|Home | About | Journals | Submit | Contact Us | Français|
Nutrient limitation causes Bacillus subtilis to develop into two different cell types, a mother cell and a spore. SpoIIID is a key regulator of transcription in the mother cell and positively or negatively regulates more than 100 genes, in many cases by binding to the promoter region. SpoIIID was predicted to have a helix-turn-helix motif for sequence-specific DNA binding, and a 10-bp consensus sequence was recognized in binding sites, but some strong binding sites were observed to contain more than one match to the consensus sequence, suggesting that SpoIIID might bind as a dimer or cooperatively as monomers. Here we show that SpoIIID binds with high affinity as a monomer to a single copy of its recognition sequence. Using charge reversal substitutions of residues likely to be exposed on the surface of SpoIIID and assays for transcriptional activation in vivo and for DNA binding in vitro, we identify two regions essential for DNA binding, the putative recognition helix of the predicted helix-turn-helix motif and a basic region near the C terminus. SpoIIID is unusual among prokaryotic DNA-binding proteins with a single helix-turn-helix motif in its ability to bind DNA monomerically with high affinity. We propose that the C-terminal basic region of SpoIIID makes additional contacts with DNA, analogous to the N-terminal arm of eukaryotic homeodomain proteins and the “wings” of winged-helix proteins, but structurally distinct. SpoIIID is highly conserved only among bacteria that form endospores, including several important human pathogens. The need to conserve biosynthetic capacity during endospore formation might have favored the evolution of a small transcription factor capable of high-affinity binding to DNA as a monomer, and this unusual mode of DNA binding could provide a target for drug design.
In response to nutrient limitation, the gram-positive bacterium Bacillus subtilis undergoes a process of endospore formation to produce a progeny cell that can survive until conditions favorable for growth return (20). The sporulation process involves the creation of two distinct cell types by polar septation; a larger mother cell and a smaller forespore (see Fig. S1 in the supplemental material). The mother cell engulfs the forespore, so it is surrounded by two membranes. A peptidoglycan cortex forms between the two membranes, and a protein coat assembles on the surface of the nascent spore, which upon maturation can withstand a variety of environmental stresses. The process is completed when the mother cell lyses to release the mature spore, which can survive in a metabolically inert state for many years. Bacilli and clostridia related to B. subtilis also form endospores, which can pose a threat to human health.
For a complex process like sporulation to be successful, a tightly controlled system of temporal and spatial gene regulation is essential. In B. subtilis, this is achieved in part by a cascade of cell-type-specific σ subunits of RNA polymerase (RNAP). σF becomes active first, in the forespore, followed by σE in the mother cell, then σG in the forespore, and finally σK in the mother cell (20, 24) (see Fig. S1 in the supplemental material). Activation of the forespore σ factors is coupled to morphogenesis, with polar septation leading to activation of σF and with completion of engulfment signaling activation of σG. The mother cell σ factors are activated in response to signals from the forespore, with σF and σG activities required to activate σE and σK, respectively.
In the mother cell, temporal regulation of transcription is further controlled by three transcription factors: GerR, SpoIIID, and GerE (see Fig. S2 in the supplemental material). GerR appears to act only as a transcriptional repressor of genes in the σE regulon early during sporulation (8). SpoIIID has been shown to activate and repress transcription of genes in both the σE and σK regulons, although most of its effects are on genes in the σE regulon (8, 13, 19, 37). Genome-wide transcriptional profiling of a spoIIID mutant in combination with approaches to detect and predict SpoIIID-binding sites in the genome suggests that nearly half of the 272 genes in the σE regulon are up- or downregulated by SpoIIID, in many cases by binding of SpoIIID to the promoter region (8). The appearance and disappearance of SpoIIID during sporulation are tightly regulated (12, 38-39), and circumventing this regulation in a way that causes the SpoIIID level to remain high late into sporulation results in spore defects (34). GerE has been shown to activate or repress the transcription of genes in the σK regulon later during sporulation (8, 18-19, 40-41). GerE crystallized as a dimer, and its strongest binding sites in DNA contain inverted repeats matching the consensus sequence RWWTRGGYNNYY (R means A or G, W means A or T, Y means C or T, and N means A, C, G, or T) (7). Each monomer of the GerE dimer has a helix-turn-helix (HTH) motif whose recognition helix is predicted to make contacts in the major groove of the DNA. Thus, GerE appears to achieve high-affinity binding to DNA by a mechanism similar to that of many other HTH-containing DNA-binding proteins (26).
How does the 93-residue SpoIIID protein achieve high-affinity binding to DNA? SpoIIID has a putative HTH motif (22) but appeared to be monomeric when purified from sporulating B. subtilis (B. Zhang and L. Kroos, unpublished data). Since some of the strongest binding sites for SpoIIID in DNA contain inverted repeats matching the consensus sequence WWRRACARNY, it was suggested that monomers might bind cooperatively at these sites (13). However, here we show that SpoIIID forms a single complex of discrete mobility in gel electrophoretic shift assays with DNA containing three matches to its consensus sequence and that a SpoIIID monomer is capable of binding to a single copy of the consensus sequence with high affinity using two regions. One region is in the predicted HTH motif near the N terminus of the protein. The other region includes basic residues located near the C terminus of the protein. We conclude that SpoIIID, a transcription factor crucial for B. subtilis sporulation and highly conserved in other bacilli and clostridia that form endospores, achieves high-affinity DNA binding by an unusual mechanism.
The plasmids used in this study are described in Tables S1 and S3 in the supplemental material, and the oligonucleotides used are listed in Table S2 in the supplemental material. Mutations were introduced into the spoIIID gene using the QuikChange site-directed mutagenesis kit (Stratagene). All cloned PCR products and all mutant spoIIID genes were sequenced at the Michigan State University Genomics Technology Support Facility to confirm that no undesired mutations were present.
