Inducible Expression, Enzymatic Activity, and Origin of Higher Plant Homologues of Bacterial RelA/SpoT Stress Proteins in Nicotiana tabacum*

All living cells possess adaptive responses to environmental stress that are essential to ensuring cell survival. For motile organisms, this can culminate in avoidance or attractile behavior, but for sessile organisms such as plants, stress adaptation is a process of success or failure within the confines of a given environment. Nearly all bacterial species possess a highly evolved system for stress adaptation, known as the “stringent response.” This ancient and ubiquitous regulatory response is mediated by production of a second messenger of general stress, the nucleotide guanosine-3′,5′-(bis)pyrophosphate (ppGpp), which mediates reprogramming of the global transcriptional output of the cell. Accumulation of ppGpp is stress-induced through the enzymatic activation of the well known bacterial ppGpp synthetases, RelA and SpoT. We have recently discovered homologues of bacterial relA/spoT genes in the model plant Nicotiana tabacum. We hypothesize that these homologues (designated RSH genes for RelA/SpoT homologues) serve a stress-adaptive function in plants analogous with their function in bacteria. In support of this hypothesis, we find 1) inducibility of tobacco RSH gene expression following treatment with jasmonic acid; 2) bona fide ppGpp synthesis activity of purified recombinant Nt-RSH2 protein, and 3) a wide spread distribution of RSH gene expression in the plant kingdom. Phylogenetic analyses identifies a distinct phylogenetic branch for the plant RSH proteins with two subgroups and supports ancient symbiosis and nuclear gene transfer as a possible origin for these stress response genes in plants. In addition, we find that Nt-RSH2 protein co-purifies with chloroplasts in subcellular fractionation experiments. Taken together, our findings implicate a direct mode of action of these ppGpp synthetases with regard to plant physiology, namely regulation of chloroplast gene expression in response to plant defense signals.

The relA and spoT genes in bacteria encode enzymes that synthesize the unusual nucleotide guanosine-3Ј,5Ј-(bis)pyro-phosphate (ppGpp), 1 which is a second messenger of the socalled "stringent response" to nutrient deprivation and environmental stress. ppGpp is the intracellular effector of the stringent response, which acts by binding directly to and inducing allosteric modification of the bacterial RNA polymerase (RNAP) (1). This results in global reprogramming of the bacterium's transcriptional activity. There is a general inhibition of transcription and halting of the production of components of the protein synthesis apparatus in order to conserve energy. Simultaneously, there is an induction of stress genes to ensure proper cell adaptation and survival (2). Until recently, it was believed that the stringent response was limited to the bacterial domain of the prokaryote kingdom; however, plant homologues to these bacterial stress enzymes were recently identified (3,4). In Arabidopsis thaliana, a relA/spoT homologue At-RSH1 was discovered in a yeast two-hybrid system using a disease resistance protein as bait (3). Two additional Arabidopsis homologues, At-RSH2 and At-RSH3, were subsequently identified upon completion of sequencing of the Arabidopsis genome (3). In another study, rice RSH genes were recovered from a jasmonic acid (JA)-treated subtractive cDNA library (4). Here we report the identification of a relA/spoT homologue in the tobacco plant, Nicotiana tabacum, which we designate Nt-RSH2 (NCBI accession number AY346377). This gene contains a 2154-bp open reading frame with 78% identity to At-RSH2 and At-RSH3, two of the three RelA/SpoT homologues present in Arabidopsis, and very low similarity (55%) to the third Arabidopsis protein At-RSH1. Tobacco Nt-RSH2 expressed at low levels in bacteria has been found to be active for synthesis of ppGpp synthesis based on two criteria: 1) genetic complementation of a bacterial relA mutant and 2) toxicity to hosts strains lacking ppGpp-degrading activity. This is similar to the behavior of the Arabidopsis homologues (3). In addition, and distinct from all previous studies, we directly demonstrate ppGpp and guanosine 3Ј(2Ј)-diphosphate 5Ј-triphosphate (pp-pGpp) (a precursor) synthetase activity of the purified Nt-RSH2 protein by in vitro biochemical assays. Northern analysis has confirmed the presence of basal level Nt-RSH2 transcripts in tobacco, and cross-species hybridization gives evidence of RSH transcripts in a wide range of plant species. Treatment of tobacco plants with JA or EtOH induced a rapid and persistent increase in accumulation of Nt-RSH2 transcripts and Nt-RSH2 protein. In addition, Nt-RSH2 protein abundance is elevated in response to infection with a bacterial pathogen. Taken to-gether, these findings suggest a central role for Nt-RSH2 in response to biotrophic pathogens and environmental stress. Phylogenetic analysis suggests an ancient symbiotic origin for RHS genes in plants, most likely inherited by horizontal transfer from the chloroplast into the nucleus. Furthermore, we have localized Nt-RSH2 protein to the chloroplasts of the tobacco plants, consistent with the horizontal transfer hypothesis and implicating Nt-RSH2 as a regulator of plastid gene expression in analogy with the role of RelA and SpoT in the bacterial cell.

EXPERIMENTAL PROCEDURES
Plant Material, Growth Conditions, and Treatments-N. tabacum (strain SR1) plants used for isolation of total RNA and for experimental treatments were germinated and grown under sterile conditions on MS medium (Duchefa) solidified with 0.8% agar in "Plantcon" growth boxes (ICN Biomedicals). Plants were grown in environmental chambers programmed for a 14-h light/10-h dark illumination cycle. For chemical treatments, 56-day-old media-grown tobacco plants were treated with 0.1 mM JA (Sigma) for 1 day or with 0.5 mM salicylic acid (SA) (Fisher) for 4 days. As controls for the JA treatment, RNA was harvested from plants treated with 0.5% ethanol (solvent for 1 mM JA). For each treatment, 5-10 plants were grown together in the same growth box. Other plant species used in this study were germinated and grown in soil in environmental chambers or in a greenhouse.
