The RofA Binding Site in Streptococcus pyogenes Is Utilized in ...

JOURNAL OF BACTERIOLOGY, Mar. 2000, p. 1529–1540 0021-9193/00/$04.00⫹0 Copyright © 2000, American Society for Microbiology. All Rights Reserved.

Vol. 182, No. 6

The RofA Binding Site in Streptococcus pyogenes Is Utilized in Multiple Transcriptional Pathways ALEXANDER B. GRANOK,1† DEREK PARSONAGE,2 R. PAUL ROSS,3

AND

MICHAEL G. CAPARON4*

1

Division of Infectious Diseases, Department of Medicine, and Department of Molecular Microbiology,4 Washington University School of Medicine, St. Louis, Missouri 63130-1093; Department of Biochemistry, Wake Forest University Medical Center, Winston-Salem, North Carolina 27157-10162; and Dairy Products Research Institute, Teagasc, Cork, Ireland3 Received 11 October 1999/Accepted 21 December 1999

Understanding the regulation of adhesins defines a pathogenic bacterium’s interaction with the local environment within the host. In certain strains of Streptococcus pyogenes, transcription of prtF, the gene which encodes the fibronectin-binding adhesin protein F, is activated by RofA under anaerobic conditions. RofA binds specifically to DNA in its target promoters and autoregulates its own expression. In this study, we have used DNase I protection assays to further investigate the interaction of RofA with its target promoters. In the region between rofA and the gene which encodes protein F (prtF), RofA binds to two distinct sites: a smaller site (17 bp) adjacent to the rofA promoter, and a larger site (40 bp) adjacent to the prtF promoter. Analysis of fusions to a novel reporter gene whose product consists of the fusion of the N-terminal secretion domain of protein F with the C-terminal enzymatic domain of the enterococcal alkaline phosphatase (PhoZ) revealed that the small RofA binding site had no direct role in control of prtF transcription but contributed to regulation of rofA. Comparison in several strains representing different patterns of prtF expression indicated that the larger site was required for activation of rofA and of prtF in all strains by both RofA-dependent and -independent pathways. Thus, it would appear that a common recognition sequence provides separate entries to a final common pathway in S. pyogenes virulence gene expression. The identification of multiple RofA-like proteins and promoters with RofA binding sites implies the existence of a widespread interacting regulatory network. protein F or Sfb (12, 37) which can mediate binding to the extracellular matrix (30) and to certain types of host cells (12, 29) and can promote invasion into these cells (17, 27). As is a common theme for variable streptococcal surface proteins (6), different alleles of protein F share a highly conserved repetitive domain located in the carboxy-terminal half of the molecule (35). However, the number of individual repeat units can vary substantially, and the N-terminal domain is conserved to a much lower extent (19, 30). Some isolates do not contain the gene which encodes protein F (prtF) (28). Of these, some possess a distinct but structurally similar fibronectin-binding protein known as protein FII (18). Other isolates contain serum opacity factor (32), a surface protein that contains the characteristic repetitive domain implicated in the fibronectin-binding properties of protein F and protein FII (18, 30). Recently, a fourth fibronectinbinding protein (PFBP) which exhibits similar repetitive domains has been identified in selected isolates (33). The genes which encode protein F, protein FII, serum opacity factor, and PFBP reside at distinct loci. Of the isolates that lack protein F, some contain both protein FII and serum opacity factor (18), and some contain no recognizable fibronectinbinding protein (28). An example of the latter is the serotype M1 strain, whose genome sequence is being determined (http://www.genome.ou.edu/strep.html). Interestingly, the locus expected to contain protein F instead contains the gene for a protein F-like protein (cpa.1) which differs from protein F primarily by the fact that it lacks the fibronectin-binding domains of protein F, including the repetitive domain (31). Associated with this diversity in the distribution of fibronectin-binding proteins is a heterogeneous pattern of expression of these proteins by different strains. For example, most strains regulate expression of protein F and protein FII in response to atmosphere and express the proteins optimally under aerobic

Considerable evidence has accumulated to suggest that fine control over the expression of virulence determinants plays an important role in the ability of many different pathogens to establish infection and cause disease. Control often involves the interactions of sophisticated regulatory networks which sense environmental cues and then process these signals for the modulation of expression of specific subsets of virulenceassociated genes (reviewed in references 25 and 26). Thus, an understanding of the molecular details of regulation can provide important insights into the dynamics of host-pathogen interplay. An interesting application of this approach has involved Streptococcus pyogenes (group A streptococcus). This grampositive bacterium is capable of causing a diverse set of human diseases that range from relatively minor and self-limiting such as pharyngitis (strep throat) to diffuse infections of soft tissues (erysipelas, cellulitis) to destructive life-threatening infections such as necrotizing fasciitis and toxic shock syndrome. Immunopathological diseases like rheumatic fever and even some types of obsessive compulsive disorder are also associated with infection by S. pyogenes (36). This wide range of diseases is reflected by diversity apparent between different isolates of S. pyogenes. Specifically, different strains express several antigenically variant surface proteins (6), vary in their complement of surface protein genes and other virulence factors (2, 38), and vary in the host cell populations and cellular receptors they recognize (29). As a typical example, most strains of S. pyogenes express a fibronectin-binding surface protein known as * Corresponding author. Mailing address: Department of Molecular Microbiology, Washington University School of Medicine, St. Louis, MO 63130-1093. Phone: (314) 362-1485. Fax: (314) 362-1232. E-mail: [email protected]. † Present address: Norumbega Medical, Bangor, ME 04401. 1529

1530

GRANOK ET AL.

conditions. Control is at the level of transcription and apparently involves a signal transduction pathway that responds to oxidative stress. Several strains which express protein F at high levels under both aerobic and anaerobic conditions have been described. Analysis of one of these strains revealed that expression under anaerobic conditions was the result of activation of transcription of rofA, which is located adjacent to prtF, but transcribed divergently, and encodes a positive activator of prtF transcription (7, 8). Interestingly, inactivation of rofA has no effect on prtF expression in aerobic environments. These data may suggest the involvement of multiple regulatory pathways. Nra, a regulator highly homologous (62% identical) to RofA has recently been described as involved in the control of transcription of the gene which encodes protein FII (prtFII) (31). In contrast to RofA, gene inactivation studies have indicated that Nra is a negative regulator of transcription of multiple genes, including prtFII and a gene which encodes a collagenbinding protein (cpa). Reminiscent of rofA and prtF, nra is located adjacent to but transcribed divergently from cpa. Also like rofA, nra does not appear to be involved in regulation in response to oxidative stress. Strains which contain rofA alone, nra alone, or both rofA and nra have been described. In the serotype M1 strain genome sequence, rofA (98% identical to rofA) is located adjacent to cpa.1 (53% identical to cpa). Thus, multiple and overlapping combinations of both targets (prtF, prtFII, cpa, and cpa.1) and regulators (rofA and nra) have been observed. RofA and Nra are not homologous to any other characterized regulatory element and thus represent a novel family of transcriptional regulators involved in bacterial virulence. Some of the details of how RofA regulates transcription have been established. Activation of RofA-regulated promoters is sensitive to the concentration of RofA, since overexpression of RofA from a multicopy plasmid leads to activation of prtF in any host and under any environmental condition. RofA acts a positive regulator of its own transcription, and consistent with the observation that both RofA and Nra contain a putative N-terminal helix-turn-helix motif, RofA has been shown to be a DNA-binding protein which can specifically bind to DNA containing the rofA and prtF promoters (7). However, other than these features, very little is understood about how RofA or Nra acts to regulate transcription. In this study, we have examined the interaction between RofA and DNA at a much higher level of resolution. Using DNase I protection analysis, we identify several binding sites for RofA adjacent to the prtF and rofA promoters. Through construction and use of a novel reporter vector for analysis of transcription in streptococci that utilizes the secreted alkaline phosphatase of Enterococcus faecalis (PhoZ), we show that these sites are important for activation of rofA and prtF transcription. Interestingly, not all sites are required for expression of prtF. Furthermore, comparisons between strains showing three distinct patterns of expression in response to environmental cues demonstrates that RofA binding sites are also required by the RofA-independent pathways for prtF activation. These data suggest that a common mechanism may underlie prtF activation in response to different regulatory pathways. MATERIALS AND METHODS Bacterial strains. S. pyogenes JRS4, HSC5, and MGAS166s have been described elsewhere (1, 34, 39). Enterococcus faecalis OGIX (18) was used as a source of phoZ. Escherichia coli DH5␣ (Gibco-BRL) was used for molecular cloning experiments. E. coli strains were cultured in Luria-Bertani broth at 37°C. S. pyogenes strains were cultured in Todd-Hewitt medium (BBL) supplemented

