mgr, a Novel Global Regulator in Staphylococcus aureus

Strains, plasmids, and growth conditions. Escherichia coli strain XL1-Blue was used as the host strain for plasmid constructions. The S. aureus strains and plasmid vectors used in this study are listed in Table 1. S. aureus RN4220 (13) was used as the recipient for plasmid electroporations. S. aureus strains were cultivated in Trypticase soy medium (Difco Laboratories, Detroit, Mich.). E. coli strains were routinely cultivated in Luria-Bertani medium (Difco). Electroporation was carried out by the procedure of Kraemer and Iandolo (12). Plasmid transduction using phage 52A was carried out as previously described (28).

Phenotypic characterization.CP8 production was determined as described before (17) with the following modification. Two milliliters of an 18-h culture adjusted to an optical density at 660 nm (OD₆₆₀) of 5.0 were pelleted and resuspended in 100 μl of phosphate-buffered saline. The suspension was treated consecutively with the following enzymes at 37°C: 100 μg of lysostaphin/ml for 15 min, 300 U of DNase I/ml for 15 min, and 100 μg of proteinase K (Sigma, St. Louis, Mo.)/ml for 1 h. The proteinase K was subsequently inactivated at 75°C for 10 min. The crude CP8 preparations were assayed by the immunoblotting method as described previously (17).

Crude extract of protein A was prepared in the same way as that of CP8 except that proteinase K treatment was omitted. Serial dilutions of the extract were blotted onto nitrocellulose membranes using a dot blot or slot blot apparatus (Bio-Rad). The membranes were incubated with a horseradish peroxidase-conjugated rabbit anti-guinea pig immunoglobulin G for 1 h, washed, and developed by color reagent (Bio-Rad). A semiquantitative assay of protein A was performed by immunoblot assay of colonies on agar plates.

Coagulase production was assayed by measuring the coagulation activities of equal-volume mixtures of whole-cell cultures and twofold serial dilutions of rabbit plasma (Difco). DNase production was assayed by reduction of the OD₂₆₀ upon incubation of herring sperm DNA with culture supernatant as described by Smeltzer et al. (29). Alpha-toxin was assayed by Western blotting of the crude cell extracts using affinity-purified anti-alpha-toxin antibody (Toxin Technology, Sarasota, Fla.) and donkey anti-sheep antibody- horseradish peroxidase conjugate (Sigma). The semiquantitative assay of alpha-toxin was performed by streaking cultures on sheep blood agar plates. The protease assay was performed essentially as described previously (29).

DNA manipulations.Standard DNA manipulations were carried out as described by Sambrook et al. (24). Rapid small-scale plasmid DNA isolation was performed according to the procedure of Holmes and Quigley (9). Qiagen (Chatsworth, Calif.) DNA isolation kits were used for general plasmid purification. Bulk chromosomal DNA from S. aureus was isolated using a chromosomal DNA purification kit (Promega, Madison, Wis.). The enzymes used in DNA manipulation were obtained from GIBCO-BRL (Gaithersburg, Md.), New England Biolabs, Inc. (Beverly, Mass.), or Promega Corp. PCR amplification was carried out with the Advantage cDNA PCR kit (Clontech, Palo Alto, Calif.). The transfer of DNA to nitrocellulose membranes was performed according to the method of Southern (30).

RNA extraction and Northern hybridization.A Blue FastRNA kit (Bio 101, Inc., Vista, Calif.) was used in RNA extraction (3). Northern hybridization was carried out as described by Sambrook et al. (24). Total RNAs were resolved in a 1% agarose gel containing formaldehyde and transferred to a nitrocellulose membrane by using a TurboBlotter (Schleicher & Schuell, Keene, N.H.). For Northern slot blotting, denatured total RNAs were blotted to nitrocellulose paper using a slot blotting apparatus. The conditions for hybridization and washing were previously described (29).

