INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Maintaining the integrity of genetic information is essential to all organisms. Consequently, all species have evolved diverse ways of maintaining the fidelity of their genetic material, particularly in response to a harsh environment (10, 30). Perhaps the most comprehensively studied system for the repair of DNA damage is the SOS regulon of Escherichia coli.

The SOS regulon of E. coli is composed of at least 20 genes scattered throughout the chromosome (10, 12). These genes all share the property that their transcription is induced or elevated by DNA damage. The mechanism whereby E. coli's SOS regulon is induced has been intensively studied with a variety of DNA-damaging agents over the past 30 years. These studies have provided a detailed picture of the molecular mechanism of SOS activation.

A variety of proteins that recognize damaged bases and effect their repair exist in the cell. Nucleotide excision repair is a common mechanism for DNA repair shared by many species (22). In E. coli and Salmonella, this method of repair does not require SOS induction (although its component enzymes are under SOS regulation) and acts on double-stranded DNA.

When the nature or level of DNA damage is such that excision repair is overwhelmed, DNA replication enzymes will encounter the damaged bases. In this event the replication enzymes can bypass the damage and reinitiate replication downstream, leaving a single-stranded gap of approximately 1,500 nucleotides called a postreplication daughter strand gap (11, 22, 23). This gap harbors the damaged bases, which must be repaired using the complementary strand on the sister chromosome (31). The formation of a postreplication daughter strand gap can lead to a double-strand break (14). This double-strand break can be repaired using the sister chromosome as a template (27).

Both the gap and the double-strand break are substrates for recombination enzymes (10). These enzymes create single-stranded DNA (ssDNA), a necessary intermediate, leading to the recombinational repair of the damaged DNA. This ssDNA is also one of the key effectors of SOS induction (2).

Once ssDNA is generated, the RecA protein polymerizes on the single strands, forming a nucleoprotein filament (13). This RecA-ssDNA filament acts as a catalyst for the cleavage of the LexA repressor protein. The LexA protein negatively regulates the transcription of the SOS regulon by binding to sites found in the 5' ends of these genes. Thus, the generation of ssDNA leads to the RecA-mediated cleavage of the LexA repressor and the consequential derepression of the SOS operon (2, 21).

The derepression of the SOS operon results in the induction of DNA repair and recombination enzymes, which effect the recombination and repair of the damaged DNA. Once recombination and repair have ceased, the concentration of ssDNA returns to normal levels, cleavage of LexA is terminated (the lexA gene is itself regulated by LexA), and homeostatic levels of LexA are reached. Thus, the SOS regulon is repressed and the cell returns to its "wild-type" physiological state.

While the SOS response in E. coli is well documented, relatively little is known about the regulon in Salmonella. It is certain that Salmonella enterica serovar Typhimurium exhibits an SOS response, since one can isolate mitomycin C-inducible LacZ fusions (J. Roth, unpublished data) and Salmonella serovar Typhimurium has homologues of E. coli SOS genes including the lexA, recA, sulA, recN, uvrABD, and umuDC genes (10, 25).

In this work, we investigate the SOS response to DNA damage using a classical inducer of SOS, mitomycin C. We present an analysis of differential gene expression using the technique of RNA fingerprinting by arbitrarily primed PCR (RAP-PCR). The results suggest that, in the 120 million years since E. coli and Salmonella serovar Typhimurium diverged from a common ancestor, their SOS regulonswhile retaining similar regulatory and enzymatic featureshave otherwise significantly diverged. In addition, we reveal that Salmonella serovar Typhimurium possesses an mitomycin C-induced, RecA-independent pathway of gene activation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, media, and reagents. Our wild-type strain of Salmonella serovar Typhimurium is DB7000 [originally from David Botstein; leuA414(Am) Fels2] and was provided by Mimi Susskind. Our recA mutant strain, TT18642 [leuA414(Am) Fels2 recA1] is described in reference 3. Standard growth medium is Luria-Bertani (LB) broth supplemented with 1× E salts and 0.2% glucose (7). The pCR2.1 cloning vector, transformation strains, methods, and supplies for cloning differentially expressed genes were provided in the Original TA Cloning Kit obtained from Invitrogen Corp. and used according to the manufacturer's protocols. DNase I (RNase free) and RNase inhibitor (human placenta) were purchased from Boehringer Mannheim. Moloney murine leukemia virus (MMLV) reverse transcriptase was purchased from Promega. The Stoffel fragment of Taq DNA polymerase was purchased from Perkin-Elmer.