J. BACTERIOL. with 0.2% yeast extract (Difco) (THY medium). Solid media were produced by the addition of Bacto Agar (Difco) to a final concentration of 1.4% (wt/vol). Unless otherwise noted, S. pyogenes was cultured at 37°C in sealed tubes without agitation. Specific culture conditions for analysis of secreted alkaline phosphatase activities are described below. Where appropriate, kanamycin was added to media at the concentrations of 500 ␮g/ml for S. pyogenes and 25 ␮g/ml for E. coli. Manipulation of DNA. Plasmid DNA was isolated by standard techniques and used to transform E. coli by the method of Kushner (20) or to transform S. pyogenes by electroporation as described previously (5). Restriction endonucleases, ligases, and polymerases were used according to the recommendations of the manufacturers. When required, DNA fragments were purified using silica gel affinity matrices following electrophoresis through agarose gels (Geneclean; Bio 101) or from PCR mixtures (QIAquick; Qiagen). Incompatible restriction fragment ends were subjected to ligation following treatment with T4 DNA polymerase (Gibco-BRL) to produce blunt fragment ends. The DNA sequences of selected PCR products and chimeric plasmids were confirmed using fluorescent dye-labeled nucleotide terminators (Big Dye) according to the recommendations of the manufacturer (PE Applied Biosystems). Selected DNA fragments were labeled by a fill-in reaction following digestion with a restriction enzyme using the Klenow fragment of DNA polymerase I and [␣-32P]dATP as recommended by the manufacturer (Gibco-BRL). DNase I protection assays. Maltose-binding protein (MBP)–RofA was purified from E. coli BL21(pMBP-RofA), and its DNA-binding activity was confirmed using an electrophoretic mobility shift assay as described previously (7). DNA substrates were prepared by PCR using primers Intergene5 (CGATT GAGAA TTCCA AGTAT TTTTC) and Intergene3 (GACTC CCTCT AGAGT GACAG CAAAT CGCC). When amplified from a pPTF8 template (12), the resulting 397-bp fragment represents the entire intergenic region between rofA and prtF from JRS4 and contains a unique terminal EcoRI site (underlined in Intergene5) and a unique terminal XbaI site (underlined in Intergene3). Digestion with EcoRI following labeling with [␣-32P]dATP as described above selectively labels the nonsense strand relative to prtF. The sense-strand relative to prtF was selectively labeled by the same method following digestion of the substrate fragment with XbaI. The ability of MBP-RofA to protect the substrate fragment from digestion with DNase I was determined by the procedure of Galas and Schmitz (9), modified as follows. Various amounts of MBP-RofA (range, 0 to 89 pmol) were incubated with 125 fmol of labeled substrate at 25°C for 30 min in a 200-␮l reaction containing 12 mM HEPES (pH 7.5), 12% glycerol, 1 mM EDTA, 60 mM KCl, 5 mM MgCl2, 3 mM CaCl2, and 0.6 mM dithiothreitol. DNase I (0.01 U; Gibco-BRL) was added, and the incubation continued for 2 min at 25°C. The reaction was terminated by precipitation of the digestion products through the addition of 50 ␮l of saturated ammonium acetate, 5 ␮g of yeast tRNA, and 645 ␮l of absolute ethanol. The precipitate was washed twice with 70% ethanol, resuspended in formamide loading buffer, and then subjected to electrophoresis through a 6% Tris-borate-EDTA–urea–6% polyacrylamide sequencing gel. DNase I cleavage patterns were compared to a DNA sequencing reaction of the labeled substrate that was prepared by the method of Maxam and Gilbert (23). Following electrophoresis, gels were dried under vacuum at 80°C and subjected to autoradiography using a phosphor exposure screen (Molecular Technologies) and Classic Blue Sensitive film (Molecular Technologies). Construction of phoZF. A novel reporter gene (phoZF) for analysis of transcription which encodes a chimeric protein consisting of the amino-terminal domains of protein F and the carboxy-terminal domains of the alkaline phosphatase (PhoZ) of Enterococcus faecalis was constructed as follows. Inspection of the phoZ sequence (GenBank accession no. AF154110) indicated that it contains a consensus signal sequence and processing site (LAGC, amino acid residues 17 to 20 of the precursor protein) characteristic of a surface-anchored lipoprotein. This information was used to design primers 5PhoBsp (GCGGG TTGTA CAAATT TATGT GCACA AAAAA GCGGC GAAAA AC) and 3PhoBsp (CGTTC TGCTT TTTGT GCACT TTGTT ATTTA CCAAT ACC), which amplified a fragment of phoZ whose 5⬘ end lacks the signal sequence, lipoprotein processing site, and 3 additional codons. Digestion of the Bsp1286I sites introduced by the primers (underlined above) allowed the fragment to be used to replace a PstI fragment of pPTF8 (13). The resulting plasmid (pMGC66 [see Fig. 3]) contains a chimeric gene in which the 5⬘ end of prtF is fused in frame following an alanine codon (bp 1127 to 1130 of the published prtF sequence; GenBank accession no. L10919) to the modified phoZ. The resulting gene encodes a chimeric protein (PhoZF) which contains the secretion domains of protein F and the enzymatic domains of PhoZ. In addition, PhoZF lacks both the lipoprotein anchoring domain of PhoZ and the carboxy-terminal fibronectinbinding and LPXTG cell wall attachment domains of protein F. Consequently, PhoZF is freely secreted from the cell and, since it retains alkaline phosphatase activity, can readily be detected in cell-free supernatants (see below). Construction of prtF promoter truncation mutations. A number of truncations of the prtF distal end of the rofA-prtF intergenic region were constructed as follows. Common primer MUT2 (CCAAA ACCGA TAGCA CCCGC G), which annealed to a site adjacent to an SphI site in prtF of JRS4 (GenBank accession no. L10919), was successively paired with a number of primers designed to anneal at various sites within the intergenic region. Digestion with SphI and BamHI (introduced by the prtF-distal primer as underlined below) allowed the fragments to be used to replace a BamHI-SphI fragment of pMGC66. The prtF-distal primers and names of the resulting plasmids are as follows: Mut-1A

VOL. 182, 2000

FIG. 1. RofA binds to two sites in the prtF and rofA promoters. The ability of MBP-RofA to protect sequences in the prtF-rofA intergenic region from digestion with DNase I is shown. A DNA fragment containing the entire intergenic region was labeled with 32P on its 3⬘ end on the noncoding strand relative to prtF. Lanes 1 and 2, Maxam-Gilbert sequencing reaction products; lanes 3 to 7, 0.125 pmol of labeled probe incubated with 0, 0.089, 0.89, 8.9, and 89 pmol of MBPRofA, respectively, in the presence of DNase I. The positions of the nucleotides included in protected regions are indicated by the brackets at the right, with the nucleotide corresponding to the prtF transcriptional start defined as position ⫹1.