Cloning of Tn917 flanking sequence by inverse PCR.Chromosomal DNA from strain CYL175 was digested with HpaII, religated, and used as a template for inverse PCR amplification as described by Ochman et al. (22). Two inverse PCR primers (5′-CACAATAGAGAGATGTCACC-3′ and 5′-GCTATGCTCGAGTGAGTACG-3′), facing away from each other and located at the right end of the Tn917 sequence, were used for PCR amplification. The amplified fragment was cloned into the pGEM-T vector (Promega) and sequenced.

Construction of chromosomal cat insertion by allele replacement.The basic strategy for allele replacement was described previously (26). In brief, upstream and downstream fragments (>0.5 kb) of the target region to be replaced were amplified by PCR and sequence verified or obtained by restriction digestion. The two fragments and the cat gene of pC194 were cloned into pCL52.2 or pLL28 in such a way that the cat gene was flanked by the two fragments in the same orientation as in the chromosome. The resultant plasmids were first electroporated into RN4220 and then transferred to strain Becker by phage transduction at 30°C. The strains containing the plasmids were subjected to temperature shift as described previously (26), and the desired mutants were verified by PCR or Southern hybridization.

Identification of a mutant with a pleiotropic phenotype.During the construction of chromosomal insertions in strain Becker using the chloramphenicol resistance gene (cat) of the S. aureus plasmid pC194 by allele replacement to study the regulation of CP8, we unexpectedly isolated a mutant, CYL5614, that produced a larger amount of CP8 and smaller amounts of alpha-toxin and protein A than the wild-type strain (not shown). To determine whether the cat insertion in strain CYL5614 was responsible for the mutant phenotype, we transduced the cat insertion back to strain Becker. The mutant phenotype did not correlate with the insertion in the backcross experiments, indicating that the cat insertion is not responsible for the pleiotropic effect. Thus, these results indicate that another mutation(s) in the chromosome of CYL5614, which probably occurred spontaneously, is responsible for the phenotype. Since the phenotype is different from those of agr, sarA, sae, ssr, and arl mutants, it is most likely that the mutation(s) represents a new global regulatory locus. We named this locus mgr for multiple gene regulator and designated the mutation in CYL5614 mgr5614.

Linking the mgr locus by transposon Tn917.To clone the mgr gene(s), we screened a plasmid library of strain Becker for complementation of protein A (Spa) or alpha-toxin (Hla) (i.e., using Spa or Hla as the phenotype for mgr, since CYL5614 produced markedly reduced protein A and alpha-toxin). However, we failed to identify any positive clone. Since there is no selection for the mgr locus, we sought an alternative approach to clone the gene by linking transposon Tn917 to mgr (11, 16). Random transposition was carried out by temperature shift using strain Becker harboring pTV1 (34), which contains Tn917 on a thermosensitive replicon. The Tn917 insertion library thus generated was transduced to strain CYL5614 by phage 52A, and the resulting Tn917-encoded erythromycin-resistant (Em^r) transductants were screened for Spa⁺ colonies representing cotransduction of Tn917 and mgr. The rationale for this experiment is that if Tn917 inserts at a location close to the wild-type mgr locus in Becker, the two markers could then be cotransduced by phage to strain CYL5614. The closer the two markers are, the higher the percentage of the cotransduction will be. By screening ∼25,000 transductants, we found that two, CYL175 and CYL180, produced wild-type levels of protein A, suggesting that these two strains have Tn917 inserted near mgr. The two transductants also produced wild-type levels of CP8, indicating that the Tn917 insertion indeed links to the mgr pleiotropic locus (data not shown). The Tn917 insertions in CYL175 and CYL180 were transduced back to CYL5614, and the transductants were screened for protein A production. These backcross experiments showed that the cotransduction frequencies for Tn917 insertion and mgr5614 using CYL175 and CYL180 as the donors were 11 and 18%, respectively. However, when Hla (alpha-toxin) was used as a marker in later experiments, we found an ∼43% link between mgr5614 and Tn917. The probable reason for this discrepancy was that Spa was rather difficult to detect by the colony immunoblotting method, which resulted in underestimating the Spa⁺ phenotype. These data indicate that we have isolated two strains with Tn917 inserted near the mgr locus. The chromosomal DNAs from CYL175 and CYL180 were purified; digested with EcoRI, Sau3A, HindIII, and HpaII; and subjected to Southern analysis using both halves of Tn917 as probes. The Southern patterns of the two strains were the same, indicating that Tn917 inserted at the same location in both strains (not shown).

