Proc. Natl. Acad. Sci. USA Vol. 88, pp. 6941-6945, August 1991 Genetics
Mutants of Agrobacterium tumefaciens with elevated vir gene expression (vir gene regulation/mutagenesis/virA mutants)
GREGORY J. PAZOUR, CHRISTOPHER N. TA, AND ANATH DAS Department of Biochemistry and Plant Molecular Genetics Institute, University of Minnesota, 1479 Gortner Avenue, St. Paul, MN 55108
Communicated by Mary-Dell Chilton, May 1, 1991 (received for review March 13, 1991)
ABSTRACT Expression of Agrobacterium tumefaciens virulence (vir) genes requires virA, virG, and a plant-derived inducing compound such as acetosyringone. To identify the critical functional domains of virA and virG, a mutational approach was used. Agrobacterium A136 harboring plasmid pGP159, which contains virA, virG, and a reporter virB: acZ gene fusion, was mutagenized with UV light or nitrosoguanidine. Survivors that formed blue colonies on a plate containing 5-bromo-4-chloro-3-indolyl fi-D-galactoside were isolated and analyzed. Quantification of 8-galactosidase activity in liquid assays identified nine mutant strains. By plasmid reconstruction and other procedures, all mutations mapped to the virA locus. These mutations caused an 11- to 560-fold increase in the vegetative level of virB: acZ reporter gene expression. DNA sequence analysis showed that the mutations are located in four regions of VirA: transmembrane domain one, the active site, a glycine-rich region with homology to ATP-binding sites, and a region at the C terminus that has homology to the N terminus of VirG. The plant pathogen Agrobacterium tumefaciens is the causative agent of crown gall tumors. The bacteria can live either as a free-living soil organism or as an opportunistic colonist of crown gall tumors. The bacteria create a crown gall niche on susceptible plants by transforming the plant cells with genes that direct the biosynthesis of phytohormones, auxin and cytokinin, and that of derivatives of basic amino acids and sugars called opines. The imbalance of hormones produces a tumor in which the bacteria live and metabolize the opines as sources of carbon, nitrogen, and energy. The genes for synthesis of the plant hormones and opines are located on a piece of DNA, the T-DNA, that is transferred from the bacterium to the plant cell, where it becomes integrated into the nuclear genome (reviewed in refs. 1 and 2). The vir genes of the bacteria mediate the processing and transfer of the T-DNA molecule from the bacterium to the plant. These genes are organized into eight operons, virA-H, that are controlled as a regulon (3-7). When the bacteria are in the free-living vegetative state, only virA and virG are expressed. Upon encountering a susceptible plant, all of the vir genes, including virA and virG, are induced (8, 9). This leads to the synthesis of proteins required for processing and transfer of the T-DNA to the plant cell. Induction of the vir regulon requires VirA, VirG, and plant phenolic compounds such as acetosyringone (AS) (8, 10). VirA and VirG are members of the family of prokaryotic two-component regulatory systems (11, 12). In twocomponent regulatory systems, one component senses the environment and then signals the second component to carry out a function (reviewed in refs. 13 and 14). VirA is thought to be the primary sensor of the plant phenolics (12). It The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
traverses the bacterial inner membrane (15, 16) and has autokinase (17, 18) and VirG phosphorylase activities (19). It is hypothesized that interaction of the phenolics with VirA activates VirA's autokinase activity. This creates VirA-P, which can then transfer the phosphate moiety to VirG. VirG-P is hypothesized to be the activated form of VirG. VirG is a sequence-specific DNA-binding protein (20, 21) and is thought to function as a transcriptional activator. The mechanism of transcriptional activation by VirG is not known. One homologue of VirG, NtrC, functions to catalyze the conversion of a closed to an open complex of RNA polymerase and promoter DNA (22). Another VirG homologue, OmpR, has been hypothesized to activate transcription simply by stabilizing the formation of the RNA polymerase-promoter binary complex (23). In addition to the plant phenolic class of inducers, several other compounds and conditions enhance vir gene induction. Monosaccharides such as glucose and galactose enhance vir gene induction when the levels of phenolic vir inducer are low. This response is mediated through the galactose binding protein ChvE. It is thought that ChvE binds the monosaccharides and interacts with the periplasmic domain of VirA to make VirA more sensitive to the phenolic inducers (24-27). Opines also enhance vir gene induction, but the mechanism by which this occurs is unknown (28). Optimum vir gene induction occurs at pH values below 5.7 (8). This effect is probably mediated through an increase in the amount of VirG in the cell as well as through changes in VirA structure. Transcription of virG is induced by acidic conditions (29); therefore, under these conditions more VirG is expected to be present, leading to increased vir gene induction. In addition to VirG, VirA is also affected by pH. Melchers et al. (15) showed that deletion of the periplasmic domain of VirA leads to reduced pH dependence. This suggests that the external (periplasmic) pH brings about conformational changes in VirA that allow it to respond to phenolic inducers.
MATERIALS AND METHODS Bacterial Strains and Plasmids. Plasmids pGP159 (30) and pGP234 contain the virA and virG genes and a virB: lacZ reporter gene on the wide host range cloning vector pTJS75 (31). Plasmid pGP234 was constructed by fusion of plasmids pGP220 and pGP228 at their unique EcoRI sites. Plasmid pGP220 was constructed by cloning a 1.2-kilobase Kpn I fragment containing virG into the unique Kpn I site of pAD1092K (30). Plasmid pGP228 is a pUC119 (32) derivative that contains the pTiA6 Kpn 11 fragment, which carries virA. To construct plasmid pGP336, pGP228 was digested with Pst I. After agarose gel electrophoresis, two fragments, 5 and 2.1 kilobases, were isolated by electroelution. These fragments contain pUC119 with the virA promoter and the coding Abbreviations: AS, acetosyringone; NG, nitrosoguanidine; P-gal, ,6-galactosidase; VirAcOn, constitutive virA mutants; T-DNA, trans-
ferred DNA.
6941
6942
Proc. Natl. Acad. Sci. USA 88 (1991)
Genetics: Pazour et al.
region for the C-terminal end of VirA, respectively. These fragments were ligated to construct pGP335; pGP335 was then fused to pGP220 to construct pGP336. This construct produces a protein identical to VirAA63/240 of Cangelosi et al. (25) and similar to that produced by pRAL7044 (pRAL7044 was constructed with virA from pTilS955) of Melchers et al. (15). The NG31 mutation was added to VirAA63/240 by site-specific mutagenesis (33) of pGP335 followed by fusion with pGP220 and was designated pGP348. Substitution mutations of virA were constructed by sitespecific mutagenesis of pGP228 and were identified by DNA sequence analysis (34). The mutant pGP228 derivatives were later fused to pGP220 to construct pGP234 derivatives. Agrobacterium strains used in this study were A136 (which lacks a Ti plasmid), its isogenic derivatives A348 (which carries the Ti plasmid pTiA6), A723 (which carries the Ti plasmid pTiB6806), and A1030 (which carries the Ti plasmid pTiB6806virA:: TnS). Mutagenesis. UV light mutagenesis was performed by exposing Agrobacterium A136/pGP159 cells plated on ABMes (pH 5.5) solid medium [lx AB salts (35), 0.2% glucose, 25 mM phosphate, and 50 mM 2-(N-morpholino)ethanesulfonic acid (Mes)] containing 5-bromo4chloro-3indolyl ,B-D-galactoside (16 ,ug/ml), tetracycline (10 ,ug/ml), and carbenicillin (50 ,ug/ml) to UV light. Cells were mutagenized to achieve a 90%o kill of the plated cells. Survivors were allowed to grow into colonies, and the blue colonies were purified and assayed for ,B-galactosidase (,(-gal) activity. Nitrosoguanidine (NG) mutagenesis was performed by growing cells (OD6w = 0.2) in ABMes liquid medium (pH 5.5) containing NG at 0.13 mg/ml for 60 min at 30°C (36). Mutagenesis was stopped by centrifugation of the cells, followed by washing in AB medium (pH 7.0) and resuspension in the same medium. Under these conditions, a 75% kill of the cells was observed. Mutated cells were plated on ABMes plates (pH 5.5) containing 5-bromo-4-chloro-3indolyl 8-D-galactoside, tetracycline, and carbenicillin to allow phenotypic expression. Blue colonies were purified and assayed for 8-gal activity. Other Procedures. (-gal activity was measured according to Miller (36). Cells were grown overnight in AB medium (pH 7.0), diluted 1:20 in ABMes medium (pH 5.5 or 6.8) containing tetracycline and carbenicillin with or without AS at 100 ,uM, and grown for 24 hr at 28°C. The pH of these media were approximately 5.0 and 6.4 at the time of assay. All results shown are averages of two independent assays.