Plasmids designed to overexpress wild-type or altered SpoIIID from a T7 RNAP promoter were transformed into Escherichia coli BL21(DE3) (Novagen), a strain that can be induced to synthesize T7 RNAP by the addition of isopropyl β-d-thiogalactopyranoside (IPTG). Transformants were grown on Luria-Bertani (LB) agar (29) containing 100 μg/ml ampicillin for 10 to 12 h at 30°C. Small, isolated colonies (as SpoIIID overexpression causes a growth defect in E. coli) were selected to inoculate 0.5 to 1 liter of LB liquid medium containing 150 μg/ml ampicillin, and the culture was incubated at 37°C with vigorous shaking until it reached an optical density at 600 nm of approximately 1, IPTG (0.5 mM) was added, and the incubation was continued for 2.5 h. Cells were then harvested by centrifugation (7,000 × g for 15 min at 4°C) and stored at −70°C.
Cells were resuspended in 20 ml of lysis buffer (50 mM Tris-HCl, pH 8.0, 10 mM MgCl2, 4 mM EDTA) to which one Complete Mini EDTA-free protease inhibitor tablet (Roche) had been added and lysed by passage through a French pressure cell at 14,000 lb/in2 four times. The lysate was cleared by low-speed centrifugation (7,000 × g for 10 min at 4°C), followed by high-speed centrifugation (approximately 175,000 × g for 1 h at 4°C). The supernatant was passed over a 1-ml SP Sepharose column (GE Healthcare) that had been equilibrated with SP column buffer (25 mM Tris-HCl, pH 8.0, 0.1% Triton X-100, to which one Complete Mini EDTA-free protease inhibitor tablet per 10 ml had been added). The column was washed with 5 ml SP column buffer, and SpoIIID was eluted with SP column buffer supplemented with a 0.2 to 0.6 M NaCl gradient. Wild-type SpoIIID and the G87Stop form eluted at ~0.5 M NaCl, while other mutant forms of SpoIIID eluted at different NaCl concentrations (K34E, K39E, and R44E at ~0.3 M; H38E and K76Stop at ~0.4 M; S33R, E43K, and D82Stop at ~0.6 M). Fractions containing SpoIIID were identified, and purity was determined by separation on sodium dodecyl sulfate (SDS)-14% Proseive polyacrylamide gels (Lonza) with Tris-Tricine electrode buffer (0.1 M Tris, 0.1 M Tricine, 0.1% SDS), followed by staining with Coomassie solution (0.1% Coomassie brilliant blue R-250, 10% acetic acid, 45% ethanol) and destaining with a 10% acetic acid-45% ethanol solution. Fractions containing SpoIIID were pooled, dispensed into 60-μl aliquots, and stored at −70°C for later use in electrophoretic mobility shift assays (EMSAs). Alternatively, for analytical ultracentrifugation, SpoIIID was purified as described above, except that the SP column buffer did not contain 0.1% Triton X-100, and fractions containing SpoIIID were pooled following purification over the SP Sepharose column and passed over a 1-ml HiTrap Heparin HP column (GE Healthcare) that had been equilibrated with 10 mM potassium phosphate buffer, pH 7.0 (buffer 1). SpoIIID was eluted with successive washes of buffer 1 supplemented with 0.6 M, 0.8 M, 1 M, 1.2 M, and 1.4 M NaCl. The SpoIIID used for analytical ultracentrifugation eluted at 1.0 M NaCl. The concentration of SpoIIID purified by either method was determined by measuring the absorbance at 280 nm and using a molar extinction coefficient of 5,960 M−1 cm−1.
Oligonucleotides corresponding to the probes listed in Table Table11 were synthesized at the Michigan State University Genomics Technology Support Facility, labeled with [γ-32P]ATP using T4 polynucleotide kinase (New England Biolabs), and annealed by being allowed to cool to room temperature after incubation in a boiling water bath for 10 min. Each annealed probe was visualized by autoradiography after electrophoresis through a 15% polyacrylamide gel, the band corresponding to the appropriate size was excised, and the labeled probe DNA was eluted by incubation at 37°C overnight in 150 μl TE (10 mM Tris-HCl, pH 8.0, 1 mM EDTA).
Labeled probe DNA (~13 nM) was incubated at 30°C for 30 min in buffer [10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 10 mM EDTA, 5% glycerol, 0.1 mM double-stranded poly(dI-dC) (Roche)] with SpoIIID at the concentrations indicated. Reaction mixtures were electrophoresed on an 8% polyacrylamide gel using 0.5× TBE (45 mM Tris-borate, 1 mM EDTA) and dried. Bands were visualized using a Storm 820 PhosphorImager (Molecular Dynamics) and quantified with ImageQuant software (GE Healthcare). The apparent dissociation constant (Kd) was determined by plotting the fraction of bound probe DNA (i.e., the signal from the shifted complex divided by the sum of the signals from the shifted complex and the unbound probe DNA) versus the concentration of SpoIIID using Origin Pro 8 (OriginLab Corporation). The data were then fitted using a nonlinear regression curve fit with the equation Fraction Bound = ([DNAtotal] + [SpoIIIDtotal] + Kd − √(([DNAtotal] + [SpoIIIDtotal] + Kd)2 − 4[DNAtotal][SpoIIIDtotal])))/(2[DNAtotal]) and solved for Kd.
Oligonucleotides identical in sequence to probe 11 (Table (Table1)1) were obtained from the Michigan State University Genomics Technology Support Facility and resuspended in water. Equimolar concentrations were combined with buffer 1 supplemented with 50 mM NaCl, placed in a boiling water bath for 10 min, and allowed to cool to room temperature. SpoIIID purified as described above using a Heparin HP column was diluted 10-fold in buffer 1 to a 0.1 M NaCl final concentration. Three different amounts of diluted SpoIIID (10.8, 12, and 13.8 nmol) were added to a constant amount of probe 11 DNA (12 nmol) prepared as described above and adjusted to 0.1 M NaCl in buffer 1, resulting in three mixtures with slightly different protein-to-DNA ratios. To each mixture, one Complete Mini EDTA-free protease inhibitor tablet was added and the mixtures were incubated at 4°C with rotation for 1 h. SpoIIID·DNA complexes were concentrated using Amicon Ultra 4 (5,000 molecular weight cutoff; Millipore) filtration devices to a final volume of 1 ml and then shipped on ice to the Colorado State University Specialized Facility for Protein Characterization, where they were dialyzed extensively versus buffer 1 supplemented with 0.1 M NaCl prior to analytical ultracentrifugation.