(GATTC GATTG ATGGA TCCAA GTATT TTTC), pPro1; Promoter226 (GATAA GTGGA TC CTATT TC), pPro2; Promoter298 (AAGTC GGATC CTTTT GAAAT AGC), pPro3; Promoter337 (CTCAA AGGAT CCTTT TCAAA AAC) pPro6; Promoter370 (TGAAA AATAG GATCC AAAAA TTGTC), pPro7; and Promoter400 (TGACC ATAAC GTGGG ATCCT CATAT ATG), pPro8. Digestion of the product of MUT-2 and Promoter298 at a DraI site followed by insertion between the SphI and BamHI sites of pMGC66 generated pPro4. Finally, pairing of primer Promoter330 (TTTAA AGATA TCTTT TCTCA AAAAA TC) and MUT2, followed by digestion with EcoRV and SphI and insertion into the BamHI-SphI fragment of pMGC66, generated pPro5. The fidelity of all constructions was determined by DNA sequencing. Structures of the intergenic regions retained by these various truncations are illustrated in Fig. 4. Construction of prtF promoter internal deletion mutations. Internal deletions were introduced into the intergenic region contained by pMGC66 by using an inverse PCR technique. An EcoRI site was engineered into each primer to allow specific annealing. The primer pairs and resulting plasmids are as follows: RofAforup (ATCGA ATTCA AAAAC AATAA TTTGG TGAAA AATAT AATCA AAAAA TTG) and RofArevup (GAGAA TTCATT GCTTT AAAGC TATTT CAAAA GGTTT CG), pPro9; RofArofdown (AATTG AATTC AAAAT ATAAT CAAAA AATTG TCTTT CTTGA CAATA ACGTGG) and RofArevdown (GTTGA ATTCA ATGAT TTTTT GAGAA AATAT

S. PYOGENES RofA BINDING SITE

1531

TGCTT TAAAG CTATT TC), pPro10. An additional internal deletion utilized RofArevup and RofAfordown (pPro11). Correct structure was confirmed by analysis of DNA sequencing reactions. Structures of the intergenic regions retained by these various internal deletions are illustrated in Fig. 4. Construction of rofA promoter mutations. For analysis of the rofA promoter, the phoZF reporter plasmid pMGC66 was modified as follows. Primers BamprtFup (ACTTG GATCC AGATC TTCCT TCAGG TTATG) and EcoprtFRBS (CATAT ATGAA TTCGA GAGGA GAGAA AATGA ATAAC AAAAT ATTTT TG) were used in an inverse PCR reaction to amplify a fragment that contains almost all of pMGC66 except a large section of the intergenic region, including the prtF and rofA promoters. Also note that the ribosome-binding site of prtF is retained. Primers BamrofAup (TTTGG GATCC ATTTT CTCTC CTCTC AAAAA CATAT ATGAG C) and EcorofAcoding (AAAAG GAATT CAGTT CCTCA CAATA ATGGT TTAGT TGTTA AAAGG) were used to amplify the rofA promoter and the entire intergenic region including the first 39 codons of rofA so that digestion of both PCR products with EcoRI and BamHI (the sites introduced by the primers are underlined above) followed by ligation generates a plasmid (pABG5 [see Fig. 7]) in which the phoZF reporter is placed under the control of the promoter for rofA. The plasmid was constructed to contain a stop codon at the end of the rofA fragment to prevent translational fusion to phoZF. A number of truncations distal to the rofA promoter were constructed as follows. Common primer EcorofAcoding (see above) was successively paired with a number of primers. Digestion with EcoRI and BamHI (introduced by the rofA-distal primer as underlined below) allowed the fragments to be used to replace a BamHI-EcoRI fragment of pABG5. The rofAdistal primers and names of the resulting plasmids are as follows: rofA5trunc (CAATT TTTGG ATCCT ATTTT TCACC AAATT AATGT TTTTG AAAAT GATTT TTTGA G), pABG5a; nobigrof (TTGGG ATCCT ATTGC TTTAA AGCTA TTTCA AAAGG TTTCG AC), pABG6; nofnrrof (GCTGG ATCCA AAGGT TTCGA CTTTT CACCA AAAAC CATTA G), pABG7; nosmallrof (CCAAA AAGGA TCCGA CTTGA TTTCT ATTTT TAGCT TAGAT AG), pABG8; rofA-10trunc (GAAAT AGGAT CCTAC ACTTA TCAAA GACTT ATTTG GC), pABG10. The fidelity of all constructions was determined by DNA sequencing. Structures of the intergenic regions retained by these various truncations are illustrated in Fig. 8. Construction of rofA promoter internal mutations. To examine the contributions of certain sequence elements while preserving the relative positions of the adjacent elements, several regions of the rofA promoter were highly mutagenized using the PCR-based sequence overlap extension technique (15). Mutagenesis of the region between bp ⫺61 and ⫺75 (see Fig. 8) involved a first round of amplifications pairing primers BamrofAup (see above) with 3rofA (GAAAT CAAAT CTAAT CGATA TAGCT CATAA GTCGA AACCT TTTG) and pairing mut2 (CCAAA ACCGA TAGCA CCCGC G) with 4rofA1 (CAAAA GGTTT CGACT TATGA GCTAT ATCGA TTAGA CTTGA TTTC). Mutagenesis of the region between bp ⫺75 and ⫺105 (see Fig. 8) employed a first round of amplifications pairing BamrofAup with 3FNR (CGAAA CCTTT TGAAA AACCA TAATA CCTAA ATTTT CTCAA AAAAT C) and pairing 4FNR (GATTT TTTGA GAAAA TTTAG GTATT ATGGT TTTTC AAAAG GTTTC G) with mut2. The two products of each first-round amplification were mixed as templates for the second round of amplifications with primers BamrofAup and mut2. The resulting final products were digested with BamHI and SphI (the BamHI site is embedded in BamrofAup [see above]; the SphI site is internal to the region amplified) and introduced between the BamHI and SphI sites of pABG5 to construct pABG11 and pABG12 (see Fig. 8). As a result, a transversion mutation is introduced at every other position (underlined in 4rofA and 4FNR) along the mutagenized region. Assay of PhoZF alkaline phosphatase activity. Preliminary experiments using JRS4(pMGC66) indicated that the maximal secreted PhoZF alkaline phosphatase activity was obtained during the early to mid-logarithmic phase of growth (data not shown). Thus, all comparisons between different promoter mutations were conducted under this condition. For anaerobic culture conditions, freshly autoclaved, previously unopened medium was purged by bubbling nitrogen gas through the liquid for 5 min in a glove box under a nitrogen atmosphere. While retained in the glove box, a 1:20 dilution of an overnight culture of the strain of interest was prepared in the freshly purged medium containing kanamycin. The culture bottle was tightly sealed and incubated statically at 37°C until the optical density at 600 reached approximately 0.4. Aerobic cultures were prepared identically except that incubation was conducted in unsealed Erlenmeyer flasks containing a volume of medium equivalent to no more than 4% of the volume of the flask and flasks were shaken at 200 rpm on orbital platform (model 3590; Lab-Line). Following culture, cells were removed by centrifugation (13,000 ⫻ g, 5 min, 25°C), and triplicate 50-␮l aliquots of cell-free supernatant from control strain JRS4(pMGC66) were added to wells of a 96-well microtiter plate. The volume of supernatant added for other samples was adjusted for any differences in cell number based on optical density at 600 nm [typically less than ⫾20% of the value for JRS4(pMGC66) control]. A 200-␮l aliquot of a 1-mg/ml solution of p-nitrophenyl phosphate (catalog no. N-9389; Sigma) in 1.0 M Tris-HCl (pH 8.0) was then added to each well. Following incubation at room temperature (typically 2 h), the absorbances were measured at 405 nm. Data are presented as a percentage of activity relative to that produced by the same number of cells of JRS4(pMGC66) grown under the identical conditions. The activity obtained from JRS4 in the absence of a plasmid containing phoZF was used to derive

1532

GRANOK ET AL.

