inhibitors of MalT (41,42,49).

values in
Previous mutagenesis of the traM gene from the nopalinetype Ti plasmid identified the two residues Val-86 and Pro-97, in the carboxyl-terminal region of the protein, as important for the inhibitory activity of TraM (27). Our directed mutagenesis approach has significantly expanded on these findings (Table  IV). Of the 12 residues where mutations significantly affect TraM inhibitory activity, five (L29A, A36G, L43A, A81G, and L93A) demonstrated a more pronounced deficiency when TraR was overexpressed (Table IV). However, seven mutants, in two distinct clusters (position 37-41 and 82-97), had null phenotypes even when TraR was expressed at low levels, suggesting more severe mechanistic defects. Binding of TraM to TraR is clearly a prerequisite to inhibition of the protein. A number of the mutants (H40A, R41A, L54A, Y72A, G94A, and P97A) demonstrate substantial reductions in binding efficiency ( Fig. 7 and Table II). Conserved residues within the region of TraM flanked by Leu-54 and Gln-82 (Gln-68, Tyr-72, Ala-81, and nonconserved Cys-64) do not appear to play important roles in the inhibition of TraR. Surprisingly, the L54A mutant manifests a more severe defect in vivo with lower levels of TraR than with larger TraR pools (Table IV).
Consistent with the role for Pro-97 reported by Hwang et al. (27), our results implicate the carboxyl-terminal region of TraM in TraR inhibition and expand this region from Gln-82 to Pro-97. Comparison of the NGR234 and A. tumefaciens TraM sequences reveals that Leu-93, Gly-94, and Pro-97 are located in a short 11-amino acid hydrophobic region of TraM shared by the two proteins. Individual mutations at all three positions in TraM cause significant deficiencies in TraR interactions, although the effect of the L93A mutation is most pronounced in the SPR experiments ( Fig. 9 and Table II).
Interestingly, the Q82A mutation, just to the amino-terminal side of the conserved hydrophobic region, results in a protein that is only slightly reduced in binding to TraR (Table II and Figs. 7 and 9) but is completely unable to inhibit its activity in vivo and in vitro (Table IV and Fig. 8). The genetic background expressing lower levels of TraR provides no suppression of this defect (Table II), yet the Q82A mutant is highly proficient for simple binding interactions. This observation suggests that the Q82A mutant is affected for some other aspect of the productive interaction between TraR and TraM and that binding is insufficient to affect TraR activity. Similar results were obtained from studies of the inhibitory effect of the E. coli MalK protein on MalT activity, in which mutations in either gene that abolished inhibition did not necessarily affect binding to MalT (41). We speculate that subsequent interactions between TraR and TraM that are facilitated by but distinct from the initial binding event act to reconfigure TraR into an inactive state and stabilize the complex. Analysis of deletion and substitution mutations in the pTiC58 traR gene led Hwang et al. (28) to propose that TraM associates with a region between residues 142 and 176 of TraR and also residues closer to the carboxyl terminus of TraR. Our findings suggest that initial binding may not be sufficient for the anti-activation complex to form. It is intriguing to speculate that subsequent interactions between the far carboxyl terminus of TraR and other regions of TraM, requiring and perhaps including Gln-82, convert TraR to a stable, inactive form.
Our results suggest that in addition to the carboxyl-terminal sequences, the region between residues 29 and 54 plays an important role in the inhibition of TraR. Deletion analysis of the nopaline-type traM gene suggested that loss of greater than 27 residues from the amino terminus significantly reduced the interaction with TraR as assessed by a yeast two-hybrid assay (27). Our own deletion analysis of TraM suggests that the carboxyl terminus of the protein is necessary and sufficient for strong binding to TraR but that truncated proteins lacking the carboxyl terminus exhibit detectable binding to immobilized TraR. 3 The H40A mutant exhibits the most severely deficient phenotype of mutants with alterations in this region. The observed complete deficiency of TraMH40A for TraR inhibition in vivo may be partially due to its instability when expressed in A. tumefaciens (Fig. 6). However, purified H 6 TraMH40A is unable to inhibit TraR activity in vitro (Fig. 8) and is severely reduced for binding to TraR (Fig. 7). SPR can detect weak interactions of TraMH40A with TraR, but this interaction is orders of magnitude weaker than that with wild type TraM ( Fig. 9 and Table  II). This suggests that residue H40A plays an important although perhaps structural role in the stable binding of TraM to TraR. The TraMR41A mutant exhibits a similar, yet less severe phenotype than the H40A mutant, suggesting a common deficiency and hence perhaps a related function for these adjacent residues. Our mutational analysis is consistent with the carboxyl terminus of TraM functioning in the initial binding of TraR but also suggests that stabilization of the complex requires interactions between other regions of the protein.
The observation of a major TraM degradation product, slightly smaller than full-length TraM (Fig. 6), in A. tumefaciens extracts may reflect proteolytic processing of the protein.
In B. subtilis the ComS inhibitor of the primary competence transcription factor ComK is turned over through recruitment of MecA (a ClpB-type chaperone) and subsequent proteolysis via the ClpP protease (44). In fact, ComS-dependent titration of the MecA-ClpP proteolysis machinery is thought to play an important role in elevating ComK pool sizes during competence. TraR that has not bound its acyl-HSL ligand is rapidly turned over by proteolysis in A. tumefaciens, whereas the acyl-HSL bound form of TraR is more stable (17). It is plausible that the physical interaction of TraM with TraR stimulates the proteolysis of both proteins, augmenting the direct inhibitory effect of TraM association.
TraM clearly acts to prevent low levels of TraR expressed under noninducing conditions from activating tra gene expression. Our results demonstrate that the anti-activation complex that forms between TraR and TraM is likely to be irreversible. Furthermore, our findings suggest that formation of a fully stable complex may require interactions between the two proteins following the initial binding event, perhaps reconfiguring the TraR into a permanently inactive state. These observations parallel the recent finding that fully folded TraR cannot associate with 3-oxo-C8-HSL but rather must fold in the presence of the pheromone to reach an active conformation (17). Therefore, in order for TraR to access its DNA targets and activate Ti plasmid conjugal transfer, it must be expressed in the presence of inducing levels of 3-oxo-C8-HSL and attain levels sufficient to titrate the effect of the TraM inhibitor.