Degenerate PCR and Isolation of NtRSH Gene Fragments-relA/spoT homologous cDNA fragments were amplified by reverse transcriptase-PCR from tobacco total RNA. To avoid bacterial contamination, RNA was isolated from plants germinated and grown under sterile conditions. Degenerate internal PCR primers were based on regions of protein sequence conserved between At-RSH1 and the Escherichia coli relA and spoT genes reported by van der Biezen et al. (3). The forward degenerate primer was GGAATTCCAARYTRGCTGAYCGRTTRCAYA-AYATG, where R represents purine (A or G), Y represents pyrimidine (C or T), M represents A or C, S represents C or G, and W represents A or T. Two reverse degenerate primers were used, CGGGATCCCGMG-TRTGSARRSWYTGRTAYCC and CGGGATCCCGKGTHCKKATYTG-RACYTC. Restriction sites for EcoRI and BamHI (underlined) were added to the forward and reverse primers, respectively, to facilitate cloning. First-strand cDNA was reverse transcribed using an oligo(dT) primer, and this was used for PCR amplification of the relA/spoT sequences. Amplifications using the forward primer plus the two reverse primers produced two bands of the expected size, ϳ550 -650 bp, whereas control reactions produced no observable bands. The fragments were cloned into pBR322; DNA sequencing confirmed homology to expected regions of relA/spoT genes (3).
cDNA Library Screening and Plasmid Rescue-PCR fragments were amplified and 32 P-end-labeled from the cloned fragments described above, gel-purified, and used to screen an N. tabacum SR1 ZAP cDNA library (Stratagene). From ϳ3 ϫ 10 5 plaques screened, five independent clones were isolated, excised to phagmids (in the pBluescript SKϩ vector) according to the manufacturer's recommendations, and sequenced. Two of the five hybridizing clones contained an identical full-length cDNA, the common open reading frame was designated Nt-RSH2, and the plasmid bearing this cDNA was designated as pNt-RSH2.
DNA Sequencing-Sequencing was performed at the University of Buffalo Center for Advanced Molecular Biology and Immunology Nucleic Acid Facility using the Visible Genetics OpenGene Long Read Tower system and Amersham Cy-Dye terminator chemistry and at the Roswell Park Cancer Institute Biopolymer Facility using the Applied Biosystems 377 Prism system with BigDye terminator chemistry. Primers corresponded to vector-borne T7 and T3 primer sites and internal sites at intervals of 500 bp or less.
RNA Extraction and Northern blot Analysis-Isolation of total RNA from leaves of tobacco and other plants and Northern analysis were performed according to methods previously described (5)(6)(7)(8). For the Nt-RSH2 probe used in Fig. 3, A and C, a 32 P-lableled gel-purified 2.1-kb EcoRI fragment from the 5Ј portion of the tobacco Nt-RSH2 cDNA clone was used. For cross-species hybridization in Fig. 3D, a 159-bp fragment from an 83% conserved region of the At-RSH2 and Nt-RSH2 genes (reverse transcriptase-PCR-amplified from Arabidopsis or tobacco total RNA, corresponding to amino acid positions 430 -483 of Fig. 1) was used. For standardization, membranes were rehybridized with a 250-bp PCRamplified DNA fragment for tobacco 18 S rRNA. Blots were visualized, and the relative intensity of hybridization to each band was quantified using a PhosphorImager (Amersham Biosciences) equipped with Image-Quant, version 4.2, software (Amersham Biosciences).
Bacterial Transfection and Complementation-E. coli strain CF1651 is derived from the prototypical wild-type E. coli strain MG1655 and bears the relA "knockout" allele relA251, in which the entire length of the relA gene is replaced with a kanamycin resistance cassette (9). The E. coli strain CF1693 is derived from CF1651 and bears a spoT knockout allele spoT207, in which the entire length of the spoT gene is replaced with a choramphenicol resistance cassette (9). Both strains were transfected with pNt-RSH2 and then plated and grown on LB medium with 100 g/ml ampicillin for ϳ30 h at 37°C. In addition, CF1651/pNt-RSH2 transfectants were screened using SMG selective medium (M9 salts minimal medium plus glucose and 50 g/ml each of serine, methionine, and glycine) with 100 g/ml ampicillin. Plate cultures were grown for 48 h at 37°C, and growth was scored visually. As controls, the strains CF1651/pBluescript SKϩ (vector control) and MG1655/pBluescript SKϩ (positive control) were simultaneously plated on SMG ϩ ampicillin medium.
His-tagged Expression Constructs and Protein Purification-Nt-RSH2 was amplified by PCR using the forward primer, GGGAATTC-CATATGGCGGTTCCGACGATAGCAC (NdeI site underlined) and the reverse primer, CCCCGCTCGAGGAACTGCCGCGACCAGCTCT (XhoI site underlined) from pNt-RSH2. This produced a 1947-bp PCR product encompassed the full-length Nt-RSH2 gene, with an added NdeI site overlapping the AUG start codon and an in frame XhoI site just prior to the natural UAG stop codon. This PCR fragment was cleaved with NdeI and XhoI, gel-purified, and then ligated into corresponding NdeI and XhoI sites of the overexpression vector pET21b (Novagen, Inc.). Ligation of the Nt-RSH2 PCR fragment into pET21b placed the Nt-RSH2 open reading frame "in frame" with six histidine codons located at the end of the protein produced upon expression of the Nt-RSH2 recombinant gene. After confirmation of the DNA sequence, the resulting construct was designated pET21b(Nt-RSH2). Overexpression of Nt-RSH2 was accomplished by transfection of the pET21b(Nt-RSH2) into the strain BL21(DE3) (Novagen, Inc.), which is conditionally inducible for expression of T7 RNAP.
The strain BL21(DE3)/pET21b(Nt-RSH2) was grown in 2 liters of Superbroth (1% yeast extract, 1.6% tryptone, and 0.5% NaCl) with 0.2% glucose, 2 mM MgSO 4 , and 100 g/ml ampicillin at 37°C. Growth was monitored turbidometrically at 600 nm, at an A 600 of 0.5, and isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 1 mM to induce T7 RNAP expression. Nt-RSH2 expression was allowed to proceed for 3 h after the addition of isopropyl-␤-D-thiogalactopyranoside. Cells were harvested by centrifugation and stored as cell pellets at Ϫ75°C. 7.0 g (dry weight) of cells were resuspended in 30 ml of lysis buffer (50 mM sodium phosphate, pH 7.0, 250 mM NaCl, 10 mM imidazole, 0.1% Triton X-100, 5 mM ␤-mercaptoethanol). The complete EDTA-free protease inhibitor mixture (Roche Applied Science) was added to the lysis buffer following manufacturer's recommendations. The cell suspension was incubated for 30 min on ice. Cells in suspension were then disrupted by nitrogen cavitation. The cellular lysate was centrifuged at 12,000 ϫ g, and insoluble (pellet) and soluble protein fractions were checked for the presence of Nt-RSH2 protein by SDS-PAGE and Coomassie Blue staining. Approximately 95% of Nt-RSH2 protein was in the insoluble pellet fraction.