Synthetic oligonucleotides. All oligonucleotides were purchased, desalted, from Genosys and resuspended in 10 mM Tris, pH 8. The sequences of the oligonucleotides used to identify regulated genes in RAP-PCR experiments are listed in Table 1. Additional oligonucleotides were used to confirm differential expression and were based on the sequence of the particular fragment. Due to the length of the latter list, we have chosen not to present it here but will gladly provide it to interested individuals.

Mitomycin C induction protocol. A fresh overnight culture of a wild-type or recA mutant strain was diluted 4,000-fold into LB broth supplemented with 1× E salts and 0.2% glucose. This culture (50 ml) was grown to a density of ~2 × 10⁷ cells/ml. The culture was then split into 2 equal volumes, to which an equal volume of LB broth at 37°C (supplemented as above) was added. The two cultures were allowed to shake at 37°C until the concentration of cells was between 3 × 10⁷ and 4 × 10⁷ cells/ml. At this point 6 ml of each culture was withdrawn, distributed into 4 × 1.5-ml microcentrifuge tubes, and microcentrifuged for 45 s; then the supernatant was aspirated off and the tube was immersed in liquid nitrogen. These samples were the time zero samples. While the time zero samples were being processed, mitomycin C was added to one of the cultures (final concentration, 0.5 or 2.0 µg/ml; the remaining culture serves as the uninduced control) and both flasks were returned to the shaking water bath. Ten minutes (t = 10) after time zero, another 6 ml was withdrawn from each flask and treated like the time zero samples. The same procedure was used for the 30-min samples. The same procedure was also used for the 60-, 90-, and 120-min samples except that only 3-ml samples were withdrawn. The frozen cell pellets were stored at 80°C until RNA extraction.

Extraction and purification of RNA. Frozen cell pellets were treated in groups of no more than six tubes. Each tube received 0.4 ml of 30 mM sodium acetate (pH 5.2) and 20 µl of 20% sodium dodecyl sulfate (SDS) (both preheated to 70°C). Each tube was vortexed vigorously and incubated in a 70°C water bath for 30 to 60 s. Each tube then received 0.4 ml of phenol (preheated to 70°C; equilibrated with 30 mM sodium acetate, pH 5.2) and was vortexed vigorously. Each tube was incubated at 70°C for an additional 10 min with occasional vortexing, then microcentrifuged for 15 min at high speed. The aqueous top layer was removed, extracted with 0.5 ml of phenol and microcentrifuged for 5 min. The aqueous layer was removed, extracted with 0.5 ml of chloroform, and microcentrifuged for 2 min. The aqueous layer was removed, brought up to a total volume of 0.4 ml with water, and combined with 45 µl of 3 M sodium acetate (pH 5.2). RNA was precipitated by addition of 0.9 ml of 100% ethanol, and the precipitated RNA was pelleted by microcentrifuging for 20 min at high speed. The nucleic acid pellet was briefly dried under a vacuum and resuspended in 10 mM Tris, pH 8, such that each time point sample was resuspended in a total of 80 µl. This preparation was then treated with DNase I.

Treatment of RNA preparations with DNase I. Each 80-µl preparation of crude RNA was treated with DNase I by combination with 20 µl of a 5× cocktail (100 mM Tris [pH 8], 50 mM MgCl₂, 1.2 U of RNase inhibitor/ml, 0.4 U of DNase I/ml) and incubation at 37°C for 40 min. The RNA was purified using RNeasy spin columns (Qiagen) according to the manufacturer's specifications. Purified RNA was eluted in water and adjusted to a final concentration of 100 ng/µl in water. RNA preparations were stored at 80°C.

General methods for the identification of differentially expressed fragments. In this work, we have used two different methods of identifying differentially expressed fragments. These methods differ in the method of synthesizing cDNA. In our original procedure, we used a 9- or 10-nucleotide primer of defined sequence to prime the reverse transcription (RT) reaction. The cDNA made from this "specific-RT" reaction was amplified (under arbitrary conditions) by the addition of a second primer. In this procedure, the cDNA amplification is accomplished by a "primer pair": the RT primer and the cDNA primer. Our alternate method (the random-hexamer RT method) uses random hexamers to prime cDNA synthesis; the cDNA is then amplified (under arbitrary conditions) by adding a gene-specific primer pair. The use of random hexamers to prime cDNA synthesis is particularly useful when one wishes to compare many different primer pair combinations (i.e., different genes) against one RT reaction.