J. BACTERIOL.

FIG. 2. Structure of the rofA-prtF intergenic region. Important features of the region between prtF and rofA are shown. The diagram at the top orients the sequence shown in the lower half relative to rofA and prtF. For clarity, only the prtF coding strand is shown, with footprint data from the noncoding strand displayed. Regions protected by RofA from DNase I digestion are indicated by the shaded boxes in the lower half. The location and orientation of a repetitive element are indicated by the arrow below the sequence. The boxes indicated by “⫺10” and “⫺35” indicate the promoters previously defined for rofA and prtF as indicated. Areas darkly shaded indicate a region of homology between the small RofA binding site adjacent to the rofA promoter and the region of the larger RofA binding site distal to the prtF promoter. The larger of the two protected regions is likely composed of two 17-bp sites in an inverted orientation (indicated by the thin arrow above the sequence) separated by 9 bp. A sequence previously noted as similar to the consensus binding for FNR is indicated by a dashed line below the sequence. The nucleotide corresponding to the first nucleotide of the prtF message is defined as position ⫹1.

background values, which were typically never greater than 1% of those of JRS4(pMGC66). Data presented represent the means and standard errors of the means for at least duplicate determinations conducted on different days.

RESULTS Identification of RofA binding sites. In S. pyogenes JRS4, prtF and rofA are located adjacent to one another but are transcribed divergently. Previous work had shown that a chimeric protein which consists of the fusion of the entire sequence of RofA to MBP bound specifically and at high affinity to a DNA segment which contained the rofA and prtF promoters. To specifically delineate the sites of RofA interaction with DNA, the ability of MBP-RofA to protect sites from cleavage by DNase I was determined. When a DNA fragment which spanned the region between rofA and prtF was labeled on the noncoding strand relative to prtF was subjected to treatment with DNase I, MBP-RofA protected two distinct regions from digestion in a concentration-dependent manner (Fig. 1). The larger of these two regions was 40 bp in size and spanned positions ⫺92 to ⫺53 relative to the prtF transcriptional start site. This locates this site 18 bp upstream of the ⫺35 region of the previously determined promoter for prtF (Fig. 2). A second, smaller site of 17 bp which spanned positions ⫺142 to ⫺126 was also protected (Fig. 1). This places this site 6 bp from the ⫺35 region of the rofA promoter (Fig. 2). Analysis of the coding strand did not reveal any additional protected regions, although both sites were shifted three base pairs closer to the prtF gene (data not shown). A closer examination of these protected regions reveals that they are delineated by a series of

repetitive elements. Specifically, an 11-bp element derived from the smaller RofA binding site is a perfect match for an 11-bp element which overlaps the end of the larger site that is located adjacent to the prtF promoter (Fig. 2). The end of the larger site distal to the prtF promoter is delineated by a nearly identical version of this repeat (matching 9 of 11 positions) located on the opposite strand relative to the other two repeat elements (Fig. 2). In addition, the entire 17-bp small protected site and the 17 bp of the end of the larger site that is distal to the prtF promoter are nearly perfect inverted repeats, matching at 14 of 17 positions (Fig. 2). Taken together, these data suggest that the minimal site recognized by RofA is 17 bp and that the larger of the protected regions consists of two 17-bp sites in an inverted orientation that are separated by 9 bp (Fig. 2). Construction of a novel reporter for analysis of transcription. Since further analysis of the roles of the various RofA binding sites in regulation required the construction and analysis of a large number of different mutations, a reporter vector was required that would offer both a simple and sensitive assay of activity and be easy to manipulate to allow the construction of an extensive library of mutations. To this end, we adapted the secreted alkaline phosphatase (PhoZ) of Enterococcus faecalis that has recently been developed as a signal sequence trap (21) as a reporter for analysis of transcription. A plasmidbased vector was developed in which a carboxy-terminal region of phoZ was introduced in an in-frame fusion with a region of prtF encoding approximately the amino-terminal one-third of the mature protein (see Materials and Methods). The resulting

VOL. 182, 2000

chimeric protein (PhoZF) contains the secretion domains of protein F and the enzymatic domain of PhoZ. In addition, the chimera lacks both the carboxy-terminal cell wall anchoring domain of protein F and the N-terminal lipoprotein anchoring domain of PhoZ. Thus, PhoZF is secreted from the cell and can be quantitated in cell-free supernatants using a number of different inexpensive, rapid, and sensitive substrates for the detection of alkaline phosphatase activity (see Materials and Methods). In preliminary experiments, PhoZF was found to be very stable and to have an active alkaline phosphatase activity (see Materials and Methods). Consistent with reports that S. pyogenes has an acid but not alkaline phosphatase activity (24) and that examination of the available genome information for S. pyogenes did not reveal a homologue of phoZ, minimal alkaline phosphatase activities were obtained from all S. pyogenes strains examined in the absence of a plasmid containing phoZF (see below). The chimeric phoZF reporter was placed under control of the promoter for prtF on an E. coli-streptococcal shuttle plasmid previously shown to express prtF in several streptococcal and enterococcal hosts (13). Consistent with previous studies of prtF transcription (7), preliminary analyses of expression of alkaline phosphatase activity from the resulting plasmid (pMGC66 [Fig. 3]) in S. pyogenes JRS4 indicated maximal expression of activity during early to mid-logarithmic phase of growth (data not shown). Studies in other streptococcal hosts revealed patterns of prtF promoter expression in response to environmental cues that were also consistent with previous data (see below). Analysis of RofA binding sites. Since pMGC66 proved to be very stable and easy to manipulate in both streptococcal and E. coli hosts, the strategy used to analyze the RofA binding sites consisted of constructing a panel of mutations, including truncation and internal deletions, in an E. coli host and then analyzing effects on expression following introduction into streptococci. The original plasmid, pMGC66, contains the entire intergenic region between rofA and prtF (Fig. 4). Analysis of a nested series of truncations in S. pyogenes JRS4 indicated that deletion of sequences up to the large RofA binding site (pPro1-pPro5 [Fig. 4]) had essentially no effect on expression of phoZF under both anaerobic (RofA-dependent) and aerobic (RofA-independent) culture conditions (compare pMGC66 to pPro1-pPro5 [Fig. 5]). Included in these deletions were the smaller of the two RofA binding sites and a site previously reported as homologous to the binding site of the fumaratenitrate response regulator (FNR) of E. coli (7) (Fig. 4). Further truncation to eliminate sequences which included the distal 11 bp of the large RofA binding site (pPro6 [Fig. 4]) had only a minimal effect on expression under aerobic conditions but had a marked effect on expression under anaerobic conditions, reducing expression to less than 10% of that obtained for the entire intergenic region (compare pMGC66 to pPro6 [Fig. 5]). Additional truncation to remove the entire large RofA binding site (pPro7 [Fig. 4]) virtually eliminated all activity under anaerobic conditions (⬍2%) while reducing expression under aerobic conditions to only about 35% of that obtained for the entire region (compare pMGC66 to pPro7 [Fig. 5]). A truncation which included the ⫺35 region and part of the ⫺10 region of the prtF promoter (pPro8 [Fig. 4]) reduced activity to background levels under all conditions tested (pPro8 [Fig. 5]). These data indicated that sequences in the large RofA binding site were critical for expression under both anaerobic and aerobic conditions. However, while high-level anaerobic expression required the entire binding site, aerobic expression was predominantly dependent on sequences in the proximal region of the site. To test this hypothesis, an internal deletion