The insoluble pellet was resuspended in 25 ml of lysis buffer plus 4 M guanidine HCl, with vigorous overnight nutation on a rocking platform. The resuspended protein pellet was then diluted 1:1 with 25 ml of lysis buffer lacking guanidine HCl before the addition of 1 ml of a 50% slurry of Ni 2ϩ -nitrilotriacetic acid-agarose (Qiagen). The protein-Ni 2ϩnitrilotriacetic acid-agarose slurry was left nutating for an additional 4 h at 4°C prior to loading on a 1-cm diameter ϫ 20-cm length column (Bio-Rad) with a stopcock flow adapter. The suspension was allowed to drip through with stopcock full open until all of the Ni 2ϩ -agarose resin was collected. Wash solution (50 mM sodium phosphate, pH 7.0, 250 mM NaCl, 35 mM imidazole, 0.1% Triton X-100, 5 mM ␤-mercaptoethanol, complete EDTA-free protease inhibitor mixture) was run through the column until no protein was detected in the elution. The His-tagged Nt-RSH2 was then eluted with wash buffer containing 500 mM imidazole, and each fraction was checked for the presence of protein until no eluting protein could be detected. Protein-containing fractions were pooled, ϳ15 ml total, and precipitated with 0.28 g/ml ammonium sulfate incubated at 4°C overnight. The protein precipitate was harvested by centrifugation at 12,000 ϫ g for 30 min. The pellet was resuspended in 2 ml of 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM ␤-mercaptoethanol and dialyzed overnight against 2 liters of 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM ␤-mercaptoethanol, 50% glycerol at 4°C, with two changes. Final protein concentration was 1 mg/ml. SDS-PAGE analysis indicated greater than 95% purity. A histidine-tagged version of the E. coli RelA protein cloned into the identical pET21b vector was purified in parallel with the Nt-RSH2 (His-tagged) protein using identical procedures.
ppGpp Synthesis Assays-Purified Nt-RSH2 protein was assayed for ppGpp and pppGpp synthesis activity as described for bacterial RelA enzyme (10). Reaction buffer consisted of 50 mM HEPES, 250 mM NaCl, 2 mM EDTA, 1 mM ␤-mercaptoethanol, 14 mM MgSO 4 , 8 mM ATP (0.2 Ci/reaction [␥-32 P]ATP), 6 mM GTP, and/or 6 mM GDP, as indicated in the legend to Fig. 2. Reactions were initiated by the addition of different amounts of Nt-RSH2 protein: 1, 2, 5, and 10 g as indicated in a final volume of 20 l. In control reactions, different concentrations of purified E. coli RelA protein (0.5, 1, 2.5, and 5 g) were added instead of Nt-RSH2. Reactions were allowed to proceed at room temperature for 4 h, and then 12 l of reaction were removed and mixed with 6 l of 3 M formic acid. The entire sample was then spotted in two applications onto the origin of a 20 ϫ 20-cm 100-m pore size polyethyleneiminecellulose flexible TLC sheets (Selecto Scientific). After drying, reaction products were resolved by thin layer chromatography using either 1.5 M potassium phosphate, pH 3.4, or 1.75 M sodium phosphate buffer, pH 3.4, for mobile phase, as indicated. Once the solvent front had reached 4 cm from the top, the TLC sheets were dried for 1 h at room temperature and autoradiographed for 4 -5 h. We note the consistent presence in commercial radioactive [␥-32 P]ATP of a radioactive contaminant that serendipitously migrated to a position located between the migration of ppGpp and pppGpp. We found this contaminant to be present in all commercially available sources of [␥- 32 P]ATP that we tested, and the amount of radioactive contaminant appeared to vary from batch to batch even from the same commercial vendor. For example, as seen in Fig Phylogenetics-We obtained Rel-like proteins from a BLAST search using Nicotiana as well as all Rel-like sequences from different genera of plants, bacteria, and cyanobacteria, with inclusion of the most divergent sequences from the major clades (11). This yielded 47 sequences (see supplementary table for accession numbers and full names) with eight proteins from plants and six from cyanobacteria. Sequences were imported into ClustalX and aligned using the default parameters and the Gonnet 250 matrix (12). An alignment of the conserved core (405 amino acids), which is available from Treebase (available on the World Wide Web at www.treebase.org; New York University, Buffalo, NY) was used for phylogenetic analyses.
Phylogenetic analyses were conducted in PAUP* 4.0 (13) and Mr. Bayes 3.0 (14,15). To estimate phylogenetic uncertainty, we used a Bayesian statistical method with Markov chain Monte Carlo sampling (14). The Markov chain Monte Carlo method samples trees from the universe of possible trees in proportion to their probability given a model of DNA evolution. We used the JTT model (16) with ␥ and covarion parameters. We sampled 100,000 generations with random starting trees and four chains. We removed the first 3000 trees (after graphical inspection) to account for variance due to convergence of the parameters on the Markov chains. The remaining trees were exposed to 50% majority rule consensus tree analysis in PAUP. The proportion of trees containing a clade represents its posterior probability or the probability of being correct given the data and model of evolution.
Nt-RSH2 Antiserum Production-Using the His-tagged purified Nt-RSH2 protein described above we contracted production of rabbit antiserum (Harlan Bioproducts for Science, Inc.). Prior to rabbit inoculation, five preimmune antisera were assayed by Western analysis for the absence of RelA/SpoT cross-reactivity using purified E. coli RelA and SpoT proteins to avoid isolating antisera with bacterial cross-reacting antigens. A single preimmune antiserum was found to be free of crossreactivity to bacterial RelA and SpoT proteins, and the rabbit from which this antiserum was tested was used for Nt-RSH2 immunization, followed by three booster immunizations performed at 1-month intervals for a 3-month period, followed by production and final bleeds 1 month following the third booster immunization.
