|Home | About | Journals | Submit | Contact Us | Français|
The plant pathogen Agrobacterium tumefaciens expresses virulence (vir) genes in response to chemical signals found at the site of a plant wound. VirA, a hybrid histidine kinase, and its cognate response regulator, VirG, regulate vir gene expression. The receiver domain at the carboxyl end of VirA has been described as an inhibitory element because its removal increased vir gene expression relative to that of full-length VirA. However, experiments that characterized the receiver region as an inhibitory element were performed in the presence of constitutively expressed virG. We show here that VirA's receiver domain is an activating factor if virG is expressed from its native promoter on the Ti plasmid. When virAΔR was expressed from a multicopy plasmid, both sugar and the phenolic inducer were essential for vir gene expression. Replacement of wild-type virA on pTi with virAΔR precluded vir gene induction, and the cells did not accumulate VirG or induce transcription of a virG-lacZ fusion in response to acetosyringone. These phenotypes were corrected if the virG copy number was increased. In addition, we show that the VirA receiver domain can interact with the VirG DNA-binding domain.
Two-component regulatory systems have a wide range of regulatory mechanisms, including hybrid histidine kinases that have an additional C-terminal domain with sequence similarity to the receiver domain of the response regulator (for a review, see reference 33). A feature of the hybrid kinase receiver domain is conservation of the aspartate that is phosphorylated in the response regulator. In some cases, the aspartate in the hybrid kinase receiver domain has been shown to be essential for activation of the cognate response regulator in conjunction with a histidine-containing phosphotransfer (HPT) element that intervenes between the kinase and the response regulator in a His-Asp-His-Asp phosphorelay. The HPT element may be a domain in the histidine kinase, located just after the receiver domain, or a separate protein. However, a survey of 156 complete microbial genomes indicated that hybrid kinases are almost five times as prevalent as identifiable HPT sequences (39). This difference may be due to limited sequence conservation among HPT domains, but another possibility is that the receiver domains of hybrid kinases have evolved so that they have diverse functions, some of which do not involve extended phosphorelay.
The VirA-VirG two-component system encoded in the tumor-inducing (Ti) plasmid (pTi) of the plant pathogen Agrobacterium tumefaciens is a well-known hybrid histidine kinase. This system responds to host signals, activating bacterial virulence genes that mediate the transfer of DNA from the pathogen to the host for tumor induction. VirA, a membrane-bound kinase, responds to chemical signals characteristic of plant wounds, including acidic pH, monosaccharides, and certain phenolic molecules. Acidic pH and phenols (e.g., acetosyringone [AS]) are essential for vir gene induction. Monosaccharides (e.g., glucose and arabinose) are not inducers themselves, but they greatly enhance vir gene expression in response to low concentrations of the phenolic inducer via the activities of the periplasmic sugar-binding protein, ChvE (15; for reviews, see references 6 and 23.
The function of VirA's receiver domain appears to differ from the functions of the receiver domains of some other hybrid kinases in a number of ways. First, chemical stability assays provided no evidence that Asp766, the residue analogous to Asp52 in the VirG receiver domain, receives a phosphate from the conserved histidine His474 (25). Second, mutation of Asp766 reduces vir the gene expression compared to the expression mediated by wild-type VirA, but it still allows significant induction (26). This observation contrasts with in vivo results indicating that mutation of homologous aspartates in LuxN, RscS, TorS, ArcB, and BvgS completely inactivates the kinase function (14, 16, 17, 19, 22, 29). Third, the VirA receiver does not include an HPT domain. While the possibility that there is a separate HPT protein that assists in transfer of the phosphate from VirA to VirG cannot be excluded, there is no evidence that such a protein exists. In fact, purified VirA rapidly phosphorylates purified VirG, suggesting that intervening phosphorelay proteins are not required for VirG activation (18). Finally, previous analyses of VirA have demonstrated that removal of the receiver domain allows vir gene expression in the absence of a phenolic inducer, provided that sugar is present and the pH is acidic (5, 8, 9, 15). These experiments led to description of the VirA receiver domain as an inhibitory element.
