The pattern of proteins produced by bacteria represents the physiological state of the organism as well as the environmental conditions encountered. Environmental stress induces the expression of several regulons encoding stress proteins. Extensive information about the proteins which constitute these regulons (or stimulons) and their control is available for very few bacteria, such as the Gram-positive Bacillus subtilis and the Gram-negative Escherichia coli (γ-proteobacteria) and is minimal for all other bacteria. Agrobacterium tumefaciens is a Gram-negative plant pathogen of the α-proteobacteria, which constitutes the main tool for plant recombinant genetics. Our previous studies on the control of chaperone-coding operons indicated that A. tumefaciens has unique features and combines regulatory elements from both B. subtilis and E. coli. Therefore, we examined the patterns of proteins induced in A. tumefaciens by environmental changes using two-dimensional gel electrophoresis and dual-channel image analysis. Shifts to high temperature, oxidative and mild acid stresses stimulated the expression of 97 proteins. The results indicate that most of these stress-induced proteins (80/97) were specific to one stress stimulon. Only 10 proteins appear to belong to a general stress regulon.
Bacteria live in environments that undergo continuous chemical and physical changes, gradual or sudden, that require suitable molecular adaptation processes. This constant requirement for adaptation resulted in the development of complex networks called stress responses that involve the increased synthesis of sets of proteins. The best studied group of stress proteins consists of heat shock proteins (Hsps) – induced by shifts to higher temperatures – and includes proteins such as GroEL (the bacterial homologue of Hsp60) or DnaK (the bacterial homologue of Hsp70). Although the genes encoding stress proteins are highly conserved, considerable variations have evolved in the regulation of the response.
The molecular basis of the stress response has been well studied in the Gram-negative Escherichia coli of the γ-proteobacteria and in the Gram-positive Bacillus subtilis. It appears that the two types of bacteria use different strategies for regulating the induction of stress proteins upon drastic chemical or physical changes [1–4]. The B. subtilis strategy involved the induction of the same set of proteins – called general stress proteins (GSPs) – upon a diverse range of stresses such as heat shock, salt stress, ethanol, and starvation for oxygen or nutrients [2,3,5]. Besides these GSPs, induced by the transcription regulator σB, each condition of stress or starvation increases the production of a specific set of proteins [2,3,6,7]. For example, heat stress induces several groups of proteins including the GSPs, the groE and dnaK operons – controlled by the global repressor HrcA that binds to a conserved regulatory inverted repeat (CIRCE) and others .
In E. coli most of the stress proteins belong to specific stress regulons, each responding to a specific stress agent. For example: the heat shock response is controlled by the transcription activator σ32 (rpoH gene product) [1,8,9], the oxidative stress response is controlled by oxyR and SoxRS [1,10–12], and the SOS regulon is controlled by lexA[1,13]. Some stress conditions induce only one stress regulon, while others – such as CdCl2– induce proteins from all these three regulons . There is no known σB and there may be a general stress regulon activated by RpoS, the stationary-starvation transcriptional activator . In contrast to the situation in B. subtilis, the E. coli stationary-starvation regulon contains fewer proteins whose concentration increases by starvation, acidic environment, heat shock but not oxidative stress.
In other Gram-negative bacteria only limited information is available on the stress regulons. Most studies concentrated on two operons, encoding the synthesis of the GroE and DnaK chaperones . However, even from these limited data it has become clear that the heat shock response of most Gram-negative bacteria is different from that of E. coli (Agrobacterium tumefaciens belongs to the α-proteobacteria, and has previously been shown to represent the majority of Gram-negative bacteria in the control of heat shock response [14,16]). In A. tumefaciens the transcription of genes encoding chaperones is regulated by a σ32-like dependent transcriptional activator, while for the groE operon there are additional controls by a CIRCE-HrcA repression and by post-transcriptional RNA processing [14–21]. There are several experiments suggesting that the VirA-VirG dependent transcription of virulence genes is also induced by several stress agents, such as acidic pH, CdCl2 or mitomycin C and is at least partly independent of the heat shock and SOS response . The complex regulation of the heat shock response in A. tumefaciens and the possible correlation of the stress response with the induction of virulence-associated proteins make this bacterium a good model system for studying stress response and its involvement in bacterial–host interactions.
Significant information on stress-induced stimulons was obtained by analysis of two-dimensional (2-D) protein gels [1,2,7,23–25]. This analysis is based on a dual-channel image analysis that enables the visualization of the content and synthesis rate of a whole set of bacterial proteins on a single electropherogram. By pulse-labeling with L-[35S]methionine, the protein synthesis pattern (computer-stained in red color) can be directly compared with the protein level pattern (computer-stained in green color). Because the total proteins and the newly synthesized proteins (i.e. stress-induced proteins) are on the same gel there is no need for matching several gels and proteins that belong to different stimulons or regulons can be identified .