Fractionation and Analysis of Purified Tobacco Chloroplast with Nt-RSH2 Antiserum-Intact tobacco chloroplasts were isolated from leaf tissue and purified on Percoll gradients as described (17). Soluble and insoluble plastid protein fractions were resuspended in 1 ml each of lysis buffer. Samples (20, 10, 5, and 2.5 l) of these two fractions were mixed with SDS-sample buffer and fractionated by SDS-PAGE. Proteins were transferred to activated polyvinylidene difluoride membrane and probed by conventional Western blot analysis with rabbit Nt-RSH2 antiserum followed by goat anti-rabbit IgG antibody conjugated with horseradish peroxidase. Nt-RSH2 antibodies were visualized by chemi-luminescence using the Renaissance Enhanced Luminol Kit (PerkinElmer Life Sciences) and exposure to x-ray film.
Erwinia Infection Protocols-Infections were performed by point inoculations with Erwinia carotovora carotovora using a pipette tip to produce a small wound on leaves and delivery of 5 l of bacterial suspensions. Controls were mock-inoculated using a sterile tip and saline. Tissue samples were harvested at 20 h postinfections. Harvesting was carried out using a 1-cm diameter cork borer to punch out a disc centered on the inoculation site. The discs were ground in 75 l of protein extraction buffer (50 mM Tris-HCl, pH 7.4, 5 mM EDTA, 10 mM ␤-mercaptoethanol), to which had been added one-tenth volume 5ϫ radioimmune precipitation assay buffer (50 mM Tris-HCl, pH 7.4, 750 mM NaCl, 0.5% SDS, 5% sodium deoxycholate, 5% Triton X-100, 25 mM EDTA).

RESULTS
Degenerate PCR and Isolation of the Nt-RSH2 Gene-For initial isolation, two internal relA/spoT homologous fragments of the expected size were PCR-amplified from tobacco leaf RNA, using degenerate oligonucleotide primers that were based on a region of conserved protein sequences of the E. coli relA and spoT genes and the Arabidopsis At-RSH1, At-RSH2, and At-RSH3 (3). After confirmation of relA/spoT homologous DNA sequences, these PCR fragments were used to probe a commercial tobacco leaf ZAP cDNA library. Five rescued phagmids containing relA/spoT homologues were isolated. Two of these contained identical sequences bearing a full-length open reading frame and untranslated flanking regions and were designated Nt-RSH2. The remaining cDNAs bore partial sequences that differed slightly and may represent separate Nt-RSH genes.
The Nt-RSH2 mRNA (accession number AY346377) is 2551 bp in length and contains an open reading frame for a 718-amino acid polypeptide that spans nt positions 276 -2430. The predicted Nt-RSH2 protein is highly similar (79%) along its entire length to the At-RSH2 and At-RSH3 proteins (698 and 695 amino acids, respectively) of Arabidopsis (3) (Fig. 1). Nt-RSH2 protein shows much lower similarity to E. coli RelA and SpoT proteins, with only 50 and 55% overall similarity, respectively, over the 327 amino acids that span the central ppGpp synthetase domain. This cross-species similarity is approximately the same as that found between the two cytoplasmic Arabidopsis At-RSH proteins and E. coli RelA and SpoT proteins (3). Similarity of Nt-RSH2 to the Arabidopsis At-RSH1 is considerably less, 55% over the 351 central amino acids. Nt-RSH2 shows highest similarity to At-RSH2 and At-RSH3 of Arabidopsis ( Fig. 1) and is more divergent from At-RSH1. Based on this observation, and on an overview of RSH sequences present in GenBank TM , we propose the classification of plant RSH genes into two distinct groupings, RSH1 and RSH2. The RSH1 group would contain genes such as At-RSH1 (and Nt-RSH1, recently entered into GenBank TM as accession number AB095098), which lack a plastid transit sequence and appear to be membrane-associated. The RSH2 group would contain proteins similar to Nt-RSH2, At-RSH2, and At-RSH3, which possess characteristics of soluble, plastid-localized proteins.
Several aspects of the full-length cDNA indicate that Nt-RSH2 is a true nuclear-encoded plant gene and did not originate from a bacterial contaminant or from organellar transcripts. First, a computer search confirmed that the tobacco chloroplast genome (accession number NC_001879) does not contain any Nt-RSH sequences. The Nt-RSH2 cDNA also has a long 5Ј-untranslated region of 276 nucleotides and 3Ј-untranslated region of 316 nucleotides, which is very atypical for bacterial or organellar mRNAs. The Nt-RSH2 5Ј-untranslated region is highly AU-rich, another characteristic typical of nuclear encoded plant mRNAs (18). The initiation codon is in good context (AUGGC) for efficient translation initiation in plants (19). Taken together, these results suggest that Nt-RSH2, At-RSH2, and At-RSH3 all represent a highly conserved group of soluble nuclear encoded RelA/SpoT-like proteins, distinct from the putative membrane-associated Arabidopsis At-RSH1 protein and distinct from the bacterial RelA and SpoT proteins.
The N-terminal portion of Nt-RSH2 has several features associated with plastid transit sequences (20). These include 1) the presence of an amino-terminal MA dipeptide, 2) no charged residues in the amino-terminal 15 amino acids, and 3) an abundance of serine residues (27%) and few Asp (2.9%), Glu (0%), or Tyr (4.3%) residues within the first 90 amino acids. The TargetP version 1.0 program (available on the World Wide Web at www.cbs.dtu.dk/services/TargetP/) (21,22) predicts that Nt-RSH2 has a 0.712 probability of containing an Nterminal chloroplast transit sequence, with a possible cleavage site at amino acid 93. The At-RSH2 and At-RSH3 proteins gave similar results with this program and probably also possess plastid transit peptides. Nt-RSH2, At-RSH2, and At-RSH3 share several conserved regions within their N-terminal portions (positions 1-107 in Fig. 1) that are not found in At-RSH1 or in the SpoT protein from E. coli. These latter two proteins were not predicted to bear chloroplast transit sequences.
