Characterization of the SarA virulence gene regulator of Staphylococcus aureus

Oligomeric state and secondary structure of SarA

SarA was expressed in E. coli and purified as described in Experimental procedures. Stoichiometric binding analysis with the P3 DNA fragment (described in Fig. 1) revealed that our SarA preparations are 90–95% active (data not shown). The concentration of dimeric (described below) active SarA was used in all subsequent experiments.

The stoichiometry of the predominant SarA complex was established by using two independent techniques: chemical cross-linking and dynamic light scattering (DLS). For the chemical cross-linking experiments, SarA was exposed to increasing concentrations of cross-linking agents (see Experimental procedures for details). The reactions were quenched, and the products were resolved by tricine-SDS–PAGE. In 2Fig. 2A a representative gel after detection of SarA by staining with Coomassie blue is shown. Increasing the concentration of the cross-linking agent increased the relative amount of a species that was the appropriate molecular weight to be a SarA dimer (calculated 29.4 kDa). The data in 2Fig. 2A were generated with DSS (disuccinimidyl suberate); however, other cross-linking agents (DSP and BS³) gave equivalent results. No other species were detected, even at the highest concentration of DSS when over 50% of SarA had been cross-linked. The same results were observed in the presence of saturating concentrations of P2–P3 promoter DNA and Mg²⁺, indicating that the species probably binding DNA is a dimer (data not shown). In addition, cross-linking for longer periods of time and under different conditions did not result in the detection of higher molecular weight species.

Dynamic light scattering (DLS) was also used to examine the oligomeric state of SarA. DLS reveals the homogeneity and oligomeric state of proteins based on the diffraction of visible light. The oligomerization state of SarA at two concentrations (3.3 and 11.7 mg ml⁻¹) was analysed using a 2001 DynaPro DLS instrument and the attendant software dynamics. The bimodal analysis of the scattering revealed a monodisperse solution with a macromolecular species of molecular weight 34 kDa. This is consistent with scattering from a slightly elongated SarA dimer (calculated 29.4 kDa). Taking the cross-linking and DLS experiments together, the predominant form of SarA is a dimer. Therefore, SarA concentration is expressed in terms of dimer concentration for all quantitative analyses in this study.

To begin to examine the structural aspects of SarA, circular dichroism was used to determine the secondary structure elements. The circular dichroism spectrum was taken from 260 nm to 180 nm. The spectrum showed large negative ellipticity at 208 nm and 220 nm, indicative of a high helical content (Fig. 2B). Analysis of the spectrum from 260 nm to 190 nm revealed SarA to be largely alpha-helical (55%), with very little β-strand (6%) and modest amounts of turn (15%) and random coil (27%). Thus, it was concluded that the structure of SarA consists predominantly of alpha-helices.

DNase I footprint analysis of SarA binding sites in the P2–P3 region

The results of genetic experiments from Arvidson's group indicated that ≈60 bp upstream of the P3 promoter was sufficient for SarA-dependent expression of RNA III (Morfeldt et al., 1996). However, footprinting results from Cheung's group indicated a primary SarA binding site that did not overlap the putative regulatory region (Chien and Cheung, 1998). To begin to clarify the exact SarA binding site(s) and to resolve the discrepancy described above, we performed DNase I footprinting with purified SarA. The DNA target for footprinting studies, the entire interpromoter region, was polymerase chain reaction (PCR) amplified from a plasmid clone (pBKH50) using ³²P-labelled −40 and reverse M13 primers. Representative results of the DNase I analysis in the presence of increasing concentrations of SarA are shown in 3Fig. 3A. The data were analysed digitally after phosphorimaging. Analysis of both strands revealed the same protected regions, and a summary of these results is presented in Fig. 1. Three sets of footprints were revealed by this analysis. Each set consisted of two footprints of ≈18 bp separated by 4–5 bp, which have an identical pattern of DNase I protection. The concomitant protection of the pairs of sites suggests that each set (A, B and C) consists of two half-sites. Sites B1 and B2 were fully protected at the lowest concentration of SarA, 6.4 nM, which is indicative of the highest affinity binding site. The C1/C2 and A1/A2 sites were fully protected at a slightly higher SarA concentration, 12.8 nM. Hypersensitive sites were observed routinely on the outermost portion of the DNA fragments tested. The sequences of the six regions protected were aligned using pileup, and a consensus sequence was derived using PRETTY, programs resident in the GCG software package (Devereux et al., 1984). The results of this analysis are shown in 3Fig. 3B. All six DNA sequences are over 79% AT.