For the use of this technology it is essential to determine and optimize the conditions for the extraction of the proteins and the 2-D separation. In this communication we describe the conditions for separating the proteins of A. tumefaciens on 2-D gels, and use of dual-channel imaging to analyze the proteins induced by heat shock, oxidative stress and mild acid stress. These conditions are important in the life of A. tumefaciens, that encounter temperature shifts during the day, and oxidative–acidic environments when encountering the plant tissue.
Ninety-seven proteins appear to be induced under the conditions tested, most of them belonging to one stimulon only. The results reported here can provide the basis for future 2-D studies of A. tumefaciens and for analysis of environmental effects on its gene expression.
2. Materials and methods
2.1 Bacteria, media and growth conditions
A. tumefaciens strain C58 GV3101  was grown at 25°C in RK  minimal medium supplemented with 1% mannitol, or H4  minimal salts medium supplemented with 0.2% glucose. A growth rate maximal for these media (doubling time about 2 h) was reached before the various treatments and labeling began. Heat shock was achieved by transferring exponentially growing cells from 25 to 42°C. Other stress conditions were oxidative stress (2 mM H2O2) and pH 5.5, achieved by addition of HCl.
2.2 Protein labeling and extraction
Ten ml of exponentially growing cultures (OD540 nm=0.4, about 4×108 cells ml−1) was pulse-labeled for 15 min with 10 μCi ml−1L-[35S]methionine, 1 Ci mmol−1 (NEN Life Science). For analyzing stress proteins, the label was added 5 min after applying the stress conditions. Following labeling the cultures were quickly cooled to 0°C and excess unlabeled methionine was added. The cells were centrifuged and washed twice with TE-PMSF (10 mM Tris pH 7.5, 1 mM EDTA, 1.4 mM PMSF). The washed cells were resuspended in 0.5 ml of TE-PMSF and disrupted by sonication, three 100-W pulses of 2 min with 1-min intervals, or until fully disrupted. Cell debris was removed by centrifugation at 20 000×g for 30 min at 4°C. The radioactivity and protein concentration were determined using Packard liquid scintillation counter and the Bradford method , respectively. The supernatants containing the proteins were lyophilized.
2.3 Isoelectric focusing and polyacrylamide gel electrophoresis (PAGE)
Crude lyophilized extracts containing 100 μg protein were dissolved for 1 h in rehydration solution (8 M urea; 2 M thiourea; 5.2 μl ml−1 Pharmalites (pH 3–10); 10 mg ml−1 CHAPS (Sigma Chemicals Co.) and 2 mg ml−1 dithiothreitol). IPG (immobilized pH gradient) strips (18 cm, pI 4–7) were loaded and rehydrated for isoelectrical focusing by incubation in the protein-containing rehydration solution for 24 h. The isoelectric focusing was performed in four steps: (1) 0–500 V gradient for 2500 Vh; (2) a constant potential of 500 V for 2500 Vh; (3) 500–3500 V gradient for 10 000 Vh and (4) a constant potential of 3500 V for 35 000 Vh . The second dimension was electrophoresed according to Bernhardt et al. . The gels were silver-stained (sensitive silver stain ), dried and scanned by a light-scanner (Amersham-Pharmacia Biotech), and a phosphorimager (Molecular Dynamics). The data were processed to create dual-channel images according to Bernhardt et al. . Each experiment was performed at least three times and only spots that were consistently induced in all experiments were marked as ‘stress-induced’.
2.4 Protein identification
For the determination of the N-terminal amino acid sequences, proteins were cut from Coomassie blue-stained 2-D gels and eluted onto an Immobiline membrane. Sequencing was performed using an Applied Biosystems A473 sequencer. Matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry was used to identify the sequenced GroEL.
3.1 Analysis of A. tumefaciens proteins by 2-D gel electrophoresis
The 2-D gel electrophoresis analysis was adapted to study expression of A. tumefaciens proteins under various growth conditions. In the first dimension proteins were separated by isoelectric focusing in a linear pH 4–7 gradient and in the second dimension, by SDS–PAGE. Two minimal media, RK and H4, were used to grow the bacteria. The RK medium was found unsuitable, as the electric current in the first isoelectric focusing step was higher than 40 μA/IPG strip for long periods and the proteins were under-focused (data not shown). In contrast, good separation was obtained from cultures grown in H4 medium – the current in the first isoelectric focusing step was always below 40 μA/IPG strip and the proteins could be focused (Fig. 1). In preparations from these cultures we could detect close to 1000 proteins on each gel.
