Mutants of Agrobacterium V1rA That Activate vir Gene Expression in the Absence of the Inducer Acetosyringone"

In the presence of inducer molecules produced by wounded plants, the VlrANirG two-component positive regulatory system of Agrobacten'um tumfaciens ini-tiates transcription of virulence genes required for crown gall tumor formation. Exactly how this system enables the bacterium to respond to an environmental signal is not known, but phosphorylation of VirA and VirG plays a role. To analyze further the function of Vir& we chemically mutagenized the virA gene. Two mutants that activate vir transcription without the plant inducer acetosyringone were found; these mutants alter VirA function by distinct mechanisms. One mutant functions entirely independently of acetosyringone, whereas the activity of the second mutant is enhanced by acetosyringone. Both mutants function best at acid pH, but respond differently to specific monosaccharides that stimulate induction by wild-type Virk Both mutant phenotypes are dominant over wild-type Vir& and both need the conserved histidine at the autophosphorylation site for strong inducer-independent vir transcrip- tion. Protein kinases regulate many diverse biological processes such as cell division, chemotaxis, and differentiation. Protein phosphorylation appears to be a signaling mechanism conserved between prokaryotes and eukaryotes. In prokaryotes, the best characterized protein kinases belong to a family of regulatory systems, termed two-component regulatory sys-tems, in which typically two proteins transform an environmental

In the presence of inducer molecules produced by wounded plants, the VlrANirG two-component positive regulatory system of Agrobacten'um tumfaciens initiates transcription of virulence genes required for crown gall tumor formation. Exactly how this system enables the bacterium to respond to an environmental signal is not known, but phosphorylation of VirA and VirG plays a role. To analyze further the function of Vir& we chemically mutagenized the virA gene. T w o mutants that activate vir transcription without the plant inducer acetosyringone were found; these mutants alter VirA function by distinct mechanisms. One mutant functions entirely independently of acetosyringone, whereas the activity of the second mutant is enhanced by acetosyringone. Both mutants function best at acid pH, but respond differently to specific monosaccharides that stimulate induction by wild-type Virk Both mutant phenotypes are dominant over wild-type Vir& and both need the conserved histidine at the autophosphorylation site for strong inducer-independent vir transcription.
Protein kinases regulate many diverse biological processes such as cell division, chemotaxis, and differentiation. Protein phosphorylation appears to be a signaling mechanism conserved between prokaryotes and eukaryotes. In prokaryotes, the best characterized protein kinases belong to a family of regulatory systems, termed two-component regulatory systems, in which typically two proteins transform a n environmental signal into a metabolic response (1-3). When responding to an environmental cue, the sensor component functions as a protein kinase by autophosphorylating a conserved histidine residue and then transferring a phosphoryl group to an invariant aspartate residue on the second component, the response regulator protein. The response regulator component then activates transcription of other genes to allow the organism to adapt to environmental change.
The VirANirG two-component regulatory system on the tumor-inducing plasmid of Agrobacterium tumefaciens enables this soil bacterium to cause tumors in dicotyledonous plants (4-6). When wounded, plant cells release specific phenolic signal molecules such as acetosyringone (AS)' that induce the * This work was supported by United States Department of Agnculture Grant 90-37262-5291 (to P. C. Z.) and National Institutes of Health Postdoctoral Grant F-32-GM14124 (to B. G. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Tel.: 510-643-9204; Fax: 510-642-4995.
VirANirG system (7, 8); monosaccharides and acid pH in the surrounding environment further potentiate expression by VirA and VirG (8-13). The induction of VkA and VirG leads to transcription of vir genes, and the resulting products generate and transfer a defined segment of DNA from the tumor-inducing plasmid to the plant nuclear genome, transforming the plant cell and leading to tumorous growth (4, 14-16). As the sensor component, VirA autophosphorylates a histidine residue and transfers a phosphate to an aspartate residue of VirG (17-19). The response regulator VirG binds to specific DNA motifs, vir boxes, upstream of vir promoters, and the phosphorylation of VirG by VirA is thought to somehow allow VirG to activate vir gene transcription (20)(21)(22)(23)(24)(25).
