The Agrobacterium tumefaciens vir gene transcriptional activator virG is transcriptionally induced by acid pH and other stress stimuli

A set of Agrobacterium tumefaciens operons required for pathogenesis is coordinately induced during plant infection by the VirA and VirG proteins. The intracellular concentration of VirG increases in response to acidic media, and this response was proposed to be regulated at the level of transcription at a promoter (P2) that resembles the Escherichia coli heat shock promoters. To test this hypothesis, we first constructed a virG-lacZ transcriptional fusion. A strain containing this fusion had higher levels of beta-galactosidase activity in acidic media than in media at neutral pH. Second, primer extension analysis of virG indicated that acidic media stimulated the transcription of this promoter. To determine whether P2 is a member of a heat shock-like regulon in A. tumefaciens, five agents that induce E. coli heat shock genes were tested for their abilities to induce a P2-lacZ fusion in A. tumefaciens. P2 was most strongly induced by low pH, was moderately stimulated by CdCl2 or mitomycin C, and was slightly induced by P2 as measured by beta-galactosidase activity and primer extension analysis. Induction by these treatments did not require any Ti plasmid-encoded function or the chromosomally encoded RecA protein. We also pulse-labeled cellular proteins after a shift to low pH and detected several proteins whose synthesis was induced by these conditions. We conclude that P2 is primarily induced by acid pH and secondarily by certain other stimuli, each of which is stressful to cell growth. This stress induction is at least partly independent of the heat shock and SOS responses.

extension analysis. Induction by these treatments did not require any Ti plasmid-encoded function or the chromosomally encoded RecA protein. We also pulse-labeled cellular proteins after a shift to low pH and detected several proteins whose synthesis was induced by these conditions. We conclude that P2 is primarily induced by acid pH and secondarily by certain other stimuli, each of which is stressful to cell growth. This stress induction is at least partly independent of the heat shock and SOS responses.
Establishment of crown gall tumors on plant hosts by Agrobacterium tumefaciens requires the recognition of a plant wound susceptible to infection, transfer of a discrete segment of DNA (T-DNA) to the plant cell nucleus, and integration of T-DNA into the plant genome (for reviews, see references 2, 12, 23, 43, and 66). The genes required for T-DNA transfer (vir genes) are located on a large plasmid, called the Ti plasmid. The vir genes are transcriptionally induced in response to a family of phenolic compounds produced by wounded plant cells (50,51). Induction also requires acidic growth medium (52) and is potentiated by certain monosaccharides (3,47).
Two members of the vir regulon, virA and virG, are required for vir gene induction (44,53,63) and encode proteins that are members of the family of bacterial twocomponent regulatory systems (28,37,42,62). For some vir promoters, VirA and VirG are the only known regulatory proteins (34,64), while the virC and virD promoters are also controlled by a repressor encoded by the chromosomal ros gene (5,7). VirA is a transmembrane protein kinase that can phosphorylate itself and VirG (19,21,22,36). VirG specifically binds cis-acting regulatory sequences involved in transcriptional activation of the virulence genes (22,40,41). Recent evidence indicates that vir expression is rate limited by the pool size of VirG (4,20,59).
The pool size of VirG protein is influenced by three classes of stimuli: plant wound-released phenolics, phosphate starvation, and acidic medium (53,58,63). Transcription of virG is initiated at two promoters, called P1 and P2. The upstream promoter, P1, is induced by phenolic compounds in a * Corresponding author.
VirA-VirG-dependent manner and by phosphate starvation (1,53,61). Although P2 was originally thought to be expressed constitutively (53), deletions which remove this promoter abolish induction of virG by acidic growth media (61), suggesting that P2 may be transcriptionally induced by this stimulus. In addition, the sequence of P2 does not resemble those of the other vir promoters (9) but is very similar at its -10 region to the consensus sequence of Escherichia coli heat shock promoters (8,61).