The difference between the two media is probably due to the finding that A. tumefaciens excretes extracellular polysaccharides (EPSs), depending on the carbon source it utilizes. Growth on RK media, using mannitol as a carbon source, results in excretion of large amounts of EPS that contains negatively charged sugars . The charged sugars and the cations bound to them are probably the source for small electrolytes that migrate towards the electrodes in the first isoelectric focusing and result in high current.
3.2 Protein synthesis in exponentially growing cells
A. tumefaciens proteins from exponentially growing cells were extracted and separated as described above. The dual-channel image of proteins from exponentially growing cells (Fig. 2) revealed that most of the accumulated cellular proteins were synthesized in the culture during the labeling (yellow spots). Several proteins were detected as green spots, indicating that they were not synthesized during the labeling. This may reflect a very slow synthesis rate or a very low methionine content. The image also shows red spots, representing proteins whose synthesis during the labeling was high although their cellular level was too low to be detected by silver stain. These proteins may react poorly in silver staining or may be present in levels that can only be detected by the sensitive autoradiogram. However, they may also represent proteins with a short half-life.
3.3 Heat shock-induced proteins
The exposure of exponentially growing A. tumefaciens cells to heat shock conditions, by shifting from 25°C to 42°C, induced the synthesis of a large set of proteins (Fig. 3) and repressed the synthesis of most other proteins. Proteins defined as ‘induced’ represented spots that were undetected in the control and appeared following the treatment, or spots that increased at least 5-fold by the treatment. Fifty-six proteins (pI range 4–7) were induced, 41 out of which were heat shock specific proteins (induced only by heat shock), while the other 16 proteins were also induced under H2O2 oxidative stress or pH 5.5 mild acidic stress, or both. The proteins induced by heat shock were labeled as H# (e.g. H1, H2). Spots numbered H-A are proteins induced by heat shock and mild acid stress and spots numbered H-O are proteins induced by heat shock and H2O2 oxidative stress. H26-G is a protein induced in all three types of stress. One of the vegetative proteins (V1) that retained a relatively constant expression level was N-terminal sequenced and is marked on the gels (Figs. 3–5).
Several Hsps in A. tumefaciens were previously identified, and include GroEL, GroES, DnaK, DnaJ, RpoH and ClpB [17,19–21]. The proteins GroEL and GroES were identified as H2 and H51, respectively (Fig. 3 and Table 1), and, as expected, were found to be induced under heat shock conditions. GroEL was identified by MALDI-TOF analysis of tryptic digested peptides of the protein (data not shown). GroES (H51 in Fig. 3) was identified by N-terminal sequencing (Table 1).
Table 1. Stress-induced proteins identified in this work
Proteins were cut out of Coomassie-stained SDS–PAGE gels and were identified by N-terminal sequencing or MALDI-TOF mass spectrometry following tryptic digestion.
ribosomal protein L7/L12
93% identity to Brucella abortus
super oxide dismutase
94% identity to Sinorhizobium meliloti
periplasmic binding protein of Rhizobium
84% identity to R. leguminosarum
The protein determined as H22-A was identified by N-terminal sequencing (Table 1) as a homologue of ribosomal protein L7/L12. It should be noted that the expression of this protein was already shown to be affected by stress – in B. subtilis it is induced by cold shock  and in Rhizobium by salt stress. In A. tumefaciens the protein was induced under heat shock and by mild acidic stress (pH 5.5) (Fig. 5) but not under H2O2 oxidative stress (Fig. 4).
The vegetative protein V1 is a homologue (84% identity in the 19 amino acid sequenced) of the periplasmic binding protein of Rhizobium leguminosarum (accession number CAB75553) that constitutes part of an ABC uptake system (Table 1). The cellular concentration of this protein was slightly decreased during heat shock but was unaffected by any of the other types of stress.
3.4 Proteins induced by H2O2 oxidative stress and by mild acid (pH 5.5) stress
The addition of H2O2 (2 mM) to exponentially growing A. tumefaciens cells induced 36 proteins in pI range 4–7 (Fig. 4). Twenty-eight proteins were H2O2 specific, six proteins were also induced by heat shock, one protein was induced by pH 5.5 mild acidic stress (O12-A) and one protein appears to be a GSP as it was induced in all the stress conditions examined in this study (H26-G). The protein determined as O10 was identified by N-terminal sequencing (Table 1) as a homologue of super oxide dismutase.