By its location in the inner membrane, the 829-amino acid VirA from the octopine-type tumor-inducing plasmid can be divided into two structural regions: a periplasmic region (flanked by two hydrophobic potentially transmembrane domains, TM1 and TM2) and a cytoplasmic region (13,261. Based on sequence analysis and limited mutational analysis, VirA can be further divided into at least four "functional" domains: periplasmic, linker, receiver, and kinase (see Fig. 1) (27). The kinase is the best characterized of the four proposed functional domains and shares homology with histidine protein kinase domains of other sensor components (1-3). In vitro, the cytoplasmic region of VirA that contains the kinase and receiver domains autophosphorylates a histidine in VirA and transfers phosphate to an aspartate in the substrate VirG (17-19).
One strategy to separate and identify the various functions of VirA involves the production and characterization of mutations in VirA that allow constitutive expression of the vir genes. The use of "constitutive" or inducer-independent mutations permits the characterization of VirA function independent of induction by plant phenolics such AS. Consequently, attention is focused on the role of VirA in activating vir gene transcription and not on its role in responding to inducer compounds. Two laboratories have used this approach to isolate mutations in VirA (28,29). However, relative to virA on the tumor-inducing plasmid, these VirA mutants were expressed at a higher copy number. In our system, we maintained virA at 1-2 copiedagrobacterium to mimic virA copy number on a tumor-inducing plasmid. Here, we describe the production and characterization of inducerindependent mutations in VirA, designated VirA"" mutants. We address the effects of various media and environmental conditions on the mutant phenotype and find that the two mutations activate VirA in very different ways.
MATERIALS AND METHODS Plasmids-The plasmid used for mutagenesis of uirA (pMutA) contained the virA and vi& genes, an Ri origin of replication for propagation in Agrobacterium, a ColEl origin of replication for propagation in Escherichia coli, and an fl origin for production of single-strand DNAin E. coli (13,17,27). The locations of the  mutants (G471E and G665D) isolated amino acid substitutions in the VirA"" here are shown.

PERIPLASM
containing octopine uirG was ligated into the XbaI and PstI sites of this 10.9-kbp linear fragment, creating pMutG. Finally, to make pMutA, the 4.5-kbp KpnI fragment containing octopine uirA was inserted at the unique KpnI site of pMutG. The reporter plasmids used to assay uir transcription, pBlacZ and PABZQCZ, each contain a uirB::lacZ translational fusion. pBlacZ is derived from pSM405cd, which contains a uirB::lacZ translational fusion in the carbenicillin-and kanamycin-resistant wide host range plasmid pVCKlO2 (4, 31); pABlacZ is derived from pSM402, which contains uirA, vi&, and a uirB::lacZ translational fusion also in pVCK102 (4, 31). Digestion with BamHI removed a fragment containing carbenicillin resistance from pSM405cd and both carbenicillin resistance and uirG from pSM402; religation formed the reporter plasmids pBlacZ and pA-B h Z , respectively. The reporter plasmids then are maintained by kanamycin resistance, whereas pMutA is maintained by carbenicillin resistance. In addition to the uirB::lacZ translational fusion, pABlacZ also retains wild-type uirA. Each reporter plasmid was electroporated into Agrobacterium A136, a strain lacking a tumor-inducing plasmid, to form either strain Al36(pBlacZ) or A136(pABlacZ). Electroporation stocks of each reporter strain were prepared and stored at -70 "C until use as recipients of chemically mutagenized DNA.
Mutagenesis-A modification of the random chemical mutagenesis method (32) was used to mutagenize uirA. Approximately 2-5 pg of pMutA single-strand DNA was mutagenized with dimethyl sulfate or formic acid. Then, single-strand linear DNA complementary to all of the plasmid except the DNA targeted for mutagenesis (uirA) was annealed to the mutagenized single-strand DNA. In some experiments, the entire uirA coding sequence was left single-stranded; since DNA polymerase has no template to direct repair, the enzyme replaces a chemically mutagenized base nonspecifically. In other experiments, DNA representing either the 5'or 3'-half of uirA was annealed to the treated DNA to direct mutations to the single-strand half of uirA. The partially double-strand plasmids were then electroporated into Agrobacterium strainA136(pBlacZ) either to determine the extent of mutagenesis or to isolate uirA mutations. The cells were plated on ABNES pH 5.5 medium (1 x AB salts, 0.4% glucose, 1-2.5 m~ phosphate, 25-50 n m MES, 40 pg/ml5-bromo-4-chloro-3-indolyl @-galactoside, 50 pdml kanamycin, and 100 pg/ml carbenicillin) and incubated at 28 "C for 2-3 days. In one experiment, Sequenase (United States Biochemical Corp.) was used to fill in the partially double-strand DNA prior to electroporation into Agrobacterium; this step, however, decreased the number of transformants and was used only once.