The DNA sequence flanking one side of the Tn917 insertion was obtained by inverse PCR amplification as described in Materials and Methods. An ∼500-bp fragment (as expected from the Southern analysis) was obtained and cloned into the pGEM-T vector. Sequencing of the cloned insert revealed that the PCR fragment contained the expected Tn917 sequences and a 444-bp Becker DNA sequence adjacent to the right side of the Tn917 insertion site.

Fine genetic mapping of mgr by transduction.Since Tn917 was inserted near the mgr locus, to map the position of mgr5614 in reference to the Tn917 insertion site by three-factor transduction and to subsequently clone the mgr gene(s), we need to isolate large DNA fragments encompassing this region of the chromosome. To this end, the 500-bp DNA fragment from the inverse PCR experiment was used as a probe to screen a cosmid library of strain Becker constructed in E. coli (25) using a low-copy-number cosmid vector, pLAFR3. Several cosmid clones with overlapping inserts spanning an ∼55-kb region were obtained, and their restriction maps were determined. The partial composite map is shown in Fig. 1. In addition, the Tn917 insertion site was mapped between the BglII and SalI sites by Southern blotting.

• Open in new tab
• Download powerpoint

FIG. 1.

Genetic mapping of the mgr locus. The insertion sites of Tn917 (in CYL175) and cat (in CYL402 and CYL412) are indicated by triangles. The rates of cotransduction are indicated as percentages. B, BamHI; Bg, BglII; E, EcoRI; N, NcoI; S, SalI; Ss, SstI; X, XbaI.

To precisely map the mgr locus, we employed fine genetic mapping by three-factor transductional crosses. To perform such experiments, however, an additional marker is needed. To provide this new marker, we inserted the cat gene of pC194 into the chromosome of strain Becker ∼8.0 kb to the left of the Tn917 insertion site (of strain CYL175) by allele replacement, which resulted in strain CYL402 (i.e., Becker::cat402). In addition, another strain, CYL412 (i.e., Becker::cat412), with a cat insertion ∼14 kb to the right of the Tn917 insertion (of strain CYL175), was also constructed. The cat insertion sites of both CYL402 and CYL412 are shown in Fig. 1. These two strains, which are phenotypically Em^s Hla⁺ Cm^r, were then used as the recipients for the three-factor cross experiments described below. However, we did not have a donor strain with the Em^r Hla⁻ Cm^s phenotype suitable for the transduction experiments. Therefore, we transduced Tn917 from CYL175 (containing Tn917 encoding Em^r) to CYL5614 with the selection of Em^r and then screened for the Hla⁻ phenotype (i.e., the Mgr5614 phenotype). The resultant strain, CYL183, with the Tn917 insertion and the mgr5614 genotype, was used as the donor for the transduction experiments. It should be noted that CYL183 is in fact Cm^r due to a cat insertion in the chromosome of CYL5614. However, this cat insertion is unlikely to have interfered with our transduction experiments because the insertion is too far away to be cotransduced with Tn917 (∼630 kb apart, according to the published S. aureus genome sequence [14]).