Interactions of SarA with the entire region between the P2 and P3 promoters

In the EMSA data reported by Morfeldt et al. (1996) using large DNA fragments between the P2 and P3 promoters, an unusual ladder pattern of SarA–DNA complexes was observed. As the SarA-containing protein sample used in that study was a crude lysate of S. aureus, we asked whether our recombinant SarA had similar activity. The 180 bp region used in the DNase I experiments described above was amplified by PCR with ³²P-labelled primers and used in EMSA experiments. The unbound and SarA-bound DNA fragments were quantified by phosphorimaging. A representative gel is shown in Fig. 4. Indeed, a ladder pattern of protein–DNA complexes was generated, very similar to that shown by Morfeldt et al. (1996). One conclusion from this experiment is that the native SarA and the recombinant SarA have very similar activities in EMSA. The high concentration of DNA used in these experiments (200 nM) allows us to determine the stoichiometry of binding. The concentration of SarA required for 50% and 100% of the DNA to be bound was observed to be a protein–DNA ratio of 3:1 (300 nM and 600 nM respectively). This result supports the concept of one SarA dimer binding to each set of footprints identified in the DNase I analysis. The total number of resolvable bands was 12, with the first seven bands having a uniform laddering pattern and shifting preferentially. Given a 3:1 SarA–DNA stoichiometry at saturation, the simplest interpretation of this result is that the bands observed in this analysis represent different conformations of SarA–DNA complexes. However, a more complicated model for interactions of SarA with this DNA fragment might explain the data better. One such model, originally described by Morfeldt et al. (1996), would postulate that SarA alters the superhelicity of the bound DNA (described fully in Discussion ).

Determination of equilibrium binding constants for SarA binding site fragments using EMSA

Whereas the DNase I footprint analysis described above indicates the relative affinity of three proposed SarA binding sites, we performed quantitative EMSAs to derive the equilibrium binding constants (K_D) for synthetic DNA fragments containing those sites, as well as sites described by Chien and Cheung (1998) and Morfeldt et al. (1996). The promoter region was divided into five fragments of DNA (P3, P2, AC, A1/A2 and B1/B2), as shown in Fig. 1. A limiting concentration of ³²P-labelled DNA (< 10 pM) was mixed with various concentrations of purified SarA and subjected to native gel electrophoresis. The amount of DNA in the unbound and bound state was determined by phosphorimaging. Representative gels of SarA with the fragments are shown in 5Fig. 5A. The SarA–DNA complexes with fragments P2, P3 and AC were observed to be a unique species in the gels. However, the complexes formed with either the A1/A2 or B1/B2 fragments yielded two distinct species. The significance of this result is addressed in the Discussion. In 5Fig. 5B, the percentage of the total DNA bound by SarA is plotted against the concentration of SarA present in the mixture. The data were fitted using the non-linear least squares program bioeqs as described in Experimental procedures. The data fitted best to a model of one stable protein species (dimer) binding to each of the 45 bp DNA fragments. However, a monomer–dimer equilibrium is most probably involved in SarA binding to the A1/A2 and B1/B2 DNAs (addressed in the Discussion ). A summary of the calculated values for ΔG and K_D are presented in Table 1. SarA binding affinity for the DNA fragments correlates with the number of half-sites present in the DNA fragment tested. The DNA fragments containing a pair of intact half-sites, A1/A2 and B1/B2, were bound with highest affinity (K_D = 10 and 7 pM respectively). The P3 and P2 fragments have 1.5 half-sites (K_D = 220 pM), and the AC fragment has one half-site (K_D = 1 nM). As a negative control, SarA binding affinity for a heterologous DNA fragment (the multiple cloning site of pUC118) was tested using EMSA. No specific SarA–DNA complexes were resolved, and the apparent K_D for the interaction was at least 3 nM, which is 300- to 400-fold higher than the A1/A2 or B1/B2 fragments (data not shown).