In similar experiments, a rapid shift from pH 7.2 to pH 5.5 resulted in induction of 23 proteins (Fig. 5). In addition to proteins also induced under other conditions – nine proteins also induced by heat shock, one protein induced by H2O2 oxidative stress (O12-A) and the GSP (H26-G) – 12 proteins (A1–A12) were specific to mild acid stress.
A. tumefaciens is a plant pathogen that is important both as a plant pathogen and as a tool for plant genetic engineering. In addition, it represents a significant group of bacteria, the group of α-proteobacteria. This group includes environmentally important bacteria such as the nitrogen fixing bacteria Rhizobium, Bradyrhizobium, Rhodospirillum and Azospirillum (purple non-sulfur bacteria) and bacteria important in breakdown of pollutants, such as members of the Sphingomonas group. In addition, the α-proteobacteria contain several important intracellular pathogens such as the Brucella group and the Rickettsia group. Yet, only fragmented information is available on the stress control of these organisms [17–19,21,34–37]. The information presented in this study is the basis for extensive examination of proteomes in A. tumefaciens. Studies on this bacterium, that can be grown in controlled media and controlled conditions in the laboratory, will provide essential information for understanding regulatory aspects of the whole group of bacteria, including these members that are difficult to grow and study in laboratory conditions.
The data presented in this paper are based on 2-D gel analysis of proteins in A. tumefaciens. These bacteria produce EPSs that interfere with the electrophoretic procedures. Growth conditions were obtained in which the level of polysaccharides was reduced and electrophoretic separation was enabled. The availability of proteomic tools for the physiological studies of A. tumefaciens is of special importance, as DNA sequence is as of now not available. In addition, transcription analysis may not be representative of the expression of protein, as it has been shown that this bacterium has at least two post-transcriptional control mechanisms – control of transcript stability  and specific, stress-induced transcript cleavage .
Three stress conditions were examined – a shift to higher temperature (25°C to 42°C), oxidative stress (2 mM H2O2) and mild acid (pH 5.5). The stress conditions were studied because of their relevance to the life of the bacterium. Since it is a plant pathogen it encounters daily temperature changes, and when interacting with the host it encounters oxidative stress and acidic conditions. Ninety-seven proteins were over-expressed following stress. Each stress resulted in the induction of a unique group of proteins – a shift in temperature induced the expression of 56 proteins (Fig. 3), 35 were induced by oxidative stress (Fig. 4) and 23 by mild acid (Fig. 5). Only a small portion of the stress proteins were induced by more than one stress condition (Fig. 6). Only one protein (H26-G) was expressed as a GSP and was induced under all three stress conditions (Figs. 3–5).
Among the nine proteins that were induced by a shift to higher temperatures and by acid stress, the protein marked as H22-A was identified as a homologue of the ribosomal protein L7/L12 (Table 1). This ribosomal protein was previously shown to be induced under different stress conditions in different bacteria: cold shock in B. subtilis, heat shock in Bradyrhizobium japonicum and salt stress in Rhizobium (unpublished results). As the ribosomal protein L7/L12 has an important role in the mRNA–ribosome interaction and in binding of several factors essential for accurate translation , its induction in several different stress conditions may reflect the need for rapid synthesis of a specific set of protective proteins. Alternatively, the finding of a ribosomal protein induced by shifts to high temperatures supports the idea  that ribosomes play a role in the sensing of environmental changes.
The results presented here are compatible with the previous information obtained from experiments in A. tumefaciens, indicating the existence of several stress dependent regulators and a high degree of complexity. As in other Gram-negative bacteria, such as E. coli, there is a σ32-like positive transcription factor involved in the upregulation of Hsps encoded by the dnaKJ operon, the groESL operon, clpB and grpE[15–18,20,21]. Yet, A. tumefaciens shares at least one control element with the Gram-positive bacteria (HrcA-CIRCE) which represses the groESL operon under non-heat shock conditions . Another regulon, induced under different stress conditions, is the vir regulon that is regulated by the VirA-VirG two component system  and appears to be involved only in A. tumefaciens pathogenicity. In this paper we described sets of proteins that are induced by oxidative and acid stresses (Figs. 4 and 5) indicating the existence of additional regulators, still to be identified. The finding of nine proteins that are induced by heat shock and acid stress, but not by oxidative stress, suggests the potential existence of an additional regulon – a general stress response – functionally homologous to the B. subtilis regulon controlled by σB. However, in contrast to the B. subtilis system, the general stress response in A. tumefaciens includes a significantly smaller number of proteins and could be functionally minor.
We are grateful to Jörg Bernhardt for his help with the dual-channel imaging. This work was partially supported by a FEMS fellowship (R.R.) and by the Manja and Morris Leigh Chair (E.Z.R.) for Biophysics and Biotechnology.