RESULTS Selection of Mutants with Elevated Vegetative Levels of vir Gene Expression. Expression of A. tumefaciens vir genes is positively regulated by VirA and VirG in response to plant phenolics such as AS (8, 10). A mutant screen was undertaken to isolate strains of bacteria that exhibit elevated vegetative levels of vir gene expression. These mutants are expected to help us identify critical functional domains and/or amino acid residues of the proteins. This screen made use of a virB: lacZ gene fusion as a reporter gene to identify strains with increased virB expression. Mutants exhibiting elevated virB expression will form blue colonies on 5-bromo4-chloro-3-indolyl ,3-D-galactoside-containing plates while wild-type colonies will be white. Ti-plasmidless A. tumefaciens A136 containing plasmid pGP159 was used in this screen. Plasmid pGP159 carries pTiA6 virA and virG genes as well as the reporter virB: lacZ gene. Examination of 100,000 survivors from UV light-induced mutagenesis yielded 1 blue colony, whereas =300 blue colonies were obtained from examination of 500,000 survivors from NG mutagenesis. The single blue colony from UV light-induced mutagenesis and 70 blue colonies from NG
mutagenesis were assayed for 83-gal activity after growth in liquid culture lacking AS. The majority of these strains showed no elevation of (3-gal activity and were not pursued further. However, nine strains including the one from UV light-induced mutagenesis had elevated levels of P-gal activity. To determine whether the mutations were plasmid borne, plasmid DNA was isolated from the mutant strains and was reintroduced into unmutagenized Agrobacterium. In all cases the mutation mapped on pGP159, indicating that the mutations were in virA, virG, or the virB promoter-regulatory region. To determine which of these genes was altered, the individual genes were subcloned, combined with unmutagenized components, and reintroduced into Agrobacterium. In all cases the mutation was localized within virA. The DNA sequence of each of these constitutive virA mutants (VirAcon) was determined to locate the site(s) of mutation. Each mutant, except NG10, had a single base change within the coding region of virA. NG10 had two base changes, but one of these was a silent change. The locations of the mutations and the resultant amino acid substitutions are shown in Fig. 1. Four strains (NG1, NG8, NG34, and NG37) had an identical base change, which led to a leucine to phenylalanine substitution at codon 24. This residue lies within the first putative transmembrane domain (residues 18-39). Two of the other virA mutations mapped very close to the active site histidine (residue 474). One of these, UV3, had a glycine to arginine substitution at position 471, whereas the other, NG53, had an alanine to valine substitution at position 469. Mutant NG31 had a leucine to phenylalanine substitution at codon 658. This residue is located within a glycine-rich region which is conserved among all the histidine kinase members of two-component regulatory systems (region III; ref. 14). This region is homologous to the 40-amino-acid motif GXGX2GXK that is found in protein kinases and is thought to make up the nucleotide-binding site (37). In VirA this sequence reads GGTGLGX35K, which is altered to GGTGEGX35K in NG31. Mutants NG9 and NG10 have an alanine to valine substitution at position 735. This is the most C-terminal of all the mutations and is located within a region of VirA homologous to VirG and other response regulator members of twocomponent regulatory systems (labeled RR in Fig. 1). This mutation changes a nonconserved residue within an a-helical region, designated aA by Stock et al. (14). VirAc Mutants Exhibit 11- to 560-Fold Increases in the Vegetative Level of virB Expression. The phenotypes of the VirAll mutants were quantitated by assaying (3-gal activity in the presence and absence of the vir-inducing compound AS (Table 1). When compared with wild type (pGP234), the increase in virB: lacZ expression in the absence of AS ranged from 11-fold (UV3) to 562-fold (NG9). The expression level TMI
TM2
NG1, NG8, NG34, NG37 (L24F)
His-P (H474)
UV3 (647 1 R)
NG53
RR N91, NG1O (A735V) 'NG31 (L658F)
(A469V)
FIG. 1. VirA structure and location of mutations. The location of the mutations isolated in this study is shown on a schematic view of VirA along with the amino acid substitutions they cause. Transmembrane domain 1 (TM1), transmembrane domain 2 (TM2), response regulator homologous domain (RR), and the site of autophosphorylation (histidine-474) are also shown. One-letter amino acid codes indicate amino acid substitutions. Residue 1 is the methionine codon of Leroux et al. (11).
Genetics: Pazour et al. Table 1. Characterization of VirA mutations at pH 5.5 virB: IacZ expression, ,8-gal units Plasmid/strain - AS + AS pGP234/A136 9 4170 UV3/A136 103 4188 NG1/A136 1560 13000 NG9/A136 5060 16500 NG31/A136 419 9500 NG53/A136 1270 9270
observed in NG9 (5060 units) is higher than that observed in wild type (pGP234) induced by AS (4170 units). All of the mutants could be further induced by AS; all (except UV3) exhibited synergistic effects of AS and the mutation. Several VirA' Mutants Are Codominant to VirA. If a VirAcon mutant encodes a receptor locked in an activated configuration, such a mutant should be dominant to wild-type VirA. To examine dominance of the VirAcon mutants, we introduced the virA mutant plasmids into A. tumefaciens A348 (harbors Ti plasmid pTiA6) and analyzed their phenotypes (Table 2). The mutants showed a 2- to 7-fold lower level of uninduced virB expression when a Ti plasmid was present in the host bacterium (- AS column,-Table 1 vs. Table 2, experiment 1). To determine whether virA on the Ti plasmid was responsible for the phenotypic suppression of the mutant phenotype, we introduced the NG9 mutation into A. tumefaciens A1030, a strain bearing a virA:: Tn5 insertion, and its isogenic parent A723. Results presented in Table 2, experiment 2, show that the presence of the Ti plasmid in A723 caused phenotypic suppression of the NG9 mutation [a 5-fold decrease compared to that in the absence of a Ti plasmid (A136)]. This suppression was partially relieved by the disruption of the virA gene in A1030. These results indicate that the presence of a wild-type copy of virA can affect the phenotype of the VirAcwn mutants, but other factors on the Ti plasmid may also be involved. VirA' Phenotype Is pH and Temperature Dependent. Induction of the vir genes is pH dependent (8); very little induction is usually observed if the pH of the growth medium exceeds 6.0. A possible explanation for this phenomenon is that lower pH is required for interaction of the vir-inducing Table 2. Effect of the presence of a Ti plasmid on the VirAcOn phenotype at pH 5.5 virB: lacZ expression, p-gal units Plasmid/strain - AS + AS Experiment I pGP234/A348 4 3,380 UV3/A348 35 3,520 NG1/A348 482 4,370 NG9/A348 753 4,290 NG31/A348 175 4,570 NG53/A348 706 5,530 Experiment 2 pGP234/A136 9 4,170 NG9/A136 5060 16,500 pGP234/A723 7 4,530 NG9/A723 1080 4,190 pGP234/A1030 6 3,610 NG9/A1030 2030 3,979 A348, A723, and A1030 contain the Ti plasmids pTiA6, pTiB6806, and pTiB6806virA::Tn5, respectively.