All experiments were performed in a Beckman XL-I using the absorbance optical system and a four-hole AN60-Ti rotor. Sedimentation velocity (SV) was performed in a 1.2-cm two-sector EPON centerpiece, while sedimentation equilibrium (EQ) was performed in a 1.2-cm six-sector EPON centerpiece. For SV, 400 μl of sample at an A260 of 0.5 was sedimented at 55,000 rpm (244,000 × g) for 4 h at 22°C, with a radial step size of 0.002 cm in the continuous-scanning mode. A total of 53 scans were analyzed using the method of Demeler and van Holde (6) to yield the diffusion-corrected, integral distribution of S over the boundary [G(s)] within Ultrascan (version 9.4 for Linux). Sedimentation coefficients (s) were corrected to that in water at 20°C (s20,w). C(s) fitting to determine the hydrodynamic properties and, ultimately, model the molecular mass of the complex was performed within Ultrascan. The solvent density and viscosity were calculated within Ultrascan (1.0033 g/cm3 and 1.0095 cP, respectively). The partial specific volumes (v-bar) for the SpoIIID protein (0.747 cm3/g), the DNA (0.55 cm3/g), and the SpoIIID-DNA complex (0.68 cm3/g) were also estimated using Ultrascan.
Based on the hydrodynamic properties returned from the SV experiments, the conditions for performing EQ (i.e., the appropriate rotor speeds for achieving equilibrium at σ = 1 to 4 and the approximate time to equilibrium) were modeled using Ultrascan. For EQ, 100-μl samples with 0.2, 0.5, and 0.7 absorbance units at 230, 260, and 280 nm, respectively, were loaded into the centerpiece. The samples were centrifuged to equilibrium (as evidenced by overlaying absorbance scans collected 4 to 8 h apart) at 28,000, 34,600, 41,300, and 48,000 rpm (63,000, 96,500, 137,500, and 185,800 × g). Ten scans at a 0.001-cm radial step size were averaged for each data set. A total of 72 data sets were available for global fitting. The data were globally fitted to various models within Ultrascan (molecular weight distribution model, single ideal species model, and two-component, noninteracting-species model) as described in Results. The fit of the data to the models was judged by the randomness of the residuals and the variance (with respect to the degrees of freedom that each model poses).
Competent B. subtilis cells were prepared by the Gronigen method as described previously (15). Strains BK395 (containing the spoIIID83 mutation) (23) and wild-type PY79 (as a control) (36) were transformed with pPH7, and transformants were selected on LB agar containing 5 μg/ml neomycin to create PH1001 and PH1003, respectively, with the PspoIVCA-gusA fusion inserted by double crossover into the amyE gene as determined by loss of amylase activity (15). Competent PH1001 cells were further transformed with pPH1 (creating strain PH2001) or derivatives in which the spoIIID allele had been modified by site-directed mutagenesis (Stratagene) (see Table S3 in the supplemental material for strain designations) or with pAK3 as a control (creating strain PH2000). Transformants were selected on LB agar containing 100 μg/ml spectinomycin and 5 μg/ml neomycin, chromosomal DNA was prepared as described previously (15), and PCRs with primers LK2189 and LK2190 (bordering thrC) and with primers LK2234 and LK2235 (bordering the mutant spoIIID83 allele at the native locus and not complementary to sequences present in pPH1 and its derivatives) were used to identify strains with spoIIID from pPH1 or its derivatives inserted by double crossover into the thrC gene and with no insertion in spoIIID83 at the native locus.
B. subtilis strains were induced to sporulate by nutrient exhaustion as described previously (15), 1-ml samples were collected at hourly intervals by centrifugation (14,000 × g for 1 min), the supernatant was decanted, the cell pellet was rinsed with 50 mM Tris-HCl (pH 8.0), centrifugation and decanting were repeated, and the cell pellet was stored at −20°C. Whole-cell extracts were prepared as described previously (16), except the lysis buffer did not contain phenylmethylsulfonyl fluoride or DNase I. After the addition of 1 volume of 2× sample buffer [50 mM Tris-HCl (pH 6.8), 4% SDS, 20% (vol/vol) glycerol, 200 mM dithiothreitol, 0.03% bromophenol blue] and boiling for 3 min, samples were subjected to Western blot analysis as described previously (21). Anti-SpoIIID antiserum (12) was used at a 1:10,000 dilution. Signals were detected using an LAS-3000 imager (Fujifilm) and analyzed with Multigauge version 3.0 software (Fujifilm).
B. subtilis strains were induced to sporulate, and samples were collected as described above. Cells were resuspended, treated with lysozyme, permeabilized with toluene, and assayed for enzyme activity as described previously (25), except that 4-nitrophenyl-β-d-glucuronide served as the substrate. One unit of enzyme hydrolyzes 1 μmol of substrate/min per A595 unit of the original cell density. The background activity (as determined by the level of β-glucuronidase activity in a strain lacking spoIIID assayed in the same experiment) was subtracted from each sample, and the values for mutants were determined as a percentage of the value for a strain bearing wild-type spoIIID measured in the same experiment. The average of three biological replicates was determined for each mutant.
To examine the binding of SpoIIID to DNA, recombinant SpoIIID was overproduced in E. coli and purified (Fig. (Fig.1A).1A). The purified SpoIIID was then tested for its ability to bind to DNA containing three matches to the SpoIIID binding site consensus sequence (probe 1, Table Table1).1). This sequence was chosen because it resembles a sequence in the B. subtilis chromosome (68 to 93 bp downstream of the sigK transcriptional start site) that was shown previously to be protected from DNase I digestion in footprinting assays with SpoIIID at a low concentration, indicating the presence of one or more high-affinity binding sites (13). The probe 1 sequence differs from that of the B. subtilis chromosome at position 12 from its 5′ end, where a G-to-C change in the second match to the SpoIIID binding site consensus sequence eliminated its only mismatch. Using EMSAs, SpoIIID and probe 1 formed a single complex of discrete mobility (Fig. (Fig.1B)1B) with an apparent Kd of 8 ± 2 nM (Table (Table1).1). A second complex indicative of SpoIIID binding simultaneously to two sites was not observed, even at high SpoIIID concentrations (>1 μM) (data not shown). Since SpoIIID purified from sporulating B. subtilis is primarily in a monomeric state (Zhang and Kroos, unpublished), it is likely that recombinant SpoIIID is monomeric and binds to only one of the three matches to the SpoIIID binding site consensus sequence in probe 1 at a given time.