1533

FIG. 3. A novel reporter gene for analysis of transcription in gram-positive organisms. Plasmid pMGC66 is an E. coli-streptococcal shuttle vector that harbors phoZF, a chimeric reporter gene that encodes a protein composed of the N-terminal secretion domains of protein F and the C-terminal enzymatic domain of the alkaline phosphate (PhoZ) of Enterococcus faecalis. The product of this chimeric gene (PhoZF) is freely secreted from the bacterial cell and retains an active alkaline phosphatase activity. The regions of the chimeric gene derived from prtF (prtFⴱ) and phoZ (ⴱphoZ) are indicated by the thick black bar and the open bar and arrow, respectively. In pMGC66, phoZF is under the control of the promoter for prtF (PprtF) contained on a region of DNA that included the entire prtF-rofA intergenic region (thick black bar). The locations of genes encoding resistance to kanamycin (Km) and chloramphenicol (Cm) and for replication (repR) of the plasmid are shown, as is the origin of replication (ori) and several restriction sites. Restriction sites enclosed by parentheses were inactivated during construction. For additional details of the construction of phoZF, see Materials and Methods.

was constructed which removed the distal end of the large binding site that is highly homologous to the small binding site (pPro9 [Fig. 4]). Consistent with the model, substantial aerobic but minimal anaerobic activities were observed (pPro9 [Fig. 5]). An unexpected result was obtained with the converse deletion which removed the proximal half of the large binding site while retaining the distal half (pPro10 [Fig. 4]). While anaerobic activity was minimal, we observed high-level aerobic activity that was actually comparable to that obtained for the entire intergenic region (compare pMGC66 to pPro10 [Fig. 5]). Inspection of sequences immediately downstream of the large binding site (Fig. 2) revealed little similarity to the proximal region of large binding site, suggesting that construction of the deletion did not result in the fortuitous regeneration of the binding site. An internal deletion which removed the entire large binding site (pPro11 [Fig. 4]) confirmed that sequences essential for both aerobic and anaerobic expression are contained in this binding site (pPro11 [Fig. 5]). Activity in strains with other patterns of prtF expression. Studies with JRS4 indicate that while anaerobic expression requires the entire large RofA binding site, either half of the site is sufficient for expression under aerobic conditions. Since several different patterns of prtF expression in response to environmental cues have been described for various isolates of S. pyogenes, it was of interest to examine the contribution of the RofA binding sites to expression in other strains. Unlike JRS4, most strains express prtF at high levels only under aerobic conditions. Previous analysis of a strain representative of this

1534

GRANOK ET AL.

J. BACTERIOL.

FIG. 4. Structures of various prtF promoter mutants. A number of deletions and truncations of the prtF promoter region in pMGC66 were generated as described in Materials and Methods. Relevant features of the sequence as defined in Fig. 2 are shown at the top. The bars below the top line indicate the structure of various derivatives of pMGC66. The numbers and arrows at the top and adjacent to the bars at the bottom correspond to the 5⬘ termini of the various derivatives, whose names are listed to the left of the bars. Regions of internal deletion are indicated by the thin lines above the indicated bars. Numbering of the sequence is as for Fig. 1 and 2.

expression pattern (HSC5) has suggested that differences in expression are not due to any variation in sequence of the prtF or rofA promoter or in rofA itself. Expression of the phoZF reporter in HSC5 under control of the entire intergenic region demonstrated the characteristic expression pattern of highlevel expression under aerobic conditions and reduced expression under anaerobic conditions (pMGC66 [Fig. 6A]). Similar to JRS4, truncation to remove sequences distal to the large

FIG. 5. Activity of various mutant prtF promoters. PhoZF activities in cellfree supernatants of various derivatives of pMGC66 in JRS4 are shown as a percentage of the activity obtained with unaltered pMGC66 under anaerobic (hatched bars) and aerobic (solid bars) conditions. An asterisk indicates a value of less that 1% of that of JRS4 containing pMGC66 under the identical condition, and “No plasmid” indicates the background activity of a plasmid-free JRS4 host. For the structure of each indicated plasmid, see Fig. 4. Alkaline phosphatase activities were determined as described in Materials and Methods. Data represent the means and standard errors of the means for at least three independent experiments.

RofA binding site, including the small RofA binding site, had no effect on expression (pPro5 [Fig. 6A]) and an internal deletion which eliminated the entire large RofA binding site reduced activity to near background levels under all conditions (pPro11 [Fig. 6A]). However, in contrast to JRS4, where either half of the large site could support aerobic expression, elimination of the distal (pPro9) or proximal (pPro10) regions of the large site resulted in a dramatic decrease in aerobic expression (Fig. 6A). Serotype M1 strains typically do not contain prtF, although they do have rofA and will express prtF at high levels under all conditions when prtF is introduced on a plasmid. Examination of expression of the phoZF reporter in a serotype M1 strain (MGAS166s) revealed that the entire intergenic region directed levels of expression that were higher than even those obtained from JRS4 under all conditions tested. Expression under aerobic conditions was particularly high compared to that of JRS4 (approximately fourfold increased) and also to that observed in MGAS166s itself under anaerobic conditions (approximately fourfold increased; pMGC66 [Fig. 6B]). Similar to both JRS4 and HSC5, truncation up to, but not including, the large RofA binding site had no noticeable effect on expression (pPro5 [Fig. 6B]), and an internal deletion which removed the entire large binding site reduced expression to background levels (pPro11 [Fig. 6B]). However, as for HSC5, neither the proximal (pPro9) nor the distal (pPro10) region of the site was sufficient by itself to support expression (Fig. 6B). Taken together, these data indicate that sequences contained within the large RofA binding site are utilized to input information from multiple environmental signals. Consistent with multiple expression patterns, different strains demonstrate variation in how these sequences are used to input information. Role of RofA binding sites in expression of rofA. Previous studies have indicated that rofA is autoregulated and that RofA is a positive activator of transcription of rofA (7). This

VOL. 182, 2000


1535

FIG. 7. A versatile vector for analysis of transcription. The phoZF-containing plasmid pMGC66 (Fig. 3) was modified to increase its utility for analysis of the promoter for rofA or for any other promoter of interest by replacing the promoter for prtF with a DNA segment that contains the entire rofA-prtF intergenic region (thin grey bar) including 39 codons from the rofA coding region (thick grey box). In the resulting plasmid (pABG5), the promoter for rofA (ProfA) is oriented to transcribe phoZF, which retains the ribosome binding site of prtF as indicated. The region containing the rofA promoter between the BamHI and EcoRI sites can be removed and replaced with any promoter of interest. Other features are identical to those described in Fig. 3.

FIG. 6. Expression is heterogeneous in different strains of S. pyogenes. HSC5 (A) and MGAS166s (B) demonstrate two distinct patterns of prtF expression. These hosts were used to analyze the PhoZF activity of selected prtF promoter mutants (Fig. 4). The bars indicate activity expressed as a percentage of that obtained from JRS4(pMGC66), which contains the entire rofA-prtF intergenic region, when the indicated strains were cultured aerobically (solid bars) or anaerobically (hatched bars). “No Plasmid” indicates the background activity of the host under analysis in the absence of any plasmid. Note that in contrast to JRS4, expression of prtF is stimulated by aerobic growth in both HSC5 and MGAS166s. However, while prtF expression is preferentially stimulated only under aerobic conditions in HSC5, high levels of expression are obtained under both anaerobic and aerobic conditions in MGAS166s, with levels considerably higher than those demonstrated by JRS4 under either condition. Despite these differences and in contrast to JRS4, elimination of either the proximal or distal region of the large RofA binding site (pPro9 or pPro10, respectively) greatly reduces expression in both HSC5 and MGAS166s under all conditions analyzed. Data represent the means and standard errors of the means for at least three independent experiments, each conducted in duplicate.