Proc. Natl. Acad. Sci. USA 88 (1991)
6943
Table 3. Characterization of VirA mutations at pH 6.8 virB: lacZ
expression, pl-gal units
Plasmid/strain pGP234/A136 UV3/A136 NG1/A136 NG9/A136 NG31/A136 NG53/A136
- AS
+ AS
12 16 80 69 41 54
1150 490 3130 1460 2160 1940
compound with VirA. If this is the only reason for a reduced pH requirement, then the VirAcon phenotype should be expressed at pH 6.8. This, however, was not found to be true (Table 3). The mutants exhibited only a small (7-fold maximum) increase in virB: lacZ expression over wild type in the absence of AS at pH 6.8. In the presence of AS at pH 6.8, the level of expression of the mutants (except UV3) was comparable to that of the wild type, a level 3- to 4-fold lower than that observed at pH 5.5. Previous studies that demonstrated an absolute pH dependence of vir gene expression used the single-copy Ti plasmid as the source of vir genes (8). When we used an RK2-derived plasmid containing virA and virG, significant induction of the vir genes was observed at pH 6.8 (Table 4). RK2-derived plasmids are multicopy in Agrobacterium; thus, it appears that an increase in copy number can partially relieve the pH dependence of vir gene induction. Similar results have been reported by Melchers et al. (15). Melchers et al. (15) reported that deletion of a portion of the periplasmic domain of VirA leads to reduced pH depen-
dence. We sought to determine whether we could obtain a pH-independent VirAcon phenotype by combining a similar
deletion with the NG31 mutation. The results presented in Table 4 confirm that deletion of residues 63-240 in the periplasmic domain of VirA leads to the relief of pH dependence of vir gene induction. The level of AS-induced virB expression by VirAA63/240 at pH 6.8 was comparable to that at pH 5.5. Combination of VirAA63/240 with the NG31 mutation yields a VirAcon phenotype at either pH. To study the effect of temperature, cells were grown at 30°C, 33°C, or 37°C. In conformity with the studies of Melchers et al. (15), we observed that strains containing wild-type virA are inducible at 30°C or 33°C but not at 37°C. However, the mutant phenotypes were manifested only at 30°C; a marginal increase in induction (2- to 5-fold over wild type) was observed at 33°C or 37°C (data not shown). Effect of Amino Acid Substitutions in the First Hydrophobic Domain of VirA. One of the VirAcOn mutants, NG1, mapped within the first hydrophobic domain of VirA. This result was unexpected in view of the fact that Melchers et al. (15) earlier reported that this segment of the protein could be replaced by Table 4. Effect of pH on VirAcon phenotype virB: lacZ expression, (3-gal units pH5.5 pH 6.8 Plasmid Genotype - AS + AS - AS + AS pGP234 Wild type 9 4170 11 1150 NG31 VirANG31 419 9500 41 2160 pGP336 VirAA63/240 25 5840 14 5110 pGP348 VirAA63/240,NG31 230 9550 290 7460 P-gal activity of Agrobacterium A136 harboring the indicated plasmid was assayed at pH 5.5 or 6.8 with or without 100 ,uM AS.
Genetics: Pazour et al.
6944
Proc. Natl. Acad. Sci. USA 88 (1991)
analogous but nonhomologous region of the Escherichia coli aspartic acid receptor (Tar). These observations led the authors to believe that this segment of the protein was probably not playing a specific role in the detection of the vir-inducing compounds. To gain an understanding of the role of the NG1 mutation in the signaling process, we examined the phenotype of mutants with substitutions of different amino acids at the NG1 position. NG1 causes a leucine to phenylalanine substitution at residue 24 (Fig. 1). The effects of different substitutions at position 24 on the signaling process are summarized in Fig. 2. These mutations were introduced in the wild-type gene by site-specific mutagenesis according to Kunkel (33). Alteration of leucine to phenylalanine is a re-creation of the NG1 mutation, and the mutant exhibits the VirAcon phenotype as observed earlier (Table 1). A substitution of isoleucine or aspartic acid at residue 24 showed a marginal VirAcon phenotype (3- to 4-fold over wild type), whereas all other substitutions had no effect on the phenotype. Interestingly, mutation of this residue to tyrosine, a close homologue of phenylalanine, did not alter the phenotype from wild type. None of the substitutions abolished VirA activity. These substitutions included such residues as proline and the charged residues aspartic acid, lysine, arginine, and histidine. Mutational Analysis of the Response Regulator Domain of VirA. The C-terminal end of VirA is homologous to CheY and the N terminus of VirG (14). The aspartic acid residue that is phosphorylated in VirG and CheY is conserved in this domain of VirA. To determine what effect mutation of this residue would have on vir gene induction, the aspartic acid residue at position 766 was replaced with asparagine (VirAD766N). This substitution causes a 2-fold increase in the basal level and a 5-fold decrease in the induced level of virB expression (Table 5). This indicates that intramolecular phosphotransfer is probably not required for vir gene induction but does not rule out other unknown pathways.
an
DISCUSSION In this study we identified mutations in the virA structural gene that cause vir gene expression in the absence of virinducing compounds. Under noninducing conditions these mutant proteins function similarly to the wild-type protein in the presence of the inducer, indicating that they may exist in conformations that mimic the structure of VirA bound to the vir inducer. These mutations map in four regions of the 1
5
c:lo
0) 0
a)
No
1
Leu Phe Arg His Ser Asp Lys ProThr Gly WT NG1
TyrPhe Met Val lle
FIG. 2. Effect of substitution at the leucine-24 codon on virA activity. virB: lacZ expression was measured in Agrobacterium A136 at pH 5.5. Black bars represent activity in the absence of inducer, and the open bars represent activity in the presence of 100 ,uM AS. Bar 1 (leucine) is wild type (WT), bar 2 is the NG1 mutant, the other phenylalanine (bar 12) is a re-creation of this mutation.
Table 5. Mutational analysis of the response regulator domain in VirA virB: lacZ expression, /3-gal units - AS + AS Genotype Plasmid 9 4170 Wild type pGP234 20 800 VirAD766N pGP377 harboring the indicated activity of Agrobacterium A136 ,p-gal plasmid was assayed at pH 5.5 with or without 100 ,uM AS.
protein: transmembrane domain one, the active site, a glycine-rich region homologous to nucleotide-binding sites, and a region homologous to CheY and the N terminus of VirG. VirA has two hydrophobic domains, which have been proposed to be membrane spanning (TM1 and TM2; Fig. 1). One VirAc`n mutant, NG1, mapped to a site within the TM1 region. Previous studies showed that this region could be replaced by an analogous but nonhomologous region of the E. coli Tar receptor (15). Thus it was thought that this domain could not be playing a major role in the sensing ofthe phenolic compounds. Its role in the transmembrane signaling process is less clear. Genetic studies imply that in the Tar receptor this region (TM1) interacts with the second transmembrane domain and with a cytoplasmic domain. It has been proposed that these interactions are involved in the transmembrane signaling process (38). It is possible that the TM1 domain of VirA also functions in a similar manner, which would indicate that in the Tar-VirA chimeric protein the Tar sequence is able to provide the transmembrane signaling. Two amino acid substitutions near the active site histidine caused the VirAc`n phenotype. These mutations are likely to change the active site conformation to mimic the activated state. An additional mutation maps to a glycine-rich region C-terminal to the active site. This region is conserved in all histidine kinase members of the two-component regulatory systems. It bears homology to the glycine-rich sequence that make up nucleotide-binding sites of many protein kinases. If this site is a nucleotide-binding site, then it must be folded in close to the active site. The final region in which a mutation was found was in a domain at the C terminus of VirA, which has homology to CheY and the N terminus of VirG. This domain is unusual in histidine kinases. FrzE, which also has this structure, is thought to have both histidine kinase and response regulator functions in the same polypeptide (39). FrzE autophosphorylates and the chemical stability of the phosphate moiety resembles a phosphoaspartic acid, indicating that an intramolecular transfer of the phosphate from the active site histidine to the aspartic acid may occur in this case. It is not known whether this type of intramolecular transfer occurs in VirA. Studies ofJin et al. (17) indicated that phosphohistidine was present in VirA-P, but these studies could not rule out the possibility that phosphoaspartic acid was also present. In the present study we show that mutation of aspartic acid at residue 766 in VirA, homologous to the aspartic acid residue that is phosphorylated in CheY (40) and VirG (19), caused a small increase in the basal level and a 5-fold decrease in the induced level of vir gene expression. These results indicate that an intramolecular signaling pathway is probably not essential for vir gene induction, although it may be important for some unknown additional pathway. The presence of a domain homologous to VirG and CheY in the C terminus of VirA is reminiscent of the "autoinhibitory" domains of eukaryotic protein kinases (37). It has been established that the low basal level of many of these kinases is due to inhibitory interactions between the active site and a domain elsewhere in the kinase that has homology to the
Genetics: Pazour et al. substrate. It is hypothesized that the allosteric activators of these kinases cause a conformational change that disrupts the interaction between the regulatory domain and the active site and allows for productive interaction between the active site and the substrate. It is possible that this may be the role played by this domain in VirA; binding of AS causes a conformational change that allows VirG access to the active site so that it can become phosphorylated and activated. Alternatively, other signals such as low pH, sugars, or opines could be causing the conformational change to enhance the activation of VirA and work synergistically with AS. The VirAcon mutants isolated in this study show codominance with VirA. It would be expected that if these mutants were simply receptors locked in the activated configuration, then they should be dominant to wild type. There are several possibilities to explain the suppression of phenotype when wild-type VirA is present. (i) There are a limited number of sites in the membrane that VirA can occupy. Since both wild type and VirAcon can compete for these sites, less VirAcon will be present in the membrane and the phenotype will be suppressed. (ii) The presence of a Ti plasmid, on which the wild-type copy is located, is contributing a large number of VirG binding sites such that less activated VirG is available to induce expression of the reporter gene, leading to phenotypic suppression. (iii) VirA in the unactivated state can bind VirG and remove it from the pool making it unavailable for activation by VirAcon. (iv) VirA is multimeric and heteromultimers of VirA and VirAcon have wild-type or intermediate phenotypes, thus causing phenotypic suppression. Disruption of virA on the Ti plasmid by a Tn5 insertion partially relieves the phenotypic suppression, indicating that codominance is caused by both the presence of the protein and an additional factor(s) on the Ti plasmid. These mutants should prove useful in further studies on the vir gene induction pathway, such as in determining the step at which noninducing chromosomal mutations are blocked. In addition, if the model that phosphorylation is the activating step is correct, then these mutants should have elevated rates of VirG phosphorylation. At present we have not been able to detect any alteration in phosphorylation using the cytoplasmic portion of VirA as the kinase (G.J.P., unpublished results). These mutants may also prove useful in the transformation of recalcitrant plant species because they do not need vir-inducing compounds and because they hyperinduce in the presence of AS. Note Added in Proof. In a related study Hess et al. (41) report a detailed analysis of the interaction between VirA and AS.