In contrast to the studies described above, the experiments described here show that VirA's receiver domain behaves as an enhancing element. When virAΔR was expressed from a multicopy plasmid, there was no response to the phenolic inducer unless sugar was also present. Our studies utilized cells carrying wild-type virG on the Ti plasmid under control of its own promoter, while previous studies describing the VirA receiver domain as an inhibitory element utilized strains expressing virG from a constitutive promoter. By using a similar expression system, we were able to obtain results similar to the previous results showing that VirAΔR could activate vir gene transcription in response to sugar, even in the absence of the normally essential phenol inducer. Using experiments that varied the copy number and/or expression of virA and virG, we showed that the VirA receiver domain was required for efficient vir gene expression, including the transcriptional induction of virG, unless the cellular concentration of VirG was artificially increased.
Finally, a bacterial two-hybrid assay demonstrated that the VirA receiver segment could interact with the DNA-binding region of VirG. In the absence of evidence for a phosphorelay path, this observation suggested that we should consider a model in which VirA's receiver domain acts as a recruitment and/or alignment factor that increases the availability of VirG for transfer of phosphate from VirA's kinase region to the VirG receiver domain.
Bacterial strains and plasmids used in this study are listed in Table Table1.1. Escherichia coli XL1blue (Stratagene) and SU202 were grown at 37°C in LB medium supplemented with suitable antibiotics. A. tumefaciens strains were cultured at 25°C in MG/L or AB*I medium (38) with the appropriate antibiotics for plasmid maintenance.
Standard methods were used for DNA ligation, PCR, plasmid isolation, and DNA analysis (2). The sequences of primers used for plasmid and strain construction are listed in Table Table2.2. Sequence analysis confirmed the correctness of all PCR products. The virAΔ707-829 receiver truncation was constructed by performing sequential PCRs (2). In two initial PCRs with primers 2985-(5′) and 3500stop-(3′) and primers 3500stop-(5′) and 4120-(3′) pAW19 was used as the template. In a third PCR the self-annealing products of the first two reactions were used to create a PCR product, which was used to replace the corresponding wild-type region of pAW19 to create pAW82, which carries two stop codons at positions 708 and 709 that effectively remove the receiver domain. Plasmid pAW102 was constructed by placing virAΔR from pAW82 at the KpnI site of pAW10. Plasmid pAW103 was constructed by placing virAΔR at the KpnI site of pYW47. pAW10 and pYW47 are similar in that they are both derivatives of pYW10 (31, 37). Derivatives of these plasmids are expected to be present in Agrobacterium at levels between 15 and 20 copies per pTi (37). Plasmid pAW168 was constructed by first moving the 7-kb HpaI insert from pVRA6 to pAW24. The 7-kb insert provided additional homology to the region flanking virA on pTi. The Kanr cassette at the KpnI site in the 7-kb insert of pVRA6 was replaced with the KpnI fragment from pAW82 that carried virAΔR. The Kanr cassette was then added back by replacing a short EcoRV fragment downstream from the stop codons that deleted the receiver domain.
The 4.6-kb KpnI fragments carrying the virA alleles of pAW18 and pAW82 were cloned into pRG109 (15) to create pAW100 and pAW107, respectively. In addition to a virA allele, these plasmids carry PN25-virG (31) and PvirB-lacZ. pAW118 derived from pAW19 and pFF5 carried a FLAG-tagged version of PN25-virA at the KpnI site of pAW10. The lacIq gene was amplified from pMAL-c2x (New England Biolabs) by performing PCR with primers lacIQKpn1 and lacIQHIII. Following digestion with KpnI and HindIII, the PCR product was cloned into pAW10 to create pAW106. Plasmid pAW132 was constructed by subcloning wild-type virG with its native promoter region from pSW103 (34) into pAW50. pAW132 includes P1 and P2 and the complete virG regulatory region, as previously defined (36). The copy numbers of pAW50 derivatives are expected to be four to six times the copy number of pTi (37).