Table 1. . Summary of binding constants. a. Number in parentheses represents co-operativity values, ρ values, from analysis of binding curves.b. Ratio of SarA dimer mol⁻¹ DNA.

In addition, we performed EMSA of SarA binding to these fragments under stoichiometric binding conditions (> 10 nM DNA; data not shown). In each case, the analysis revealed a stoichiometry of 1:1. These data lend further support to a model of one SarA dimer binding to two half-sites. The data were also analysed using a sigmoidal power logistic function resident in origin software, which yields a value, ρ, reflecting the slope of the curve or co-operativity of binding (Czernik et al., 1996). Fragments P2, P3 and AC have ρ values of 1, indicating no co-operativity in SarA binding. Whereas the fragments bound with the highest affinity, A1/A2 and B1/B2, had ρ values of 2, reflecting considerable apparent positive co-operativity. As the stoichiometry of SarA–DNA for all fragments tested was 1:1, we conclude that, at the very low concentrations of SarA needed to bind the A1/A2 and B1/B2 fragments, there must be considerable SarA monomer present. The presence of DNA serves to drive the equilibrium towards dimer formation.

Expression and purification of SarA

The SarA coding region was amplified using PCR from S. aureus strain DB with primers incorporating NdeI and BamHI restriction enzyme sites:

NdeI 5′-GGGAGGTTTTACATATGGCAATTACAAAAATC-3′;

BamHI 5′-GTTTAATAGAATGGATCCTCTATCAAACTTCACC-3′.

S. aureus chromosomal DNA was isolated as described by Smeltzer et al. (1996). The PCR products were restricted with NdeI and BamHI and ligated into similarly restricted pET9A (Novagen) to yield pET-DB. The fidelity of the construct was determined by sequence analysis.

Recombinant SarA expression was induced in E. coli strain BL21(DE3)pLysS when the OD of the culture was 0.4 at 600 nm for various times using a final concentration of 1 mM IPTG in LB broth at 37°C. The levels of expression were checked by SDS–PAGE of crude lysates. Maximal expression occurred within 3 h of induction. Cells were collected by centrifugation (5000 × g, 10 min, 4°C), and cell pellets were frozen at −20°C. Cells were lysed in 50 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol (DTT), 1 mM EDTA, 1 mM phenylmethylsulphonyl fluoride (PMSF) on ice. Chromosomal DNA was sheared with ultrasound treatment, and the insoluble debris was removed by centrifugation (15 000 × g, 30 min, 4°C). The cleared lysate was brought to 70% saturation with solid ammonium sulphate, and the insoluble material was removed by centrifugation (15 000 × g, 30 min, 4°C). The soluble supernatant was twice subjected to dialysis against at least 100 volumes of HSB-150 (HSB = 20 mM Tris-HCl, pH = 7.6, 1 mM EDTA, 1 mM DTT and 150 mM NaCl). The resultant solution was loaded on a heparin-Sepharose column (50 ml bed volume) and washed with at least 300 ml of HSB-150. Proteins were eluted with a linear gradient of HSB-150 to HSB-1500 (HSB with 1.5 M NaCl). The total volume of the gradient was 400 ml. Column fractions were analysed for the presence of SarA using tricine-SDS–PAGE, and peak fractions were pooled. Purified SarA was stored at −20°C and is stable for at least 6 months without noticeable loss of activity. The typical yield of SarA was 1–3 mg l⁻¹ of culture. The concentration of SarA was determined spectrophotometrically at 280 nm using an extinction coefficient of 7740 M⁻¹ (Gill and von Hippel, 1989).