The E. coli heat shock promoters are recognized by RNA polymerase containing o32, the sigma factor encoded by the rpoH (htpR) gene (8,13). The genes transcribed by Cr32 constitute a regulon that encodes at least 17 proteins whose syntheses are induced by a shift to high temperature (38). Other forms of cellular stress have been shown to induce the synthesis of some of the heat shock proteins. These include treatment with ethanol, heavy metals (57), DNA-damaging agents (25,57), or extreme pH shifts (16,17,56). Recent work suggests that the A. tumefaciens heat shock response is similar to the E. coli response and that A. tumefaciens has a cr32-like sigma factor (31).
Here we use a virG-lacZ transcriptional fusion and primer extension analysis to measure virG expression in response to acidic pH. Because P2 resembles the E. coli heat shock promoters (61), we tested the hypothesis that virG may be a member of a heat shock-like regulon in A. tumefaciens. We used the previously constructed P2-lacZ fusion (61) to measure gene expression in response to five agents that induce the synthesis of the heat shock proteins in E. coli. Our results indicate that P2 is regulated at the level of transcription primarily by low pH and secondarily by other stress stimuli and that this regulation is at least in part distinct from the heat shock and SOS responses.
Strains and plasmids. A. tumefaciens A348 is a derivative of A136 containing the octopine plasmid pTiA6 (46). A136 is a rifampin-resistant derivative of strain C58 lacking the nopaline Ti plasmid pTiC58. A. tumefaciens 363MX contains a Tn3HoHol insertion in the virG gene on the Ti plasmid (51). Agrobacterium strain UIA143 is a derivative of NT-1 containing an erythromycin cassette in the recA gene (10) and was obtained from S. K. Farrand (University of Illinois). E. coli MC4100 (AlacVJ69 avaD thiA rspL relA) was obtained from C. Manoil (University of Washington).
The broad-host-range plasmid pUCD2 (6) was obtained from C. Kado (University of California, Davis). The plasmid vector pRS415, containing a promoterless lacZ operon and termination codons in all three reading frames upstream of the lacZ reading frame (48), was obtained from V. Stewart (Cornell University). The virG-lacZ transcriptional fusion was constructed by ligating the EcoRI-BamHI fragment containing the 270-bp upstream region of the virG TTG translational start site and the 630-bp coding region of virG from pSW104 (62) into pRS415. The 7.8-kb PstI-SalI fragment containing the virG-lacZ fusion was cloned into pUCD2, creating pNM102. Construction of pSW174 and pSW264 (61) and pCC101 (14) have been described previously. Plasmids were introduced into Agrobacterium strains by transformation (18) or electroporation (32). virG induction assays. A. tumefaciens strains were grown in AB medium (29) (pH 7.0) to an optical density (OD) of 0.4 at 600 nm, centrifuged, suspended in 1 ml of water containing 15% glycerol, and frozen at -70°C. For induction assays, cells were thawed, diluted 1:2,000 into AB minimal medium (pH 7.0) in 250-ml Erlenmeyer flasks, and cultured at 30°C in a rotary aerator to an OD of 0.1 at 600 nm. Cells were dispensed into prewarmed culture tubes (125 by 25 mm) containing inducers, and 1-ml samples were removed for P-galactosidase assays (35) over a period of 6 h. For the pH shift experiments, cells were grown in AB minimal medium (pH 7.0) supplemented with 40 mM MES or 40 mM EPPS to an OD of 0.1 at 600 nm. Cells were collected by centrifugation at 8,000 x g for 10 min, suspended in AB minimal medium at the appropriate pH, and assayed as described above.
RNA isolation and primer extension. Cultures of A348 (pSW174) were grown to an OD of 0.6 at 600 nm and induced as described above. Total RNA was extracted from 10 ml of cells by lysozyme treatment in the presence of diethyl pyrocarbonate (54) and quantitated by spectrophotometry at 260 nm. Ten picomoles of a primer complementary to the virG coding sequence was labeled at the 5' end by using [_y-32P]ATP and T4 polynucleotide kinase as described elsewhere (30). Efficiency of the kinase reaction was determined by trichloroacetic acid precipitation (30).