Plasmid pAW104 served as source of the virG receiver domain (codons 2 to 120) of virG, which was cloned in pSR658 to create pQF365. The virA receiver domain (codons 712 to 829) was amplified by performing PCR with primers FFE-(SacI-5′) and FFF-(KpnI-3′) and with pFF5 as the template and was cloned into pSR658 to create pQF366. Primers FFI-(SalI-5′) and FFJ-(HindIII-3′) were used to amplify the DNA-binding domain (codons 132 to 240) of virG, which was cloned into pYW15b to create pQF79. The virG DNA-binding domain was removed from pQF79 for insertion into pSR659 to make pQF368. DNA fragments carrying the ompR receiver and DNA-binding domains were generated by performing PCR with genomic DNA isolated from A. tumefaciens C58 as the template based on sequence information available at http://www.genome.jp/kegg/kegg2.html. Plasmid pYL262 carried the ompR receiver domain (codons 2 to 132) in pSR658 and was constructed using primers YL359-(SacI-5′) and YL360-(KpnI-3′). Plasmid pYL263 carried the ompR DNA-binding domain (codons 132 to 239) in pSR659 and was constructed using primers YL361-(XhoI-5′) and YL362-(KpnI-3′). Immunoblot analysis confirmed the expression of the lexA fusions in pQF365, pQF366, pQF368, pYL262, and pYL263.
Strain AB400 was constructed by transforming electrocompetent strain A348 with 0.5 μg linearized pAW168 and selecting for resistance to kanamycin using a gene replacement method similar to that used to construct A348-3 (20). Replacement of wild-type virA with virAΔR was confirmed by performing PCR with total DNA isolated from the transformants and A348 (as a control) using an AquaPure genomic DNA isolation kit (Bio-Rad) and primer 3470m-(5′) with primer 903-(3′), primer 3790m-(5′) with primer 4400c-(3′), primer ΔRxm-(5′) with primer 3700c-(3′), and primer 3720m-(5′) with primer 4120-(3′). Gene replacement was confirmed to have occurred in approximately 20% of the transformants.
For vir gene induction, A. tumefaciens strains carrying either a PvirB-lacZ fusion (pSW209Ω, pAW100, or pAW107) or a PvirG-lacZ fusion (pSW174) were inoculated into AB*I medium containing glycerol, arabinose, or glucose as the carbon source. Arabinose and glucose have similar effects on maximizing vir gene expression and increasing sensitivity to acetosyringone (AS), a phenolic inducer (1, 27). When isopropyl-β-d-thiogalactoside (IPTG) was included, the amounts specified below were added. Following overnight incubation with shaking at 25°C, β-galactosidase activity was determined by the method described by Miller (24).
Leaf explants (16 mm2) cut from a sterilized Nicotiana tabacum cv. H425 leaf were briefly incubated in a culture (optical density at 600 nm [OD600], 0.5) of A. tumefaciens strain A348, AB400, or A348-3 that carried either pAW10 (vector) or pYW47 (PN25-virG). Leaf pieces were then placed on hormone-free MS cocultivation medium (4) containing no AS or 3 μM AS. The cultivation plates were incubated at 25°C for 48 h. Leaf squares were then washed in MG/L broth containing 200 μg/ml of timentin and placed on hormone-free MS selection medium containing 200 μg/ml of the same antibiotic to inhibit bacterial growth. The preparations were incubated at 25°C for 10 days before tumors were counted.