We thank Ann Vogel and John Larkin for a critical reading of the manuscript. This work was supported by grants from the National Institutes of Health (GM 37555), the American Cancer Society (JFRA 170), and a McKnight Land Grant Professorship Award from the University of Minnesota. G.J.P. is a Graduate School Dissertation Fellow of the University of Minnesota. 1. 2. 3. 4.
Zambryski, P. (1988) Annu. Rev. Genet. 22, 1-30. Ream, W. (1989) Annu. Rev. Phytopathol. 27, 583-618. Garfinkel, D. & Nester, E. (1980) J. Bacteriol. 144, 732-743. Ooms, G., Kalpwijk, P., Poulis, J. & Schilperoort, R. (1980) J. Bacteriol. 144, 82-91. 5. Stachel, S. & Nester, E. (1986) EMBO J. 5, 1445-1454. 6. Kanemoto, R., Powell, A., Akiyosh, D., Regier, D., Kerstetter,
Proc. Natl. Acad. Sci. USA 88 (1991)
6945
R., Nester, E., Hawes, M. & Gordon, M. (1989) J. Bacteriol. 171, 2506-2512. 7. Melchers, L., Maroney, M., Dulk-Ras, A., Thompson, D., van Vuuren, H., Schilperoort, R. & Hooykaas, P. (1990) Plant Mol. Biol. 14, 249-259. 8. Stachel, S. & Zambryski, P. (1986) Cell 46, 325-333. 9. Winans, S., Kerstetter, R. & Nester, E. (1988) J. Bacteriol. 170, 4047-4054. 10. Stachel, S., Messens, E., Van Montagu, M. & Zambryski, P. (1985) Nature (London) 318, 624-629. 11. Leroux, B., Yanofsky, M., Winans, S., Ward, J., Ziegler, S. & Nester, E. (1987) EMBO J. 6, 849-856. 12. Winans, S., Ebert, P., Stachel, S., Gordon, M. & Nester, E. (1986) Proc. Natl. Acad. Sci. USA 83, 8278-8282. 13. Ronson, C., Nixon, B. & Ausubel, F. (1987) Cell 49, 579-581. 14. Stock, J., Ninfa, A. & Stock, A. (1989) Microbiol. Rev. 53, 450-490. 15. Melchers, L. S., Regensburg-Tuink, T., Bourret, R., Sedee, J., Schilperoort, R. & Hooykaas, P. (1989) EMBO J. 8, 1919-1925. 16. Winans, S., Kerstetter, R., Ward, J. & Nester, E. (1989) J. Bacteriol. 171, 1616-1622. 17. Jin, S., Roitsch, T., Ankenbauer, R., Gordon, M. & Nester, E. (1990) J. Bacteriol. 172, 525-530. 18. Haung, Y., Morel, P., Powell, B. & Kado, C. (1990) J. Bacteriol. 172, 1142-1144. 19. Jin, S., Prusti, R., Roitsch, T., Ankenbauer, R. & Nester, E. (1990) J. Bacteriol. 172, 4945-4950. 20. Jin, S., Roitsch, T., Christie, P. & Nester, E. (1990) J. Bacte-
riol. 172, 531-537. 21. Pazour, G. & Das, A. (1990) J. Bacteriol. 172, 1241-1249. 22. Popham, D., Szeto, D., Keener, J. & Kustu, S. (1989) Science 243, 629-635. 23. Tsung, K., Brissette, R. & Inouye, M. (1990) Proc. Natl. Acad. Sci. USA 87, 5940-5944. 24. Huang, M., Cangelosi, G., Halperin, W. & Nester, E. (1990) J. Bacteriol. 172, 1814-1822. 25. Cangelosi, G., Ankenbauer, R. & Nester, E. (1990) Proc. Natl. Acad. Sci. USA 87, 6708-6712. 26. Ankenbauer, R. & Nester, E. (1990) J. Bacteriol. 172, 64426446. 27. Shimoda, N., Toyoda-Yamamoto, A., Kangamine, J., Usami, S., Katayama, M., Sakagami, Y. & Machida, Y. (1990) Proc. Natl. Acad. Sci. USA 87, 6684-6688. 28. Veluthambi, K., Krishnan, M., Smith, R. & Gelvin, S. (1989) J. Bacteriol. 171, 3696-3703. 29. Veluthambi, K., Jayaswal, R. K. & Gelvin, S. (1987) Proc. Natl. Acad. Sci. USA 84, 1881-1885. 30. Das, A. & Pazour, G. (1989) Nucleic Acids Res. 17, 4541-4550. 31. Schmidhauser, T. & Helinski, D. (1985) J. Bacteriol. 164, 446-455. 32. Vieira, J. & Messing, J. (1987) Methods Enzymol. 153, 3-11. 33. Kunkel, T. (1985) Proc. Natl. Acad. Sci. USA 82, 488-492. 34. Sanger, F., Nicklen, S. & Coulson, A. (1977) Proc. Natl. Acad. Sci. USA 74, 5463-5467. 35. Chilton, M.-D., Currier, T., Farrand, S., Bendich, A., Gordon, M. & Nester, E. (1974) Proc. Natl. Acad. Sci. USA 71, 3672-3676. 36. Miller, J. H. (1972) Experiments in Molecular Genetics (Cold Spring Harbor Lab., Cold Spring Harbor, NY). 37. Soderling, T. (1990) J. Biol. Chem. 265, 1823-1826. 38. Oosawa, K. & Simon, M. (1986) Proc. Natl. Acad. Sci. USA 83, 6930-6934. 39. McCleary, W. & Zusman, D. (1990) J. Bacteriol. 172, 6661-
6668.
40. Sanders, D., Gillece-Castro, B., Stock, A., Burlingame, A. & Koshland, D. (1989) J. Biol. Chem. 264, 21770-21778. 41. Hess, K., Dudley, M., Lynn, D., Joerger, R. & Binns, A. (1991) Proc. Natl. Acad. Sci. USA, in press.