After the donor and recipient strains were constructed, genetic crosses were carried out by phage 52A transduction from CYL183 to CYL402 and CYL412 with Em^r selection. The transductants were then scored for Hla and Cm phenotypes. The results in Table 2 show that the frequency of cotransduction for all three markers, mgr5614, Tn917, and cat402, was 25% (Em^r Hla⁻ Cm^s phenotype [row 2 in transduction 1]). In contrast, the frequency of cotransduction for mgr5614, Tn917, and cat412 was 9% (row 2 in transduction 2). These results indicate that mgr is closer to cat402 than to cat412, suggesting that mgr is located to the left of Tn917. To determine the order of mgr5614, cat402, and Tn917, we examined the data in Table 2, transduction 1. The fact that the frequency of Em^r Hla⁺ Cm^s transductants was the lowest (2%) among the four possible phenotypes indicates that obtaining transductants with this phenotype (i.e., crossover of Tn917 and cat but not mgr5614 markers) requires the most (at least four) crossovers. These data, therefore, unambiguously positioned mgr5614 to the right of the cat402 insertion. Thus, the order of these markers was, from left to right, cat402-mgr5614-Tn917-cat412, as shown in Fig. 1. Also shown in Fig. 1 are the cotransduction frequencies for these markers: 39 to 46% between mgr5614 and Tn917 (25% plus 14% [Table 2, transduction 1] and 9% plus 37% [Table 2, transduction 2]), 27% between cat402 and Tn917 (2% plus 25% [Table 2, transduction 1]), and 18% (9% plus 9% [Table 2, transduction 2]) between Tn917 and cat412. The cotransduction frequencies for cat402 and Tn917 and those for cat412 and Tn917 correspond well to the actual physical distances between the markers, indicating that there is little, if any, marker effect due to differences in the transduction rates of individual markers.

Cloning of the mgr gene.The results of the genetic crosses described above indicate that the mgr locus is located to the left of Tn917 near the cat402 insertions. To clone the mgr gene(s), we subcloned the DNA fragments that most likely contained the mgr locus from cosmid clones to pCL95, an E. coli-S. aureus multicopy shuttle vector (10), or to pCL83, a single-copy vector (15), with tetracycline resistance selection in S. aureus. These subclones were used to complement CYL5614 for its ability to produce alpha-toxin. Surprisingly, none of the inserts in either the multicopy or single-copy vector complemented the mutant (not shown). Clones containing inserts further to the left or right also showed no complementation. These results were rather unexpected and were consistent with our failed efforts earlier to clone the gene by screening a plasmid library by complementation. To search further for the mgr gene(s), we deleted several regions, including a 9-kb region that encompassed both Tn917 and cat402 insertion sites. To our surprise, the 9-kb knockout strain, CYL776, exhibited a phenotype entirely opposite to that of CYL5614 (i.e., slightly decreased CP8, slightly increased protein A, and moderately increased alpha-toxin). The backcrossed strain also showed the same phenotype, indicating that the resulting phenotype is due to the 9-kb deletion (data not shown). We next sequenced an ∼5-kb region that most likely contained the mgr gene(s), based on genetic mapping. By a sequence comparison between Becker and CYL5614, we found one point mutation with a T-to-C transition in a 227-bp intergenic noncoding region between two divergent ORFs (444- and 927-bp ORFs corresponding to SA0641 and SA0642 of the sequenced N315 strain [14], respectively) (see Fig. 3A). The mutation is located 155 bp upstream of the SA641 ORF and 72 bp upstream of the SA642 ORF. One possible mechanism leading to these results could be that the mgr5614 mutation in CYL5614 is an up mutation that increases mgr expression instead of a down mutation. In this scenario, we would not be able to complement the mutant and a deletion of the mgr gene(s) would result in a reverse phenotype of the mutant. To test this hypothesis, Northern slot blots of the total RNAs isolated from Becker and CYL5614 using the internal fragments of both SA0641 and SA0642 as probes were compared. The results showed that the mutant strain produced four- to eightfold more message of SA641 than the wild-type Becker, whereas there was no difference in the amount of SA642 message (Fig. 2). These results strongly indicate that the 444-bp SA641 is the mgr gene responsible for the phenotype found in CYL5614. We named this ORF mgrA.

• Open in new tab
• Download powerpoint

FIG. 2.

Northern slot blot of twofold serial dilutions of Becker and CYL5614 total RNAs probed with mgrA or SA642. Blots of 16S rRNA were used as controls.