To determine the activity of the purified SarA, we used stoichiometric binding conditions in EMSA with DNA fragment P3 (Fig. 1). At DNA concentrations that are very high relative to the equilibrium dissociation constant (K_D), the amount of protein required to bind 50% of the available DNA is used to determine the concentration of active protein using the equation K_D = [P]_1/2 − 1/2[DNA]_o, where [P]_1/2 is the protein concentration at 50% saturation, and [DNA]_o is the total DNA concentration (Riggs et al., 1970; Hurlburt and Yanofsky, 1990). Under these conditions ([DNA] > 10 nM), K_D is insignificant, and the equation simplifies to [P]_1/2 = 1/2[DNA]_o. Assuming that SarA binds to the P3 DNA fragment as a dimer (described in Results ), the activity of SarA in our preparations is routinely 90–95% active of the value determined spectrophotometrically. The concentration of SarA reported in our data reflects the concentration of active protein.

Chemical cross-linking of SarA

SarA in 50 mM sodium phosphate (pH 8.0), 150 mM NaCl, 1 mM EDTA and 1 mM DTT was treated with protein cross-linking agents bis-(sulphosuccinimidyl)suberate (BS³), disuccinimidyl suberate (DSS) and dithiobis(disuccinimidylpropionate) (DSP). In the case of DSP, DTT was excluded from the buffer. The reaction mixture contained 10 μM SarA and cross-linker in a total reaction volume of 30 μl. Reactions were allowed to proceed for 1 min at 4°C and quenched by the addition of 10 μl of 1 M Tris-Cl (pH 8.0). Denaturing loading dye was added to each reaction and incubated at 95°C for 20 min. Reaction products were analysed on a 12% tricine-SDS–polyacrylamide gel and visualized by Coomassie blue staining. A product consistent with a SarA dimer was observed. In the optimization of the reaction conditions, concentrations of cross-linker (5–500 μM), time (1 min to 1 h) and temperature were varied. No oligomeric species were detected exceeding a dimer. In addition, 5 mM MgCl₂ and 15 μM DNA containing the P2 and P3 promoter regions were included in the reaction mixture. Again, the only product detected was a SarA dimer.

Dynamic light scattering

The oligomerization state of purified SarA was examined in the presence and absence of 10 mM MgCl₂ by dynamic light scattering (DLS) using a 2001 DynaPro Dynamic Light Scattering Instrument and analysis software, dynamics, version 3.30. In each experiment, SarA (3.3 mg ml⁻¹) was buffered by solutions containing 50 mM Tris-HCl, pH 7.5, 2 mM DTT, 500 mM NaCl and 1 mM EDTA. Twenty measurements were taken at 22°C for each analysis. The bimodal analysis of the scattering revealed a monodisperse solution with a macromolecular weight of 34 kDa. This is consistent with the scattering from a slightly elongated SarA dimer (calculated molecular weight of 29.4 kDa).

Circular dichroism

The protein concentration used for this study was 1.8 mg ml⁻¹ in a solution of 50 mM potassium phosphate, pH 7.5. The circular dichroism spectrum was taken from 260 nm to 180 nm. The spectrum showed large negative ellipticity at 208 nm and 220 nm, indicative of a high helical content. Using the spectrum from 260 nm to 190 nm, the percentage of each secondary structure element was calculated.