The LexA-based bacterial protein interaction system used in this work has been described preciously (10, 12). E. coli strain SU202 carrying pQF368 was transformed with pQF365, pQF366, or pYL262. SU202 carrying pYL263 was transformed with pQF365, pQL366, and pYL262. Overnight LB medium cultures were diluted using LB medium containing 1 mM IPTG and antibiotics for plasmid maintenance. Western blot analysis confirmed expression for all pSR658 and pSR659 derivatives (not shown). β-Galactosidase activity was determined at an OD600 between 0.4 and 0.6 using the Miller method (24). Stable interactions between the DNA-binding domain of VirG and the receiver segments were monitored by examining repression of the sulA::lacZ reporter system. Repression was calculated by using the β-galactosidase activity produced by the strain carrying the pSR658 and pSR659 vectors as the maximal activity, as follows: repression = 1 − (β-galactosidase activity of sample)/(maximal activity).
A. tumefaciens strains were grown as described above for vir gene induction assays with the amounts of AS indicated. The OD600 values of cell samples were standardized before the samples were used in all immunoblots and in the β-galactosidase assay whose results are shown in Fig. Fig.2B.2B. For immunoblots, equivalent volumes of whole cells were resuspended in sample buffer (3% sodium dodecyl sulfate [SDS], 5% β-mercaptoethanol, 10% glycerol; pH 6.8) and heated for 5 min at 95°C before they were loaded on an SDS-polyacrylamide gel. VirG was visualized by Western blot analysis with rabbit anti-VirG antibody (see Fig. Fig.2A2A and Fig. Fig.4)4) or mouse anti-His antibody (see Fig. Fig.2B).2B). For visualization we used either anti-rabbit or anti-mouse horseradish peroxidase-conjugated IgG secondary antibody with ECL Plus (Amersham Biosciences).
Previously, the function of the VirA receiver domain was examined using strains that expressed virG from a constitutive promoter (5, 8, 9, 15). We reexamined the effect of deleting the receiver domain on vir gene expression when virG was encoded on pTi by following the expression of a virB-lacZ fusion in strain A348-3 (ΔvirA). In the absence of sugar (Fig. (Fig.1A),1A), VirAΔ707-829 (VirAΔR) expressed from pAW102 behaved like a null mutant even with a relatively high concentration (300 μM) of the phenolic inducer. In contrast, when constitutively expressed virG (PN25-virG) was included in the same plasmid (pAW103), virB-lacZ expression was activated by VirAΔR at lower concentrations of AS, and the maximum activity was higher than that in cells containing wild-type VirA expressed from pAW16. In the absence of a virA allele, constitutively expressed virG (pYW47) did not activate virB-lacZ expression. When sugar was included in the induction medium (Fig. (Fig.1B),1B), VirAΔR was capable of activating VirG, but the maximum level of vir gene expression was low compared to that determined by wild-type VirA. In this medium, constitutive expression of virG strongly enhanced the activity of VirAΔR and allowed VirAΔR to induce vir gene expression in the absence of the phenolic inducer. The latter result corresponds to observations that led to characterization of the receiver domain as an inhibitory element (5, 8, 9, 15). Experiments by other workers (8) have shown that VirAΔ712-829 is stable. Our experiments showed that constitutive expression of a FLAG-tagged version of virAΔ707-829 produces a protein with a predicted molecular mass of 78 kDa which is stable in the absence of VirG or AS (see Fig. S1 in the supplemental material). Furthermore, the induction profile of VirAΔ707-829 is similar to that of VirAΔ712-829 (Fig. (Fig.1;1; also, see Fig. S2 in the supplemental material) (8).
The observation that the activity of VirAΔR depended on whether cells carried the PN25-virG construct suggested that signal transduction mediated by VirAΔR might require a higher concentration of VirG than signaling which depends on full-length VirA. Furthermore, the interdependent, autoregulated nature of virA and virG expression (35) suggests that the availability of VirG at the initial stages of induction might be an important factor in the ability of VirAΔR to upregulate both virA and virG expression and hence activate vir gene expression. We compared the amount of VirG produced from a PN25-virG construct with the amount produced from wild-type virG controlled by its native promoter on pTi under noninducing conditions (Fig. (Fig.2A).2A). In this experiment, cells also carried constitutively expressed virA (pAW118) to rule out the possibility of any VirA-derived regulatory effects on VirG accumulation. We confirmed that the PN25-virG construct produced a high basal level of VirG that was not available in cells that carried virG in the Ti plasmid.