To further characterize the binding of SpoIIID to probe 1, the contribution of each match to the SpoIIID binding site consensus sequence was tested individually by altering the sequence of the other two matches (Table (Table1,1, probes 2 to 4). Binding of SpoIIID to each of these probes was assayed using EMSAs. Probe 2, with only the third match to the consensus sequence intact, allowed the highest-affinity binding, exhibiting only a threefold increase in the apparent Kd compared with probe 1 (Table (Table11 and Fig. Fig.1B).1B). Probe 3, with only the second match to the consensus intact, exhibited considerably lower binding affinity (higher apparent Kd) than probes 1 and 2 (Table (Table11 and Fig. Fig.1B).1B). Probe 4, with only the first match (containing two mismatches to the consensus), did not exhibit binding, even at the highest concentration of SpoIIID tested (Table (Table11 and Fig. Fig.1B).1B). These results suggest that the third match to the consensus sequence is the major determinant of SpoIIID binding to probe 1. The second match to the consensus sequence also plays a role in binding, while the first match appears to be dispensable.
Since the second and third matches to the consensus sequence appeared to contribute weakly and strongly, respectively, to the binding of SpoIIID to probe 1, we considered the possibility that more molecules of SpoIIID bind to probe 1 than to probe 2 or 3, although this did not appear to be the case based on comparison of the migration of shifted complexes on separate gels (Fig. (Fig.1B).1B). When compared on the same gel, probes 1, 2, and 3 formed complexes with SpoIIID that comigrated (Fig. (Fig.2A).2A). Therefore, EMSAs provided no evidence that SpoIIID binds simultaneously to the second and third matches to the consensus sequence in probe 1.
If SpoIIID binds with high affinity to the third match to the consensus sequence in probes 1 and 2, as suggested by the results presented above, it might be possible to shorten those probes and still observe high-affinity binding (i.e., comparable to that observed for probe 2 in which the first and second matches to the consensus were altered). Probes 1 and 2 were shortened to include only three base pairs on each side of the third match to the consensus sequence. No matter whether the first three base pairs were identical to those in probe 2 (probe 5, Table Table1)1) or probe 1 (probe 6, Table Table1),1), the shortened probes exhibited apparent Kd values similar to that of probe 2 (Table (Table1).1). These results demonstrate that a single match to the consensus sequence is sufficient for high-affinity binding of SpoIIID to DNA.
An examination of the sequences in 20 SpoIIID binding sites mapped by DNase I footprinting in several studies revealed the most common nucleotide for each position of the binding site consensus sequence (8, 13, 19, 37). We call this sequence, 5′-TAGGACAAGC-3′, the idealized consensus sequence. When this sequence was substituted for the third match to the consensus sequence in probe 1 (probe 7, Table Table1),1), probe 2 (probe 8, Table Table1),1), or probe 6 (probe 9, Table Table1),1), no significant change in the apparent Kd for SpoIIID binding was observed (Table (Table1).1). In these contexts, the two sequences exhibited similar affinities for SpoIIID.
We serendipitously discovered that the sequence context of the match to the consensus sequence can affect SpoIIID binding affinity. When two of the three base pairs on each side of the idealized consensus were changed in probe 9, the binding affinity increased twofold (probe 10, Table Table1).1). The rationale for this change came from studies with probe 11, which was created to allow potential base pair interactions at the ends to form repeats of the DNA sequence (Table (Table1).1). Probe 10 was designed to have the same two base pairs at each end of the idealized consensus sequence as probe 11, and both of these probes were bound by SpoIIID with about twofold higher affinity than was probe 9 (Table (Table1).1). However, changing the context of the third match to the consensus sequence in this way (probe 12, Table Table1)1) did not significantly change the binding affinity (compare probe 6, Table Table1).1). We conclude that the sequence surrounding the consensus sequence can have a small effect on the affinity of SpoIIID for certain short DNA probes. This “context effect” might explain the two-fold to threefold higher affinity of SpoIIID for probes 1 and 7 than for probes 2 and 8, respectively, although other explanations are possible (see Discussion).
While probe 11, with single base overhangs at its 5′ ends, was designed to potentially allow a DNA repeat with bound SpoIIID to form, the migration of the complex that formed was indistinguishable from the complex formed with probe 10, which has blunt ends (Fig. (Fig.2B).2B). EMSAs provided no evidence for the formation of a DNA repeat with SpoIIID bound. Rather, it appears that a single molecule of SpoIIID (likely a monomer) can bind to a single match to the consensus sequence with high affinity.
To determine the composition of the complex formed between SpoIIID and DNA containing a single copy of the idealized consensus sequence, we used analytical ultracentrifugation. SpoIIID was mixed with probe 11 (Table (Table1)1) at slightly different ratios, and the homogeneity of the complex was assessed using absorbance-detected SV experiments. We monitored the sedimentation of the complex by detecting both the DNA component (at 260 nm) and the protein component (at 230 nm). The data were analyzed using the boundary analysis method of Demeler and van Holde (6), which results in the diffusion-corrected, integral distribution of S over the entire boundary. A homogeneous sample will have a vertical or nearly vertical G(s) plot, while a heterogeneous sample will have a G(s) plot with one or more positive slopes. As shown in Fig. Fig.3A,3A, the plots which result from analysis of the data from detection at 260 nm were nearly vertical, and >80% of each sample sedimented at ~2.5S. Less than 20% of a smaller, slower-sedimenting material was also detected. Analysis of the data from detection of the sedimentation of the protein component of the complex at 230 nm yielded similar plots and also detected a small fraction of slower-sedimenting material (~1S; data not shown). Further analysis of the data from the 1:1 mixture of probe 11 with SpoIIID, using C(s) analysis (31), resulted in the observation of two components: ~12% was a 1.4S and 9,000-Da component, and ~83% was a 2.45S and 21,000-Da component. This suggested that the majority of the sample exhibited sedimentation properties consistent with a complex made up of one duplex of probe 11 DNA and one monomer of SpoIIID (calculated molecular mass of complex, 19,946 Da). However, while the C(s) analysis is a good first step toward analyzing a macromolecular complex in solution, it assumes that all of the components have similar frictional coefficients, so the mass values returned are only accurate for a truly homogeneous sample. Nonetheless, despite the small fraction of slower-sedimenting material in the sample, G(s) and C(s) analyses showed the 1:1 mixture of probe 11 with SpoIIID to be an excellent candidate for sedimentation equilibrium analysis to rigorously determine the molecular mass of the complex.