To assay the extent of mutagenesis, A136(pWacZ+pMutA) cells were plated on ABNES pH 5.5 minimal medium supplemented with 100 p~ AS, and the plates were scored for VirA function by looking for blue colonies after 2-3 days at 28 "C. Of the colonies from dimethyl sulfatetreated DNA, 10% were white, suggesting that 10% of the population had mutation(s) that inactivated VlrA, of the colonies from formic acidtreated DNA, 20% of the population had a mutation(s) that inactivated VirA. To identify mutations that allow constitutive expression of the uirB::lacZ fusion, A136(pBlacZ+pMutA) cells were plated on AB/MES pH 5.5 medium with no AS. After 3 days at 28 "C, blue colonies were purified. pMutA plasmid DNA was isolated from these colonies, transformed into and isolated out ofE. coli, and then electroporated back into Al36(pBlacZ) to confirm that lac2 expression was caused by a mutation in pMutA.
p-Galactosidase Assays-P-Galactosidase activity in liquid cultures was measured according as described (7). Agrobacteria were grown for 48 h in YEB medium (0.5% beef extract, 0.1% yeast extract, 0.5% peptone, 0.5% sucrose, 2 m~ MgSO,) at 28 "C. The cells were then pelleted and resuspended to Am = 0.1 in ABhfES pH 5.5 minimal medium CYTOPLASM containing kanamycin and carbenicillin with or without As; modifications to the medium are described "Results" and in the figure legends. Cells were grown in minimal medium for 14 h at 28 "C and assayed for P-galactosidase activity. All results are averages of at least two independent assays.
Sequencing-Plasmids were sequenced with a Sequenase version 2.0 kit (United States Biochemical Corp.) using double-strand DNA produced in E. coli.

Isolation of VirA Mutations-To identify residues crucial for
VirA activation of vir transcription, the virA gene was chemically mutagenized. Mutations were scored by the ability of VirA protein to function in the absence of inducer AS. Expression of the virB::lacZ fusion in the presence of mutagenized pMutA was assayed on plates containing the chromogenic substrate 5-bromo-4-chloro-3-indolyl p-D-galactoside. In the absence of AS, colonies containing wild-type or unmutagenized virA remain white; blue colonies indicate that the virB promoter is activated in the absence of inducer and suggest that virA is constitutively activating vir transcription. Screening of 50,000 colonies revealed four independent blue colonies. Mutations in DNA from all four colonies mapped to the mutagenized pMutA plasmid. Mutagenized virA and the rest of the plasmid were cloned separately to test for constitutive expression of the reporter. In all cases, only virA caused constitutive expression of the virB::lacZ fusion.
Sequence analysis of the four mutants, designated VirA"" for VirA "onn in the absence of inducer, shows that they localize to 2 different amino acids in the proposed kinase domain of VirA ( Fig. 1). Mutant G471E encodes a glycine to glutamic acid change at amino acid 471,3 residues from the autophosphorylation site of VirA at histidine 474. Using different methods for mutagenesis, Ankenbauer et al. (28) observed the same glycine to glutamic acid change, whereas Pazour et al. (29) found an arginine substituted for this glycine residue. The second mutant isolated from our mutagenesis (mutant G665D) has not been found previously and substitutes aspartic acid for glycine at amino acid 665. Three VirA"" mutants contain this mutation.
The ability of the VirA"" mutants to activate vir transcription was quantified in liquid cultures by measuring p-galactosidase activity of the virB::lacZ reporter fusion in the absence and presence of AS. For initial characterization, cells were induced in ABMES pH 5.5 minimal medium containing 0.4% glucose (Fig. 2). In the absence of AS, the level of virB expression activated by the VirA"" mutants is 900 (G471E) to 1500-fold (G665D) higher than that seen with uninduced wild-type VirA. In the presence of AS, wild-type a n d VirA"" mutants increase expression to the same level. Thus, AS does not seem to hyperinduce vir transcription by either VirA"" mutant.