We next used virG expressed from the PN25 promoter and regulated by the LacI repressor to examine the relationship between vir gene expression mediated by VirAΔR and the amount of VirG in the cells. Plasmid derivatives of pRG109 (15) that carried PN25-virG, a virB-lacZ fusion, and either wild-type virA (pAW100) or virAΔ707-829 (pAW107) were introduced into strain A136. A136 does not contain the Ti plasmid and, therefore, does not contain the wild-type virA or virG alleles. Transcription from PN25 (which contains the lac operator) is normally constitutive in A. tumefaciens strains which lack the LacI repressor. We added pAW106, which carries the lacIq gene, to control virG expression.
Cells that did not carry lacIq had high levels of VirG, and as expected, the receiver truncation mutant expressed the virB-lacZ fusion at a high level (Fig. (Fig.2B,2B, lanes 1 and 2). The presence of LacI in the cells significantly reduced the levels of both VirG and vir gene expression independent of the version of VirA in the absence of IPTG (lanes 3 and 4). Induction of the cells that contained LacI in the presence of 25 μM IPTG modestly increased the cellular VirG content and increased the vir gene expression mediated by wild-type VirA 6-fold, while vir gene expression that depended on VirAΔR remained suppressed (lanes 5 and 6). Culturing these strains in the presence of 300 μM IPTG further increased the VirG levels and vir gene expression for both strains (lanes 7 and 8). This result indicates that the ability of VirAΔR to activate vir gene expression is relatively poor but can be corrected by sufficiently overexpressing virG.
We replaced wild-type virA in pTi with virAΔR, creating strain AB400, so that we could observe the receiver truncation effect while the natural virA/virG ratio was maintained. In contrast to the multicopy results, vir gene expression was not supported in AB400, regardless of the presence of glucose and high levels of the phenolic inducer. Addition of pYW47 (PN25-virG) to AB400 (Fig. 3A and B) resulted in high levels of vir gene expression, demonstrating that VirAΔR was functional. In this case, we again observed vir gene expression in response to glucose when the phenolic inducer was not present, revealing the inhibitory effect of the VirA receiver.
We next placed virG, with its natural promoter, in a low-copy-number (4 to 6 copies per pTi plasmid) vector which replicates in Agrobacterium using a portion of the origin of replication from the root-inducing plasmid of Agrobacterium rhizogenes (37). Expression of virG from the low-copy-number plasmid (pAW132) was sufficient to activate vir gene expression in AB400 (Fig. (Fig.3C),3C), although the expression in glycerol media was somewhat lower than that in a strain containing full-length wild-type virA in pTi. In media containing glucose (Fig. (Fig.3D),3D), the vir gene induction mediated by virAΔR was similar to that in the wild-type background. However, the induction in response to glucose in the absence of AS was minimal with pAW132 (VirAΔR, 86 U; VirA, 21 U). vir gene expression was supported by full-length virA (pAW52) (Fig. (Fig.3D),3D), confirming the functionality of wild-type virG expressed from the Ti plasmid in AB400.
We tested the ability of VirAΔR to promote tumor formation by cocultivating bacteria with tobacco leaf explants in the absence of acetosyringone (Fig. (Fig.4).4). In the absence of AS, leaf pieces incubated with A348 (wild-type virA and virG in pTi) carrying pAW10 (vector) formed approximately seven tumors per leaf, and the number of tumors per leaf increased to 13 if the strain carried pYW47 (PN25-virG). AB400 (virAΔR and wild-type virG in pTi) carrying pAW10 showed very poor tumor formation, with a mean of less than 1 tumor per leaf piece. AB400/pYW47 exhibited good tumor formation (18.8 tumors per leaf piece). Addition of 3 μM AS to cultivation plates modestly increased tumor formation (about 30%) for the A348 strains, but it did not significantly affect tumor formation by the AB400 strains (data not shown). A348-3 (ΔvirA) carrying either pAW10 or pYW47 did not form tumors regardless of whether AS was included in the cocultivation plates.