For primer extension analysis, RNA (10 ,ug) and labeled primer (0.5 pmol) were coprecipitated in ethanol and suspended in 17 ,ul of a buffer containing 50 mM Tris HCl (pH 8.3), 70 mM KCl, and 3 mM MgCI2. The samples were heated to 80°C for 10 min and then cooled slowly to 30°C (1.5 h). A 33-,ul volume of a mix containing 50 mM Tris HCl (pH 8.3); 70 mM KCl; 3 mM MgCl2; 10 mM dithiothreitol; 1 mM each dGTP, dATP, dCTP, and dTTP; 0.5 U of RNAse Block II; and 200 U of Mololoney murine leukemia virus reverse transcriptase was added, and the samples were placed at 37°C for 45 min. The reaction was stopped by the addition of 25 mM EDTA (pH 8.0) and then extracted with 2 volumes of phenol-chloroform (1:1), ethanol precipitated, and suspended in 4 ,ul of Tris-EDTA (pH 7.4) and 6 ,lI of Stop buffer obtained with the Sequenase DNA sequencing kit. The samples were heated to 85°C for 5 min, and 4 1±l of each sample was size fractionated by electrophoresis through a 6% polyacrylamide gel. The gel was transferred to filter paper, vacuum dried, and exposed on Kodak X-OMAT XAR-5 diagnostic film at -70°C with an intensifying screen. A sequencing ladder was generated by using Sequenase according to procedures recommended by the manufacturer, the primer used in the primer extension reaction, and singlestranded pSW167 (61).
Pulse-labeling experiments. Cultures of A348 were grown as described above to an OD of 0.4 at 600 nm. The cells were then centrifuged at 8,000 x g for 10 min and suspended in an original volume of AB minimal medium at pH 5.0 or 7.0. At appropriate time intervals, a 1-ml sample of cells was transferred to a prewarmed test tube (13 by 100 mm), labeled with 5 ,uCi of [L-35S]methionine for 3 min, chased with nonradioactive L-methionine (to a final concentration of 2 mM) for 3 min, and then transferred to 1.5-ml Eppendorf tubes containing 0.5 ml of 5% trichloroacetic acid. Cells were harvested by centrifugation at 10,000 x g for 3 min, washed once with acetone, vacuum dried, and suspended in sodium dodecyl sulfate (SDS) sample buffer (0.12 M Tris base, 0.07 M SDS, 20% [vol/vol] glycerol, 2% [vol/vol] 2-mercaptoethanol, 0.01% [wt/vol] bromphenol blue; pH 6.8). All samples were heated to 100°C for 5 min prior to electrophoresis. Electrophoresis was through 10% polyacrylamide-SDS gels overlaid with stacking gels as described by Laemmli (26). Gels were stained with Coomassie brilliant blue dye, vacuum dried, and exposed on Kodak X-OMAT XAR-5 diagnostic film.

RESULTS
Measurement of virG transcription in response to low pH. It was previously shown by using virG-lacZ translational fusions that the region containing the P2 promoter was required for induction of virG by acidic pH (61). To test whether this regulation occurred at the level of transcription, we used vector pRS415 (48) to construct a plasmid containing a virG-lacZ operon fusion. This plasmid (pNM102) was introduced into A. tumefaciens strains containing or lacking the Ti plasmid (strains A348 and A136, respectively), and the ,-galactosidase total activities of these strains were measured after a shift to low pH. We found that the amount of ,B-galactosidase activity of strain A348(pNM102) increased as the pH of the medium decreased (Fig. 1). These results agree with those observed when a virG-lacZ translational  (61). We did not test cells grown in media at pHs below 5.0. Induction in strain A136(pNM102) was similar to that obtained in strain A348(pNM102) (data not shown), confirming that the Ti plasmid was not required for induction (63).