Interestingly, AS affects expression by G471E, but not by G665D. Since our standard assay reflects p-galactosidase ac- cumulation over 14 h, we cannot detect differences in the kinetics of AS induction. Consequently, cells were grown with or without AS, and aliquots were taken every 2 h to assay for @-galactosidase activity. As shown in Fig. 3, AS has no effect on vir transcription by G665D. Compared with G665D, vir transcription by G471E seems to experience a lag phase, after which G471E cultures with AS initially show a sharper increase in expression than cultures without AS. Following this initial increase, the presence ofAS in G471E cultures results in consistently higher levels of vir transcription. Increased copy number of vi& (5-10 copiedcell) does not prevent the G471E lag phase, indicating that the phenotype of G471E is not a reflection of limiting VirG protein (data not shown). Although G665D and G471E promote similar rates of vir expression, G665D is fully active without AS and appears unresponsive to AS, whereas G471E is partially active and still responsive to AS. Thus, these results (Figs. 2 and 3) suggest that the mutations in G471E and G665D modify VirA in significantly different ways.
Effects of Monosaccharides on Mutant Phenotype-In addition to plant phenolics, vir gene transcription is affected by environmental stimuli such as specific monosaccharides and acid pH (8)(9)(10)(11)15). Several laboratories have shown that in the presence of low levels of phenolics (1-10 p~) , monosaccharides such as glucose enhance vir gene induction, whereas compounds such as sucrose and glycerol do not; at high concentrations of AS (100 p~), monosaccharides play no role in vir induction (9)(10)(11). This effect of specific monosaccharides on vir induction is mediated by VirA and ChvE, a periplasmic sugarbinding protein (9,33).
To examine the effects of monosaccharides on vir gene induction, the VirA"" mutants were assayed in minimal medium containing either glucose or glycerol as the carbon source (Fig.  4A). Just as shown in Fig. 2, in the presence of glucose, G471E activates virB expression without AS, whereas addition of AS increases expression. Interestingly, when 0.5% glycerol is substituted for 0.4% glucose, G471E almost completely reverts back to a wild-type phenotype. With glycerol, G471E resembles wild-type VirA both in the absence ofAS and in the presence of high levels of AS (100 p~); at low concentrations of AS (1 p~) , however, virB expression is 3-fold higher with G471E than with wild-type VirA. The second mutant (G665D) responds differently. With glucose, G665D increases expression of VirB without AS 1500-fold higher than wild-type VirA, and addition ofAS does not significantly increase vir induction. With glycerol and no AS, G665D increases virB expression 12-fold over wild-type VirA; with low AS concentrations, 12-15-fold over wild-type Vir& and with high AS concentrations, 3-fold over wild-type VirA. Both mutants then best activate vir transcription in medium containing glucose. Substitution of glycerol for glucose represses the constitutive phenotype of G471E as well as decreases the maximal induction by wild-type VirA, G471E, and, to a lesser extent, G665D.
To test whether the different carbon sources specifically influence v i r induction versus general cell growth rates, cells were grown without or with 1 p~ AS in minimal medium containing 0.5% glycerol and either 10 m~ sucrose or glucose as metabolic sources (Fig. 4B). With sucrose, both the activity and the response to AS of each mutant resemble those seen with glycerol alone. Thus, in the absence of AS, G471E requires monosaccharides for constitutive expression of the virB fusion, and a t low AS concentrations, it induces virB to a greater extent than wild-type VirA, indicating that the mutant is still more active than wild-type VirA. Unlike G471E, G665D does not require specific monosaccharides for constitutive expression of virB, although monosaccharides enhance expression by increasing the maximal levels of induction. These distinct responses to monosaccharides further illustrate how the mutants differ.
Effects of pH on Mutant Phenotype-Another environmental factor known to influence vir induction is pH (8,12). If the pH of the growth medium exceeds pH 6.0, little induction of vir gene transcription occurs. Using deletion mutants, Chang and Winans (27) suggested that the VirA cytoplasmic linker domain plays an undefined yet required role in sensing both acid environment and phenolic compounds such as AS. The point mutations identified in our mutants do not map to this domain. Still, since the VirA"" mutants function without needing to sense AS, they also may not need an acid pH. Thus, to determine the effects of the mutations on pH sensitivity, minimal media were buffered to either pH 5.5 or 7.0 with BisTris (Fig. 5). At pH 7.0, both mutants expressed virB 2-2.5-fold higher than wild-type VirAin the absence ofAS. In the presence ofAS, wild-type VirA did not induce virB expression at pH 7.0; yet, relative to wildtype VirA, the VirA"" mutants increased expression 2-3-fold with G665D, inducing vir transcription more strongly than G471E. Nevertheless, compared with expression at pH 5.5, overall expression at pH 7.0 was dramatically lower. Thus, both VirA"" mutants function best in an acid environment.