Mutants of Agrobacterium tumefaciens with elevated vir gene expression (vir gene regulation/mutagenesis/virA mutants)
GREGORY J. PAZOUR, CHRISTOPHER N. TA, AND ANATH DAS Department of Biochemistry and Plant Molecular Genetics Institute, University of Minnesota, 1479 Gortner Avenue, St. Paul, MN 55108
Communicated by Mary-Dell Chilton, May 1, 1991 (received for review March 13, 1991)
ABSTRACT Expression of Agrobacterium tumefaciens virulence (vir) genes requires virA, virG, and a plant-derived inducing compound such as acetosyringone. To identify the critical functional domains of virA and virG, a mutational approach was used. Agrobacterium A136 harboring plasmid pGP159, which contains virA, virG, and a reporter virB: acZ gene fusion, was mutagenized with UV light or nitrosoguanidine. Survivors that formed blue colonies on a plate containing 5-bromo-4-chloro-3-indolyl fi-D-galactoside were isolated and analyzed. Quantification of 8-galactosidase activity in liquid assays identified nine mutant strains. By plasmid reconstruction and other procedures, all mutations mapped to the virA locus. These mutations caused an 11- to 560-fold increase in the vegetative level of virB: acZ reporter gene expression. DNA sequence analysis showed that the mutations are located in four regions of VirA: transmembrane domain one, the active site, a glycine-rich region with homology to ATP-binding sites, and a region at the C terminus that has homology to the N terminus of VirG. The plant pathogen Agrobacterium tumefaciens is the causative agent of crown gall tumors. The bacteria can live either as a free-living soil organism or as an opportunistic colonist of crown gall tumors. The bacteria create a crown gall niche on susceptible plants by transforming the plant cells with genes that direct the biosynthesis of phytohormones, auxin and cytokinin, and that of derivatives of basic amino acids and sugars called opines. The imbalance of hormones produces a tumor in which the bacteria live and metabolize the opines as sources of carbon, nitrogen, and energy. The genes for synthesis of the plant hormones and opines are located on a piece of DNA, the T-DNA, that is transferred from the bacterium to the plant cell, where it becomes integrated into the nuclear genome (reviewed in refs. 1 and 2). The vir genes of the bacteria mediate the processing and transfer of the T-DNA molecule from the bacterium to the plant. These genes are organized into eight operons, virA-H, that are controlled as a regulon (3-7). When the bacteria are in the free-living vegetative state, only virA and virG are expressed. Upon encountering a susceptible plant, all of the vir genes, including virA and virG, are induced (8, 9). This leads to the synthesis of proteins required for processing and transfer of the T-DNA to the plant cell. Induction of the vir regulon requires VirA, VirG, and plant phenolic compounds such as acetosyringone (AS) (8, 10). VirA and VirG are members of the family of prokaryotic two-component regulatory systems (11, 12). In twocomponent regulatory systems, one component senses the environment and then signals the second component to carry out a function (reviewed in refs. 13 and 14). VirA is thought to be the primary sensor of the plant phenolics (12). It The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
traverses the bacterial inner membrane (15, 16) and has autokinase (17, 18) and VirG phosphorylase activities (19). It is hypothesized that interaction of the phenolics with VirA activates VirA's autokinase activity. This creates VirA-P, which can then transfer the phosphate moiety to VirG. VirG-P is hypothesized to be the activated form of VirG. VirG is a sequence-specific DNA-binding protein (20, 21) and is thought to function as a transcriptional activator. The mechanism of transcriptional activation by VirG is not known. One homologue of VirG, NtrC, functions to catalyze the conversion of a closed to an open complex of RNA polymerase and promoter DNA (22). Another VirG homologue, OmpR, has been hypothesized to activate transcription simply by stabilizing the formation of the RNA polymerase-promoter binary complex (23). In addition to the plant phenolic class of inducers, several other compounds and conditions enhance vir gene induction. Monosaccharides such as glucose and galactose enhance vir gene induction when the levels of phenolic vir inducer are low. This response is mediated through the galactose binding protein ChvE. It is thought that ChvE binds the monosaccharides and interacts with the periplasmic domain of VirA to make VirA more sensitive to the phenolic inducers (24-27). Opines also enhance vir gene induction, but the mechanism by which this occurs is unknown (28). Optimum vir gene induction occurs at pH values below 5.7 (8). This effect is probably mediated through an increase in the amount of VirG in the cell as well as through changes in VirA structure. Transcription of virG is induced by acidic conditions (29); therefore, under these conditions more VirG is expected to be present, leading to increased vir gene induction. In addition to VirG, VirA is also affected by pH. Melchers et al. (15) showed that deletion of the periplasmic domain of VirA leads to reduced pH dependence. This suggests that the external (periplasmic) pH brings about conformational changes in VirA that allow it to respond to phenolic inducers.
MATERIALS AND METHODS Bacterial Strains and Plasmids. Plasmids pGP159 (30) and pGP234 contain the virA and virG genes and a virB: lacZ reporter gene on the wide host range cloning vector pTJS75 (31). Plasmid pGP234 was constructed by fusion of plasmids pGP220 and pGP228 at their unique EcoRI sites. Plasmid pGP220 was constructed by cloning a 1.2-kilobase Kpn I fragment containing virG into the unique Kpn I site of pAD1092K (30). Plasmid pGP228 is a pUC119 (32) derivative that contains the pTiA6 Kpn 11 fragment, which carries virA. To construct plasmid pGP336, pGP228 was digested with Pst I. After agarose gel electrophoresis, two fragments, 5 and 2.1 kilobases, were isolated by electroelution. These fragments contain pUC119 with the virA promoter and the coding Abbreviations: AS, acetosyringone; NG, nitrosoguanidine; P-gal, ,6-galactosidase; VirAcOn, constitutive virA mutants; T-DNA, trans-
ferred DNA.
6941
6942
Proc. Natl. Acad. Sci. USA 88 (1991)
Genetics: Pazour et al.
region for the C-terminal end of VirA, respectively. These fragments were ligated to construct pGP335; pGP335 was then fused to pGP220 to construct pGP336. This construct produces a protein identical to VirAA63/240 of Cangelosi et al. (25) and similar to that produced by pRAL7044 (pRAL7044 was constructed with virA from pTilS955) of Melchers et al. (15). The NG31 mutation was added to VirAA63/240 by site-specific mutagenesis (33) of pGP335 followed by fusion with pGP220 and was designated pGP348. Substitution mutations of virA were constructed by sitespecific mutagenesis of pGP228 and were identified by DNA sequence analysis (34). The mutant pGP228 derivatives were later fused to pGP220 to construct pGP234 derivatives. Agrobacterium strains used in this study were A136 (which lacks a Ti plasmid), its isogenic derivatives A348 (which carries the Ti plasmid pTiA6), A723 (which carries the Ti plasmid pTiB6806), and A1030 (which carries the Ti plasmid pTiB6806virA:: TnS). Mutagenesis. UV light mutagenesis was performed by exposing Agrobacterium A136/pGP159 cells plated on ABMes (pH 5.5) solid medium [lx AB salts (35), 0.2% glucose, 25 mM phosphate, and 50 mM 2-(N-morpholino)ethanesulfonic acid (Mes)] containing 5-bromo4chloro-3indolyl ,B-D-galactoside (16 ,ug/ml), tetracycline (10 ,ug/ml), and carbenicillin (50 ,ug/ml) to UV light. Cells were mutagenized to achieve a 90%o kill of the plated cells. Survivors were allowed to grow into colonies, and the blue colonies were purified and assayed for ,B-galactosidase (,(-gal) activity. Nitrosoguanidine (NG) mutagenesis was performed by growing cells (OD6w = 0.2) in ABMes liquid medium (pH 5.5) containing NG at 0.13 mg/ml for 60 min at 30°C (36). Mutagenesis was stopped by centrifugation of the cells, followed by washing in AB medium (pH 7.0) and resuspension in the same medium. Under these conditions, a 75% kill of the cells was observed. Mutated cells were plated on ABMes plates (pH 5.5) containing 5-bromo-4-chloro-3indolyl 8-D-galactoside, tetracycline, and carbenicillin to allow phenotypic expression. Blue colonies were purified and assayed for 8-gal activity. Other Procedures. (-gal activity was measured according to Miller (36). Cells were grown overnight in AB medium (pH 7.0), diluted 1:20 in ABMes medium (pH 5.5 or 6.8) containing tetracycline and carbenicillin with or without AS at 100 ,uM, and grown for 24 hr at 28°C. The pH of these media were approximately 5.0 and 6.4 at the time of assay. All results shown are averages of two independent assays.