Cross-species Complementation of E. coli with NtRSH2-The Nt-RSH2 cDNA borne on the phagemid pNt-RSH2 is fortuitously oriented such that expression of Nt-RSH2 is possible in E. coli by virtue of the presence of the E. coli lac promoter upstream of the Nt-RSH2 gene. Translation in E. coli is possible via an "in frame" fusion of Nt-RSH2 to the Lac ␣-peptide, which overlaps the multiple cloning site of the pBluescript SKϩ vector. Taking advantage of these circumstances, the pNt-RSH2 plasmid was checked for phenotypic complementation in the E. coli strain, CF1651 (9), which is deleted for the relA gene. This relA null strain displays no obvious phenotypic manifestations of the loss of relA, except with respect to growth on specialized selective growth media (2). For example, relA null mutants cannot grow on minimal salt glucose medium supplemented with the amino acids serine, methionine, and glycine (SMG media). Inability of relA mutants to grow on SMG medium is due to the phenomenon that this particular combination of amino acids invokes a metabolic imbalance that results in isoleucine starvation due to suppression of expression of the isoleucine biosynthetic operon, ilvIH. Induction of high level accumulation of ppGpp by RelA following isoleucine starvation induced by growth on SMG medium in turn induces high level expression of ilvIH, leading the cell to overcome SMGinduced isoleucine starvation. In the absence of the RelA stringent factor, growth on SMG medium is restricted due to inability to accumulate high levels of ppGpp needed to prompt adequate high level expression of ilvIH (23). To check for relA phenotypic complementation in E. coli, the relA null strain CF1651 was transfected with the pNt-RSH2 plasmid and checked for growth on SMG medium, alongside the parental strain CF1651(negative control) and grandparental "wild-type" E. coli strain MG1655 (relAϩ spoTϩ; positive control), as well as a pBluescript SKϩ vector transfectant (vector control). After overnight growth at 37°C, growth was scored by visual inspection for two consecutive days (Table I). Obvious growth (indicative of adaptation) of the E. coli relA mutant strains bearing the tobacco Nt-RSH2 gene on a multicopy plasmid on SMG medium was apparent after 1 day of incubation at 37°C. In comparison, growth of even the wild-type E. coli strain was not apparent until 2 days incubation. Thus, it appears that in the presence of the plasmid-borne tobacco Nt-RSH2 gene, unregulated elevated levels of ppGpp are present, which allows for "preadaptation" and a more rapid apparent response to isoleucine starvation due to elevated ppGpp levels leading to elevated constitutive expression of ilvIH. Consistent with this interpretation, we noted that wild-type E. coli strains bearing the pNt-RSH2 plasmid display a slow growth phenotype (data not shown), similar to what is observed for E. coli spoT mutants that accumulate high constitutive intracellular ppGpp levels (2).
Additional evidence for the ability of Nt-RSH2 to synthesize ppGpp in vivo was found when we attempted to transfect relA and spoT double-deleted E. coli strain, CF1693 (9), with pNt-RSH2. We were unable to obtain any stable transfectants of CF1693 with pNt-RSH2, whereas parallel transfections with equal amounts of the vector plasmid pBluescript SKϩ resulted in confluent growth (Ͼ10,000 colonies) on selective plates. We suspect that the reason no transfectants were obtained with pNt-RSH2 is due to the fact that the deletion of spoT together with relA, while it deletes both ppGpp synthetases in the E. coli cell, also deletes the only ppGpp degradase activity contained in the dual function SpoT protein. Therefore, the introduction of a sole ppGpp synthetase activity into the relA spoT doubledeleted strain leads to a situation of "runaway" accumulation of ppGpp, due to a lack of ability to turn over ppGpp, with subsequent toxic effects on cell growth and loss of cell viability (2). This probably explains our failure to obtain stable CF1693/pNt-RSH2 transfectants and is consistent with our hypothesis that Nt-RSH2 is a bona fide ppGpp synthetase.
In Vitro ppGpp Synthesis by Nt-RSH2 Protein-To directly determine the ability of the Nt-RSH2 protein to synthesize ppGpp, we cloned the tobacco Nt-RSH2 gene and the E. coli RelA gene into a T7 promoter-based overexpression plasmid system, fusing them at their C-terminal coding region with a hexahistidine tag for conventional Ni 2ϩ -agarose purification. Both His-tagged proteins were purified in parallel using identical protocols. We had previously found that the presence of a C-terminal hexahistidine tag on the E. coli RelA protein fortuitously activates it for constitutive ppGpp synthetic activity independently of ribosomes. 2 Thus, the His-tagged recombinant E. coli RelA is used here for direct comparison of ppGpp synthesis without inclusion of the normal array of co-factors (ribosomes, tRNA, and mRNA) normally required by the bacterial RelA proteins for activation of ppGpp synthesis activity (2). In vitro reactions contained [␥-32 P]ATP and GTP and/or GDP to assay for pyrophosphate transfer of the ␥and ␤-phosphates from ATP to GTP and/or GDP. Reaction products were analyzed by ascending TLC as described by Cashel (10). Assayed in parallel with the recombinant constitutively active His-tagged bacterial RelA protein, the tobacco Nt-RSH2 protein was apparently able to condense ATP and either GTP or GDP into pppGpp or ppGpp, respectively, in a roughly protein concentration-dependent manner with an apparent higher abundance of ppGpp compared with pppGpp ( Fig. 2A).
Some batches of [␥Ϫ 32 P]ATP used in these studies bore radioactive contaminants that fortuitously chromatographed to a position close to that of ppGpp or pppGpp (observed clearly in Fig. 2B, no protein control). However, alternative chromatographic conditions (Fig. 2C) confirm that the radioactive contaminant is distinct from bona fide ppGpp or pppGpp. Levels of radioactivity at the ppGpp and pppGpp positions increased in manner that requires the addition of RSH2 protein in Fig. 2,  2. In vitro activities of His-tagged purified tobacco RSH2 protein. C-terminal histidine-tagged Nt-RSH2 protein was overexpressed in E. coli and purified to homogeneity. In vitro (p)ppGpp synthesis assays were performed as described under "Experimental Procedures" unless otherwise indicated. A, comparison of ppGpp and pppGpp synthetic activities in the presence of GTP and GDP as pyrophosphate acceptors between E. coli RelA (ECO:RelA) and N. tabacum RSH2 protein (TBCO:RSH2) with increasing amounts of protein as indicated; the positions of migration of the expected products are indicated. B, kinetics of (p)ppGpp accumulation in the presence of GTP and GDP as pyrophosphate acceptors with the Nt-RSH2 protein (TBCO:RSH2) over a 23-h period. Note the presence of a radioactive contaminant present in the no protein control samples (lane 1 and 9) with similar mobility to pppGpp. C, reactions were performed in the presence of increasing concentrations of only GDP or GTP or both (central lane) as pyrophosphotransfer acceptor leading to the accumulation of only ppGpp or pppGpp, respectively. Samples were chromatographed using 1.75 M NaH 2 PO 4 , pH 3.4, as mobile phase to better resolve products from contaminants. Note the different mobility of radioactive contaminants (lanes 1 and 9) compared with ppGpp and pppGpp.