Effects of Wild-type VirA on Mutant Phenotype-% determine if the presence of wild-type VirA altered the phenotype of the VirA"" mutants, the mutants were expressed in A136(pABlacZ) containing both the virB::lacZ fusion and wildtype virA (Fig. 6). @-Galactosidase assays show that wild-type VirA has no significant effect on either VirA"" phenotype. These results differ from those of Pazour et al. (29), who found that the phenotype of their VirA inducer-independent mutants was codoininant with wild-type VirA the presence of a tumor-inducing plasmid encoding wild-type VirA suppressed the constitutive mutant phenotype in @-galactosidase assays. Based on these data, they hypothesized that VirA functions as a dimer. Pazour et al. also found that disruption of virA on the tumorinducing plasmid only partially restored the VirA mutant phenotype, suggesting that other factors on or encoded by the tumor-inducing plasmid affect this phenotype (29). In our assays, the plasmid encoding wild-type virA did not have any tumor-inducing plasmid genes other than a virB::lacZ translational fusion.
Modification of Phosphorylation Site in Mutants-Like all characterized sensors of two-component systems, VirA autophosphorylates a conserved histidine (VirA residue 474) (17,18). The constitutive phenotype of the VirA"" mutants may then result from changes in autophosphorylation at histidine 474 or in phosphotransfer to VirG or possibly by phosphorylation at other sites in VirA. In vitro assays using the cytoplasmic portion of VirA did not demonstrate changes in VirA autophos-This negative result prompted us to compare our strategy with phorylation rates between wild-type and mutant VirA, and at-that used previously to isolate VirG"" mutants (34)(35)(36). For this tempts to assay phosphorylation differences between wild-type comparison, a previously identified VirG"" mutant, N54D, was and mutant VirA contained in Agrobacterium inner membrane constructed using oligonucleotide-directed mutagenesis to reextracts were not successful (data not shown). To determine if place asparagine at codon 54 with aspartic acid; when present the VirA"" mutants G471E and G665D allow phosphorylation on a multicopy plasmid, the N54D mutant causes a VirG"" at amino acids other than the conserved histidine, histidine 474 phenotype, i.e. induction of vir gene transcription in the abwas changed to glutamine, which cannot be phosphorylated sence of VirA and AS (38)(39)(40). Potentially, the low copy number and produces biologically inactive VirA (17). p-Galactosidase (1-2 copiedcell) of pMutG, the plasmid encoding virG, preassays show that histidine 474 is required by both G471E and vented isolation of VirG"" mutants in our system, and indeed, G665D for significant levels of expression of virB (data not pMutG containing the N54D mutant did not exhibit a VirG"" shown). However, both with and without AS, G665D with the phenotype (Fig. 7). In addition to the standard assay using a mutated histidine 474 slightly induces expression, 2-fold over virB::lacZ reporter, the lack of the VirG"" phenotype in N54D at wild-type VirA, suggesting that an alternative phosphorylation low copy number was confirmed by a more sensitive antibiotic site may be inefficiently used in the mutant. challenge using a uirB::neomycin phosphotransferase reporter Use of Low Copy Number Plasmid for Isolation and Charm-plasmid. Thus, N54D in the low copy system appears inactive terization of VirG Mutants-A strategy similar to that used for (Fig. 7); that this mutant is active in the multicopy system may the isolation of VirA"" mutants was applied to the isolation of imply that artificially high levels of N54D are needed to acti-VirG"" mutants. The virG coding region, contained on the vate vir transcription. pMutG plasmid, was chemically mutagenized no VirG"" mutants were found after screening 120,000 colonies, 80,000 DISCUSSION A136(pBZacZ+pMutG) and 40,000 A136(pABZacZ+pMutG).