To provide an independent test for transcriptional induction of P2 and to determine the approximate start site of the P2 transcript, we used primer extension analysis (33). To facilitate detection of virG mRNA, we used strain A348 (pSW174), which contains the Ti plasmid and a virG-lacZ translational fusion on a multicopy plasmid (61). We isolated total RNA from this strain at 10 and 50 min after a shift from pH 7.0 to 5.0. A faint reverse transcript whose 3' end mapped to the P2 promoter was observed after primer extension analysis of RNA from cells cultured at pH 7.0 (Fig. 2, lanes 1 and 4). A strong transcript from this promoter was observed after bacteria were cultured for 50 min at pH Primer extension analysis of promoter P2. A log-phase culture of A348(pSW174) at pH 7.0 and 30°C was shifted to medium at high temperature or low pH and incubated for 10 min (lanes 1, 2, and 3) or 50 min (lanes 4, 5, and 6). Lanes 1 and 4, no treatment; lanes 2 and 5, temperature shift to 39°C; lanes 3 and 6, shift to pH 5.0. Yeast RNA (Sigma) was used as a negative control (lane C). The C and T reactions of a dideoxy sequencing reaction generated with virG template and unphosphorylated primer used in the primer extension reaction were loaded adjacent to the primer extension products. The complete sequence of the region is printed on the right, and the asterisk indicates the apparent start site of transcription. 5.0 (lane 6). These results provide further evidence that transcription initiates at P2 in response to low pH. The apparent start site of transcription is indicated by an asterisk in Fig. 2 and is within three bases of the apparent start site previously identified by Si nuclease mapping (53). The actual start site could lie one or two bases upstream of the apparent start site, because the primer extension experiments were conducted with a phosphorylated primer, while the sequence ladder was made by using a nonphosphorylated primer.
Induction of P2 by other stress stimuli. We hypothesized that P2 might be inducible by agents which induce the synthesis of E. coli heat shock proteins, because the -10 region of P2 strongly resembles the consensus sequence of E. coli heat shock promoters (61). To test this, we measured ,B-galactosidase activity from strain A348(pSW264) after treatments which induce the heat shock proteins of E. coli. These treatments include growth at elevated temperatures, at extremes of pH, or after addition of ethanol, CdCl2, or mitomycin C. Samples were collected at 2-h intervals and assayed for ,-galactosidase activity. pSW264 contains a P2-lacZ translational fusion (61), while sequences upstream of P2 have been deleted. virG was strongly induced by low pH (Fig. 3D), significantly induced by CdCl2 (panel C) or mitomycin C (panel F) treatment, and mildly induced by alkaline pH (panel E) or ethanol (panel B). In contrast, virG was not detectably induced by temperature upshift (panel A). The degree of induction increased with increasing acidity or alkalinity or increasing amounts of CdCl2, mitomycin C, or ethanol. Because samples were collected and assayed for ,B-galactosidase activity at 2-h intervals, it can be seen that each of these treatments inhibited cell growth, suggesting that environmental stress could play some role in induction. However, virG expression was not simply growth rate dependent, because equivalent growth inhibition by different treatments did not cause equivalent virG expression (compare panels A and D). The kinetics of expression were rapid in response to low pH and slower in response to CdCl2, mitomycin C, and ethanol. Similar results were observed when strain A136(pSW264), which does not contain the Ti plasmid, was used (data not shown).