RESULTS Selection of Mutants with Elevated Vegetative Levels of vir Gene Expression. Expression of A. tumefaciens vir genes is positively regulated by VirA and VirG in response to plant phenolics such as AS (8, 10). A mutant screen was undertaken to isolate strains of bacteria that exhibit elevated vegetative levels of vir gene expression. These mutants are expected to help us identify critical functional domains and/or amino acid residues of the proteins. This screen made use of a virB: lacZ gene fusion as a reporter gene to identify strains with increased virB expression. Mutants exhibiting elevated virB expression will form blue colonies on 5-bromo4-chloro-3-indolyl ,3-D-galactoside-containing plates while wild-type colonies will be white. Ti-plasmidless A. tumefaciens A136 containing plasmid pGP159 was used in this screen. Plasmid pGP159 carries pTiA6 virA and virG genes as well as the reporter virB: lacZ gene. Examination of 100,000 survivors from UV light-induced mutagenesis yielded 1 blue colony, whereas =300 blue colonies were obtained from examination of 500,000 survivors from NG mutagenesis. The single blue colony from UV light-induced mutagenesis and 70 blue colonies from NG
mutagenesis were assayed for 83-gal activity after growth in liquid culture lacking AS. The majority of these strains showed no elevation of (3-gal activity and were not pursued further. However, nine strains including the one from UV light-induced mutagenesis had elevated levels of P-gal activity. To determine whether the mutations were plasmid borne, plasmid DNA was isolated from the mutant strains and was reintroduced into unmutagenized Agrobacterium. In all cases the mutation mapped on pGP159, indicating that the mutations were in virA, virG, or the virB promoter-regulatory region. To determine which of these genes was altered, the individual genes were subcloned, combined with unmutagenized components, and reintroduced into Agrobacterium. In all cases the mutation was localized within virA. The DNA sequence of each of these constitutive virA mutants (VirAcon) was determined to locate the site(s) of mutation. Each mutant, except NG10, had a single base change within the coding region of virA. NG10 had two base changes, but one of these was a silent change. The locations of the mutations and the resultant amino acid substitutions are shown in Fig. 1. Four strains (NG1, NG8, NG34, and NG37) had an identical base change, which led to a leucine to phenylalanine substitution at codon 24. This residue lies within the first putative transmembrane domain (residues 18-39). Two of the other virA mutations mapped very close to the active site histidine (residue 474). One of these, UV3, had a glycine to arginine substitution at position 471, whereas the other, NG53, had an alanine to valine substitution at position 469. Mutant NG31 had a leucine to phenylalanine substitution at codon 658. This residue is located within a glycine-rich region which is conserved among all the histidine kinase members of two-component regulatory systems (region III; ref. 14). This region is homologous to the 40-amino-acid motif GXGX2GXK that is found in protein kinases and is thought to make up the nucleotide-binding site (37). In VirA this sequence reads GGTGLGX35K, which is altered to GGTGEGX35K in NG31. Mutants NG9 and NG10 have an alanine to valine substitution at position 735. This is the most C-terminal of all the mutations and is located within a region of VirA homologous to VirG and other response regulator members of twocomponent regulatory systems (labeled RR in Fig. 1). This mutation changes a nonconserved residue within an a-helical region, designated aA by Stock et al. (14). VirAc Mutants Exhibit 11- to 560-Fold Increases in the Vegetative Level of virB Expression. The phenotypes of the VirAll mutants were quantitated by assaying (3-gal activity in the presence and absence of the vir-inducing compound AS (Table 1). When compared with wild type (pGP234), the increase in virB: lacZ expression in the absence of AS ranged from 11-fold (UV3) to 562-fold (NG9). The expression level TMI
TM2
NG1, NG8, NG34, NG37 (L24F)
His-P (H474)
UV3 (647 1 R)
NG53
RR N91, NG1O (A735V) 'NG31 (L658F)
(A469V)
FIG. 1. VirA structure and location of mutations. The location of the mutations isolated in this study is shown on a schematic view of VirA along with the amino acid substitutions they cause. Transmembrane domain 1 (TM1), transmembrane domain 2 (TM2), response regulator homologous domain (RR), and the site of autophosphorylation (histidine-474) are also shown. One-letter amino acid codes indicate amino acid substitutions. Residue 1 is the methionine codon of Leroux et al. (11).
Genetics: Pazour et al. Table 1. Characterization of VirA mutations at pH 5.5 virB: IacZ expression, ,8-gal units Plasmid/strain - AS + AS pGP234/A136 9 4170 UV3/A136 103 4188 NG1/A136 1560 13000 NG9/A136 5060 16500 NG31/A136 419 9500 NG53/A136 1270 9270
observed in NG9 (5060 units) is higher than that observed in wild type (pGP234) induced by AS (4170 units). All of the mutants could be further induced by AS; all (except UV3) exhibited synergistic effects of AS and the mutation. Several VirA' Mutants Are Codominant to VirA. If a VirAcon mutant encodes a receptor locked in an activated configuration, such a mutant should be dominant to wild-type VirA. To examine dominance of the VirAcon mutants, we introduced the virA mutant plasmids into A. tumefaciens A348 (harbors Ti plasmid pTiA6) and analyzed their phenotypes (Table 2). The mutants showed a 2- to 7-fold lower level of uninduced virB expression when a Ti plasmid was present in the host bacterium (- AS column,-Table 1 vs. Table 2, experiment 1). To determine whether virA on the Ti plasmid was responsible for the phenotypic suppression of the mutant phenotype, we introduced the NG9 mutation into A. tumefaciens A1030, a strain bearing a virA:: Tn5 insertion, and its isogenic parent A723. Results presented in Table 2, experiment 2, show that the presence of the Ti plasmid in A723 caused phenotypic suppression of the NG9 mutation [a 5-fold decrease compared to that in the absence of a Ti plasmid (A136)]. This suppression was partially relieved by the disruption of the virA gene in A1030. These results indicate that the presence of a wild-type copy of virA can affect the phenotype of the VirAcwn mutants, but other factors on the Ti plasmid may also be involved. VirA' Phenotype Is pH and Temperature Dependent. Induction of the vir genes is pH dependent (8); very little induction is usually observed if the pH of the growth medium exceeds 6.0. A possible explanation for this phenomenon is that lower pH is required for interaction of the vir-inducing Table 2. Effect of the presence of a Ti plasmid on the VirAcOn phenotype at pH 5.5 virB: lacZ expression, p-gal units Plasmid/strain - AS + AS Experiment I pGP234/A348 4 3,380 UV3/A348 35 3,520 NG1/A348 482 4,370 NG9/A348 753 4,290 NG31/A348 175 4,570 NG53/A348 706 5,530 Experiment 2 pGP234/A136 9 4,170 NG9/A136 5060 16,500 pGP234/A723 7 4,530 NG9/A723 1080 4,190 pGP234/A1030 6 3,610 NG9/A1030 2030 3,979 A348, A723, and A1030 contain the Ti plasmids pTiA6, pTiB6806, and pTiB6806virA::Tn5, respectively.