RT. Final reaction conditions for the RT reaction were 4 U of MMLV reverse transcriptase/ml, 2 µM arbitrary primer (or random hexamers), 50 mM Tris (pH 8.3; at 37°C), 50 mM KCl, and 4 mM MgCl₂. Total amounts of RNA in a 10-µl reaction volume ranged from 1 to 0.2 µg. The reaction mixtures were assembled by spotting a 5-µl aliquot of a 2× cocktail containing everything except RNA and 5 µl of RNA on the sides of a 0.2-ml PCR tube. The reaction was initiated by briefly centrifuging the tubes; the reaction tubes were incubated at 37°C for 1 h, heated to 95°C for 10 min, diluted with a 3× volume of water, and stored at 20°C. This mixture was used directly in the cDNA amplification reaction.

cDNA amplification. The cDNA amplification procedure has been described elsewhere (16, 17).

Identification, isolation, purification, cloning, and sequencing of candidate differentially expressed fragments. The identification, isolation, and purification of differentially expressed genes have been described in detail previously (15, 16). Briefly, 3 µl of a RAP-PCR mixture was diluted with 12 µl formamide of loading dye (95% formamide, 10 mM EDTA, 0.1% bromophenol blue, 0.1% xylene cyanol) and denatured at 94°C for 5 min, and 2 µl was loaded onto a 7 M urea-5% polyacrylamide gel (Accugel; National Diagnostics) and run for 2 h at a constant 50 W. The gel was dried and exposed to film, and differentially expressed fragments were identified. If a differentially expressed fragment was observed, the remainder of the reaction mix was run on a preparative gel; the band was excised from the gel and the DNA was eluted by soaking the gel slice in 10 mM Tris (pH 8) for 5 to 24 h at 50°C. The eluted nucleic acid was precipitated with sodium acetate and ethanol, pelleted, dried, and resuspended in 10 µl of SSCP running buffer (10 mM NaOH, 95% formamide, 0.1% bromophenol blue, 0.1% xylene cyanol). The sample was denatured for 3 min at 95°C, cooled on ice, loaded onto a SSCP gel (FMC Bioproducts) (15), and run overnight at 8 W (constant). The gel was dried, the fragment was visualized by autoradiography, and the band was excised and eluted by soaking (as before) in 150 µl of 10 mM Tris (pH 8). An aliquot of this mixture was amplified using the appropriate primers and PCR was run on a preparative 2% agarose (SeaPlaque; FMC Bioproducts) gel. The amplified fragment was visualized with ethidium bromide and UV lighting, excised, and purified with the QIAquick Gel Extraction Kit (Qiagen). This nucleic acid was directly cloned into the vector pCR2.1 supplied with the Original TA Cloning Kit (Invitrogen). Clones harboring inserts were identified and sequenced on a Licor 4200 automated sequencing apparatus. Fragments were sequenced on both strands using the vector's M13 priming sites. All sequence data have been deposited with GenBank. Accession numbers are listed in Table 2.

Verification of differential expression. Once a candidate differentially expressed fragment was sequenced, complementary primers 18 to 25 nucleotides in length were designed which specifically amplified a portion of the cloned sequence. Differential expression of the sequence in question was then confirmed with those primer pairs by the following procedure. RAP-PCR was performed on cDNA (primed from random hexamers) using a gene-specific primer pair under the following conditions: 25 cycles of 30 s at 95°C, 45 s at 35°C, and 60 s at 72°C, followed by 10 min at 72°C. Synthesized DNA was quantified by ³²P incorporation. If the appropriately sized band showed the expected differential expression, this was considered confirmation of differential expression. Each fragment identified in Table 2 was shown to be differentially expressed at two different concentrations of mitomycin C (0.5 and 2.0 µg/ml), although we present data only for a mitomycin C concentration of 0.5 µg/ml (see Fig. 2 and Table 3). In addition, differential expression at each concentration of mitomycin C was confirmed on RNA derived from minimum of three independent induction experiments.