To determine whether the agents we tested for Fig. 3 might act nonspecifically on gene expression, strain A348 (pCC101) was assayed for P-galactosidase activity after the same treatments. pCC101 contains a Plac-lacZ fusion (4). The Plac promoter was not detectably induced by these treatments (data not shown). Also, because pNM102 and pSW264 are multicopy plasmids, we wanted to determine whether virG was induced by these treatments when it was present on the Ti plasmid. Strain 363MX contains a Tn3Ho Hol insertion in the virG gene on the Ti plasmid (51). We tested induction of ,-galactosidase in strain 363MX in response to mitomycin C, CdCl2, and ethanol. The rate of virG expression was similar to that observed with multicopy plasmids (data not shown), though the level of expression was lower, probably because of the lower number of copies of this gene fusion.
The results in Fig. 3 indicate that virG was not induced by prolonged growth at elevated temperatures. However, transcription of the heat shock genes in E. coli is rapid and transient after temperature upshift (38), and a similarly transient induction of P2 might be difficult to detect by using a lac fusion. We therefore used primer extension analysis to measure P2 transcription after heat shock. There was no detectable stimulation of transcription of P2 after 10 or 50 >; min of growth at 39°C (Fig. 2, lanes 2 and 5). These conditions are sufficient to induce A. tumefaciens major heat shock proteins (31). This result provides further evidence that P2 is not induced by a shift to high temperature. virG induction in a recA background. DNA-damaging agents like mitomycin C are primarily inducers of the SOS regulon in E. coli and are dependent on the recA gene product for this activation (60). virG was induced by mito-mycin C, and we wanted to determine whether RecA was required for this induction. We transformed pSW264 into a recA strain of Agrobacterium (10) and assayed ,-galactosidase activity in response to low pH or mitomycin C. The results were similar to those seen in a wild-type background (Fig. 4), indicating that the SOS system is not involved in virG regulation. Protein  was the most efficient inducer of P2, we asked whether the synthesis of other proteins was induced by these conditions. We pulse-labeled strain A348 with [L-35S]methionine at pH 7.0 or at various intervals after a shift to pH 5.0, and an equivalent amount of cells from each treatment was subjected to SDS-polyacrylamide gel electrophoresis (PAGE) (Fig. 5). Total protein synthesis was reduced after the pH shift (in Fig. 5, compare odd-numbered lanes with evennumbered lanes). However, induction of several new proteins was observed. Among these, the most strongly expressed were a 35-kDa protein whose synthesis was greatest between 20 and 30 min (lanes 5 and 7) and a 45-kDa protein induced by 20 min (lane 5). These proteins were distinct from proteins induced by shifting cultures to high temperature (lane 13). These results indicate the preferential synthesis of a distinct set of proteins in response to low pH. We did not determine whether the synthesis of these proteins required the presence of the Ti plasmid. None of the proteins induced by low pH had a molecular weight similar to that of VirG, which migrates as a 25-kDa protein on SDS-PAGE (22).

DISCUSSION
P2 was originally referred to as "P c for "constitutive," since Si nuclease protection assays suggested that this promoter was expressed constitutively (53). However, deletion analysis of the upstream region of virG using virG-lacZ translational fusions indicated that the P2 promoter was required for virG expression in response to low pH (61). In this paper, we extend these findings and provide direct 1 2 3 4 5 6 7 8 9 10 11  evidence that P2 is induced at the level of transcription in response to low pH. First, virG-lacZ transcriptional fusions were induced by low-pH media (Fig. 1). Second, primer extension experiments showed that acidification of growth media caused a rapid increase in the levels of transcripts initiating at P2 (Fig. 2). Expression of virG in response to acidic pH does not require VirG protein or any other Ti plasmid-encoded function, indicating that the gene regulating P2 must be located on the chromosome (63). The sequence of P2 is different from those of other identified A. tumefaciens promoters (9), suggesting that transcription of P2 may require an alternative sigma factor.