DNase I protection assay

A 180 bp DNA fragment, which spans the −10 regions of the P2 and P3 promoters, was used as the target for DNase I protection studies. This region of DNA (shown in Fig. 1) was cloned from the genomic DNA of the S. aureus strain DB by PCR amplification. The primers contained BamHI sites, and the PCR product was cloned into the BamHI site of the plasmid pUC118 to form pBKH50. A 240 bp DNA fragment containing the P2–P3 promoter region was PCR amplified from the plasmid pBKH50 with the ³²P end-labelled primers, −40 and M13 reverse. SarA was allowed to bind to the DNA in a reaction mixture containing 40 pM ³²P-labelled DNA, 10 mM HEPES (pH 7.6), 5 mM MgCl₂, 1 mM CaCl₂, 1 mM DTT and 100 mM KCl at room temperature for 30 min. An amount of DNase I that produced ≈50% non-nicked DNA was added to the reaction mixture and incubated for 2 min. The reaction was quenched by the addition of stop solution (80% EtOH, 1 μg ml⁻¹ tRNA, 0.3 M NH₄OAc) and immediately put into a dry ice/EtOH bath for 30 min. A visible pellet was obtained after centrifugation and washed with 70% ethanol. After denaturing gel electrophoresis, the DNA fragments were detected and quantified by phosphorimaging. The radioactivity in the bands for samples containing SarA was subtracted from the no-protein control using Microsoft Excel. The concentration of SarA required to protect a cleavage site fully was determined. To identify the regions of protection within the interpromoter region, purine sequencing was included in the analysis. Chemical sequencing was performed according to standard procedures.

Electrophoretic mobility shift assay (EMSA)

The sequences of DNAs used in EMSA are illustrated in Fig. 1. DNAs P2, P3, A1/A2 and B1/B2 were designed to test the importance of the heptad repeats and the footprints revealed in this study respectively. The AC fragment is exactly the same as a fragment used by Chien and Cheung (1998) and was used to test the importance of the footprint reported in that study. The oligonucleotides were synthesized, end-labelled with ³²P and purified after denaturing polyacrylamide gel electrophoresis. Complimentary DNAs were annealed, purified by native gel electrophoresis and used in the binding experiments. Quantitative EMSA was performed with these synthetic DNA fragments to determine both the stoichiometry of SarA binding and equilibrium dissociation constants (K_D). DNA concentrations greater than 10 nM were used for stoichiometric binding analysis. Limiting concentrations of DNA (< 10 pM) were used when determining K_D values. Various concentrations of SarA were incubated in a 20 μl reaction buffer with ³²P-labelled DNA and in buffer containing 10 mM HEPES, pH 7.6, 1 mM EDTA, 2 mM DTT, 50 mM KCl, 0.05% Triton X-100 and 5% glycerol. Binding reactions were allowed to equilibrate for 30 min before electrophoresis. Bound products were separated from free DNA on 6% native polyacrylamide (50:1 acrylamide–bisacrylamide) in 0.5 × Tris borate–EDTA. Gels were run at 200 V, and temperature was maintained at 16°C with a circulating water bath. Resolved gels were dried and the products quantified by phosphorimaging. The amount of bound DNA was calculated from the reduction in the unbound DNA and plotted against the respective SarA concentrations. Stoichiometric determinations were performed as described above. For K_D determinations, the resulting data were fitted to various potential models of SarA–DNA interactions including: SarA SarA₂ SarA₂-DNA; SarA₂ SarA₂-DNA; 2SarA₂ 2SarA₂-DNA, using the numerically based algorithm bioeqs (Royer et al., 1990; 1993; Royer and Beechem, 1992). Based on the Gibbs free energies of dissociation (ΔG), bioeqs solves directly for the species population. The equilibrium dissociation constants (K_D) were calculated from the ΔG values using the formula, ΔG = −RTlnK_D. Co-operativity values were determined as described by Czernik et al. (1996) using the sigmoidal power logistic function resident in origin (Microcal). The ρ value is indicative of the slope of the curve. A ρ value greater than 1 indicates positive co-operativity, and less than 1, negative co-operativity. If ρ = 1, then no co-operativity is apparent.