Quantification of differential expression. The differential expression of the gene fragments listed in Table 2 was quantified using a Molecular Dynamics Storm PhosphorImager. For each time point, the differentially expressed band was normalized to a constant band or region within each lane. The same background band or region is used across all time points for a particular fragment. Each of these "internally" normalized time points is then normalized to the time zero point. The data presented in Table 3 are the results of quantifying a typical induction experiment for a mitomycin C concentration of 0.5 µg/ml. Data for a mitomycin C concentration of 2.0 µg/ml were also generated with similar results (data not shown).

Identification of fragment sequences. The sequence of each differentially expressed fragment was compared to the National Institutes of Health (NIH) nonredundant (nr) sequence database using the BLASTN and BLASTX programs (1). The results of the BLASTN search were used to classify the sequence as a novel or known sequence. A sequence was classified as known if our sequence had a greater than 98% match with the database sequence with an expectation value (E) of 10¹⁰. This criterion allowed an unambiguous classification of the data. If there was no significant nucleotide homology with the database, a BLASTX search was performed. The BLASTX program translates the query sequence in all six reading frames and compares this to a protein database. The highest-scoring result(s) of the BLASTX search is reported in Table 2 if the expectation value is 10³.

Nucleotide sequence accession numbers. The sequences of the following differentially expressed fragments have been deposited in GenBank under the accession numbers given in parentheses: 10-1p257 (AF206676), 12p257 (AF206677), 1p257 (AF206678), 1p85 (AF206679), 1p90 (AF206680), 2p257 (AF206681), 4p85 (AF206682), 4p90 (AF206683), 5p257 (AF206684), 6p257 (AF206685), 7p257 (AF206686), 8p257 (AF206687), ap88 (AF206688), ep43 (AF206689), fp94g1 (AF206690), fp94g2 (AF206691), jp94 (AF206692), lp94 (AF206693), rw1 (AF206694), and x1p203 (AF206695).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of genes induced by treatment of Salmonella serovar Typhimurium with mitomycin C. A fresh culture of Salmonella serovar Typhimurium was treated with two different concentrations of mitomycin C as described in Materials and Methods. An untreated control for each time point was also taken, although time points other than time zero were not included in the initial experiments. RNA fingerprinting reactions were run on the RNA isolated from treated and untreated cells. A typical RNA fingerprinting gel is shown in Fig. 1. This figure shows an RNA fingerprint derived from the EA10-OPN22 primer pair.

View larger version (44K): [in this window] [in a new window]   FIG. 1.   Standard RAP gel for a mitomycin C treatment of 2.0 µg/ml. In this experiment, the RT primer is EA10 (Table 1) and the cDNA amplification primer is OPN22. The time zero point is used as the untreated control. At each time point, two RNA samples were analyzed, using a total of 500 µg (Hi) or 250 µg (Lo) for each time point. In this particular experiment, bands A, B, and C were extracted and isolated as described in Materials and Methods.

Confirmation and quantification of differential expression. Once a candidate gene fragment such as that shown in Fig. 1 was identified, it was extracted from the gel, purified, cloned, and sequenced. From the sequence data, we then designed a specific, homologous PCR primer pair that would amplify a portion of the fragment. This specific PCR primer pair was employed to amplify first-strand cDNA made using random hexamers. These specific PCRs were used to independently confirm the observation of differential expression. Figure 2 shows the results of such a confirmation experiment for two different concentrations of mitomycin C. 

View larger version (93K): [in this window] [in a new window]   FIG. 2.   Confirmation of differential expression using a standard quantitative RAP-PCR gel. In this experiment, RNA was reverse transcribed by priming with random hexamers, and the subsequent PCR amplification was accomplished with a gene-specific primer pair. Molecular weight markers are on the extreme left and in the middle of the gel. The experiment for which results are shown on the left (with respect to the central molecular weight marker) was done on wild-type Salmonella serovar Typhimurium at 0, 0.5, and 2.0 µg of mitomycin C (MC)/ml. Samples were withdrawn and analyzed at the indicated time points. The right side of the figure shows the results of the same protocol carried out on a recA mutant strain. The data clearly indicate that the gene is not induced in the recA mutant, while other background bands remain unaffected.