suggests that the various RofA binding sites play a role in expression of rofA. To test this hypothesis, it was first necessary to modify the phoZF-containing reporter vector to place the reporter gene under control of the rofA promoter (see Materials and Methods). The modifications were designed to introduce restriction sites so that the resulting plasmid (pABG5) can be further modified to introduce any promoter of interest as a general reporter vector for analysis of gene regulation (Fig. 7). Preliminary studies comparing expression between wild-type and rofA mutant strains indicated that consistent with previous results, expression of phoZF under the control of the rofA promoter was dependent on a functional copy of rofA and that rofA could function in trans (data not shown). Similar to the analysis described above for prtF, the initial construct

(pABG5) contained the entire prtF-rofA intergenic region (Fig. 8) and the rofA promoter was about as active as the prtF promoter in directing expression of the phoZF reporter (compare pMGC66 to pABG5 [Fig. 9]). Analysis of a series of truncations revealed that removal of the region which includes the promoter for prtF (pABG5A [Fig. 8]) resulted in about a twofold increase in expression (pABG5a [Fig. 9]). Further truncation resulting in elimination of the large RofA binding site (pABG6 [Fig. 8]) resulted in about a fourfold decrease in expression relative to the entire intergenic region (compare pABG5 to pABG6 [Fig. 9]). No further decrease was observed following removal of bases up to but not including the small RofA binding site (pABG7 [Fig. 8]; compare pABG7 to pABG6 [Fig. 9]). However, elimination of the small binding site (pABG8 [Fig. 8]) resulted in an additional threefold decrease in expression to a level of only about 8% of that obtained with the full-length intergenic region (compare pABG8 to pABG5 [Fig. 9]). A deletion which eliminates the ⫺35 and ⫺10 regions of the rofA promoter (pABG10 [Fig. 8]) reduced expression to background levels (pABG10 [Fig. 9]). To examine the small site and the region between the two RofA binding sites (⫺105 to ⫺75) more specifically, the overall architecture of the entire intergenic region was preserved, but the sites were extensively mutagenized by the introduction of a transversion substitution at every other position across the two sites. Analysis of the resulting plasmids (pABG11 and pABG12 [Fig. 8]) indicated that activity of the rofA promoter was reduced by approximately 40% by mutation of the region between the two RofA binding sites (pABG11 [Fig. 9]) and was not affected by mutation of the small RofA binding site (pABG12 [Fig. 9]). Taken together, these data suggest that

1536

GRANOK ET AL.

J. BACTERIOL.

FIG. 8. Structure of various rofA promoter mutants. A number of truncations of the rofA promoter region in pABG5 were generated as described in Materials and Methods. The diagram at the top illustrates the relevant features of the sequence as defined in Fig. 2, except that the orientation is reversed and the rofA transcriptional start site is defined as position ⫹1. The bars below the top line indicate the structures of various derivatives of pABG5. The numbers and arrows at the top and adjacent to the bars at the bottom correspond to the 5⬘ termini of the various derivatives, whose names are listed to the left of the bars. The vertical bars represent the regions mutagenized but not deleted in pABG11 and pABG12. Mutagenesis consisted of the introduction of transversion substitutions at alternating positions along the indicated regions.

both the large and small RofA binding sites and the region between the two sites contribute to activation of the promoter for rofA under the conditions used in this study; however the major contribution is made by the large RofA binding site. DISCUSSION In this study, we have examined the interaction between RofA and its target promoters by a combination of DNase I footprinting and analysis of mutants using a novel vector for examination of transcriptional regulation in gram-positive bacteria. This analysis revealed that RofA recognizes two sites in the region between prtF and rofA, that these sites are bounded by a conserved sequence, and that the larger of the two sites may actually consist of two adjacent sites. Furthermore, while both the large and small sites may contribute to expression of

rofA, only the larger site is required for expression of prtF. Consistent with multiple patterns of regulation of prtF between different strains of S. pyogenes, different strains exhibited heterogeneity in their requirements for different regions of the large site for prtF expression. Interestingly, the large binding site was also required for expression of prtF under conditions where RofA itself is not, suggesting that a common mechanism of activation underlies regulation of fibronectin binding in response to multiple pathways and environmental signals. Careful examination of the regions protected from DNase I digestion reveals a number of striking features. First, the regions are bounded by nearly identical 11-bp sequences that are in an inverted orientation in the larger site. While the beginning of this repeat in the prtF proximal copy was not expressly protected from DNase I digestion of the noncoding strand

FIG. 9. Activities of various mutant rofA promoters. PhoZF activity in cell-free supernatants of various derivatives of pABG5 in JRS4 are shown as a percentage of the activity obtained with unaltered pABG5 under anaerobic conditions. For the structure of each indicated plasmid, see Fig. 8. Alkaline phosphatase activities were determined as described in Materials and Methods. Data represent the means and standard errors of the means for at least three independent experiments, each conducted in duplicate.

VOL. 182, 2000

(Fig. 1), it was protected on the coding strand, as both protected regions were shifted by three base pairs toward prtF (data not shown). Also, since the small site and the prtF distal 17 bases of the large site are a highly conserved inverted repeat (matching at 14 of 17 positions) and similar to the prtF proximal end of the large site, it is likely that (i) the small site represents the minimal binding site for RofA and (ii) the large site is actually two inverted sites separated by nine bases. From comparison of these three 17-bp sites, it is possible to derive a preliminary consensus RofA binding site of TTTTCACCAAA AANCAT (where consensus was defined by the presence of a nucleotide in at least two of the sequences and the “N” indicates that a consensus nucleotide could not be determined by this criterion). Characterization of the RofA binding site made it possible to survey the genome for other genes potentially regulated by RofA. An obvious candidate is cpa.1, the gene for a putative surface protein located adjacent to rofA in the serotype M1 strain that has been sequenced (31). The region between rofA and cpa.1 is highly similar to the prtF-rofA intergenic regions of strains JRS4 and HSC5 and the RofA binding site themselves are identical to those of the prtF-containing strains. A preliminary search of the rest of the genome revealed at least one other gene that is potentially regulated by RofA. The putative promoter region for an open reading frame bearing significant homology to the stress-induced ClpX chaperone (a region consisting of approximately three-fourths of the amino acid sequence of the putative gene is 66% identical and 82% similar to ClpX of Bacillus subtilis [GenBank accession no. X953O6) contains a RofA box identical in sequence to the RofA binding site adjacent to the promoter for rofA. Furthermore, the box is in the identical orientation and position relative to the open reading frame and putative promoter of clpX as compared to rofA. This observation is interesting for several reasons. In gram-negative organisms, clpX is located in an operon with the gene which encodes the ClpP protease and is under the control of the heat shock alternative sigma factor ␴32 (11). In grampositive organisms, clpX and clpP reside at unlinked loci, and clpP, but not clpX, is regulated by the recently described transcriptional repressor CtsR (4, 10). The regulation of clpX is not understood, but the participation of RofA is suggested by the presence of the RofA box. If true, this would provide an additional link between regulation of stress responses, adhesin expression and virulence. Some support for this idea has recently come from several studies using signature-tagged mutagenesis in gram-positive bacteria which have independently identified clpX as important for survival of bacteria during infection (3, 24). Manipulation of the RofA binding sites provided some interesting contrasts and comparisons among the various strains tested. In all strains, the smaller RofA site was not required for prtF expression, nor was the region between the small and large sites. In contrast, the larger RofA site was required for expression of prtF under anaerobic conditions in all strains. While the entire large site was required for prtF expression in MGAS166s, either half of the larger RofA site was sufficient for aerobic expression in JRS4. Since previous data suggest aerobic activation in JRS4 involves a trans-acting regulatory element distinct from RofA, these data imply that this regulatory element recognizes the same site that is bound by RofA. In addition, since the two sites that comprise the large binding site are in an inverted configuration, the observation that either site is sufficient may suggest that the ability of this element to activate transcription is not dependent on the conformation of the bound regulator relative to prtF promoter. Since aerobic expression in MGAS166s required the entire large binding site,