Agrobacterium VirA Mutants Activate vir Gene Expression
In this study, we isolated two VirA mutants that activate vir transcription in the absence of the plant phenolic inducer AS. Mutant G665D, identified in three isolates, causes vir transcription in the absence of AS at the same high levels as wildtype VirA in the presence of AS. Mutant G471E causes vir transcription in the absence of AS to -50% of maximal induction and is further induced to high levels in the presence ofAS. Thus, these two mutants appear to activate VirA differently.
Effects of Mutations on VirA-Mutant G471E maps 3 residues from the phosphorylation site at histidine 474 and replaces glycine at position 471 with glutamic acid. Without AS, G471E requires monosaccharides for strong constitutive vir transcription; when isolated and assayed by Ankenbauer et al. (28), however, this mutant did not require monosaccharides to promote vir transcription. Differences in monosaccharide, culture medium, and method of analysis may account for these contrasting results. For example, Ankenbauer et al. compared the response of wild-type and mutant VirA in 2.5 p~ AS to varying concentrations of D-glucuronic acid after 24 h; we compared the response of wild-type and mutant VirA in 1 p~ AS to one concentration of glucose (10 m~) after 14 h (28). As reported previously, G471E functions best in an acid environment and induces little vir transcription a t pH 7.0 (28). The second mutation replaces glycine at position 665 with aspartic acid. Unlike G471E, this mutant (G665D) does not require monosaccharides for vir transcription, but activation is enhanced by monosaccharides. Like G471E, G665D functions best at acid pH, but some activation of vir transcription is observed at pH 7.0.
The mutations in the VirA"" mutants may cause a conformational change that mimics the action of AS. Hess et al. (37) postulated that AS may be protonated when bound by its receptor, possibly VirA, resulting in the now-activated phenol protonating a basic site on the receptor to induce a conformational change. This conformation then activates the receptor. Alternatively, if AS does not bind VirA as has been suggested, the AS-binding protein itself may interact with VirA and induce a conformational change (38). Introduction of the negative charges of glutamic acid at position 471 or of aspartic acid at position 665 in the VirA"" mutants may resemble the AS-induced conformation of VirA. Addition of AS to G471E may induce a more appropriate activated conformation and further induce vir transcription. Interestingly, AS does not have a significant effect on G665D. Although amino acid 665 is not in the region presumed to interact with AS, the conformation induced by this negatively charged amino acid may reproduce exactly the AS-induced conformational change.
Potential Effects of Mutations on Phosphorylation-Zn vitro phosphorylation assays did not detect obvious differences between wild-type VirA and either VirA"" mutant. We then used a genetic approach to show the importance of the phosphorylation site at histidine 474 for constitutive vir transcription by the mutants. In the absence of histidine 474, both G471E and wild-type VirA cannot activate vir transcription. However, mutant G665D appears to activate vir transcription at very low levels; this minimal expression may result from weak autophosphorylation activity or from phosphorylation by another kinase in the bacterium. Thus, both VirA"" mutants still require histidine 474 for constitutive activation of vir expression, suggesting that activation of VirA involves both conformational changes and phosphotransfer.
In agrobacteria containing both wild-type VirA and VirA"", the mutant phenotype dominates. Whereas this dominance may simply result from a much higher rate of autophosphorylation or phosphotransfer by VirA"" to VirG, phosphotransfer between mutant and wild-type VirA molecules also may occur. This mechanism may be analogous to transphosphorylation between shortened CheA and kinase-deficient CheA variants that allows the variants to phosphorylate CheY, a receiver response component (39). Potentially, then, transphosphorylation of wild-type VirA by the VirA"" mutants may activate both species of VirA to maintain the AS-independent phenotype. Our results differ from those of Pazour et al. (29), who found that wild-type VirA suppressed the phenotype of their VirA"" mutants. They proposed that VirA functions as a dimer and that components on the tumor-inducing plasmid further influence VirA activity. Our data neither support nor contradict the hypothesis that active VirA functions as a dimer. For instance, if VirA is a dimer, the dominance of our mutant phenotypes suggests that the mutants can transactivate wild-type VirA, alternatively, lack of inhibition of the mutant phenotype by wildtype VirA is consistent with the protein functioning as a monomer.