A-C, indicating an Nt-RSH2-dependent increases in the production of (p)ppGpp in this in vitro assay. It cannot be ruled out however, that some degree of the radioactive contaminant, given its mobility, contributes to the apparent radioactivity quantified as either ppGpp or pppGpp.
The kinetics of (p)ppGpp accumulation (Fig. 2B) indicate a slow but steady increase in (p)ppGpp accumulation over a 23-h period. In addition, it appears that ppGpp synthesis occurs more rapidly and for a longer period of linearity than pppGpp (Fig. 2B), accounting most likely for the higher abundance of ppGpp over pppGpp apparent in Fig. 2A. To confirm the identity of the radioactive spots observed in Fig. 2A as ppGpp and pppGpp, respectively, reactions were performed in the presence of either GTP or GDP alone, giving only the possibility of either pppGpp or ppGpp synthesis, respectively. Samples were run together by ascending TLC along with a single reaction sample containing both GTP and GDP as potential pyrophosphate transfer acceptors (Fig. 2C). This result clearly shows that the observed radioactive spots probably correspond to the products ppGpp and pppGpp, respectively, as assigned.
The Nt-RSH2 protein sequence contains the highly conserved HD domain at amino acid residue 268 (Fig. 1, black box), indicative of metal phosphatases, which define a superfamily of metal-dependent phosphohydrolases (11). This domain is strongly associated with the SpoT-like proteins that can degrade as well as synthesize ppGpp (11). In the case of the E. coli SpoT protein, the ppGpp hydrolase activity carries out the following reaction: (p)ppGpp 3 GT(D)P ϩ PP i . We made several attempts to assess the ability of purified Nt-RSH2 to degrade (p)ppGpp but failed to obtained definitive results. Experiments to assay for (p)ppGpp turnover remain ongoing and will be reported elsewhere.
Inducible Nt-RSH2 Gene Expression in N. tabacum-Northern analysis was performed to determine whether the tobacco Nt-RSH2 is transcribed in intact nonstressed plants and whether Nt-RSH2 mRNA levels are affected by known inducers (JA or SA) of plant defense response pathways (24,25).
To identify and quantify any possible induction of Nt-RSH2 mRNA accumulation by JA, a series of treatments were conducted (Fig. 3). Three separate "Plantcon" growth boxes containing control untreated plants (Fig. 3A, C 1 , C 2 , and C 3 ; 5 plants/box), three boxes containing 0.5% EtOH-treated plants (Fig. 3A, E 1 , E 2 , and E 3 ; 5 plants/box), and three boxes containing JA-treated plants each (Fig. 3A, J 1 , J 2 , and J 3 ; 5 plants/box) were used. For each treatment, leaves from all five plants grown together in a given box were pooled for RNA extraction 24 h after treatment. In all three untreated control groups, the Nt-RSH2 probe hybridized to a transcript of ϳ2.5 kb in size. In the three experimental groups treated with JA for 24 h, there is a 3-4-fold increase in levels of NtRSH2 mRNA relative to the untreated control plants (Fig. 3A, compare lanes C 1 , C 2 , and C 3 with lanes J 1 , J 1 , and J 1 ). Fig. 3B shows the average levels of JA induction for Nt-RSH2 mRNA and, furthermore, that these levels were similar for each of the three repeats. These findings are in contrast to those of a similar study with the At-RSH genes of Arabidopsis, where no induction by JA was observed (3).
It is noteworthy that exposure of tobacco plants to ethanol (0.5% final concentration in medium, to control for the ethanol used to solubilize JA) for 1 day was in itself sufficient to induce Nt-RSH2 mRNA accumulation ϳ2-fold 24 h after treatment (Fig. 3, A and B). Thus, Nt-RSH2 mRNA levels appear to also increase in response to EtOH exposure, a condition associated with stress conditions such as dehydration, osmotic shock, and hypoxia (29 -31).
In a separate experiment (Fig. 3C), exposure of the tobacco plants to SA appeared to have no effect on levels of Nt-RSH2 mRNA, relative to levels already present in untreated tobacco plants (Fig. 3C, compare control lane and lane SA), although plants grown at the same time and under the same conditions did respond to treatment with JA (lane JA). This is consistent with the lack of any SA response reported for At-RSH genes of Arabidopsis (3).
Taken together, it appears that the Nt-RSH2 transcript accumulates constitutively in nonstressed plants. Treatment of these plants with SA, a known inducer of plant defense responses (25,27), had no observable effect on the accumulation of these transcripts. JA, another signaling compound that plays a role in defense and senescence (24,25,27,28), was associated with a consistent 3-4-fold increase in levels of Nt-RSH2 mRNA accumulation (Fig. 3, B and C). This increase occurred as early as 8 h of treatment and persisted through 48 h (data not shown). The inducibility of mRNA induction for both the JA and EtOH treatments, relative to nontreated controls, were quantitatively reproducible in independently grown groups of plants ( Fig. 3A; see error bars from average of all experiments in Fig. 3B). Thus, variables such as potential differences in evaporation of EtOH, placement in the growth chamber, or other unknown factors, had little or no influence on our findings. In addition, using immunoanalysis, we found that in parallel with the 3-4-fold increase in Nt-RSH2 mRNA levels, Nt-RSH2 protein synthesis rates were increased 3-4-fold following JA treatment (data not shown). Thus, increases in mRNA levels leads directly to equal increases in Nt-RSH2 protein levels following JA treatment.