To determine whether mgrA indeed encodes the mgr regulatory factor, we constructed a mutant strain with an mgrA-specific deletion by allele replacement. The mutant strain CYL1040, with a cat gene replacing a 269-bp internal fragment of the mgrA ORF, showed exactly the same phenotype as the 9-kb deletion strain, CYL776 (Fig. 3). To further confirm that the mutant phenotype is due to the deletion of mgrA, we attempted to complement CYL1040 with the 0.9-kb EcoRI-HindIII fragment containing the intact mgrA ORF and with the 0.6-kb EcoRI-AccI fragment containing the truncated mgrA. Our results (Fig. 3) showed that, as expected, the plasmid pTL2996 with the 0.6-kb EcoRI-AccI fragment containing the truncated mgrA did not complement CYL1040. However, pTL2989, with a 0.9-kb EcoRI-HindIII fragment containing the intact ORF mgrA, complemented the mutant CYL1040 excessively, particularly with respect to CP8 and protein A production [i.e., CYL1040(pTL2989) produced much less protein A and much more CP8 than the wild type (Fig. 3B)]. The phenotype of CYL1040(pTL2989) could be explained by the fact that mgrA was cloned in a multiple-copy plasmid vector, which likely resulted in overexpression of mgrA. These results clearly showed that mgrA was involved in mgr regulation. A BLAST search showed that MgrA had homology with the MarR family of helix-turn-helix transcriptional regulators, of which the multiple-antibiotic resistance regulatory gene marR of E. coli is the prototype (1, 31). The homologous region contains a potential helix-turn-helix DNA-binding motif from amino acid 56 to 76 of the MgrA protein (Fig. 4). MgrA (SA641) of the recently sequenced S. aureus strain N315 was designated a conserved hypothetical protein (14).

• Open in new tab
• Download powerpoint

FIG. 3.

Localization of the mgr locus. (A) Specific segments of the chromosomal region containing the mgr locus were deleted by allele replacement with the cat gene. The corresponding mutants are indicated. The bottom two lines indicate the DNA fragments used for complementation of strain CYL1040. The phenotypes of each mutant and complemented strain are shown on the right. ++, strong producer; +, moderate producer; −, nonproducer. Note that the phenotypic characteristics of CYL1040(pTL2989) are the same as those of Becker, although pTL2989 excessively complemented CYL1040 due to a multiple-copy effect (see the text). The T-to-C transition in CYL5614 is indicated by a small circle. H, HindIII; A, AccI. (For the other abbreviations, see the legend to Fig. 1.) (B) Production of alpha-toxin (Hla) from strains shown in panel A was semiquantitatively assayed on sheep blood agar. The production of protein A (Spa) and CP8 by thestrains used in the complementation tests shown in panel A was analyzed by Western dot blot analysis of samples serially diluted as indicated above the blots.

• Open in new tab
• Download powerpoint

FIG. 4.

Comparison of MgrA sequence and MarR family consensus sequence. The helix-turn-helix motif (HTH) is indicated. Colons and periods indicate identical and similar residues, respectively.

Defining the mgr locus.From the N315 genome, we found that the third ORF (SA638) to the right of mgrA (Fig. 3A) is the bcaA gene involved in bacitracin resistance, indicating that this gene is not likely to be involved in mgr regulation. The second ORF (SA643) to the left of mgrA had been disrupted by the cat402 insertion without altering the Mgr phenotype, suggesting that it is not part of the mgr locus either. Thus, from these results, we believe that the mgr regulatory locus consists of at most four ORFs (SA642, mgrA, SA640, and SA639). To define the boundaries and the number of genes in the mgr locus, we constructed chromosomal deletion mutants of these three uncharacterized genes in this region by replacing the deletion regions with the pC194 cat gene. Strain CYL1065, with a 610-bp deletion within SA0642, and strain CYL896, with a 2.7-kb deletion of both SA640 and SA639, showed no change in Mgr phenotype (Fig. 3). These results indicate that the three genes flanking mgrA are not part of the mgr regulatory system. Thus, the mgr locus appears to consist of only the mgrA gene.