Figure Figure55 compares the amounts of VirG present in A348 and AB400 when cells carried the pAW50 vector or pAW132 (virG expressed from its native promoter). In the absence of AS, VirG was not detected in cells that carried pAW50 (Fig. (Fig.5,5, 0 μM AS, lanes 1 and 3), while a low level of VirG was observed in cells that carried pAW132 (0 μM AS, lanes 2 and 4). After addition of 150 μM AS, the amount of VirG increased substantially in the wild-type A348 background with either the vector or pAW132 (Fig. (Fig.5,5, 150 μM AS, lanes 1 and 2). In contrast, significant accumulation of pTi-encoded VirG was not observed in AB400 carrying pAW50 (150 μM AS, lane 3), while the presence of pAW132 in AB400 allowed production of VirG at levels close to the levels seen in cells that carried full-length virA (150 μM AS, lane 4).
The failure of AB400 carrying an empty vector to accumulate VirG under inducing conditions (Fig. (Fig.5,5, lanes 3) suggested that AB400 might be defective in transcriptional induction of virG. To test this hypothesis, we assayed virG transcription in A348, AB400, and A136 (Fig. (Fig.6).6). For this work we used pSW174, a plasmid that carries the P1 and P2 virG promoters and the upstream regulatory region fused to lacZ (35, 36). In the wild-type strain (A348), virG transcription was strongly induced by addition of AS. In AB400 (virAΔR), addition of AS did not increase the β-galactosidase activity above a basal level of approximately 50 U, which was similar to the results obtained for A348 in the absence of AS and for A136, which does not contain virA or virG.
Structural studies of some response regulators have indicated that their activity may be regulated through interdomain contacts between the effector domain (often a DNA-binding domain) and the receiver region that contains a phosphorylatable aspartate (3, 11, 28). Homology between the VirA and VirG receiver domains suggests that if the receiver domain of VirG interacts with its effector domain, then the receiver domain of VirA might also interact with this domain. We used a bacterial two-hybrid system based on the LexA repressor (10, 12) to analyze interactions between the VirG DNA-binding domain and the VirA and VirG receiver domains. As control for specificity, we also looked at the interactions of VirA and VirG domains with the domains of the A. tumefaciens OmpR protein. For this experiment, the receiver portions of virA, virG, and ompR were individually fused to the DNA-binding region of wild-type lexA carried on pSR658, while the DNA-binding domains of virG and ompR were fused to a mutated version of the DNA-binding region of lexA carried on pSR659. Heterodimer formation that brings together the mutant and wild type DNA-binding portions of LexA represses transcription of a LexA-regulated PsulA-lacZ fusion. The presence of the empty “bait” and “prey” vectors sets the unrepressed level of β-galactosidase activity in one strain, while a decline in β-galactosidase activity in cells that express the LexA fusion to VirG's effector domain indicates heterodimer formation between that protein and a LexA-receiver fusion expressed from pSR658. In an initial test of the protein interaction assay we used a leucine zipper (LZ) fusion to the VirA linker domain cloned into both the bait and prey vectors. The level of repression was 99% for the LZ-linker constructs (data not shown), in agreement with the strong self-interaction previously observed for the LZ-linker construct (30).