Proc. Natl. Acad. Sci. USA 88 (1991)
6943
Table 3. Characterization of VirA mutations at pH 6.8 virB: lacZ
expression, pl-gal units
Plasmid/strain pGP234/A136 UV3/A136 NG1/A136 NG9/A136 NG31/A136 NG53/A136
- AS
+ AS
12 16 80 69 41 54
1150 490 3130 1460 2160 1940
compound with VirA. If this is the only reason for a reduced pH requirement, then the VirAcon phenotype should be expressed at pH 6.8. This, however, was not found to be true (Table 3). The mutants exhibited only a small (7-fold maximum) increase in virB: lacZ expression over wild type in the absence of AS at pH 6.8. In the presence of AS at pH 6.8, the level of expression of the mutants (except UV3) was comparable to that of the wild type, a level 3- to 4-fold lower than that observed at pH 5.5. Previous studies that demonstrated an absolute pH dependence of vir gene expression used the single-copy Ti plasmid as the source of vir genes (8). When we used an RK2-derived plasmid containing virA and virG, significant induction of the vir genes was observed at pH 6.8 (Table 4). RK2-derived plasmids are multicopy in Agrobacterium; thus, it appears that an increase in copy number can partially relieve the pH dependence of vir gene induction. Similar results have been reported by Melchers et al. (15). Melchers et al. (15) reported that deletion of a portion of the periplasmic domain of VirA leads to reduced pH depen-
dence. We sought to determine whether we could obtain a pH-independent VirAcon phenotype by combining a similar
deletion with the NG31 mutation. The results presented in Table 4 confirm that deletion of residues 63-240 in the periplasmic domain of VirA leads to the relief of pH dependence of vir gene induction. The level of AS-induced virB expression by VirAA63/240 at pH 6.8 was comparable to that at pH 5.5. Combination of VirAA63/240 with the NG31 mutation yields a VirAcon phenotype at either pH. To study the effect of temperature, cells were grown at 30°C, 33°C, or 37°C. In conformity with the studies of Melchers et al. (15), we observed that strains containing wild-type virA are inducible at 30°C or 33°C but not at 37°C. However, the mutant phenotypes were manifested only at 30°C; a marginal increase in induction (2- to 5-fold over wild type) was observed at 33°C or 37°C (data not shown). Effect of Amino Acid Substitutions in the First Hydrophobic Domain of VirA. One of the VirAcOn mutants, NG1, mapped within the first hydrophobic domain of VirA. This result was unexpected in view of the fact that Melchers et al. (15) earlier reported that this segment of the protein could be replaced by Table 4. Effect of pH on VirAcon phenotype virB: lacZ expression, (3-gal units pH5.5 pH 6.8 Plasmid Genotype - AS + AS - AS + AS pGP234 Wild type 9 4170 11 1150 NG31 VirANG31 419 9500 41 2160 pGP336 VirAA63/240 25 5840 14 5110 pGP348 VirAA63/240,NG31 230 9550 290 7460 P-gal activity of Agrobacterium A136 harboring the indicated plasmid was assayed at pH 5.5 or 6.8 with or without 100 ,uM AS.
Genetics: Pazour et al.
6944
Proc. Natl. Acad. Sci. USA 88 (1991)
analogous but nonhomologous region of the Escherichia coli aspartic acid receptor (Tar). These observations led the authors to believe that this segment of the protein was probably not playing a specific role in the detection of the vir-inducing compounds. To gain an understanding of the role of the NG1 mutation in the signaling process, we examined the phenotype of mutants with substitutions of different amino acids at the NG1 position. NG1 causes a leucine to phenylalanine substitution at residue 24 (Fig. 1). The effects of different substitutions at position 24 on the signaling process are summarized in Fig. 2. These mutations were introduced in the wild-type gene by site-specific mutagenesis according to Kunkel (33). Alteration of leucine to phenylalanine is a re-creation of the NG1 mutation, and the mutant exhibits the VirAcon phenotype as observed earlier (Table 1). A substitution of isoleucine or aspartic acid at residue 24 showed a marginal VirAcon phenotype (3- to 4-fold over wild type), whereas all other substitutions had no effect on the phenotype. Interestingly, mutation of this residue to tyrosine, a close homologue of phenylalanine, did not alter the phenotype from wild type. None of the substitutions abolished VirA activity. These substitutions included such residues as proline and the charged residues aspartic acid, lysine, arginine, and histidine. Mutational Analysis of the Response Regulator Domain of VirA. The C-terminal end of VirA is homologous to CheY and the N terminus of VirG (14). The aspartic acid residue that is phosphorylated in VirG and CheY is conserved in this domain of VirA. To determine what effect mutation of this residue would have on vir gene induction, the aspartic acid residue at position 766 was replaced with asparagine (VirAD766N). This substitution causes a 2-fold increase in the basal level and a 5-fold decrease in the induced level of virB expression (Table 5). This indicates that intramolecular phosphotransfer is probably not required for vir gene induction but does not rule out other unknown pathways.
an
DISCUSSION In this study we identified mutations in the virA structural gene that cause vir gene expression in the absence of virinducing compounds. Under noninducing conditions these mutant proteins function similarly to the wild-type protein in the presence of the inducer, indicating that they may exist in conformations that mimic the structure of VirA bound to the vir inducer. These mutations map in four regions of the 1
5
c:lo
0) 0
a)
No
1
Leu Phe Arg His Ser Asp Lys ProThr Gly WT NG1
TyrPhe Met Val lle
FIG. 2. Effect of substitution at the leucine-24 codon on virA activity. virB: lacZ expression was measured in Agrobacterium A136 at pH 5.5. Black bars represent activity in the absence of inducer, and the open bars represent activity in the presence of 100 ,uM AS. Bar 1 (leucine) is wild type (WT), bar 2 is the NG1 mutant, the other phenylalanine (bar 12) is a re-creation of this mutation.
Table 5. Mutational analysis of the response regulator domain in VirA virB: lacZ expression, /3-gal units - AS + AS Genotype Plasmid 9 4170 Wild type pGP234 20 800 VirAD766N pGP377 harboring the indicated activity of Agrobacterium A136 ,p-gal plasmid was assayed at pH 5.5 with or without 100 ,uM AS.
protein: transmembrane domain one, the active site, a glycine-rich region homologous to nucleotide-binding sites, and a region homologous to CheY and the N terminus of VirG. VirA has two hydrophobic domains, which have been proposed to be membrane spanning (TM1 and TM2; Fig. 1). One VirAc`n mutant, NG1, mapped to a site within the TM1 region. Previous studies showed that this region could be replaced by an analogous but nonhomologous region of the E. coli Tar receptor (15). Thus it was thought that this domain could not be playing a major role in the sensing ofthe phenolic compounds. Its role in the transmembrane signaling process is less clear. Genetic studies imply that in the Tar receptor this region (TM1) interacts with the second transmembrane domain and with a cytoplasmic domain. It has been proposed that these interactions are involved in the transmembrane signaling process (38). It is possible that the TM1 domain of VirA also functions in a similar manner, which would indicate that in the Tar-VirA chimeric protein the Tar sequence is able to provide the transmembrane signaling. Two amino acid substitutions near the active site histidine caused the VirAc`n phenotype. These mutations are likely to change the active site conformation to mimic the activated state. An additional mutation maps to a glycine-rich region C-terminal to the active site. This region is conserved in all histidine kinase members of the two-component regulatory systems. It bears homology to the glycine-rich sequence that make up nucleotide-binding sites of many protein kinases. If this site is a nucleotide-binding site, then it must be folded in close to the active site. The final region in which a mutation was found was in a domain at the C terminus of VirA, which has homology to CheY and the N terminus of VirG. This domain is unusual in histidine kinases. FrzE, which also has this structure, is thought to have both histidine kinase and response regulator functions in the same polypeptide (39). FrzE autophosphorylates and the chemical stability of the phosphate moiety resembles a phosphoaspartic acid, indicating that an intramolecular transfer of the phosphate from the active site histidine to the aspartic acid may occur in this case. It is not known whether this type of intramolecular transfer occurs in VirA. Studies ofJin et al. (17) indicated that phosphohistidine was present in VirA-P, but these studies could not rule out the possibility that phosphoaspartic acid was also present. In the present study we show that mutation of aspartic acid at residue 766 in VirA, homologous to the aspartic acid residue that is phosphorylated in CheY (40) and VirG (19), caused a small increase in the basal level and a 5-fold decrease in the induced level of vir gene expression. These results indicate that an intramolecular signaling pathway is probably not essential for vir gene induction, although it may be important for some unknown additional pathway. The presence of a domain homologous to VirG and CheY in the C terminus of VirA is reminiscent of the "autoinhibitory" domains of eukaryotic protein kinases (37). It has been established that the low basal level of many of these kinases is due to inhibitory interactions between the active site and a domain elsewhere in the kinase that has homology to the
Genetics: Pazour et al. substrate. It is hypothesized that the allosteric activators of these kinases cause a conformational change that disrupts the interaction between the regulatory domain and the active site and allows for productive interaction between the active site and the substrate. It is possible that this may be the role played by this domain in VirA; binding of AS causes a conformational change that allows VirG access to the active site so that it can become phosphorylated and activated. Alternatively, other signals such as low pH, sugars, or opines could be causing the conformational change to enhance the activation of VirA and work synergistically with AS. The VirAcon mutants isolated in this study show codominance with VirA. It would be expected that if these mutants were simply receptors locked in the activated configuration, then they should be dominant to wild type. There are several possibilities to explain the suppression of phenotype when wild-type VirA is present. (i) There are a limited number of sites in the membrane that VirA can occupy. Since both wild type and VirAcon can compete for these sites, less VirAcon will be present in the membrane and the phenotype will be suppressed. (ii) The presence of a Ti plasmid, on which the wild-type copy is located, is contributing a large number of VirG binding sites such that less activated VirG is available to induce expression of the reporter gene, leading to phenotypic suppression. (iii) VirA in the unactivated state can bind VirG and remove it from the pool making it unavailable for activation by VirAcon. (iv) VirA is multimeric and heteromultimers of VirA and VirAcon have wild-type or intermediate phenotypes, thus causing phenotypic suppression. Disruption of virA on the Ti plasmid by a Tn5 insertion partially relieves the phenotypic suppression, indicating that codominance is caused by both the presence of the protein and an additional factor(s) on the Ti plasmid. These mutants should prove useful in further studies on the vir gene induction pathway, such as in determining the step at which noninducing chromosomal mutations are blocked. In addition, if the model that phosphorylation is the activating step is correct, then these mutants should have elevated rates of VirG phosphorylation. At present we have not been able to detect any alteration in phosphorylation using the cytoplasmic portion of VirA as the kinase (G.J.P., unpublished results). These mutants may also prove useful in the transformation of recalcitrant plant species because they do not need vir-inducing compounds and because they hyperinduce in the presence of AS. Note Added in Proof. In a related study Hess et al. (41) report a detailed analysis of the interaction between VirA and AS.