1537

this strain may not use the same activators of transcription as does JRS4 under aerobic conditions. Alternatively, JRS4 may utilize additional transcriptional activators which are more permissive in their interaction with DNA and RNA polymerase. Regardless, expression in MGAS166s resembles that in JRS4, in that they both do require sites that are recognized by RofA. Heterogeneity among strains in requirement for different regions of the large RofA binding site is consistent with the observation that different strains demonstrate different patterns of expression of prtF and differ in the degree to which they require RofA under certain conditions (7, 8). Available data suggest that this heterogeneity is due not to any difference in RofA itself but rather to the contribution of other regulatory elements (7). The identification of the virulence regulator Nra, with its high level of homology to RofA, and an apparent overlap of regulatory targets (31), combined with the observation that the RofA binding sites are important for regulation of prtF and rofA in response to multiple pathways, suggests that Nra and RofA may have cooperative roles. Arguing against this idea are the findings that many rofA-containing strains do not have nra and that some nra-containing strains do not have rofA (31). However, examination of the serotype M1 genome reveals two additional open reading frames highly homologous to rofA (which we have provisionally designated RALP, for RofA-like proteins [Fig. 10]), which suggests a more extensive and potentially interactive network of regulators. An additional RALP was found in the partial genome information available for Streptococcus pneumoniae (RALP5 [Fig. 10]) (http://www.tigr.org). Overall, the RALPs are similar in size (502 to 512 residues), are on average 52% similar and 29% identical to RofA, and contain multiple regions of near identity (Fig. 10). Several cysteine residues are well, but not universally, conserved, and this family contains an unusually high overall content of leucine residues (14 to 16%). Not well conserved is the region corresponding to a putative helix-turn-helix region of RofA (Fig. 10). This may suggest that these proteins recognize unique target DNA sequences. However, it has not been established that this region of RofA is involved in DNA binding or that the other members of this family are DNA-binding proteins. However, the observation that the homologue in the pneumococcal genome is located in the vicinity of two open reading frames with homology to a known fibronectin and fibrinogen-binding protein (SspA of Streptococcus gordonii) (14) suggests a general role for RALPs in regulation of adhesin expression. Indeed, the observation that RofA and Nra are involved in regulating similar genes raises the possibility that one or more of the RALPs might also be involved in this regulatory network. Additional work will be required to elucidate the molecular details concerning the structure and function of this emerging and novel family of proteins. The phoZF reporter gene and its adaptation for use on an easy to manipulate E. coli-streptococcal shuttle vector provided a valuable tool to study transcription of streptococcal genes. The plasmid, the chimeric gene and secreted PhoZF polypeptide all proved to be very stable and in addition, PhoZF possesses a robust alkaline phosphatase activity. The wide variety of rapid and simple assays and the large number of substrates that are available for alkaline phosphatase, including colorimetric, fluorescent, and light-emitting varieties, in combination with the ready access that substrate has to the freely secreted PhoZF makes this a versatile reporter that should have wide application for analyses of gene expression in S. pyogenes. The broad host range of the pLZ12 plasmid vector

FIG. 10. The emerging RofA-like family of proteins (RALPs). A multiple alignment of several RALPs relative to RofA is shown. Nra, a repressor of several virulence-associated genes in S. pyogenes, has been described elsewhere (31), and RALP3 and RALP4 were identified in the available serotype M1 genome information (http://www.genome.ou.edu/strep.html). RALP5 was identified in the partial genome information available for S. pneumoniae (http://www.tigr.org). Identical amino acids are shaded in black, and similar amino acids shared by at least 3 of proteins and putative proteins are highlighted in grey. The region containing the putative helix-turn-helix DNA-binding domain previously noted in RofA is underlined. An asterisk and an arrowhead indicate cysteine residues conserved between at least two and at least three open reading frames, respectively. Note the high degree of conservation of leucine residues throughout the proteins.

1538 GRANOK ET AL. J. BACTERIOL.


VOL. 182, 2000

extends the utility of PhoZF to many other gram-positive bacterial species. Potential drawbacks to the use of this vector include its high copy number, which may mask subtle regulatory phenomena or result in the titration of certain regulatory elements. A copy number effect could explain preliminary data which indicated that a rofA::phoZF construct did not demonstrate the same degree of environmental regulation that has been observed in previous studies. However, this could also reflect the fact that due to the secreted nature of PhoZF, it was not possible to quantitate alkaline phosphatase activities under the same highly aerobic culture conditions used in previous studies (culture on the surface of solid medium). A number of other observations suggested that expression of the reporter allele did resemble expression of the chromosomal allele, including that consistent with prior studies, expression of the plasmidborne reporter allele demonstrated an absolute dependence on an intact chromosomal copy of rofA (data not shown) and that expression of rofA::phoZF required the RofA binding sites. Thus, this reporter should make a valuable contribution to the techniques available for analysis of gene expression in S. pyogenes and other pathogenic gram-positive bacterial species. Further analysis of RofA will continue to provide important insights into this novel and expanding family of regulators and into regulation of virulence in S. pyogenes.

13.

14.

15. 16. 17. 18. 19.

20.

21.

ACKNOWLEDGMENTS We thank George C. Fogg for purification the MBP-RofA fusion protein and Neil Barg for his gift of MGAS166s. We also thank the Streptococcus pyogenes genome sequencing project at the University of Oklahoma Health Sciences Center and TIGR Microbial Database for genomes in progress. This work was supported by Public Health Service grant AI38273 and training grant 5 T32 AI07172 from the National Institutes of Health. M.G.C. is an Established Investigator of the American Heart Association.

22. 23. 24. 25. 26.

REFERENCES 1. Betschel, S. D., S. M. Borgia, N. L. Barg, D. E. Low, and J. C. S. De Azavedo. 1998. Reduced virulence of group A streptococcal Tn916 mutants that do not produce streptolysin S. Infect. Immun. 66:1671–1679. 2. Chaussee, M. S., J. Liu, D. L. Stevens, and J. J. Ferretti. 1996. Genetic and phenotypic diversity among isolates of Streptococcus pyogenes from invasive infections. J. Infect. Dis. 173:901–908. 3. Coulter, S. N., W. R. Schwan, E. Y. W. Ng, M. H. Langhorne, H. D. Ritchie, S. Westbrock-Wadman, W. O. Hufnagle, K. R. Folger, A. S. Bayer, and C. K. Stover. 1998. Staphylococcus aureus genetic loci impacting growth and survival in multiple infection environments. Mol. Microbiol. 30:393–404. 4. Derre´, I., G. Rapoport, and T. Mdadek. 1999. CtsR, a novel regulator of stress and heat shock response, controls clp and molecular chaperone gene expression in Gram-positive bacteria. Mol. Microbiol. 31:117–131. 5. Dunny, G. M., L. N. Lee, and D. J. LeBlanc. 1991. Improved electroporation and cloning vector system for gram-positive bacteria. Appl. Environ. Microbiol. 57:1194–1201. 6. Fischetti, V. A. 1989. Streptococcal M protein: molecular design and biological behavior. Clin. Microbiol. Rev. 2:285–314. 7. Fogg, G. C., and M. G. Caparon. 1997. Constitutive expression of fibronectin binding in Streptococcus pyogenes as a result of anaerobic activation of rofA. J. Bacteriol. 179:6172–6180. 8. Fogg, G. C., C. M. Gibson, and M. G. Caparon. 1993. Identification of rofA, a positive-acting regulatory component of prtF expression: use of a m␥␦based shuttle mutagenesis strategy in Streptococcus pyogenes. Mol. Microbiol. 11:671–684. 9. Galas, D. J., and A. Schmitz. 1978. DNase footprinting: a simple method for the detection of protein-DNA binding specificity. Nucleic Acids Res. 5:3157– 3170. 10. Gerth, U., A. Wipat, C. R. Harwood, N. Carter, P. T. Emmerson, and M. Hecker. 1996. Sequence analysis and transcriptional analysis of clpX, a classIII heat-shock gene of Bacillus subtilis. Gene 181:77–83. 11. Gottesman, S., W. P. Clark, V. de Crecy-Lagard, and M. R. Maurizi. 1993. ClpX, an alternative subunit for the ATP-dependent Clp protease of Escherichia coli. Sequence and in vivo activities. J. Biol. Chem. 268:22618–22626. 12. Hanski, E., and M. G. Caparon. 1992. Protein F, a fibronectin-binding

27.