For sedimentation equilibrium analysis, a total of 72 data sets were collected using four different rotor speeds and nine sample concentrations (from 1.1 to 6.3 μM complex) with detection at 230, 260, and 280 nm. The data were best fitted (variance = 2.93 × 10−5) using a two-component, noninteracting-species model (Fig. (Fig.3B),3B), as indicated by the SV analysis presented above. The two fitted components exhibited molecular masses of 9,579 Da and 20,310 Da (partial specific volumes of 0.55 cm3/g and 0.68 cm3/g, respectively), the latter of which is in excellent agreement (within 1.8%) with the calculated molecular weight of a 1:1 complex of probe 11 DNA and an SpoIIID monomer. The smaller component most closely approximated the mass of the unliganded DNA (within 4.5%), though this analysis cannot preclude the possibility that it represents the unliganded SpoIIID monomer (within 12%). Taken together, the analytical ultracentrifugation analyses indicate that a SpoIIID monomer forms a 1:1 complex with probe 11 DNA, which contains a single copy of the idealized consensus sequence.
SpoIIID has been predicted to contain an HTH DNA-binding motif (residues 23 to 42) (22). In agreement with this prediction, modeling of SpoIIID based primarily on the similarity of its predicted secondary structure to structures from the Protein Data Bank resulted in tertiary structure predictions for SpoIIID based on the structures of two HTH DNA-binding proteins, cI repressor protein of bacteriophage λ (W. Wedemeyer, personal communication) and SinR of B. subtilis (data not shown). Both predictions depict a protein with an N-terminal core composed of four α helices, including an HTH motif (helices 2 and 3). The predictions differ at the C terminus, where the structure based on cI predicts a fifth α helix (residues 76 to 85) extending away from the rest of the protein, while the SinR-based structure predicts that the region is disordered. The SinR and cI DNA-binding domains exhibit a high degree of structural similarity, and both proteins form oligomers (in contrast to SpoIIID).
To begin testing the validity of the structure predictions for SpoIIID and to identify residues important for the function of SpoIIID, we established a convenient system for analysis of spoIIID mutations. To measure the ability of SpoIIID to bind DNA and activate transcription, we constructed a spoIVCA-gusA transcriptional fusion reporter. Transcription from the spoIVCA promoter by σE RNA polymerase was shown previously to be activated by SpoIIID in vitro (13), and spoIVCA failed to be expressed in spoIIID mutant cells during sporulation (23, 30). To allow expression of mutant spoIIID alleles, we used a vector plasmid designed to allow genes to be recombined into the B. subtilis chromosome at an ectopic site (thrC), which plays no known role in sporulation gene expression. A DNA fragment encompassing spoIIID and its promoter region was cloned into the vector plasmid. The resulting plasmid (pPH1) was transformed into B. subtilis strain PH1001, which contains the spoIVCA-gusA reporter integrated at the amyE locus in a spoIIID mutant background. Recombination between the plasmid and the chromosome resulted in replacement of the thrC gene with a copy interrupted by spoIIID and a gene coding for neomycin resistance (used for selection), creating strain PH2001. As expected, the ectopic copy of spoIIID was expressed normally during sporulation (Fig. (Fig.4A).4A). SpoIIID began to accumulate by 3 h into sporulation, and its concentration rose by 4 h and then began to decline by 6 h, as observed previously for wild-type B. subtilis expressing SpoIIID from the native spoIIID locus (12). As a negative control, the vector plasmid (with no spoIIID gene) was transformed into B. subtilis strain PH1001, creating strain PH2000 with only the neomycin resistance gene at the ectopic site. This strain failed to accumulate SpoIIID during sporulation, although a weak signal presumably due to a protein that comigrated with SpoIIID and cross-reacted with antibodies against SpoIIID was observed (Fig. (Fig.4A4A).
The two B. subtilis strains described above that were subjected to Western blot analysis (Fig. (Fig.4A)4A) were assayed for GusA (β-glucuronidase) activity from the spoIVCA-gusA reporter during sporulation (Fig. (Fig.4B).4B). For comparison, activity from the spoIVCA-gusA reporter in an otherwise wild-type background (B. subtilis strain PH1003 expressing SpoIIID from the native spoIIID locus) was measured. GusA activity increased similarly in B. subtilis strain PH2001, which expresses spoIIID ectopically, and in wild-type strain PH1003. The increase in GusA activity (Fig. (Fig.4B)4B) correlated with the accumulation of SpoIIID (Fig. (Fig.4A)4A) for PH2001. In the negative control strain lacking spoIIID at the ectopic site, PH2000, spoIVCA-gusA failed to be expressed (Fig. (Fig.4B).4B). These results demonstrate that spoIVCA-gusA expression provides an assay for functional SpoIIID, establishing a system for mutational analysis of spoIIID.