RSH Genes Are Expressed in a Variety of Plant Species-
To gain insight about the distribution and expression of RSH genes throughout the plant kingdom, Northern analysis was performed using total RNA isolated from a taxonomic crosssection of green plants. These included Chara longifolia (lane Cl), a eukaryotic algae considered to be closely related to the line that gave rise to higher plants (31); Arabidopsis, a C 3 dicot (lane At); Flaveria bidentis, a C 4 dicot (lane Fb); and Zea mays, a C 4 monocot (lane Zm). Hybridization was conducted using a PCR probe corresponding to a conserved region of the Nt-RSH2 and At-RSH2 genes (this 159-bp region is ϳ83% conserved at the nucleotide level) and amplified from Arabidopsis RNA (Fig.  3D, At-RSH probe). This probe gave somewhat fainter bands with heterologous plant mRNAs than with the Arabidopsis sample as expected, despite approximately equal loading and transfer of the samples (as indicated by hybridization to the 18 S rRNA probe (Fig. 3D, 18SrRNA)). This most likely reflects differences in DNA sequences of RSH genes between widely divergent species rather than relative differences in transcript abundance. Hybridization with the At-RSH PCR probe produced faint images of bands in lanes Cl, Fb, Zm, and Nt. All of the higher plant species showed a hybridization band of approximately the same size (2.6 -2.7 kb), as expected based on previous reports (3) and cDNA sequences reported in the data base. A larger band (ϳ3.0 kb) was observed in the C. longifolia lane. Interestingly, an RSH cDNA and corresponding transcript identified from the eukaryotic algae Chlamydomonas reinhardtii was also found to be larger than those of the higher plant Nt-RSH2, At-RSH2, and At-RSH3 genes (32). Based on this diverse sampling, we expect that RelA/SpoT homologues are widespread and show constitutive basal expression at the level of mRNA accumulation throughout the plant kingdom.
Phylogenetic Analysis of Plant RSH and Bacterial RelA/ SpoT Proteins-Since it appears that the plant RelA/SpoT homologues may constitute a separate branch of the RelA/SpoT superfamily, we conducted phylogenetic analysis to test this possibility. Consistent with this notion, an initial BLAST analysis with the Nt-RSH2 protein sequence against the translated nonredundant data base gives highest alignment similarity scores to plant RelA/SpoT homologue or evolutionarily closely related organisms. For example, the highest scoring alignments are in the following order: red pepper Capsicum annuum RSH gene, Arabidopsis AtRSH2 gene and AtRSH3 genes, and Synechocystis sp.
(blue-green biosynthetic cyanobacteria) RelA/SpoT. Phylogenetic analysis revealed that the plant RSH genes appear to constitute two separate branches within the RelA/ SpoT superfamily. The optimal tree (Fig. 4) grouped the plant sequences together in a clade with Deinoccocus, which in turn was grouped with a monophyletic cyanobacteria clade. Posterior probabilities indicated moderate support for a plant/cyanobacteria clade but strong support for the plant clade and for the dicots containing RSH2-like proteins. Nicotiana grouped strongly with the Capsicum sequence.
Localization of Nt-RSH2 Protein-To follow up on the prediction above that the Nt-RSH2 protein appears to bear a chloroplast-targeting sequence at its N terminus, we assayed by Western analysis for the presence of Nt-RSH2 protein in tobacco chloroplast using subcellular fractionation. Intact tobacco chloroplast were purified on Percoll gradients as described under "Experimental Procedures." Immunoreactivity with anti-Nt-RSH2 antiserum was highly specific for the chloroplast-enriched fractions of tobacco leaf tissue (data not shown). These purified chloroplasts were further fractionated into soluble and insoluble fractions following osmotic lysis and assayed by Western analysis for the presence of Nt-RSH2 protein (Fig. 5). As observed in Fig. 5, all Nt-RSH2 reactivity was found to occur only in the insoluble chloroplast fraction, which includes membranes, membrane-associated ribosomes, and other insoluble constituents. In addition, the chloroplast reactivity corresponding to Nt-RSH2 is smaller than the predicted size of full-length Nt-RSH2 of 718 amino acid residues. This is probably accounted for by removal of the N-terminal 93 amino acids during chloroplast translocalization.
Induction of Nt-RSH2 Protein by Microbial Infection-To further investigate the JA-associated increase in Nt-RSH2 gene expression observed in Fig. 3, and given the fact that JA functions as a messenger that is involved in the activation of some plant stress and defense systems (25), we examined Nt-RSH2 induction following microbial infection of N. tabacum. Tobacco leaves were spot-infected with the bacterial pathogen E. cartovora cartovora, as described under "Experimental Procedures." 20 h postinfection, leaf discs were harvested from infected areas, and protein samples were prepared. When these samples were assayed by Western analyses, levels of Nt-RSH2 protein were found to increase from barely detectable amounts to clearly observable levels (ϳ10-fold) by 20 h postinfection (data not shown). These results are the first direct evidence that a plant RSH protein is induced in response to pathogen infection. Taken together with our findings of JA-inducible Nt-RSH2 gene expression, our results strongly suggest that plant RSH proteins play an active role in a plant-defense response and in plant-pathogen interactions.

DISCUSSION
Nt-RSH2 is a plant homologue of the bacterial stress genes relA and spoT in the higher plant Nicotiana tabacum. Similar homologues have very recently been identified in a small number of other plant species, including the higher plants Arabidopsis (3), rice (4), C. annuum (AY043214.1), and the eukaryotic algae Chlamydomonas reinhardtii (32). By Northern analysis, we have found that transcripts that have homology with and similar size as Nt-RSH2 accumulate in a divergent sampling of higher plant species, including two other dicots; a monocot, C 3 , and C 4 plant species; and a eukaryotic algae. The occurrence of these homologues or their transcripts in a diverse cross-section of plant species indicates a high probability that relA/spoT homologues are present and expressed in most, if not all, higher plant species.
Distinct from the RSH genes of Arabidopsis (3), we observed a rapid and persistent JA-induced increase in the accumulation of Nt-RSH2 mRNA in tobacco (ϳ3-4-fold) and in parallel with an equal increase in Nt-RSH2 protein abundance. Similar to findings in Arabidopsis, Nt-RSH2 transcript levels did not increase in response to treatment with SA. In addition, we observed Ն10-fold induction of Nt-RSH2 protein levels in response to infection with a bacterial pathogen. These findings indicate that Nt-RSH2 gene expression is responsive to at least some defense and stress-related pathways in tobacco but may be induced in response to different signals in Arabidopsis. The distinct SA-induced defense pathway (25) does not appear to be associated with modulation of these transcripts in either plant. In agreement with findings presented here, a recent report identified a relA/spoT homologue as one of a group of genes inducible by JA in the monocot Oryza sativa (4). Based on these observations, we suspect that signals or molecular processes associated with regulation of RSH gene expression will show some variation among different plant species.