The bacterial two-hybrid experiment (Fig. (Fig.7)7) indicated that both the VirA and VirG receiver domains can interact with the DNA-binding domain of VirG, and 77 and 66% repression was observed. The VirA receiver and the VirG receiver also exhibited some interaction with the OmpR DNA-binding domain, and the VirA receiver exhibited 36% repression in this case. The OmpR receiver showed very little interaction with the VirG DNA-binding domain (5% repression). The level of the OmpR receiver domain's interaction with its own DNA-binding domain appeared to be rather low (33%) and was about one-half that of the interaction between the VirG receiver domain and the VirG DNA-binding domain. Additional protein interaction assays further suggested that the VirA and VirG receiver domains do not interact with each other and that the VirA kinase domain does not interact with the VirG DNA-binding domain (data not shown).
We demonstrated that the receiver domain of the hybrid histidine kinase, VirA, is an activating factor for vir gene expression in A. tumefaciens. When virG was encoded solely on the Ti plasmid, a form of VirA lacking the receiver domain and expressed from a high-copy-number plasmid (15 to 20 copies per pTi plasmid) exhibited an absolute requirement for the auxiliary sugar signal. The signal-transducing activity of VirAΔR(707-829) in the presence of sugar and a phenolic inducer was generally lower, with a maximum activity between 30 and 50% that of full-length VirA (Fig. (Fig.1).1). In contrast, if the cells included constitutively expressed virG, VirAΔR did not require sugar, but if sugar was present, the protein was active in the absence of a phenolic inducer. The latter result is similar to the findings obtained using a slightly longer version of VirAΔR, VirAΔR(712-829), and led to characterization of the VirA receiver domain as inhibitory (5, 8, 9, 15).
Using a system that allowed us to control the amount of VirG in the cell, we found that activation of vir gene expression by VirAΔR required a higher cellular content of VirG than activation by wild-type VirA required (Fig. (Fig.2).2). This suggests that the abundant VirG available when virG is constitutively expressed facilitates effective interaction between VirAΔR and VirG and demonstrates that the receiver domain of VirA is necessary for vir gene expression when the amount of VirG is limited, as would be the case in the early stages of induction if virG were present solely on the pTi plasmid.
Gene expression and virulence assays indicated that the receiver domain has divergent functions and may activate or inhibit different aspects of the protein's signal-transducing ability. For example, AB400 (virAΔR) carrying an empty vector was unable to induce vir gene expression or form tumors on tobacco explants (Fig. (Fig.33 and and4).4). Yet addition of constitutively expressed virG to AB400 increased both the virulence and vir gene expression compared to the virulence and vir gene expression of a strain that carried wild-type virA. Interpretation of these seemingly contradictory observations is aided by an analysis of A. rhizogenes VirA (13). An in vitro phosphorylation study was performed with a version of the protein purified from E. coli that lacked the periplasmic and linker regions. The authors found that an additional deletion in the receiver domain increased the protein's rate of autophosphorylation. Faster autophosphorylation might be expected to increase the protein's signal-transducing activity. However, it was also shown that the same receiver deletion reduced the rate of phosphotransfer to VirG. Thus, the VirA receiver may regulate phosphotransfer from ATP to the conserved histidine and from the histidine to VirG differently. An increase in the rate of VirAΔR autophosphorylation in the presence of abundant constitutively expressed VirG may account for the high activity of VirAΔR compared to the activity of full-length VirA when the cells also carry PN25-virG (Fig. (Fig.11 and and33).