We thank Ann Vogel and John Larkin for a critical reading of the manuscript. This work was supported by grants from the National Institutes of Health (GM 37555), the American Cancer Society (JFRA 170), and a McKnight Land Grant Professorship Award from the University of Minnesota. G.J.P. is a Graduate School Dissertation Fellow of the University of Minnesota. 1. 2. 3. 4.
Zambryski, P. (1988) Annu. Rev. Genet. 22, 1-30. Ream, W. (1989) Annu. Rev. Phytopathol. 27, 583-618. Garfinkel, D. & Nester, E. (1980) J. Bacteriol. 144, 732-743. Ooms, G., Kalpwijk, P., Poulis, J. & Schilperoort, R. (1980) J. Bacteriol. 144, 82-91. 5. Stachel, S. & Nester, E. (1986) EMBO J. 5, 1445-1454. 6. Kanemoto, R., Powell, A., Akiyosh, D., Regier, D., Kerstetter,
Proc. Natl. Acad. Sci. USA 88 (1991)
6945
R., Nester, E., Hawes, M. & Gordon, M. (1989) J. Bacteriol. 171, 2506-2512. 7. Melchers, L., Maroney, M., Dulk-Ras, A., Thompson, D., van Vuuren, H., Schilperoort, R. & Hooykaas, P. (1990) Plant Mol. Biol. 14, 249-259. 8. Stachel, S. & Zambryski, P. (1986) Cell 46, 325-333. 9. Winans, S., Kerstetter, R. & Nester, E. (1988) J. Bacteriol. 170, 4047-4054. 10. Stachel, S., Messens, E., Van Montagu, M. & Zambryski, P. (1985) Nature (London) 318, 624-629. 11. Leroux, B., Yanofsky, M., Winans, S., Ward, J., Ziegler, S. & Nester, E. (1987) EMBO J. 6, 849-856. 12. Winans, S., Ebert, P., Stachel, S., Gordon, M. & Nester, E. (1986) Proc. Natl. Acad. Sci. USA 83, 8278-8282. 13. Ronson, C., Nixon, B. & Ausubel, F. (1987) Cell 49, 579-581. 14. Stock, J., Ninfa, A. & Stock, A. (1989) Microbiol. Rev. 53, 450-490. 15. Melchers, L. S., Regensburg-Tuink, T., Bourret, R., Sedee, J., Schilperoort, R. & Hooykaas, P. (1989) EMBO J. 8, 1919-1925. 16. Winans, S., Kerstetter, R., Ward, J. & Nester, E. (1989) J. Bacteriol. 171, 1616-1622. 17. Jin, S., Roitsch, T., Ankenbauer, R., Gordon, M. & Nester, E. (1990) J. Bacteriol. 172, 525-530. 18. Haung, Y., Morel, P., Powell, B. & Kado, C. (1990) J. Bacteriol. 172, 1142-1144. 19. Jin, S., Prusti, R., Roitsch, T., Ankenbauer, R. & Nester, E. (1990) J. Bacteriol. 172, 4945-4950. 20. Jin, S., Roitsch, T., Christie, P. & Nester, E. (1990) J. Bacte-
riol. 172, 531-537. 21. Pazour, G. & Das, A. (1990) J. Bacteriol. 172, 1241-1249. 22. Popham, D., Szeto, D., Keener, J. & Kustu, S. (1989) Science 243, 629-635. 23. Tsung, K., Brissette, R. & Inouye, M. (1990) Proc. Natl. Acad. Sci. USA 87, 5940-5944. 24. Huang, M., Cangelosi, G., Halperin, W. & Nester, E. (1990) J. Bacteriol. 172, 1814-1822. 25. Cangelosi, G., Ankenbauer, R. & Nester, E. (1990) Proc. Natl. Acad. Sci. USA 87, 6708-6712. 26. Ankenbauer, R. & Nester, E. (1990) J. Bacteriol. 172, 64426446. 27. Shimoda, N., Toyoda-Yamamoto, A., Kangamine, J., Usami, S., Katayama, M., Sakagami, Y. & Machida, Y. (1990) Proc. Natl. Acad. Sci. USA 87, 6684-6688. 28. Veluthambi, K., Krishnan, M., Smith, R. & Gelvin, S. (1989) J. Bacteriol. 171, 3696-3703. 29. Veluthambi, K., Jayaswal, R. K. & Gelvin, S. (1987) Proc. Natl. Acad. Sci. USA 84, 1881-1885. 30. Das, A. & Pazour, G. (1989) Nucleic Acids Res. 17, 4541-4550. 31. Schmidhauser, T. & Helinski, D. (1985) J. Bacteriol. 164, 446-455. 32. Vieira, J. & Messing, J. (1987) Methods Enzymol. 153, 3-11. 33. Kunkel, T. (1985) Proc. Natl. Acad. Sci. USA 82, 488-492. 34. Sanger, F., Nicklen, S. & Coulson, A. (1977) Proc. Natl. Acad. Sci. USA 74, 5463-5467. 35. Chilton, M.-D., Currier, T., Farrand, S., Bendich, A., Gordon, M. & Nester, E. (1974) Proc. Natl. Acad. Sci. USA 71, 3672-3676. 36. Miller, J. H. (1972) Experiments in Molecular Genetics (Cold Spring Harbor Lab., Cold Spring Harbor, NY). 37. Soderling, T. (1990) J. Biol. Chem. 265, 1823-1826. 38. Oosawa, K. & Simon, M. (1986) Proc. Natl. Acad. Sci. USA 83, 6930-6934. 39. McCleary, W. & Zusman, D. (1990) J. Bacteriol. 172, 6661-
6668.
40. Sanders, D., Gillece-Castro, B., Stock, A., Burlingame, A. & Koshland, D. (1989) J. Biol. Chem. 264, 21770-21778. 41. Hess, K., Dudley, M., Lynn, D., Joerger, R. & Binns, A. (1991) Proc. Natl. Acad. Sci. USA, in press.