Direct biochemical as well as genetic evidence indicates that the Nt-RSH2 protein is a fully functional ppGpp synthetase, in analogy to the bacterial RelA enzyme (Fig. 2). The purified Nt-RSH2 enzyme was found to be capable of bona fide ppGpp synthesis in a highly purified in vitro system and is reported here for the first time for any plant RSH protein. These find- FIG. 5. Western analysis of fractionated purified chloroplast with Nt-RSH2 antisera. Intact tobacco chloroplasts were isolated from leaf tissue and purified on Percoll gradients as described (16). Soluble and insoluble plastid protein fractions were resuspended in 1 ml each of lysis buffer. Samples (20, 10, 5, and 2.5 l) of these two fractions were mixed with SDS-sample buffer and fractionated by SDS-PAGE. Proteins were transferred to activated polyvinylidene difluoride membrane and probed by conventional Western blot analysis with rabbit Nt-RSH2 antiserum followed by goat anti-rabbit IgG antibody conjugated with horseradish peroxidase. Nt-RSH2 antibodies were visualized by chemiluminescence using the Renaissance Enhanced Luminol kit (PerkinElmer Life Sciences) and exposure to x-ray film. Different amounts of soluble and insoluble chloroplast protein extracts are indicated above each lane. Positions of prestained molecular weight markers are indicated at the left.
ings support a role for the unique signal-transducing nucleotide, ppGpp, in the regulation of plant gene expression.
Nt-RSH2, At-RSH2, and At-RSH3 (Fig. 1) all have properties that identify them as likely plastid-targeted soluble proteins. Our subcellular fractionation experiments (Fig. 5) provide additional evidence that the Nt-RSH2 protein is localized to the chloroplast. Taken together with recent findings that an RSH protein from the single cell eukaryotic algae C. reinhardtii localizes to algal chloroplasts in vitro (32), these observations provide insights as to the potential function of RSH protein in plants. We hypothesize that in plant chloroplasts, as in bacteria, Nt-RSH2 produces ppGpp as a regulatory signal in response to plant stress. This unique nucleotide then binds to and modifies chloroplast RNA polymerase, thus reprogramming its transcriptional output. In support of this hypothesis, it has been noted that the chloroplast RNA polymerase is highly homologous to E. coli RNAP (33), where ppGpp is proposed to directly bind and regulate transcription by allosteric modification of E. coli RNAP (1). It is notable that Nt-RSH2 co-fractionates exclusively with the insoluble fraction of the chloroplast extracts, since this fraction contains the majority of translating plastid ribosomes (34). This result is consistent with the interaction of Nt-RSH2 with choroplast ribosomes, again in analogy with the operation of bacterial RelA proteins that are ribosomeassociated. However, in vitro we have demonstrated ppGpp/ pppGpp synthetic activity without additional or accessory protein factors (Fig. 2). Thus, it may be that Nt-RSH2 operates independently of ribosomes and fortuitously co-fractionates with the insoluble chloroplast extract fraction.
The structural and enzymatic similarities between the prokaryotic RelA/SpoT proteins and their plant homologues and our finding that Nt-RSH2 expression is inducible by JA suggest a stress-signaling role for these proteins in plants. The fact that the Arabidopsis At-RSH1 was isolated based on its interactions with RPP5, a known plant R-protein, implies a pathogenesisrelated role for At-RSH1. An RSH gene from the halophyte Suaeda japonica (Sj-RSH) conferred enhanced salt tolerance when expressed in E. coli and in yeast, suggesting that plant RSH proteins might be capable of activating conserved stressresponse genes in these organisms (35). Our findings that NtRSH protein levels are increased dramatically in response to infection of plants with the bacterial pathogen E. carotovora give stronger support to this hypothesis.
Little is known about the regulation of chloroplast genes in response to JA treatment or pathogen attack. JA, which increases during senescence and in response to necrotrophic pathogens, is known to repress plastid gene expression at the levels of transcription and translation (36,37). Changes in chloroplast function, viability, and morphology are also characteristics associated with programmed cell death during senescence and in response to disease (38). Chlorophyll breakdown products that originate in the chloroplasts during these events have been implicated in the modulation of cell death and disease resistance (39,40). Thus, one possible function for RSH genes in plants may be to alter the expression of chloroplastencoded genes in order to prepare the plastid for a role in stress response, disease resistance, or senescence.
Of the major groups on the tree of life, only bacteria and plants possess RelA/SpoT proteins. Our results (Fig. 4) suggest that a lateral gene transfer from bacteria to plants occurred early in plant evolution, because representative RSH proteins from the major plant groups (monocots, eudicots, and unicellular green alga) form a single evolutionary group. A possible route for lateral transfer is from the primordial chloroplast, which is generally considered to have originated from an endosymbiotic cyanobacteria (41). Our findings that higher plant RSH2 proteins contain a putative plastid transit sequence and our observation that Nt-RSH2 protein is localized to tobacco chloroplast support the endosymbiotic transfer hypothesis. Our phylogenetic results are consistent with a cyanobacterial origin for plant Rel-like plant proteins. Nevertheless, there is only modest branch support for this result. This is unsurprising, given the number of processes that make recovering truly ancient events like endosymbiotic transfer from short sequences of extant organisms difficult (41). Our analysis and that of Yamada et al. (35) indicate that the plant RSH family is further divided between sequences similar to At-RSH1 and Nt-RSH1 (the RSH1 group) and those more akin to Nt-RSH2, At-RSH2, and At-RSH3 (the RSH2 group). The former group appears to have association domains for the plasma membrane, whereas the latter group appears to possess plastid transit sequences. The grouping of Deinococcus with the RSH plant proteins could indicate either an ancient transfer from cyanobacteria or from plants; several stress-related proteins of the Deinococcus genome are believed to have been laterally transferred from plants (42). Now that genes encoding RelA/SpoT proteins have been definitively identified in higher plants and in eukaryotic algae and their inducible expression has been confirmed, it is important to determine what role these unique proteins play in plant pathology, stress adaptation, and possibly even development. Our evidence suggests that these genes were transferred from the genome of an ancient symbiote and then retained over a long period of evolutionary time as highly conserved members of plant nuclear genomes. It is therefore likely that RSH genes serve an essential function, possibly conserved from their bacterial origins, that is essential for the adaptability and survival of all plants, possibly operating via regulation of chloroplastid gene expression and, thus, the regulation of plastid metabolism.