To identify residues important for the function of SpoIIID, we subjected the spoIIID gene to site-directed mutagenesis and used the system described above to express the mutant allele and measure the effect on spoIVCA-gusA expression during sporulation. Since previous work had shown that SpoIIID positively autoregulates by an unknown mechanism (22), we reasoned that our system would provide a sensitive screen for reduced function of altered SpoIIID (i.e., reduced function would result in less expression of the altered SpoIIID protein, which might result in much less expression of the spoIVCA-gusA reporter). Our mutagenesis strategy was to create charge reversals throughout SpoIIID since it has many charged residues and these are likely to be surface exposed. In addition, we mutated every residue in the predicted HTH DNA-binding motif (residues 23 to 42). Each mutant allele was placed at the ectopic thrC site in the chromosome of the B. subtilis strain (PH1001) that contains the spoIVCA-gusA reporter integrated at the amyE locus in an spoIIID mutant background. Expression of spoIVCA-gusA was measured at 5 h into sporulation, which was the time when expression in a strain with the wild-type spoIIID gene at the ectopic site reached a plateau (Fig. (Fig.4B).4B). As shown in Fig. Fig.5,5, charge reversals in the region encompassing the first 21 residues of SpoIIID had modest effects (less-than-threefold change) on expression of the reporter, with the exception of the R8E substitution. Substitutions in the predicted first helix (residues 23 to 30) of the HTH changed reporter expression less than twofold, with the exceptions of V23K, R24A, R24E, and I26E. Some substitutions in the predicted turn (residues 31 to 33) of the HTH reduced reporter expression more than fourfold (V32E and S33R), but others had a less-than-twofold effect (G31E and S33A). Strikingly, most substitutions in the predicted second helix (residues 34 to 42) of the HTH, which is the predicted “recognition helix” that would interact directly with DNA in the major groove, dramatically reduced or abolished expression of the spoIVCA-gusA reporter. Charge reversals in the C-terminal half of SpoIIID had modest effects (less-than-threefold change) on the expression of the reporter, with the exceptions of the R44E, D51K, H63E, K64E, H68E, K76E, K78E, and D82K substitutions. We also made C-terminal truncations of SpoIIID by substituting a stop codon for codons that normally specify residue 76 or 82 of the 93-residue protein. Elimination of 12 residues (D82Stop) from the C-terminal end of SpoIIID had little effect on reporter expression, but elimination of 18 residues (K76Stop) reduced reporter expression about fourfold. In summary, our mutational analysis revealed primarily two regions that are important for producing functional SpoIIID; the predicted DNA recognition helix (residues 34 to 42) and a basic region near the C terminus (residues 63 to 81). Charge reversals at a few other positions (R8, R24, R44, and D51) dramatically reduced or abolished expression of the SpoIIID-dependent reporter.
Expression of the spoIVCA-gusA reporter presumably requires that SpoIIID accumulate, bind to the spoIVCA promoter region, and activate transcription by σE RNA polymerase (13, 23, 30). To try to distinguish between mutations that prevent SpoIIID accumulation versus those that prevent DNA binding and/or transcriptional activation, we performed Western blot analyses with antibodies against SpoIIID. For several strains with reduced or abolished reporter expression (i.e., strains designed to express the R24A, R24E, S33R, H63E, and K64E substitutions), we could not detect the altered SpoIIID reliably at 5 h into sporulation, whereas for several strains with normal or elevated reporter expression (i.e., strains designed to express the K14E, K28E, E29K, S33A, and D61K substitutions), the altered SpoIIID protein was detected (data not shown). Presumably, the mutations that allow reduced reporter expression do allow the altered SpoIIID protein to accumulate at a low concentration, but we could not detect the altered SpoIIID protein in these strains reliably above the background signal in the negative control strain (PH2000), which lacks spoIIID at the ectopic site (Fig. (Fig.4A).4A). As noted above, previous work had shown that SpoIIID positively autoregulates by an unknown mechanism (22) and this autoregulatory effect might be advantageous for detecting mutations that impair SpoIIID function. However, SpoIIID autoregulation appears to make it difficult to measure the accumulation of altered SpoIIID proteins with reduced function in sporulating B. subtilis. Therefore, we utilized the information in Fig. Fig.5,5, and our structure predictions, to identify several altered SpoIIID proteins for characterization of their abilities to accumulate when expressed in E. coli and, after purification, to bind DNA.
Our mutational analysis of spoIIID (Fig. (Fig.5)5) revealed primarily the predicted DNA recognition helix (residues 34 to 42) and a basic region near the C terminus (residues 63 to 81) as important for the function of SpoIIID. To investigate how these two regions contribute to SpoIIID function, we characterized altered SpoIIID proteins with substitutions in or adjacent to the predicted DNA recognition helix, and we characterized two C-terminally truncated proteins (Table (Table2).2). Expression of each protein was engineered in E. coli. All of the proteins accumulated normally in soluble form; however, the D40K substitution caused SpoIIID to be unstable during purification. The D40K substitution might alter the structure of SpoIIID in a way that makes it more susceptible to E. coli proteases. The other altered proteins were stable during purification, suggesting that they are not unfolded.
As a step toward distinguishing between altered SpoIIID proteins that are defective for DNA binding and those specifically impaired in transcriptional activation, their ability to bind to DNA containing the idealized consensus sequence (probe 10, Table Table1)1) was measured using EMSAs. Figure Figure66 shows representative results, and Table Table22 lists the average apparent Kd of altered SpoIIID proteins. All of the proteins were defective for DNA binding, but some were more defective than others, and the magnitude of the defect generally correlated with the effect on expression of the spoIVCA-gusA reporter in sporulating B. subtilis. The K34E (Fig. (Fig.6C),6C), H38E, and K39E substitutions in the predicted DNA recognition helix abolished detectable binding, as did the R44E substitution and the K76Stop truncation, and the S33R substitution (Fig. (Fig.6B)6B) in the predicted turn of the HTH increased the apparent Kd more than 15-fold (Table (Table2).2). The corresponding mutations in spoIIID, when tested in sporulating B. subtilis as described above, reduced spoIVCA-gusA expression at least fourfold (Fig. (Fig.5).5). The E43K substitution in the residue following the predicted recognition helix reduced reporter expression about 2-fold (Fig. (Fig.5)5) and increased the apparent Kd about 10-fold (Table (Table2).2). The D82Stop truncation had little effect on reporter expression (Fig. (Fig.5)5) and increased the apparent Kd about threefold (Fig. (Fig.6D6D and Table Table2).2). Because reporter expression presumably requires that SpoIIID accumulate (which is influenced by SpoIIID autoregulation, as noted above), bind to the spoIVCA promoter region (a different sequence than in probe 10, which was used in the EMSAs), and activate transcription by σE RNA polymerase in sporulating B. subtilis (13, 23, 30), we did not expect reporter expression to correlate quantitatively with changes in the apparent Kd measured using EMSAs. The comparison indicates that a considerable defect in DNA binding as measured in vitro (e.g., 10-fold higher apparent Kd for the E43K substitution, Table Table2)2) can nevertheless allow substantial reporter expression in vivo (about 50% of the wild-type level, Fig. Fig.5).5). However, greater defects in DNA binding in vitro (i.e., the 15-fold higher apparent Kd for the S33R substitution and the loss of detectable binding for several proteins, Table Table2)2) reduced reporter expression at least fourfold (Fig. (Fig.5).5). Taken together, our data show that charge reversals in or adjacent to the predicted DNA recognition helix impair DNA binding and that one or more residues in the C-terminal region spanning from K76 to K81 are critical for DNA binding by SpoIIID.