Because the sequence of P2 is similar to the consensus sequence of the E. coli heat shock promoters, we tested the hypothesis that P2 is a member of a heat shock-like regulon in A. tumefaciens. It was recently reported that the A. tumefaciens heat shock response is similar to the E. coli response and that A. tumefaciens has a u32-like sigma factor (31). The results in this paper indicate that P2 regulation is at least in part independent of the heat shock response. First, we did not detect an increase in virG-lacZ activity after prolonged growth at 39°C (Fig. 3). Others have measured heat shock gene expression by using lacZ fusions (39,65), suggesting that the apparent absence of virG expression is not attributable to our use of lac fusions to measure expression. Second, we used primer extension analysis to measure P2 expression immediately after a shift to high temperature. There was no detectable stimulation of P2 transcription after a shift from 30 to 39°C (Fig. 2). These conditions are sufficient to induce the major heat shock proteins of A. tumefaciens (31). Although virG was not induced by temperature upshift, we cannot rule out the possibility that transcription of P2 may require a u32-like sigma factor in addition to other regulatory factors.
However, certain treatments which induce the heat shock response of E. coli did cause moderate or slight induction of virG (Fig. 3). At least some of these treatments, including those with CdCl2 (57), mitomycin C (25), and alkaline pH (45,56), induce multiple stress regulons in E. coli. In contrast, heat shock and ethanol, which induced virG poorly or not at all, induce only the heat shock proteins in E. coli (57). Low  and heat had similar abilities to inhibit cell growth but not similar abilities to induce virG (Fig. 3). From these data, we conclude that P2 is primarily induced by low pH and is secondarily responsive to certain other stress stimuli, especially those that induce a broad spectrum of stress responses in E. coli. virG was shown to be induced by stimuli that induce the SOS regulon in E. coli, including mitomycin C (60), CdCl2 (57), and alkaline pH (45). There is precedence for genes associated with virulence in other phytopathogens to be induced by SOS inducers (67). However, a recA mutation had no effect on the induction of P2 by acid pH or mitomycin C (Fig. 4), indicating that P2 is induced by these stimuli independent of the SOS response. This result was expected, because P2 does not contain a sequence that resembles a LexA-like binding site (60). Furthermore, the A. tumefaciens recA strain is fully virulent (10).
What is the mechanism of P2 regulation? We hypothesize that P2 might be a member of a pH-inducible regulon. First, because induction of P2 by low pH is mediated by chromosomal genes, one might speculate that this regulatory system could control the expression of additional genes, at least some of which might be located on the chromosome. Second, acidification of growth media resulted in the preferential expression of specific proteins (Fig. 5), and genes encoding these proteins could be coregulated with P2. It was not determined whether these proteins were encoded on the chromosome or the Ti plasmid. Acidic-pH-inducible genes and proteins have been identified in E. coli and Salmonella spp. (11,16,17,49). Whether these genes constitute a low-pH regulon has not been determined.
At least two of the stimuli that induce P2 are associated with plant wounds. Plant wound sites are generally thought to be acidic (24). In addition, many plants produce antimicrobial compounds called phytoalexins in response to mechanical wounding or infection (27). Some phytoalexins have DNA-damaging properties (55). Heberlein and Lippincott reported that treatment of A. tumefaciens with mild doses of mitomycin C or UV prior to infection enhanced tumorigenesis severalfold (14,15). These observations could now be in part explained by virG expression.
The pool size of VirG can limit the efficiency of vir gene induction (4,20,59). We view virG induction as occurring in two steps. The first step involves induction of virG by means independent of VirG protein, specifically, the induction of P2 in response to stimuli described in this paper as well as induction of P1 in response to phosphate starvation (61). These can be thought of as "pump-priming" stimuli which raise the intracellular concentration of VirG protein to a sufficient level that it, in conjunction with VirA, can further induce its own gene as well as inducing all other vir genes. The second step is a positive autoregulatory loop, which could have the effect of strongly committing bacteria to plant infection (63). It is interesting that all these pump-priming stimuli are stressful to bacteria, and this suggests that A. tumefaciens may be more pathogenic when experiencing environmental stress than at other times.