28. 29. 30.

31. 32. 33. 34. 35. 36.

1539

protein, is an adhesin of the group A streptococcus, Streptococcus pyogenes. Proc. Natl. Acad. Sci. USA 89:6172–6176. Hanski, E., P. A. Horwitz, and M. G. Caparon. 1992. Expression of protein F, the fibronectin-binding protein of Streptococcus pyogenes JRS4, in heterologous streptococcal and enterococcal strains promotes their adherence to respiratory epithelial cells. Infect. Immun. 60:5119–5125. Holmes, A. R., C. Gilbert, J. M. Wells, and H. F. Jenkinson. 1999. Binding properties of Streptococcus gordonii SspA and SspB (antigen I/II family) polypeptides expressed on the cell surface of Lactococcus lactis MG1363. Infect. Immun. 66:4633–4639. Horton, R. M., H. D. Hunt, S. N. Ho, J. K. Pullen, and L. R. Pease. 1989. Engineering hybrid genes without the use of restriction enzymes: gene splicing by overlap extension. Gene 77:61–68. Ike, Y., R. A. Craig, B. A. White, Y. Yagi, and D. B. Clewell. 1983. Modification of Streptococcus faecalis sex pheromones after acquisition of plasmid DNA. Proc. Natl. Acad. Sci. USA 80:5369–5373. Jadoun, J., V. Ozeri, E. Burstein, E. Skutelsky, E. Hanski, and S. Sela. 1998. Protein F1 is required for efficient entry of Streptococcus pyogenes into epithelial cells. J. Infect. Dis. 178:147–158. Jaffe, J., S. Natanson-Yaron, M. G. Caparon, and E. Hanski. 1996. Protein F2, a novel fibronectin-binding protein from Streptococcus pyogenes, possesses two binding domains. Mol. Microbiol. 21:373–384. Katerov, V., A. Andreev, C. Schalen, and A. A. Totolian. 1998. Protein F, a fibronectin-binding protein of Streptococcus pyogenes, also binds human fibrinogen: isolation of the protein and mapping of the binding region. Microbiology 144:119–126. Kushner, S. R. 1978. An improved method for transformation of Escherichia coli with ColEl-derived plasmids, p. 17–23. In H. W. Boyer and S. Micosia (ed.), Genetic engineering. Elsevier/North Holland Biomedical Press, New York, N.Y. Lee, M. H., A. Nittayajarn, R. P. Ross, C. B. Rothschild, D. Parsonage, A. Claiborne, and C. E. Rubens. 1999. Characterization of Enterococcus faecalis alkaline phosphatase and use in identifying Streptococcus agalactiae secreted proteins. J. Bacteriol. 181:5790–5799. Malke, H. 1998. Cytoplasmic membrane lipoprotein LppC of Streptococcus equisimilis functions as an acid phosphatase. Appl. Environ. Microbiol. 64: 2439–2442. Maxam, A., and W. Gilbert. 1980. Sequencing end-labeled DNA with basespecific chemical cleavages. Methods Enzymol. 65:499–560. Mei, J.-M., F. Nourbakhsh, C. W. Ford, and D. W. Holden. 1997. Identification of Staphylococcus aureus virulence genes in a murine model of bacteraemia using signature-tagged mutagenesis. Mol. Microbiol. 26:399–407. Mekalanos, J. J. 1992. Environmental signals controlling expression of virulence determinants in bacteria. J. Bacteriol. 174:1–7. Miller, J. F., J. J. Mekalanos, and S. Falkow. 1989. Coordinate regulation and signal transduction in the control of bacterial virulence. Science 243: 916–921. Molinari, G., S. R. Talay, P. Valentin-Weigand, M. Rohde, and G. S. Chhatwal. 1997. The fibronectin-binding protein of Streptococcus pyogenes, SfbI, is involved in the internalization of group A streptococci by epithelial cells. Infect. Immun. 65:1357–1363. Natanson, S., S. Sela, A. Moses, J. M. Musser, M. G. Caparon, and E. Hanski. 1995. Distribution of fibronectin-binding proteins among group A streptococci of different M types. J. Infect. Dis. 171:871–878. Okada, N., A. Pentland, P. Falk, and M. Caparon. 1994. M protein and protein F act as important determinants of cell-specific tropism of Streptococcus pyogenes in skin tissue. J. Clin. Investig. 94:965–977. Ozeri, V., A. Tovi, I. Burstein, S. Natanson-Yaron, M. G. Caparon, K. M. Yamada, S. K. Akiyama, I. Vlodavsky, and E. Hanski. 1996. A two-domain mechanism for group A streptococcal adherence through protein F to the extracellular matrix. EMBO J. 15:898–998. Podbielski, A., M. Woischnik, B. A. B. Leonard, and K.-H. Schmidt. 1999. Characterization of nra, a global negative regulator gene in group A streptococci. Mol. Microbiol. 31:1051–1064. Rakonjac, J. V., J. C. Robbins, and V. A. Fischetti. 1995. DNA sequence of the serum opacity factor of group A streptococci: identification of a fibronectin-binding repeat domain. Infect. Immun. 63:622–631. Rocha, C. L., and V. A. Fischetti. 1999. Identification and characterization of a novel fibronectin-binding protein on the surface of group A streptococci. Infect. Immun. 67:2720–2728. Scott, J. R., P. C. Guenther, L. M. Malone, and V. A. Fischetti. 1986. Conversion of an M⫹ group A streptococcus to M⫹ by transfer of a plasmid containing an M6 gene. J. Exp. Med. 164:1641–1651. Sela, S., A. Arvik, I. Burstein, A. Tovi, M. G. Caparon, and E. Hanski. 1994. Protein F, an adhesin of Streptococcus pyogenes, binds fibronectin via two distinct domains. Mol. Microbiol. 10:1049–1055. Swedo, S. E., H. L. Leonard, M. Garvey, B. Mittleman, A. J. Allen, S. Perlmutter, S. Dow, J. Zamkoff, B. K. Dubbert, and L. Lougee. 1997. Pediatric autoimmune neuropsychiatric disorders associated with streptococcal infections: clinical description of the first 50 cases. Am. J. Psychiatry 155: 264–271.

1540

GRANOK ET AL.

37. Talay, S. R., P. Valentin-Weigand, P. G. Jerlstro¨m, K. N. Timmis, and G. S. Chhatwal. 1992. Fibronectin-binding protein of Streptococcus pyogenes: sequence of the binding domain involved in adherence of streptococci to epithelial cells. Infect. Immun. 60:3837–3844. 38. Talkington, D. F., B. Schwartz, C. M. Black, J. K. Todd, J. Elliot, R. F. Breiman, and R. Facklum. 1993. Association of phenotypic and genotypic

J. BACTERIOL. characteristics of invasive Streptococcus pyogenes isolates with clinical components of streptococcal toxic shock syndrome. Infect. Immun. 61:3369– 3374. 39. VanHeyningen, T., G. Fogg, D. Yates, E. Hanski, and M. Caparon. 1993. Adherence and fibronectin-binding are environmentally regulated in the group A streptococcus. Mol. Microbiol. 9:1213–1222.