We have discovered that a highly conserved key regulator of gene expression during sporulation binds to specific sequences in DNA as a monomer with high affinity by using at least two regions. Dwindling resources in the mother cell during sporulation might have favored the evolution of a small (B. subtilis SpoIIID is 93 residues) transcription factor capable of high-affinity binding to specific sites in the chromosome as a monomer. If our structure prediction for SpoIIID is correct, as suggested by preliminary nuclear magnetic resonance (NMR) data of SpoIIID in complex with probe 11 DNA (B. Chen, P. Himes, A. Liu, H. Yan, and L. Kroos, unpublished data), SpoIIID achieves high-affinity binding by a novel mechanism involving an HTH motif (whose recognition helix presumably contacts the major groove of the DNA), followed by an additional α helix (the C-terminal basic region) that makes contacts with the DNA. Below, we discuss our findings in the context of this emerging model and we note additional studies that will be required to understand not only how SpoIIID binds DNA but how it functions as an activator and repressor of transcription.
SpoIIID is highly conserved among bacilli and clostridia related to B. subtilis that also form endospores (see Fig. S3 in the supplemental material). Non-spore-forming genera like Listeria and Staphylococcus, although closer phylogenetically to bacilli than are clostridia, do not harbor SpoIIID orthologs in their genomes, as is true for many other sporulation-specific genes (8-9). To our knowledge, the role of SpoIIID has not been investigated in any organism other than B. subtilis. However, a recent study showed that expression of spoIIID in Clostridium perfringens is not under the control of σE RNAP (14), as it is in B. subtilis (22). Strikingly, SpoIIID orthologs exhibit the highest identity in their putative recognition helix (residues 34 to 42) and in their C-terminal basic region (residues 63 to 81) (see Fig. S3 in the supplemental material). Conservation in the putative recognition helix was noted previously, and it was inferred that SpoIIID orthologs would bind to similar DNA sequences (8). This inference is strengthened by our finding that the C-terminal basic region is critical for DNA binding and the observation that this region is highly conserved. The high degree of conservation in regions shown in this study to be important for DNA binding suggests that SpoIIID orthologs likely are key regulators of gene expression during endospore formation by many different bacteria, including several human pathogens.
Most bacterial transcription factors are homodimers that bind to palindromic sites in DNA (17). HTH-containing transcription factors are generally no exception to this rule, so it was somewhat surprising that SpoIIID behaved as a monomer during gel filtration chromatography after purification from sporulating B. subtilis (Zhang and Kroos, unpublished). Members of the AraC/XylS family of transcription factors bind to DNA as monomers, but the monomer contains two HTH motifs (11) and SpoIIID is predicted to contain only one (22). Bacteriophage λ excisionase (Xis) is involved in site-specific recombination of DNA, rather than transcription, but Xis has a single, winged HTH and is a small protein that is monomeric in solution and can bind to a single site with moderate affinity and binds cooperatively to DNA containing more than one copy of its recognition sequence (1, 4, 28). We thought that SpoIIID might likewise achieve high-affinity binding via cooperative interactions. We found that SpoIIID forms a single shifted complex in EMSAs with DNA containing three matches to its binding site consensus sequence (Fig. (Fig.1B,1B, probe 1). The absence of a second shifted complex of lower mobility suggested that two SpoIIID monomers did not bind simultaneously.
Although SpoIIID did not bind cooperatively to probe 1, the threefold higher binding affinity to probe 1 compared with probe 2, which had only the third match to the consensus sequence intact (Table (Table1),1), suggests a synergistic effect of having more than one match to the consensus sequence in probe 1. Probably, the second match in probe 1 (Table (Table1,1, arrow 2) is responsible for the synergistic effect. We detected weak binding of SpoIIID to probe 3, which has only the second match intact (Table (Table1).1). However, the sequence context of the second match is different in probes 1 and 3 and we found that the sequence context can affect binding affinity (compare probes 9 and 10 in Table Table1).1). Perhaps SpoIIID binds with higher affinity to the second match in probe 1 than to that in probe 3. Alternatively or in addition, the “caging effect” of the gel during EMSAs might contribute to the synergistic effect on the binding of SpoIIID to probe 1. Caging is believed to increase reassociation by impeding escape of protein from DNA after dissociation (10). Probe 1, with the second and third matches to the consensus sequence intact, might exhibit more binding due to caging than probe 2, with only the third match intact. Further studies with shorter probes containing the second match in different sequence contexts might resolve some of these questions. Also, studies with longer probes containing two matches to the consensus sequence with different spacing and orientation might reveal cooperative binding of SpoIIID to DNA, a possibility that our results do not exclude.
Here, we focused on SpoIIID binding to the third match to the consensus sequence in probe 1 (Table (Table1,1, arrow 3; note that this is the sequence of the other strand). Shorter probes containing only the third match to the consensus sequence (probes 5, 6, and 12) were bound by SpoIIID with high affinity, similar to that of probe 2, and a 14-bp duplex containing one copy of the idealized consensus sequence (probe 11) was bound by SpoIIID with slightly higher affinity, similar to that of probe 1 (Table (Table1).1). Whether the 14-bp duplex in probe 11 constitutes a minimal high-affinity site remains to be tested with shorter duplexes. In any case, the EMSAs strongly suggested that a SpoIIID monomer can bind DNA with high affinity. We used analytical ultracentrifugation analyses to show that SpoIIID binds as a monomer to probe 11 (Fig. (Fig.33).