Phenotypic characterization of mgr.To further study the effect of mgr on its target gene products and to test whether mgr affects additional virulence factors, we quantitatively measured the production of CP8, alpha-toxin, nuclease, protein A, protease, and coagulase by strains Becker, CYL1040, and CYL5614 as described in Materials and Methods. In the alpha-toxin assay by Western analysis, to prevent protein A interference in our analysis, we constructed a spa-null mutant strain, CYL978, that produced undetectable protein A (not shown). As shown in Fig. 5A (top), the inclusion of the spa-null strain allowed us to unambiguously identify alpha-toxin in the gel. The results confirmed that MgrA repressed Hla. The results shown in Fig. 5A (middle and bottom) also confirmed that mgrA repressed Spa but activated CP8. In addition, our results (Table 3) also showed that mgr up-regulated nuclease but down-regulated protease and coagulase (although the coagulase activities in CYL1040 and Becker were the same). We found that the mgr5614 mutation seemed to have greater effects on the production of the target products than the mgrA-null mutation, although the two mutations had opposite effects. However, in general, the extent of regulation was proportional to the quantity of mgrA expression.

• Open in new tab
• Download powerpoint

FIG. 5.

Phenotypic analysis of mgr. (A) The production of alpha-toxin was performed by Western analysis using anti-alpha-toxin antibody after cellular proteins were subjected to SDS-8% PAGE. The production of protein A and CP8 was analyzed by Western dot blot analysis of samples serially diluted as indicated above the blots. (B) SDS-PAGE of whole-cell extracts and supernatants. Samples were run on a 5 to 20% gradient gel.

The results described above showed that mgr regulates CP8, protein A, and four other extracellular proteins. To estimate the extent of the effect of mgr on gene expression in S. aureus, proteins from whole-cell lysates and extracellular supernatants of broth cultures from the wild-type strain Becker, an mgr deletion mutant (CYL1040), and an mgr overproduction strain (CYL5614) were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). As shown in Fig. 5B, no major difference between the wild type and the mutant strains was observed in cell extracts. Increasing or decreasing the amount of cell extract in the gel also did not show a difference between the strains (not shown). On the other hand, major differences were found between these strains in the extracellular supernatants. The intensities of at least four bands increased drastically, and the intensities of a large number of bands with molecular masses ranging from 40 to 80 kDa decreased significantly in the mgr-overproducing strain CYL5614. In the mgr-null mutant CYL1040, the intensities of many bands increased greatly, especially at ∼70 kDa. Although we were unable to show the exact number of proteins affected by the mutations due to the low resolution of the one-directional SDS-PAGE system, these results clearly show that mgr profoundly affects the production of the secreted proteins but only slightly affects the production of the cellular proteins. Thus, mgr is likely to control the expression of a great number of extracellular proteins.

mgr affects the transcription of target genes.To determine whether mgr affects the transcription of its target genes, serial dilutions of the total RNAs from strains Becker, CYL1040, and CYL5614 were slot blotted to nitrocellulose membranes. The membranes were hybridized to gene probes specific for CP8, protein A, nuclease, or alpha-toxin. The results (Fig. 6) showed that the mgr5614 mutation and the null mutation affected the mRNA production of each of the gene products tested. The patterns were very comparable to those of the phenotypic characterization shown in Fig. 5 and Table 3, suggesting that mgr most probably regulates cap8, spa, nuc, and hla at the transcriptional level.

• Open in new tab
• Download powerpoint

FIG. 6.

Effect of mgr on expression of cap8, hla, spa, and nuc by Northern slot blot analysis. Twofold serial dilutions of the total RNAs from Becker (lanes 1), CYL5614 (lanes 2), and CYL1040 (lanes 3) were transferred to membranes and probed with specific gene probes. The loading controls were probed with either gyrA (gyrase A subunit) or 16S rRNA probes.