The genetic regulation of mitomycin C-induced operons in Salmonella serovar Typhimurium. In E. coli, the SOS response to DNA damage is under the direct control of the LexA protein. The RecA protein is the effector that passes on the signal for SOS induction by stimulating cleavage of the LexA repressor. In E. coli, recA null mutants are unable to stimulate the LexA cleavage reaction and are unable to establish the SOS response. If the mitomycin C-inducible regulon we have revealed in Salmonella serovar Typhimurium is analogous to the E. coli SOS regulon, we would predict the Salmonella regulon to be under LexA-RecA control.

View larger version (40K): [in this window] [in a new window]   FIG. 3.   An experiment similar to that described for Fig. 2 was performed for the RW1 sequence. The transcript shows induction by mitomycin C (MC) (at 0.5 and 2.0 µg/ml) in the presence and absence of a functional RecA protein. The gene is clearly induced in the recA mutant strain and therefore is the first example of an SOS-induced gene with a regulatory structure different from that of the standard MC-induced regulon. Quantified induction levels are shown in Table 3.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have conducted a survey of genes induced by the DNA-damaging agent mitomycin C in Salmonella serovar Typhimurium. The technique we have used, RAP-PCR, is capable of identifying both increased and decreased mRNA expression against a background of unchanging expression (17). We have described 20 differentially expressed genes, all of which are activated by mitomycin C. We did not find any transcripts which were repressed by mitomycin C, although these experiments revealed at least five genes whose transcripts were repressed by entry into stationary phase (data not shown).

The fact that only induced genes were found may be a bias of the RAP-PCR system. Repressed genes will be observed only among those genes that generate PCR products in the controls. In contrast, induced genes can be sampled from the PCR fragments that are not visible in the control but become visible upon induction. The class of genes not visible on the gel is obviously larger than the visible class, so the class of observed up-regulated genes will usually be larger than the class of observed down-regulated genes.

The regulation of the 20 mitomycin C-induced genes was investigated in a recA mutant background. We found that 19 of the 20 genes required the presence of an active RecA protein for induction by mitomycin C. The requirement for an active RecA protein, and the identity of some of the genes revealed, affirms that Salmonella has an SOS operon analogous to that in E. coli, as other investigators have shown (20, 25). The discovery that the RW1 gene is induced in the absence of a functional RecA protein reveals the presence of a novel, RecA-independent, SOS transcriptional activation mechanism.

These data represent the application of an approach to gene expression different from that used in E. coli. The SOS system in E. coli has been largely analyzed with genetic methods. Most of the genes found in the E. coli SOS regulon were revealed by use of Mu dlac operon fusions (reviewed in references 10 and 30). While the utility of such fusions is widely known, their potential as a gene discovery tool can be limited by the effect of the Mu dlac insertion on the viability and growth rate of the cell. In contrast, RAP-PCR is a biochemical approach whose sampling method has no genetic consequences on the cell. In principle, if a large enough spectrum of PCR primers is employed, RAP-PCR could detect every regulated gene regardless of the role that gene plays in the cell's health.

Our data show that a variety of genes are induced in Salmonella serovar Typhimurium, with a spread of induction time points and with variable induction ratios. These data, shown in Table 3, are all within the parameters for components of the SOS regulon described for E. coli (10).

Identities of gene fragments. We have classified the gene fragments found in this study into two broad categoriesgenes of known function and novel genes, including those distantly related to known genes. This information is summarized in Table 2 for each gene fragment. Those fragments whose sequence (or a homologous sequence) has been previously reported in the public databases are described here as "known" genes (5 of 20 genes). Those gene fragments for which no close match or homologue could be found are described here as novel genes (15 of 20 genes). The criterion used as the basis for these designations is discussed in Materials and Methods.

Known genes. The sequences characterized here as known Salmonella genes are sulA, stpA (h-ns), umuC, and recE. A fifth sequence corresponds to a hypothetical gene in the lysU-cadA intergenic region of E. coli and is presumed to be the Salmonella homologue. The results of BLAST searches for these sequences are shown in Table 2.

Novel genes. The majority of fragments (15 of 20) revealed in this work have poor nucleotide homology with the public sequence databases. The range of BLASTN expectation values for this group was e¹⁰ to 2.8, with two sequences having no hit whatsoever.

3

3

10