SpoIIID is unusual among prokaryotic DNA-binding proteins with a single HTH motif in its ability to bind DNA monomerically with high affinity. Eukaryotic homeodomain proteins can bind as monomers to short DNA segments because, in addition to their HTH, they have an N-terminal arm that interacts with the adjacent minor groove (33). As is typical for homeodomain proteins and other HTH proteins (2), our structural modeling of SpoIIID predicts an α helix preceding the HTH, but unlike homeodomain proteins, this helix is predicted to extend to the N terminus of SpoIIID, leaving no room for an N-terminal arm. Moreover, our mutational analysis indicated that most charge reversals in predicted helix 1 of SpoIIID had little or no effect on its function, as determined by the ability to activate the expression of a spoIVCA-gusA reporter during sporulation (Fig. (Fig.5).5). In contrast, several charge reversals in the highly conserved C-terminal basic region (residues 63 to 81) (see Fig. S3 in the supplemental material) of SpoIIID greatly reduced or abolished reporter expression, as did elimination of the 18 C-terminal residues (K76Stop) (Fig. (Fig.5).5). The truncated K76Stop SpoIIID protein did not bind to DNA in vitro (Table (Table2).2). We propose that the C-terminal basic region of SpoIIID makes additional contacts with DNA, analogous to the N-terminal arm of eukaryotic homeodomain proteins, allowing a monomer to bind with high affinity.
Another strategy that DNA-binding proteins containing a single HTH use to increase the affinity of their interaction with DNA is the winged-helix motif (3). Hepatocyte nuclear factor 3γ provided the first glimpse of this motif bound to DNA, and it was recognized that histone H5 has a similar structure (5). These monomeric αβ proteins have “wings” (loops) that, together with the recognition helix of the HTH, interact with DNA, resembling a butterfly perched on a rod. Two examples from bacteriophages, Xis (mentioned above) and MuR (a transcriptional repressor of phage Mu), have a single “wing” following their HTH that interacts with the minor groove, and these monomeric αβ proteins increase their affinity for DNA by cooperative binding (1, 4, 28, 35). We considered the possibility that SpoIIID contains two “wings” following its HTH that allow it to bind DNA with high affinity as a monomer. Recently, the C-terminal winged-helix domain of the bacterial transcription factor OmpR has been shown to bind to DNA as a monomer, although the affinity of the interaction (Kd, ≥1.5 μM) (27) was not as high as that measured for SpoIIID (Table (Table1).1). However, neither our structural modeling nor preliminary NMR data of SpoIIID in complex with probe 11 DNA (B. Chen, P. Himes, A. Liu, H. Yan, and L. Kroos, unpublished data) support the idea that SpoIIID is an αβ protein with a winged-helix motif. Rather, both the modeling and the data suggest that the HTH of SpoIIID is followed by an α helix.
We propose that the C-terminal basic region of SpoIIID is an α helix that makes additional contacts with DNA, analogous to the N-terminal arm of eukaryotic homeodomain proteins and the “wings” of winged-helix proteins but structurally distinct. The structure could be similar to that of the E. coli PurR repressor, whose HTH is followed by a “hinge” helix that becomes ordered upon ligand binding and interacts with the minor groove of DNA (32). PurR is dimeric, and other characterized members of the LacI family are dimeric or tetrameric. Their recognition sites in DNA are palindromic. SpoIIID binds to nonpalindromic DNA sequences as a monomer with high affinity (Table (Table1).1). Efforts to determine the structure of SpoIIID in complex with probe 11 DNA (B. Chen, P. Himes, A. Liu, H. Yan, and L. Kroos, unpublished data) promise to reveal the role of the C-terminal basic region and the predicted HTH in DNA binding. A novel mode of DNA binding by SpoIIID and its orthologs in pathogenic endospore formers could provide the basis for the rational design of selective inhibitors with therapeutic potential.
While SpoIIID has been implicated in up- or downregulation of 122 genes in the σE regulon by transcriptome analysis (8) and in the regulation of a few genes in the σK regulon by biochemical studies (19), only 20 SpoIIID binding sites have been mapped by DNase I footprinting (8, 13, 19, 37). The positions of the binding sites suggest that SpoIIID can repress transcription by interfering with RNAP (σE or σK) or activator (GerE) binding. The two sites from which SpoIIID has been shown to activate transcription overlap the spoIVCA and sigK −35 promoter regions, suggesting that SpoIIID contacts RNAP, most likely the sigma factor (i.e., σE at the spoIVCA promoter and both σE and σK at the sigK promoter) (13). Part of our motivation for making charge reversal substitutions throughout SpoIIID was to identify residues that might contact RNAP. Such “positive control mutants” would have mutations in spoIIID that reduce or eliminate transcriptional activation without impairing DNA binding. Our mutational analysis identified several charged residues that are likely surface exposed and are candidates for making contact with RNAP since charge reversal at these positions reduced spoIVCA-gusA expression more than threefold: R8, D51, H63, K64, H68, K76, and K78 (Fig. (Fig.5).5). Altered SpoIIID proteins with charge reversals at these positions are candidates for overexpression, purification, and EMSA studies, as shown here for K34E, H38E, K39E, E43K, and R44E, with the results showing involvement of these residues in DNA binding (Table (Table2).2). K76 and/or K78 might also be important for DNA binding, since the K76Stop protein failed to bind DNA detectably and the D82Stop protein bound DNA almost normally (Table (Table2).2). Although R24A and R24E substitutions reduced and eliminated spoIVCA-gusA expression, respectively (Fig. (Fig.5),5), the purified R24A and R24E proteins showed reduced and undetectable binding to probe 1 DNA, respectively (data not shown), so R24 is important for DNA binding and is therefore not included in the above list of candidate residues for making contact with RNAP. D82 is not a candidate residue for contacting RNAP, since D82Stop (lacking D82) activated spoIVCA-gusA expression (Fig. (Fig.5)5) and bound to DNA (Table (Table2),2), but the D82K substitution is interesting because it eliminates reporter expression (Fig. (Fig.5).5). We speculate that D82K might prevent K76 and/or K78 from contacting RNAP. Our findings provide a foundation for further studies aimed at elucidating the mechanism of transcriptional activation by SpoIIID.
We thank B. Chen, H. Yan, A. Liu, W. Wedemeyer, and B. Zhang for sharing results prior to publication, and we thank H. Yan and M. Thomashow for helpful comments on the manuscript.
This work was supported by National Institutes of Health grant GM43585 (to L.K.) and by the Michigan Agricultural Experiment Station.
Published ahead of print on 8 January 2010.
†Supplemental material for this article may be found at http://jb.asm.org/.