Effects of Copy Number on Mutagenesis-As noted, the characteristics of the VirA"" mutants identified in this study sometimes differ from those of VirAon mutants isolated by other laboratories. Unlike the multicopy plasmids used by other laboratories for mutagenesis, the copy number of the plasmid encoding our VirA"" mutants mimics the copy number of the tumor-inducing plasmid. How the VirAon mutants function in our system should then mirror how the mutants would function encoded on a tumor-inducing plasmid. When this mutagenesis strategy was applied to VirG on the same copy number plasmid, no mutations were found that allowed VirG to function without VirA and AS, although such mutations have been found using a multicopy plasmid approach (38)(39)(40). Characterization of a VirG mutation previously identified on a multicopy plasmid further illustrates the differences between approaches. When this VirG mutant (N54D) is encoded on a multicopy plasmid, it functions in the absence of VirA and AS (34)(35)(36). When expressed at 1-2 copiedcell, however, this mutant does not function even when a more sensitive antibiotic selection is substituted for the p-galactosidase screen (data not shown). This result demonstrates an important contrast between our mutagenesis approach and those used previously and confirms the importance of copy number for transcriptional activation by the VirANirG system.
Location of Mutations Near Conserved Sensor Domains-The VirA mutations in G471E and G665D map near two of the five short blocks of common sequence found variably spaced among sensor proteins (1,2). Of the five common motifs, the most variable is a 9-amino acid block (block H (VirA amino acids 472-480)) that includes the histidine autophosphorylation site (1,2). Found just outside block H is the site of G471E; this glycine in wild-type VirA is conserved among a number of sensor proteins (2). That a mutation at this site has been produced by three laboratories using three mutagenesis methods strongly implies that this site is important for VirA function (28,29). Exclusive of the conserved histidine, mutations at residues in block H in other sensor components show that the local structure around the histidine influences phosphorylatioddephosphorylation activities (40,41). In the same manner, the mutation at glycine 471, just outside of block H, may activate VirA by altering the local structure around histidine 474. Two other conserved domains in the sensor components, blocks G1 (VirA amino acids 627-635) and G2 (VirA amino acids 655-6601, resemble glycine-rich regions of a nucleotide-binding site and may function as separate nucleotide-binding domains (1,2). In sensor proteins such as EnvZ, mutations in these two conserved motifs destroy autophosphorylase activity (42,43). Mutant G665D maps just outside of block G2, and an inducer-independent mutant isolated by Pazour et al. (29) maps in the middle of block G2 at amino acid 658. Parkinson and Kofoid (1) suggested that this region could be a functional analog of the cGMP-binding site of the cGMPdependent protein kinases; this binding site helps modulate protein conformation to activate the kinase. Identification of two mutations in this region suggests that it too plays an important role in VirA activation of vir gene transcription.
As mentioned, AS affects G471E and G665D differently, suggesting that the two mutations activate VirA by two distinct mechanisms. To activate wild-type VirA, AS is proposed to somehow cause a conformational change in VirA (29,37). This activated conformation may bind nucleotides at the proposed binding sites in blocks G1 and G2, initiating phosphotransfer to VirG and derepressing potential inhibition by the VirA receiver domain. Rather than being directly involved in phosphotransfer, the nucleotide bound at block G2 may simply modulate VirA conformation to activate transphosphorylation either to VirG and/or within VirA to derepress potential autoinhibition. Since its mutation maps near block G2, the conformation of G665D may directly mimic AS-induced nucleotide binding at block G2; consequently, addition of AS does not influence the activity of G665D. However, with its mutation mapping near the autophosphorylation site, the conformation of G471E somehow transfers phosphate to VirG, but lacks the potential transphosphorylation and/or derepression necessary to fully activate VirA. Addition of AS to G471E may then induce nucleotide binding at block G2, leading to complete activation of VirA.
The mapping of both mutants (G471E and G665D) to two regions conserved among sensors demonstrates the importance of these regions for the interaction between VkA and VirG. The identification of mutations at or near these amino acids by other laboratories re-enforces that these two domains are crucial for VirA function. Notably, the response of the two VirA"" mutants (G471E and G665D) to the inducer AS differs significantly. Consequently, the mutants isolated from this study can be used to analyze two major aspects of vir expression. First, by separating environmental sensing from transcriptional activation, the mutants focus attention on interactions of the sensor VirA with both its own domains and its response regulator (VirG). Second, based on their very different responses to the inducer AS, the mutants provide tools for analyzing the specific role ofAS in VirA activation, a complex and not well understood process that gives Agrobacteriurn its ability to exploit wounded plant cells.