The failure of AB400 to activate vir gene expression in the absence of added virG (Fig. (Fig.3)3) likely reflects the interdependent and autoregulatory nature of the virA and virG genes. Normally, inducing conditions strongly upregulate expression of both virA and virG in a VirA/VirG-dependent manner (35). Transcriptional induction of virG proceeds from two promoters: the AS-inducible VirA/VirG-dependent promoter (P1) and a second promoter (P2) that produces a low level of virG transcription in response to acidic pH independent of VirA and VirG (35, 36). This basal level of uninduced virG transcription was seen in A348 (wild-type virA) in the absence of AS (Fig. (Fig.6)6) and also in AB400 (virAΔR) and A136 (no virA) whether or not AS was present. A relatively small increase in the virG copy number allowed induction of vir gene expression in AB400/pAW132 (Fig. 3C and D) and accumulation of VirG in the same strain (Fig. (Fig.5,5, 150 μM AS, lane 4). Presumably, the phenotype correction was due to a slight increase in the amount of VirG produced from the VirA/VirG-independent promoter due to the increased copy number. An increase in the amount of VirG determined by P2 would be expected to promote autoregulated induction of both virAΔR and virG from their AS-inducible promoters. The failure of AS to induce further expression of the virG-lacZ transcription in AB400/pSW174 (Fig. (Fig.6)6) revealed that the VirA receiver domain is necessary for induction of virG transcription in response to virulence-inducing conditions.
The two-hybrid experiment (Fig. (Fig.7)7) demonstrated that VirG's DNA-binding domain could interact with its own receiver domain and with that of VirA. The former result is in line with structural studies that have revealed interactions between the receiver and effector domains of other response regulators (3, 7, 11, 28). However, in numerous trials in this experiment, the strongest interaction was consistently seen between the VirA receiver domain and the VirG DNA-binding domain. We expected the two OmpR domains, included in a specificity test, to have stronger self-interaction. However, OmpR homologues may exhibit considerable variation in the extent of surface interaction between the receiver and DNA-binding domains (7, 28), and it is not known how fusion to the LexA DNA-binding domain affects the interaction. A stronger interaction between VirG and OmpR domains might also have been predicted, as the level of sequence identity between the two response regulators is 38%, compared to 21% identity for the VirA and VirG receiver domains. Yet the OmpR receiver domain exhibited significantly less ability to interact with the VirG DNA-binding domain than the VirA receiver domain exhibited (5% repression versus 77% repression).
Nevertheless, the protein interaction data should be interpreted with caution as in vivo interactions between receiver domains and DNA-binding domains may be significantly different due to concentration and localization effects. For example, the fact that the receiver and DNA-binding domains of response regulators are physically linked would be predicted to facilitate any natural tendency to interact by effectively increasing the concentration of the two domains through colocalization (21). In addition, steric interference might affect interactions between individual proteins. Still, the evidence that the VirA receiver domain can interact with the VirG effector domain hints at an explanation for the conditional inefficiency of virAΔR. Activation of VirG likely requires a fairly specific alignment between the VirA kinase region and the VirG receiver domain. In the early stages of induction, upregulation of virA and virG requires interaction between two proteins that are still present in minute amounts (35, 36) (Fig. (Fig.5).5). It is possible that interaction between the VirA receiver domain and the VirG DNA-binding domain increases the occurrence of productive interactions between the VirA kinase region and VirG's receiver domain to promote the key phosphotransfer event that allows VirG to activate vir gene transcription. We cannot rule out the possibility of a phosphorelay function for VirA's receiver domain as an explanation for VirAΔR's diminished capacity to activate vir gene expression when virG is encoded solely on pTi. However, as the results of experiments performed by other workers and us suggest, such a relay is not essential and may not occur (18, 25, 26).
The hypothesis that the VirA receiver domain acts as a recruitment and/or alignment factor for VirG requires further study. This is particularly relevant given that bioinformatic analysis of a large number of bacterial genomes suggested that the number of hybrid histidine kinases is >5-fold greater than the number of identifiable HPT domains (39). Thus, a significant number of these kinases have receiver domains with no obvious function. Recruitment of the response regulator to the cognate kinase is one role that could be considered in examinations of these systems.
This work was funded by grant NIH R01 GM47369 from the National Institutes of Health and by grant NSF 0818613 from the National Science Foundation.
We are grateful to Xiaoquin Lai for assistance with Western blots and to John Zhang and Jinlei Zhou for assistance with plant experiments.
Published ahead of print on 14 January 2010.
†Supplemental material for this article may be found at http://jb.asm.org/.